128 95 5MB
English Pages 298 [299] Year 2023
Seemings This volume presents new research on the epistemology of seemings. It features original essays by leading epistemologists on the nature and epistemic import of seemings and intuitions. Seemings and intuitions are often appealed to in philosophical theorizing. In fact, epistemological theories such as phenomenal conservatism and dogmatism give pride of place to seemings. Such views insist that seemings are of central importance to theories of epistemic justifcation. However, there are many questions about seemings that have yet to be answered satisfactorily. What kinds of seemings are there? How do seemings justify? Are seemings connected to truth? Do they play a signifcant role in inquiry? The chapters in this volume ofer a range of useful arguments and fresh ideas about seemings, the nature of justifcation and evidential support, intuitions, inquiry, and the nature of inference. Seemings: New Arguments, New Angles will be of interest to researchers and advanced students working in epistemology and philosophy of mind. Kevin McCain is Professor of Philosophy at the University of Alabama at Birmingham. He has authored and edited several works in epistemology and philosophy of science including the following from Routledge: Evidentialism and Epistemic Justifcation (2014), Epistemology: 50 Puzzles, Paradoxes, and Thought Experiments (2021), What is Scientifc Knowledge? An Introduction to Contemporary Epistemology of Science (with Kostas Kampourakis, 2019), Epistemic Duties: New Arguments, New Angles (with Scott Stapleford, 2020), and Epistemic Dilemmas: New Arguments, New Angles (with Scott Stapleford and Matthias Steup, 2021). Scott Stapleford is Professor of Philosophy at St. Thomas University in Fredericton, Canada. His publications for Routledge include Logic Works: A Rigorous Introduction to Formal Logic (with Lorne Falkenstein and Molly Kao, 2022), Hume’s Enquiry: Expanded and Explained (with Tyron Goldschmidt, 2021), Berkeley’s Principles: Expanded and Explained (with Tyron Goldschmidt, 2016), and two edited collections: Epistemic Dilemmas: New Arguments, New Angles (with Kevin McCain and Matthias Steup, 2021) and Epistemic Duties: New Arguments, New Angles (with Kevin McCain, 2020). Matthias Steup received his PhD from Brown University in 1985. He is Professor of Philosophy at the University of Colorado, Boulder, and the author of An Introduction to Contemporary Epistemology (1996) and numerous articles in epistemology. He is the editor of Knowledge, Truth and Duty (2001) and co-editor of Contemporary Debates in Epistemology (2005, 2014), A Companion to Epistemology (2010), and Epistemic Dilemmas: New Arguments, New Angles (Routledge, 2021).
Routledge Studies in Epistemology Edited by Kevin McCain
University of Alabama at Birmingham, USA and
Scott Stapleford
St. Thomas University, Canada
The Epistemology of Modality and Philosophical Methodology Edited by Anand Jayprakash Vaidya and Duško Prelević Rational Understanding From Explanation to Knowledge Miloud Belkoniene Illuminating Errors New Essays on Knowledge from Non-Knowledge Edited by Rodrigo Borges and Ian Schnee Digital Knowledge A Philosophical Investigation J. Adam Carter Seemings and the Foundations of Justifcation A Defense of Phenomenal Conservatism Blake McAllister Trust Responsibly Non-Evidential Virtue Epistemology Jakob Ohlhorst Rationality in Context Unstable Virtues in an Uncertain World Steven Bland Seemings New Arguments, New Angles Edited by Kevin McCain, Scott Stapleford, and Matthias Steup For more information about this series, please visit: www.routledge.com/Routledge-Studiesin-Epistemology/book-series/RSIE
Seemings New Arguments, New Angles
Edited by Kevin McCain, Scott Stapleford, and Matthias Steup
First published 2024 by Routledge 605 Third Avenue, New York, NY 10158 and by Routledge 4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2024 selection and editorial matter, Kevin McCain, Scott Stapleford, and Matthias Steup; individual chapters, the contributors The right of Kevin McCain, Scott Stapleford, and Matthias Steup to be identifed as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identifcation and explanation without intent to infringe. ISBN: 978-1-032-28960-1 (hbk) ISBN: 978-1-032-28963-2 (pbk) ISBN: 978-1-003-29934-9 (ebk) DOI: 10.4324/9781003299349 Typeset in Sabon by Apex CoVantage, LLC
Contents
List of Contributors Acknowledgments Introductory Note
viii ix 1
KEVIN MCCAIN, SCOTT STAPLEFORD, AND MATTHIAS STEUP
PART 1
Seemings and How They Justify 1 The Chemistry of Epistemic Justifcation
5 7
MATTHIAS STEUP
2 Seemings and Truth
23
BLAKE MCALLISTER
3 Nonsubjectivism About How Things Seem
38
MATTHEW MCGRATH
4 Against the Phenomenal View of Evidence: Disagreement and Shared Evidence
54
ELIZABETH JACKSON
5 Appearances and the Problem of Stored Beliefs
63
KEVIN MCCAIN AND SCOTT STAPLEFORD
6 Emotions as Evidence for Evaluations EARL CONEE AND RICHARD FELDMAN
75
vi
Contents
7 How to Be Irrational
94
MICHAEL HUEMER
PART 2
Seemings in Inference and Inquiry 8 Dogmatism, Seemings, and Non-deductive Inferential Justifcation
109 111
BERIT BROGAARD AND DIMITRIA ELECTRA GATZIA
9 Inference Without the Taking Condition
130
DECLAN SMITHIES
10 Zetetic Seemings and Their Role in Inquiry
147
VERENA WAGNER
11 Intuition in Philosophical Inquiry
162
JOHN BENGSON
PART 3
Seemings and Perception
185
12 Veridical Perceptual Seemings
187
ELIJAH CHUDNOFF
13 Perceptual Seemings and Perceptual Learning
202
HARMEN GHIJSEN
14 Phenomenal Explanationism and the Look of Things
217
KEVIN MCCAIN AND LUCA MORETTI
PART 4
Intellectual Seemings and Intuitions
233
15 A Priori vs. A Posteriori Justifcation: The Central Role of Rational Intuitions
235
BRUCE RUSSELL
Contents 16 Thought Experiments as Tools of Theory Clarifcation
vii 252
GRACE HELTON
17 Lessons from Commonsensism for Religious Epistemology
271
MICHAEL BERGMANN
Index
287
Contributors
John Bengson, University of Texas at Austin, USA Michael Bergmann, Purdue University, USA Berit Brogaard, University of Miami, USA Elijah Chudnoff, University of Miami, USA Earl Conee, University of Rochester, USA Richard Feldman, University of Rochester, USA Dimitria Electra Gatzia, University of Akron, USA Harmen Ghijsen, Radboud University, The Netherlands Grace Helton, Princeton University, USA Michael Huemer, University of Colorado Boulder, USA Elizabeth Jackson, Toronto Metropolitan University, Canada Blake McAllister, Hillsdale College, USA Kevin McCain, University of Alabama at Birmingham, USA Luca Moretti, University of Aberdeen, UK Bruce Russell, Wayne State University, USA Declan Smithies, Ohio State University, USA Scott Stapleford, St. Thomas University, Canada Matthias Steup, University of Colorado Boulder, USA Verena Wagner, University of Konstanz, Germany
Acknowledgments
We are grateful to Andrew Weckenmann, Rosaleah Stammler, and all the folks at Routledge for their support and help in moving this project from an idea to the present volume. And we are grateful to each of the contributors for providing the excellent chapters that make this collection what it is.
Introductory Note Kevin McCain, Scott Stapleford, and Matthias Steup
Seemings and intuitions are frequently appealed to in philosophical theorizing. This is nothing new. There is some reason to think that modern philosophers, such as Thomas Reid (1710–1796) and David Hume (1711– 1776), allowed that seemings provide evidence. In fact, a case could be made that as far back as Aristotle (384–322 BC), seemings were taken to play a signifcant role in the justifcation of beliefs.1 Discussion of seemings and their epistemic credentials is on the rise. A “seemings renaissance” was arguably triggered by the publication of James Pryor’s “The Skeptic and the Dogmatist” (2000) and Michael Huemer’s Skepticism and the Veil of Perception (2001). Pryor’s “dogmatism” and Huemer’s “phenomenal conservatism” place seemings at the center of debates surrounding epistemic justifcation. Despite the popularity of seemings-focused theories in current epistemology, many questions about seemings are thus far unresolved. What kinds of seemings are there? How do seemings justify? Are seemings connected to truth? Do they play a signifcant role in inquiry? This collection provides some answers and raises additional questions. It represents the cutting edge of new arguments and angles on seemings and intuitions. The volume falls into four parts. The chapters in Part 1 explore the nature of seemings and how they justify. Part 2 contains chapters focused on the role of seemings in inference and inquiry. The impact of seemings on perceptual justifcation is the topic of Part 3. The chapters in Part 4 close the volume with an exploration of the nature of intellectual seemings and intuitions. Part 1: Matthias Steup raises the anchor in Chapter 1 by contrasting three seemings-focused epistemological theories: dogmatism, conservatism, and credentialism. Steup argues that credentialism is the only theory of epistemic justifcation that secures a truth-connection. Discussion of truth-connections continues in Chapter 2, where Blake McAllister argues that seemings make their content likely to be true. In Chapter 3, Matthew McGrath distinguishes between subjectivist and non-subjectivist understandings of seemings. In McGrath’s view, seemings should be understood DOI: 10.4324/9781003299349-1
2
Kevin McCain, Scott Stapleford, and Matthias Steup
in the non-subjectivist sense, according to which its seeming that p amounts to a relevant body of facts supporting p. Elizabeth Jackson challenges the idea that seemings provide justifcation in Chapter 4. She insists that theories which ground justifcation in seemings face problems in accounting for disagreements and the possibility of sharing evidence. In Chapter 5, Scott Stapleford and Kevin McCain argue that the supposed problem facing internalist theories of epistemic justifcation in accounting for the justifcation of stored beliefs can be solved by recognizing the justifying role of dispositional seemings. Earl Conee and Richard Feldman explore the possibility of emotions providing independent evidence for evaluations in Chapter 6. They argue that while emotions can be part of the inductive grounds for accepting particular propositions as true, they do not provide evidence in and of themselves. This part closes in Chapter 7 with Michael Huemer’s discussion of how irrational beliefs can be formed on an internalist picture. In particular, Huemer responds to an apparent puzzle arising from consideration of how one could form a belief that is irrational from one’s own point of view. Part 2: The focus shifts in Part 2 from the nature of seemings in general to special cases, beginning in Chapter 8 with Brogaard and Gatzia’s arguments linking theories of perceptual justifcation with inferential justifcation. Specifcally, they argue that the only theory of perceptual justifcation that can plausibly be extended to account for non-deductive inferential justifcation is a form of dogmatism—one maintaining that perceptual seemings can provide immediate justifcation in virtue of their phenomenology. In Chapter 9, Declan Smithies argues that Paul Boghossian’s “Taking Condition” on inference is false. By Smithies’ lights, inferences can be justifed without any personal-level representation that a given set of premises supports a conclusion. Verena Wagner argues in Chapter 10 that seemings play a special role in inquiry: they provide reasons for performing various zetetic tasks but cannot provide reasons for concluding inquiry. John Bengson wraps up this part of the volume in Chapter 11 by exploring the role that intuitions play in philosophical inquiry. He argues that intuitions are presentational states that should be distinguished from seemings, and the former (but not the latter) are central to philosophical theorizing. Part 3: Elijah Chudnof considers the epistemic signifcance of taking veridical perceptual seemings at face value in Chapter 12. He is sympathetic to the idea that when we take these seemings at face value and lack defeaters, the result is knowledge. In Chapter 13, Harmen Ghijsen argues that the phenomenon of perceptual learning puts pressure on the idea that perceptual seemings supply immediate justifcation for their contents. Relatedly, Kevin McCain and Luca Moretti defend their preferred seemings-based theory, Phenomenal Explanationism, from a similar worry by arguing that perceptual seemings can provide immediate justifcation.
Introductory Note
3
Part 4: The last part of the volume opens with Bruce Russell’s take on a priori justifcation in Chapter 15. Contra Timothy Williamson, Russell argues that there is a signifcant diference between a priori and a posteriori justifcation, and contra Paul Boghossian, he argues there is no signifcant diference between the a priori justifcation of analytic and synthetic propositions because all a priori justifcation is based on rational intuitions. In Chapter 16, Grace Helton argues that thought experiments do not overturn philosophical theories simply by eliciting intuitions. Instead, she contends that thought experiments do their epistemic work by clarifying theories and also perhaps by prompting intuitions. The last word goes to Michael Bergmann in Chapter 17, who argues that the intuition-based commonsense responses to radical skepticism he laid out in earlier work can provide guidance for responding to religious skepticism. Note 1 See McAllister (2023: ch. 8) for discussion.
References Huemer, Michael. 2001. Skepticism and the veil of perception. Lanham: Rowman and Littlefeld. McAllister, Blake. 2023. Seemings and the foundations of justifcation: A defense of phenomenal conservatism. New York: Routledge. Pryor, James. 2000. The skeptic and the dogmatist. Noûs, 34: 517–549.
Part 1
Seemings and How They Justify
1
The Chemistry of Epistemic Justifcation Matthias Steup
1.1.
Epistemic Chemistry
A chemical reaction is a process in which one or more substances, the reactants, combine to form a new substance. For example, copper and tin create bronze; hydrogen and oxygen form water, and the combination of iron, water, and oxygen results in rust. By reacting to each other, two or more things create a new thing. If we apply this concept to epistemology, the thought is that certain non-epistemic reactants produce epistemic justifcation as a new product. But what are the non-epistemic reactants that, when combined, create epistemic justifcation? That is the mystery of epistemic chemistry. I will propose a solution to this mystery. A lucid account of epistemic chemistry can be found on the frst page of Alvin Goldman’s seminal paper, “What Is Justifed Belief?” Goldman himself didn’t use the term “epistemic chemistry”, but what I have in mind is exactly the kind of theory he describes: The term “justifed”, I presume, is an evaluative term, a term of appraisal. Any correct defnition or synonym of it would also feature evaluative terms. I assume that such defnitions or synonyms might be given, but I am not interested in them . . . I want a theory of justifed belief to specify in non-epistemic terms when a belief is justifed.1 About the kind of theory he seeks, Goldman makes two important points. First, the goal is not to defne the concept of epistemic justifcation but instead to specify the conditions when a belief is justifed. Second, these conditions must be stated without the use of epistemic terms. Generalizing, we can talk of normative chemistry: normative status—something’s being good or bad in some respect—does not emerge out of nowhere. Instead, it is grounded in non-evaluate, purely factual conditions—non-normative reactants—that make a thing either good or bad. Accordingly, the task of
DOI: 10.4324/9781003299349-3
8
Matthias Steup
epistemic chemistry is not merely to identify the epistemic reactants in nonepistemic terms but in terms that are not evaluative in any way. A theory of justifcation can aim at two targets. The frst is doxastic justifcation. In that case, the analysandum is “S is justifed in believing that p.” The second is propositional justifcation. The analysandum is then “S has justifcation for believing that p.” Goldman was concerned with the former, and process reliabilism was his proposal. There are familiar objections to reliabilism: the new evil demon problem, Laurence BonJour’s reliable clairvoyant, and the generality problem. Since I take at least the frst two objections to be decisive, I do not think reliabilism is a correct theory of epistemic chemistry. My proposal is that the correct solution to the mystery of epistemic chemistry can be found within the framework of classical internalist foundationalism. 1.2.
Internalist Foundationalism
The key idea of classical internalist foundationalism is that all beliefs are ultimately justifed by the Given. What is the Given? An infuential approach, championed by James Pyror and Michael Huemer, identifes the Given with seemings: experiences or appearances produced by faculties such as perception, introspection, memory, reasoning, and a priori intuition. Seemings have four primary characteristics. First, they are mental states with propositional content. A seeming is always a seeming that p, where p is the seeming’s content. Second, seemings represent their content assertively. They are, so to speak, internal “testifers” making claims about what reality is like. Third, seemings are not beliefs. Fourth, seemings are not the kind of thing that can be either justifed or unjustifed. Since seemings are not beliefs, they are exactly what foundationalists need to make sense of the idea of basic justifcation. Foundationalism is the thesis that a subject’s set of justifed beliefs is divided into basic and non-basic beliefs. Whereas non-basic beliefs owe their justifcation to other beliefs, basic beliefs are justifed without receiving any of their justifcation from other beliefs. When a seeming that p confers justifcation on a belief that p, the belief is justifed by a mental state that is not in turn another belief. This point is essential. If seemings were belief states, they couldn’t be the kind of thing that stops the justifcatory regress and thus makes a justifed belief basic. Of course, while seemings are not beliefs, they typically give rise to beliefs. For example, when you believe that there is a red book on the table, the cause of your belief is typically a visual seeming. The fourth feature is of particular importance. A seeming that p couldn’t possibly be justifed or unjustifed, yet it can justify a belief that p. When you have a visual seeming of a red book on a table, this perceptual experience is not the sort of thing that “has” justifcation. Yet it can justify
The Chemistry of Epistemic Justifcation
9
your belief that there is a red book on the table. To generalize, seemings can justify without having justifcation. They do not transmit justifcation but create it. That is what qualifes them to play a key role in epistemic chemistry. Next, consider a prominent objection to internalist foundationalism. Claiming that the Given is a myth, critics have confronted foundationalists with a dilemma.2 Seemings either have propositional content or they don’t. If they do, they are belief-like, and thus can confer justifcation on beliefs only if they are justifed themselves. They are then unable to stop the regression of justifcation. If, on the other hand, seemings do not have propositional content, then they are not, as Wilfrid Sellars would have put it, located in the space of reasons. They are then epistemically impotent, incapable of conferring any justifcation on beliefs. Either way, the attempt to ground justifcation in the phenomenally given fails. The foundationalist reply to this argument is to deny the alleged problem of the frst horn. It is true that, due to having propositional content, seemings are, in one respect, belief-like. But, in another respect, seemings are unlike beliefs. Whereas beliefs can be justifed or unjustifed, seemings cannot. Therefore, seemings can justify without being justifed themselves. While this is an efective reply, it moves to center stage the aforementioned mystery of epistemic chemistry. Why is it that a seeming can confer justifcation without having any? Where does a seeming’s power to create justifcation—its J-power—come from? Among phenomenal foundationalists, that’s a huge bone of contention. There are two competing solutions to this mystery. According to the frst, seemings have J-power because of their phenomenal character, their inherent claim to represent reality, the “feel” of truth that seemings generate. This answer has an important consequence. A single seeming that p by itself can make a belief that p justifed. According to the second solution, a particular seeming’s J-power always comes from additional elements within the total set of the reactants of epistemic chemistry. By itself, a single seeming cannot provide any justifcation for believing its content. To have J-power, seemings need proper epistemic credentials: evidence of their reliability. Next, I will discuss three theories of justifcation, all of which are competing versions of internalist foundationalism. The frst two champion the frst solution. The third endorses the second solution. 1.3. Three Versions of Phenomenal Foundationalism The frst theory is dogmatism. According to it, a seeming’s J-power does not depend on any qualifying conditions. Dogmatism:
Sp → you have J for Bp.3
10
Matthias Steup
Compare dogmatism with Conservatism:
(Sp + no defeater) → you have J for Bp.4
By adding a no-defeater clause to the antecedent, conservatism imposes a qualifying condition on seemings. They have J-power only if they are undefeated. Next, consider a theory I’ll call credentialism. According to it, the no-defeater clause is too weak. For a seeming to have J-power, it must be backed up by evidence of reliability: Credentialism:
(Sp + you have evidence that Sp is reliable) → you have J for Bp.
Dogmatism and credentialism are non-conservative theories, but each of them is non-conservative in its own distinctive way. What makes a theory of justifcation conservative is the application of a defeasible presumption of innocence—x is innocent unless proven guilty. When the presumption is applied to beliefs, we get doxastic conservatism. When it is applied to seemings, we get phenomenal conservatism. Neither dogmatism nor credentialism are conservative in that sense. Dogmatism is a non-conservative theory because the antecedent in “Sp → you have justifcation for believing that p” does not include a nodefeater clause. Therefore, the J-power of Sp is not subject to the “unless proven guilty” qualifcation. Dogmatism entails, then, that seemings always have J-power, no matter what.5 Even in the presence of defeat, the J-juice keeps fowing. What’s blocked is merely doxastic justifcation. For example, when you notice that a red light is shining at the red-looking table in front of you, dogmatists would certainly agree that the table’s looking red to you does not justify you in believing that the table is red. However, they would insist that the table’s looking red has J-power even then. Because the table looks red to you, you have justifcation—albeit defeated justifcation—for believing that the table is red. Like dogmatists, conservatives would of course agree that, since you have a defeater, you would not be justifed in believing that the table is red. Unlike dogmatists, though, they would not agree that the J-juice is still fowing. The red light defeater indicts the table’s looking red as unreliable. The no-defeater condition is not met. The looks-red seeming is no longer innocent. Under the circumstances (the circumstances of you noticing the light), the seeming loses its J-power. Conservatism, then, is more demanding than dogmatism. However, the degree to which the theory is demanding is importantly limited. The key idea of conservatism is that, whereas evidence of unreliability can rob a seeming of its J-power, there is no need for evidence of reliability. Arguably,
The Chemistry of Epistemic Justifcation
11
you do in fact have plenty of evidence for taking your perceptual seemings to be reliable. Conservatives view such evidence as mere icing on the cake. They allow for cases, however strange they may be, in which a perceptual seeming that p is a source of justifcation for you even though you have zero evidence that the seeming is reliable or trustworthy. For credentialists, conservatism isn’t demanding enough. They reject the key conservative idea that a seeming can have J-power even in the absence of reliability evidence. Credentialism, then, replaces the presumption of innocence with the presumption of guilt: seemings are guilty, they don’t have J-power, unless they are proven innocent, that is, unless they have proper epistemic credentials bestowed upon them by evidence of their reliability or trustworthiness. The three theories are arranged on an increasing scale of demandingness about the conditions that must be satisfed for a seeming to have J-power. According to dogmatism, seemings always have J-power. This is as nondemanding as it gets. Conservatism is moderately demanding because, according to it, seemings have J-power only if they are undefeated. Credentialism is still more demanding because it says that the J-power of seemings requires the possession of reliability evidence.6 I will now turn to problems for each of the three theories. 1.4. Illusions and Lying Witnesses Suppose it seems to you that p but you know that the seeming is misleading: you know that p is false. In such a case, does the seeming, as dogmatism would have it, have J-force? Arguably, it does not. Consider the Müller-Lyer illusion: Line A: Line B: The relevant seeming, Sp, is that line B looks longer than line A. Let “p” stand for the proposition that line B is longer than line A. Since you know that the lines are equally long, you have a defeater. Dogmatism says that, although you would not, of course, be justifed in believing that p on the basis of the looks-longer seeming, Sp nevertheless gives you some justifcation for believing that p. According to the objection, that’s not plausible. When you know that p is false, Sp is completely discredited. It doesn’t have any J-power at all. The second objection makes an analogous point about testimony. The point is that testimony you know to be misleading doesn’t have any J-force.
12
Matthias Steup
Assume testifer Ted, a Republican wearing a red MAGA hat, asserts that p, where “p” stands for the proposition that the 2020 election was stolen from Donald Trump. You know that p is false. Ted is either deluded or lying. When we apply dogmatism to testimony, we get the result that, although you would of course not be justifed in believing that p, Ted’s assertion nevertheless has J-power. It supplies you with a reason, albeit a weak reason that is in fact defeated, to believe that the election was stolen. According to the objection, that’s not plausible. Ted’s assertion has no J-power at all.7 To objections like these, dogmatists typically reply that, although you know that the diferent-length seeming is misleading, it nevertheless retains a “little bit” of residual justifcational force. Although you are not justifed in believing that Line B is longer than Line A because you know the seeming is misleading, the fact that it seems this way to you provides you with at least a “some” justifcation for believing it. Likewise, although you know that Ted is either lying or deluded and thus wouldn’t be justifed in believing that the election was stolen, Ted’s assertion nevertheless gives you at least some very low degree of (propositional) justifcation for believing what he is saying. My response is that the idea of having “some” justifcation is unclear. Let “X” stand for whatever gives you justifcation for believing that p. X either raises the probability of p above 0.5 or it does not. If it does, you’re having “some” justifcation can mean either (a) X gives you some degree of justifcation located somewhere on the scale between a probability just above 0.5 and 1, or (b) X gives you some degree of justifcation in the low range of, say, a probability between 0.51 and 0.7. Consider a conference room in which everybody present is a Republican. Logically speaking, it is then true that there are some Republicans in the room. This sense of “some” is analogous to the sense of “some justifcation” in (a). Even if p is certain to you, where certainty signifes the highest degree of justifcation, you count as having “some” degree of justifcation. Of course, that is not the sense in which “some” is used in ordinary language. To say that there are some Republicans in the room carries the conversational implicature that most people in the room are not Republicans. This sense of “some” corresponds to the sense of “some justifcation” captured by (b). To have “some” justifcation means that the degree of justifcation in question is somewhere within the low range. If X confers a degree of probability on p higher than, say, 0.7, you do not merely have “some” justifcation but “complete” justifcation.
The Chemistry of Epistemic Justifcation
13
Either one of these senses of “some” justifcation looks acceptable to me. However, neither (a) nor (b) supports the claim that defeated seemings have minimal justifcational force. If you know that the two lines are of the same length, then surely the diferent length seeming fails to raise the probability of p above 0.5. Rather, the probability would be at the extremely low end of the scale, perhaps in the vicinity of 0.001. So, you wouldn’t have “some” justifcation in the sense of (b). Now consider the alternative case in which X does not raise the probability of p above 0.5. Assume X raises the probability of p to some value below 0.5. Does X give you “some” justifcation for believing that p? I don’t see that it does. I think X gives you zero justifcation for believing that p is true. Instead, it gives you justifcation for believing that p is false. In the example of the Müller-Lyer illusion, you know that the two lines are equally long. Relative to your knowledge, the fact that line B looks longer than line A confers on p, let’s say, a probability of 0.001. As such, it gives you zero justifcation for believing—for taking it to be true—that line B is longer than line A. Rather, it gives you a very high degree of justifcation for disbelieving that line B is longer than line A. I do not think, therefore, that it’s plausible to claim that, when you know that p is false, Sp retains some minimal justifcational force in support of believing p. 1.5.
The Problem of Easy Justifcation
Conservatives don’t have an easy time with the problem of easy knowledge, or, for that matter, easy justifcation.8 The problem they run into is that, unless they take efective countermeasures, conservatism is committed to endorsing arguments that naturally strike us as unacceptably bad. The question is whether the available countermeasures are successful. Here is an example of an argument that looks unacceptably bad. Assume Connor and Clare are in the Zoo. Connor is a conservative, Clare is not. They are looking at Zebras. Clare says: “So you think these animals are zebras? What makes you think they are not cleverly disguised mules (CDMs)?” Connor says, “That’s easy. Here is my argument.” The Zebra Argument (1) They are zebras. (2) They are zebras → they are not CDMs. Therefore: (3) They are not CDMs.
14 Matthias Steup Next, Connor attempts to strengthen his argument with a follow-up: “I’m justifed in believing the frst premise because the animals appear to be zebras, and I’m justifed in the second premise because it is an obvious entailment. The argument is valid. So, I’m justifed in believing the conclusion.” Connor’s line of reasoning refects the conservative view. According to conservatism, the visual zebra-seeming, if undefeated, is by itself sufcient to justify Connor in believing (1). Since (2) is an a priori knowable entailment, no additional perceptual evidence is needed for him to deduce, and thereby acquire justifcation for believing, the conclusion. However, Connor’s deployment of the Zebra argument looks suspiciously like begging the question and therefore invites the objection that, reversing the actual fow of justifcation, it has a circular structure and thus fails to transmit justifcation from the premises to the conclusion. Justifcation for (1) requires prior justifcation for (3). That is, for Connor’s zebra-seeming to generate justifcation, he must, to begin with, have evidence for rejecting alternative scenarios in which the zebra-seeming is misleading. Therefore, Connor cannot generate justifcation for (3) relying solely on the visual experience that the animals seem to be zebras. What do conservatives say in reply? One main move is to introduce two separate types of epistemic evaluation, corresponding to two diferent questions9: Q1 Q2
Is it possible, via the Zebra Argument, to acquire justifcation, and thereby come to know, that the animals are not CDMs? Is it possible, by deploying the Zebra Argument, to dispel the skeptical doubt that the animals, although they look like Zebras, are in fact CDMs?
Conservatives insist that, while the answer to Q2 is “no”, the answer to Q1 is “yes.” Acquiring justifcation for rejecting the skeptical alternative is one thing. Successfully rebutting doubt about it is another. But are the two questions really independent of each other? Can one really be justifed in believing that p while not having independent evidence to ward of skeptical doubt about the truth of p? Suppose I claim the yellow bird before us is a goldfnch. You ask me what justifes me in believing it’s not a canary. I reply that I don’t have any independent reason to dispel your skeptical doubt but am nevertheless justifed in believing the bird is a goldfnch and therefore justifed in believing it’s not a canary. You should not, it seems to me, accept this reply. The very point of confronting me with the skeptical canary alternative is to test whether I’m really justifed in believing the bird is a goldfnch. If I can’t cite evidence
The Chemistry of Epistemic Justifcation
15
that the bird is not a canary, then I can’t reasonably claim to be justifed in believing it’s a goldfnch. I do not think, therefore, that Q1 and Q2 are, as conservatives claim, independent of each other. 1.6.
The Truth-Connection
Epistemic justifcation is supposed to be somehow connected to truth.10 While this thought enjoys broad agreement, it is highly controversial exactly how we should think of this connection. Since there is also broad agreement that the connection is not entailment, it is natural (though not inescapable) to take it to be one of probability. Justifed beliefs must be beliefs that are, minimally, more likely true than not.11 Externalists want a “robust” connection that ensures that justifed beliefs are probably true in an “objective” sense of probability that is not relativized to the subject’s perspective. On this conception, a BIV’s belief in hand possession is unjustifed. Deceptive cognitive processes fail to make it likely that the beliefs based on them are true. Internalists object that this approach results in a technical concept of justifcation with dubious credentials. They want to preserve the intuition that BIVs are as well justifed in believing they have hands as the denizens of normal worlds. On this approach, the probability above 0.5 that justifcation must ensure is still objective, but it is relative to what the subject has to go on. This produces the result internalists are looking for. Relative to what a BIV has to go on, hand possession enjoys a high degree of probability. The three theories under consideration—dogmatism, conservatism, and credentialism—are all internalist because they consider seemings, without imposing any external constraints on them (like having a reliable origin), to be precisely the sort of thing “one has to go on” when forming beliefs about the world. Thus all three theories agree that, if you were a BIV, you would be justifed in believing that you have hands. The question, then, is: How do seemings generate likelihood of truth? I have argued that, if some X gives you justifcation for believing—rather than disbelieving—that p, then X must raise the probability of p above 0.5. Suppose we accept this. It will then be unclear how a seeming, Sp, can by itself raise the probability of p above 0.5. Surely evidence of a positive correlation between Sp and the truth of p would be needed for Sp to accomplish this feat. Assume conservatives agree that, if Bp is justifed by Sp, Sp must raise the probability of p above 0.5. They might then say that seemings by themselves can generate likelihood of truth because they have an intrinsic feature F. In the absence of a defeater, Sp confers a degree of probability above 0.5 on p due to feature F. Several candidates for F come to mind: seemings can justify without needing justifcation themselves, they come with a “feel” of truth, they assert the truth of their propositional content, and, unless
16
Matthias Steup
misleading, they present to the subject the truth-maker of their content. I don’t think any of these options are plausible. It’s not easy to see that any such feature makes it more likely than not that, given Sp, p is true. For p to have, given Sp, a probability above 0.5, what is needed is evidence that Sp is a reliable indicator of truth: evidence that there is a positive correlation between instances of Sp and the truth of p. Alternatively, conservatives might say that the propositions that are the contents of undefeated seemings always have a prior epistemic probability above 0.5. On this approach, if you have a seeming that p, the probability that p is true is, to begin with, greater than 0.5. Independent evidence conferring such a degree of probability is not needed. Here are two worries about this idea. First, the probability calculus stipulates that necessary propositions have a prior probability of 1. But why think this is an epistemic probability? A proposition’s epistemic probability is typically defned as a “measure of the degree of its justifcation.”12 Consider an extremely complicated necessary truth. Assume this proposition is completely beyond my comprehension. In that case, I have zero justifcation for believing it. It is, therefore, unclear in which sense a prior probability of 1 is bound to be “epistemic.” Second, suppose I see a little yellow bird that looks to me like a goldfnch. The proposition “This bird is a goldfnch” is contingent. In some scenarios, Sp is veridical; in others, it is misleading. I don’t think it is plausible to claim that, solely by virtue of the bird’s looking like a goldfnch, its being a goldfnch has a prior probability greater than 0.5. Rather, if the probability of the bird’s being a goldfnch is to be assessed from within my subjective perspective, that probability cannot be greater than 0.5 unless I have evidence that, in most cases in which a bird looks like a goldfnch to me, it really is a goldfnch. I have argued that recognized illusions and obviously lying witnesses pose serious problems for dogmatism. Conservatism avoids these problems but is vulnerable to the objection from easy justifcation. Moreover, it is unclear how conservatism can establish the needed link between justifcation and likelihood of truth. Next, I will discuss two problems that arise for credentialism. 1.7.
The Regress Problem
One main motivation for being either a dogmatist or a conservative is that the alternative to these views, credentialism, appears to trigger an infnite regress. Consider an extreme form of credentialism, classic coherentism. The view says: CC
Bp is justifed → you have a justifed belief that Bp originates from a reliable source.
The Chemistry of Epistemic Justifcation
17
It is obvious that, if CC were true, having a justifed belief would require an infnite number of further justifed beliefs. That’s a decisive reason for rejecting CC. An analogous problem aficts the following phenomenalist version of credentialism: SC
Your seeming that p, Sp, is J for Bp → You have a justifying metaseeming that Sp is an instance of a reliable seeming.
If SC were true, then having a seeming with J-power would require an infnite number of justifying meta-seemings. That’s a decisive reason for rejecting SC. Given the obvious problems that arise for credentialism, what motivates the view to begin with? It is the following thought. If seemings are to have J-power, they must generate likelihood of truth. They can do that in only two ways: either externalistically by virtue of enjoying de facto reliability or internalistically by virtue of being accompanied by evidence of reliability. So, if you are an internalist and you expect justifcation for Bp to make it likely that p is true, it’s hard to see how you can carry on without an evidence of reliability requirement. The question is whether it’s possible to defend such a requirement without triggering an infnite regress. What, then, is evidence of reliability? If such evidence is to be identifed with occurrent seemings of reliability, we are back at SC, which I agree should be rejected. But is it plausible to identify evidence with seemings? Let’s distinguish between evidence there is and evidence one has. The former is, roughly speaking, that which confers a degree of probability on facts, events, and propositions. Evidence one has for believing that p—the only kind of evidence that determines whether one has justifcation for believing that p—is part of what one has to go on in fguring out whether p is true. What is excluded and what is included in what one has to go on? According to Roderick Chisholm, the criterion is availability on refection.13 What you have to go on at a given time in ascertaining whether p is true are the seemings available to you when you refect on whether p is true. Consider as an example of a standing belief of mine: Helsinki is the capital of Finland. Let’s agree that this is a justifed belief. Since it is not continuously accompanied by a seeming that Helsinki is the capital of Finland, what justifes it is not an occurrent seeming. The point can be put in general terms. Seemings are by necessity occurrent mental states. They cannot, therefore, be what justifes standing beliefs.14 What, then, justifes my belief? Following Chisholm’s approach, we might say that, if I were to refect on what justifes me in believing that Helsinki is the capital of Finland, I would be presented (minimally) with the following pieces of evidence: frst, I seem to remember it; second, the memory seeming I’m relying on seems to me an instance of a reliable type of seeming.
18
Matthias Steup
Next, consider an occurrent perceptual belief. I’m in the zoo in front of an enclosure with animals I take to be zebras. I wonder whether my belief is justifed. On refection, it seems to me that the animals’ looking like zebras is an instance of a reliable type of seeming, and that a bona fde zoo like the one I’m currently visiting is unlikely to fool its visitors with CDMs. Generalizing from these examples, credentialists might suggest the following: RC
Sp gives you J for Bp → if you were to attentively refect on whether p is true, it would seem to you that Sp is reliable.15
Does RC generate an infnite regress? I’m inclined to think it does not. Whereas SC requires the actual presence of a meta-seeming ascribing reliability to the frst-order seeming, RC does not. RC merely requires that, if you are justifed in believing that p, then upon refection on whether p is true, p will seem true to you, and that seeming will appear to you to be an instance of a reliable type of seeming. Due to the “upon refection” proviso, RC does not require having an infnite number of seemings. This rather quick argument won’t completely allay the infnite regress worry. But I think it succeeds in showing that it is by no means clear that credentialism must succumb to the regress problem.16 1.8.
The Problem of Vicious Circularity
Another main motivation in support of either dogmatism or conservatism is the worry that the credentialist demand for evidence of reliability generates a vicious kind of epistemic circularity. The following Pyrrhonian argument exploits this thought.17 The Pyrrhonian Argument Perceptual seemings have J-force for you → you already have justifcation for taking perceptual seemings to be reliable. RJ You have justifcation for taking perceptual seemings to be reliable → perceptual seemings have J-force for you. NP (JR + RJ) → Neither having justifcation for your perceptual beliefs nor having justifcation for taking perceptual seemings to be reliable is possible for you. Therefore: BN Neither having justifcation for your perceptual beliefs nor having justifcation for taking perceptual seemings to be reliable is possible for you. JR
The Chemistry of Epistemic Justifcation
19
The hallmark of dogmatism and conservatism is the rejection of JR. Therefore, neither of these views is threatened by the skeptical conclusion. That is a point in favor of them. However, it would be a mistake to think that credentialists get stuck with the skeptical conclusion. They have, after all, the option of rejecting NP. On the face of it, NP looks plausible. Suppose your eyesight is really bad and you have misplaced your glasses. Finding your glasses requires having them on. Having them on requires fnding them frst. The outcome is that you are out of luck. But does the conjunction of JR and RJ really amount to a Catch-22 like that? Consider the following argument, the structure of which is analogous to that of the Pyrrhonian Argument. The Clownfsh Argument FA The clownfsh exists → the sea anemone exists. AF The sea anemone exists → the clownfsh exists. NP If FA and AF are both true, then neither one can exist. Therefore: BN Neither one can exist. The fact of the matter is that both of these creatures exist. Obviously, the argument is a dud. Yet it is true that the clownfsh depends on the sea anemone and vice versa. The culprit, then, must be NP. NP is false because the mutual dependence in question is not an obstacle, but rather a beneft, to the clownfsh and the sea anemone. Finding inspiration in the obvious failure of Clownfsh Argument, credentialists can attack the NP premise in the Pyrrhonian Argument by distinguishing between two types of dependence. Diachronic: “A depends on B” stands for a temporal relation. B must come frst. Once B is there, you can have A. Synchronic: “A depends on B” stands for a modal relation. It’s not possible to have A without B. The example in which you have misplaced your glasses illustrates diachronic dependence. The clownfsh and the sea anemone illustrate synchronic dependence. The two creatures biologically depend on each other, but that doesn’t mean for one of them to exist, the other one needs to exist frst. Rather, they both emerged together via a process of Darwinian evolution over a very long time. Credentialists can reject the NP premise in the Pyrrhonian Argument on the ground that, when it comes to perceptual justifcation and
20
Matthias Steup
evidence of perceptual reliability, the dependence relation between them isn’t diachronic but, as in the case of the clownfsh and the sea anemone, synchronic. Obviously, though, we are now looking at epistemic, not biological, dependence. The key point applies nevertheless. It’s not the case that one of the two things—either perceptual justifcation or evidence of perceptual reliability—must come frst, and then you can have the other. Rather, the point is that they emerge together, not through a process of Darwinian evolution but rather via the cognitive development of an individual mind. As infants begin to perceive the world, they acquire, at the very same time, evidence of perceptual reliability. Of course, they cannot acquire evidence of perceptual reliability without perceiving the world. One thing necessarily comes along with the other. Neither one comes frst. 1.9.
The Evolution of the Epistemic Web
What, then, is the correct epistemic chemistry? I have argued that frstorder seemings by themselves are not enough. What is needed in addition are second-order seemings that supply the frst-order seemings with proper epistemic credentials. According to classical coherentism, all justifcation comes from a web of belief. According to dogmatism and conservatism, all justifcation ultimately comes from frst-order seemings. Higher-order seemings are not necessary; they are merely icing on the cake. According to credentialism, higher-order seemings are not merely supplementary but necessary to begin with. All justifcation comes from a web of actual frst-order seemings and potential higher-order seemings available through refection. The latter involve evidence indicating that our lower-level seemings are reliable and evidence on the basis of which we can rule out local and global error possibilities. How is such a web possible? My answer is that epistemic chemistry must include three key ideas. First, seemings can justify without having justifcation themselves; they can “create” justifcation. Second, the J-power of frst-order seemings requires second-order seemings, available upon refection, to the efect that our frst-order seemings are reliable and thus make it more likely than not that what appears to us to be the case is true. Third, frst-order justifcation— seemings of various kinds—and meta-justifcation—evidence of reliability available on refection—evolve together. Neither one needs to come frst. When infants start observing the world around them, they notice that what can be seen can be touched, heard, tasted, and smelled. What was seen yesterday is seen again today. A rattle that was red in the past continues to be red in the present. As they get older, the web of mutual confrmations begins to include introspection, memory, and reasons. That, in a nutshell, is how an epistemic web evolves.18
The Chemistry of Epistemic Justifcation
21
Notes 1 Goldman 1979, p. 1. 2 See, for example, BonJour 1985, p. 69. 3 To be read as: If it seems to you that p, then you have justifcation for believing that p. The view is defended in Huemer 2001 and Pryor 2000. 4 Defended in Huemer 2007, where he revises the earlier, 2001 version of the view he calls “phenomenal conservatism.” 5 Robert Audi has suggested to me that the point can be put this way: according to dogmatism, a seeming that p indefeasibly or ineradicably provides defeasible justifcation for believing that p. He holds an analogous view about promises: if you promise to ϕ, then you have an ineradicable but overridable duty to ϕ. Cf. Audi 2004, p. 23f. 6 There are additional, even more demanding, versions of credentialism. For example, according to BonJour’s 1985 defense of coherentism, no belief is justifed unless accompanied by a meta-belief to the efect that the belief in question was formed in a truth-conducive way. 7 Theoretically, dogmatists about perception could endorse credentialism about testimonial justifcation. That strikes me as an unprincipled and undesirable option. 8 See Cohen 2002. 9 See Pryor 2004 and Markie 2005. For Cohen’s response, see his 2005. For further discussion, see Moretti 2020, chapter 5, and McCain and Moretti 2021, pp. 43–49. 10 See, for example, Alston 1985, p. 59, and BonJour 1985, p. 8. Compare Kelly 2003. 11 There are some detractors, though. See, for example, Littlejohn 2012, who argues that justifcation entails truth. 12 Pollock and Cruz 1999, p. 92. 13 See Chisholm 1977. 14 It might be argued that standing beliefs are justifed only when they are occurrent beliefs and, at that time, justifed by an occurrent seeming. That does not strike me as an attractive position. 15 What if, due to a brain lesion, attentively refecting on whether p is true causes you to faint? Well, once you have fainted, you are no longer refecting. So, I don’t think that’s a counterexample. 16 For further elaboration of the regress issue, see my 2019. 17 “NP” stands for “not possible” and “BN” for “bad news.” The “→” symbol can be read as standing for “requires” or “depends on.” The frst premise is a justifcation-version of the KR principle in Cohen 2002. While Cohen argues that denying principles like KR leads to the problem of easy knowledge, he concedes that endorsing KR results in getting stuck with another puzzle. The Pyrrhonian Argument can be seen as a way of making that puzzle explicit. 18 For helpful comments and discussion, I’m indebted to Robert Audi, Earl Conee, Mylan Engel, Richard Feldman, Kevin McCain, Alastair Norcross, Bruce Russell, Scott Stapleford, and Briana Toole.
References Alston W. (1985). ‘Concepts of Epistemic Justifcation.’ The Monist: 68, 57–89. Audi R. (2004). The Good and the Right. Princeton, NJ: Princeton University Press.
22
Matthias Steup
BonJour L. (1985). The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press. Chisholm R. (1977). Theory of Knowledge. Englewood Clifs, NJ: Prentice Hall. Cohen S. (2002). ‘Basic Knowledge and the Problem of Easy Knowledge.’ Philosophy and Phenomenological Research: 65, 309–329. Cohen S. (2005). ‘Why Basic Knowledge Is Easy Knowledge.’ Philosophy and Phenomenological Research: 70, 417–430. Goldman A. I. (1979). ‘What Is Justifed Belief?’ In George Pappas (ed), Justifcation and Knowledge, pp. 1–23. Dordrecht: Reidel. Huemer M. (2001). Skepticism and the Veil of Perception. Lanham, MD: Rowman & Littlefeld. Huemer M. (2007). ‘Compassionate Phenomenal Conservatism.’ Philosophy & Phenomenological Research: 74, 30–55. Kelly T. (2003). ‘Epistemic Rationality as Instrumental Rationality: A Critique.’ Philosophy and Phenomenological Research: 86, 612–640. Littlejohn C. (2012). Justifcation and the Truth Connection. New York: Cambridge University Press. Markie P. (2005). ‘Easy Knowledge.’ Philosophy and Phenomenological Research: 70, 406–416. McCain K. and Moretti L. (2021). Appearance and Explanation. Phenomenal Explanationism in Epistemology. Oxford: Oxford University Press. Moretti L. (2020). Seemings and Epistemic Justifcation. How Appearances Justify Beliefs. Cham: Springer. Pollock J. and Cruz J. (1999). Contemporary Theories of Knowledge. Lanham: Rowman and Littlefeld. Pryor J. (2000). ‘The Skeptic and the Dogmatist.’ Noûs: 34, 517–549. Pryor J. (2004). ‘What’s Wrong with Moore’s Argument?’ Philosophical Issues: 14, 349–378. Steup M. (2013). ‘Is Epistemic Circularity Bad?’ Res Philosophica: 90(2), 215–235. Steup M. (2019). ‘Benign Infnity.’ In B. Fitelson, R. Borges and C. Braden (eds), Themes from Klein: Knowledge, Skepticism, and Justifcation, pp. 235–257. Cham: Springer.
2
Seemings and Truth Blake McAllister
Laurence Bonjour writes, “the basic role of justifcation is that of a means to truth” (Bonjour 1985, 7). Justifed beliefs must bear some connection to truth precisely because they are justifed. This is what makes justifcation a distinctively epistemic status as opposed to a practical one. It also explains why justifcation interests us from an epistemic point of view, where the goal is (roughly) to attain true beliefs and avoid false ones. How are we to understand the nature of this truth-connection? There is, of course, the dispute between infallibilists and fallibilists over whether justifed beliefs are guaranteed to be true or only likely to be true (I will assume the latter). More pertinent is whether justifcation ensures that justifed beliefs are externally likely—that is, likely to be true in the way that reliably formed beliefs are likely to be true. If we answer in the afrmative, then we are conceiving of justifcation as an efective means to truth. Many epistemologists insist that it must be so and that justifcation would be of no epistemic signifcance were it not. An alternative conception (compatible with but separable from the frst) is to see justifcation as an intentional means to truth.1 On this conception, justifed beliefs are internally likely, meaning that the evidence within the subject’s frst-person point of view indicates, on balance, that they are true. On either rendering of the truth-connection, there is reason to question whether phenomenal conservatism can account for it. Phenomenal conservatism is the view that appearances, or seemings, confer prima facie justifcation on their content independent of any verifcation of reliability. In other words, we are permitted to implicitly trust how things seem to us until further considerations bring those seemings into question. (I’ll say more about what seemings are later.) A standard formulation of phenomenal conservatism comes from Michael Huemer: PC
If it seems to S that p, then, in the absence of defeaters, S thereby has at least some degree of justifcation for believing that p.2
DOI: 10.4324/9781003299349-4
24
Blake McAllister
Though the name “phenomenal conservatism” is new, the view itself is quite old. Arguably, restricted versions of it are endorsed by Aristotle, Epictetus, Descartes, Locke, and many others. It also enjoys a signifcant following among contemporary foundationalists, who see it as providing a simple and compelling account of immediate justifcation.3 The issue we will discuss here is whether seemings, absent defeaters, guarantee that the beliefs based on them are connected to truth in the way justifers are supposed to. Plausibly, not all of them do. On the efective rendering of the truth-connection, seemings would have to always serve as reliable grounds for believing in their content in some specifed environment (absent defeaters), but they do not. On the intentional rendering, seemings would have to always indicate the truth of their content (again, absent defeaters), but plausibly seemings do not do this without proper verifcation of their reliability. But if seemings fail to sustain the required connection to truth, then they cannot automatically confer prima facie justifcation, as PC maintains. Let’s call this the truth objection to phenomenal conservatism. The central goal of this chapter is to rebut the truth objection. I begin by explaining how phenomenal conservatism can accommodate an efective connection between justifcation and truth if necessary. However, I ultimately set this aside in favor of an exclusively intentional rendering of the truth-connection and explain how seemings can account for this connection so understood. 2.1.
Justifcation as an Efective Means to Truth
An efective truth-connection can be understood as a real connection or a would-be connection. A real, efective connection entails that justifed beliefs are in fact externally likely. A would-be efective connection entails only that justifed beliefs would be externally likely were they produced in the subject’s “home environment” (leaving open how exactly to defne this). The problem for PC is that seemings do not guarantee the external likelihood of belief in either sense, even if one lacks defeaters. For instance, the brain-in-a-vat has many undefeated perceptual seemings, but the beliefs based on those seemings are not reliably formed. This shows that seemings do not guarantee a real efective connection to truth. Moreover, if we limit ourselves to our home environments, we will still fnd many situations in which undefeated seemings do not serve as reliable grounds—say, when those seemings are unknowingly caused by bias or wishful thinking. Hence, seemings don’t guarantee a would-be efective connection to truth either. Objections of this sort to PC are rarely made explicit,4 but they lurk behind other more common objections such as the problem of cognitive
Seemings and Truth
25
penetration. This latter objection argues that seemings cannot even prima facie justify if they are brought about by bias, wishful thinking, or other tainted sources. Why not? The answer generally seems to be that the beliefs based on them are unreliable, even if one lacks defeaters. How can the proponents of PC respond? One option is to retreat from PC to a more restricted version of phenomenal conservatism, which distinguishes seemings that confer prima facie justifcation (foundational seemings) from those that don’t. Specifcally, the principle should limit foundational seemings to those that, absent defeaters, always serve as reliable grounds for belief in their content (either in the actual environment or in one’s home environment). There is an externalist way of doing this and an internalist way.5 To illustrate the externalist way, a simple process reliabilist could maintain a real or a would-be efective connection with the following sort of principle: PCPR If it seems to S that p, and believing that p on the basis of this kind of seeming is a reliable belief-forming process in S’s actual/home environment, then, in the absence of defeaters, S thereby has at least some degree of justifcation for believing that p. This principle assumes externalism about justifcation since the properties that determine whether seemings are justifers aren’t always accessible to the subject or things of which he could become aware. To remain internalist, one must give a characterization of foundational seemings that appeals only to properties that feature within the subject’s frst-person point of view. Classical foundationalists can go in this direction by defning foundational seemings as infallible seemings, where a seeming that p is infallible if and only if, necessarily, if a belief that p is properly based on that seeming, then that belief is true6: PCCF If it seems to S that p, and that seeming is an infallible seeming, then, in the absence of defeaters, S thereby has at least some degree of justifcation for believing that p. Both of these forms of phenomenal conservatism preserve an efective connection between justifcation and truth. Hence, this frst leg of the truth objection is really an objection to versions of phenomenal conservatism that do not restrict foundational seemings to those that constitute reliable grounds apart from defeaters. This includes not only unrestricted versions like PC but also restricted versions that do not design such restrictions around the guarantee of reliability. Alas, those are precisely the versions of phenomenal conservatism that I am most interested in defending. Nor am I alone in this. The contemporary
26
Blake McAllister
interest in phenomenal conservatism originated from internalists and moderate foundationalists who would be unwilling to adopt either of the above strategies of retreat.7 To these internalist proponents, the externalist strategy is unacceptable precisely because it is externalist.8 The classical foundationalist approach is also unacceptable since they think it leads to skepticism about matters of common sense (such as the existence of the external world or other minds). I would add that, from their point of view, the restriction to infallible foundations is unmotivated. That restriction is made precisely so that justifcation guarantees an efective connection to truth, but the aforementioned proponents think that, as it pertains to justifcation, the guarantee of an efective truth-connection should be relinquished. Justifcation guarantees only an intentional connection to truth. On the purely intentional approach, a justifed belief is only guaranteed to be internally likely. This means that the evidence accessible from the subject’s frst-person point of view will indicate (on balance) that the belief is true. But this does not always correlate with the external likelihood of the belief in one’s actual or home environment. Is this a plausible way of understanding the connection between justifcation and truth? There are two worries with the purely intentional approach. The frst is that the contemporary notion of justifcation descends from early moderns like Descartes, and they thought of justifcation as both an intentional and efective means to truth. On what grounds do we abandon half of this vision? The second worry is that, if we do drop the efective connection, then justifcation shouldn’t be of any epistemic interest to us. The following speaks for many: Why should we, as cognitive beings, care whether our beliefs are epistemically justifed? . . . What makes us cognitive beings at all is our capacity for belief, and the goal of our distinctively cognitive endeavors is truth: we want our beliefs to correctly and accurately depict the world. . . . And, if our standards of epistemic justifcation are appropriately chosen, bringing it about that our beliefs are epistemically justifed will also tend to bring it about . . . that they are true. If epistemic justifcation were not conducive to truth in this way, . . . then epistemic justifcation would be irrelevant to our main cognitive goal and of dubious worth. (BonJour 1985, 7–8) These are large issues that cannot be fully resolved here, but I will ofer an initial response. A well-known lesson from early modern epistemology is that we cannot ensure the truth of our beliefs no matter how carefully we reason—at least not once we step beyond the incorrigible. Less appreciated is that we cannot guarantee the external likelihood of our beliefs any more
Seemings and Truth
27
than we can guarantee their truth. No amount of contentiousness will save us if we have the misfortune of entering an epistemically hostile environment like that of Descartes’ evil demon or if we simply evolved to have unreliable faculties in the frst place. Richard Foley has, more than anyone, exhorted us to accept this fact: Epistemology cannot provide us with a recipe that if followed stringently enough will guarantee that we won’t fall into great error. The search for such assurances is doomed from the start. . . . We cannot get out of our intellectual skins to provide guarantees of our own reliability. (Foley 1993, 62) If this is right, then we must severe the two halves of Descartes’ project: the project of believing in accordance with what the evidence indicates to be true, and the project of believing in a way that reliably mirrors the world. Beliefs that succeed with respect to the former project are internally rational and enjoy an intentional connection to truth. Beliefs that succeed with respect to the latter project are externally warranted and boast an efective connection to truth. Success in one of these projects does not entail success in the other. This lack of entailment does not imply that the projects are completely unrelated. A refectively stable, rational belief system must include the belief that following the evidence is in fact a reliable way of forming beliefs.9 But this belief is consistent with the understanding that following the evidence is not necessarily reliable and that were we in diferent environments those two things might come apart. Thus, the coincidence between internal rationality and external warrant is accepted as grace, not as something that following the evidence fully merits for us. Once we draw the distinction between internal rationality and external warrant, the question arises as to which we should identify with justifcation. There are reasons, I think, to identify justifcation with the former: namely, that the contemporary notion of justifcation descends from Descartes, and Descartes thought of justifcation most fundamentally as an intentional means to truth.10 We know this because it is the only way to explain why Descartes thought that justifed beliefs were both internally rational and externally warranted. It is plausible (though ultimately mistaken) to think that following the evidence entails that one’s beliefs will be true, but no one thinks that one’s beliefs being true entails that one has followed the evidence. So, if Descartes is where we should look to fx the referent, then “justifcation” will refer to internal rationality and is properly thought of as an intentional means to truth. We simply understand, as Descartes did not, that being justifed in this way does not come with any guarantees of truth or even external probability.
28
Blake McAllister
But is justifcation epistemically interesting anymore? Yes, because from the frst-person point of view, the only intelligible way of pursuing the epistemic goal is by believing in accordance with what the evidence on balance indicates to be true, and it is from this frst-person perspective that we are forced to operate. Imagine that your goal is to get home. In what direction should you go? In the direction of your home, obviously. This is what you should do in an efective sense. But how are you to implement that advice? You cannot simply do that. If you manage to do it, it is only by virtue of trying to do it, and the only intelligible way of trying to do it is by doing that which the evidence indicates is most likely to do it. That is, go in the direction that your evidence indicates will take you home. This is what you should do in the intentional sense. What’s the alternative? To go in the direction that the evidence indicates won’t take you home? Poppycock. For similar reasons, what you should believe in the intentional sense continues to be epistemically signifcant even after we acknowledge that believing in that way may not produce true, or even reliable, beliefs. To the contrary, the signifcance of most efective conceptions of justifcation stands on shakier footing than that of intentional conceptions, and this is precisely because such views tend to drop the intentional truthconnection altogether.11 Most epistemologists today are fallibilists. They do not think of justifcation as guaranteeing truth, but only the likelihood of truth. But if truth is the goal, and the importance of a positive epistemic status is correlated solely with its efectiveness in reaching that goal, then why should we settle for the mere likelihood of truth? All we should care about is whether the belief is true, not whether it is likely. But then we should not care about justifcation since it only guarantees us the latter. The only epistemic norm that should interest us is the norm of truth. Even if we grant that reliability is usually easier to secure than truth, to even consider practicability rather than mere efectiveness is to abandon the singular fxation on efectiveness and to reintroduce consideration of intentionality. In short, when justifcation becomes an instrument for efectively getting at truth and nothing more, it becomes irrelevant. This irrelevance can be avoided if we simply admit that our interest in justifcation is about something other than, or at least more than, efectiveness. In summary, while there are versions of phenomenal conservatism that can account for an efective connection between justifcation and truth, these are not principles that most contemporary proponents of PC will fnd appealing. This is because those proponents think of justifcation, in its essence, as an exclusively intentional means to truth.12 I have ofered reasons to think not only that this is the correct way of thinking about justifcation but also that justifcation so understood should remain of great interest to epistemologists.
Seemings and Truth 2.2.
29
Justifcation as an Intentional Means to Truth
Even if we concede that justifcation guarantees only an intentional connection to truth, the question remains whether seemings, apart from any verifcation of their reliability, are sufcient to establish such a connection in the absence of defeaters. Consider a classic case from Bonjour where the subject lacks an intentional connection to truth: Norman, under certain conditions which usually obtain, is a completely reliable clairvoyant with respect to certain kinds of subject matter. He possesses no evidence or reasons of any kind for or against the general possibility of such a cognitive power or for or against the thesis that he possesses it. One day Norman comes to believe that the President is in New York City, though he has no evidence either for or against this belief. In fact the belief is true and results from his clairvoyant power under circumstances in which it is completely reliable. (BonJour 1985, 41) Norman’s belief is roundly agreed to be, from the frst-person point of view, only accidentally true. This is because Norman has no indication that the content of his belief is true. Now make the following revision: “One day it strongly seems to Norman that the President is in New York City, and he comes to believe it on this basis, though he has no (other) evidence either for or against this belief.” Arguably, the two cases are not that dissimilar. Sure, Norman has a seeming in the revised case, but he also has no evidence either for or against the reliability of this seeming—if he did, then that would constitute additional evidence either for or against his belief. Hence, Ali Hasan remarks: When we add to the Norman case that it seems to Norman that the president is in NY . . . and leave everything else in the case as is, this intuitively makes no diference to Norman’s epistemic perspective. (Hasan 2013, section 4) Jack Lyons agrees, writing: Norman the clairvoyant is appeared to the-president-is-in-New-Yorkly. . . . [D]o we—does the internalist—want to say that anyone thus appeared to is prima facie justifed in believing the president is in New York? . . . This does not strike me as the sort of view an internalist would want to endorse; nor is it very plausible in its own right. (Lyons 2013, 26)
30 Blake McAllister Andrew Moon adds that, at the very least, such cases “seem to be just as challenging to seeming internalism as they are to reliabilism” (Moon 2018, 253).13 According to Bonjour, what Norman needs is a metajustifcation vouching that his belief is based on reliable grounds. In other words, Bonjour endorses something like the following principle14: META
In order for S’s belief that p to be internally likely, S must have justifcation for believing that S’s belief that p is externally likely.15
PC fouts the requirement for metajustifcation by allowing seemings to justify themselves without evidence of reliability being in place, and so ostensibly fails to guarantee an intentional connection to truth. The frst step in defending PC is to recognize that the requirement of metajustifcation is too demanding. META says that to have justifcation for believing some proposition p requires that one frst have justifcation for believing some higher-order proposition q; but, applying META again, one has justifcation for believing q only if one has justifcation for some even higher-order proposition r; and so on. From a foundationalist perspective, this prevents one from having justifcation to believe anything. For the foundationalist, justifcation is linear, and so the justifcation of p depends on having justifcation for q. Accordingly, whatever is supposed to justify p is, considered in itself, only a conditional justifer—it justifes p only if one has prior and independent justifcation for q. And the same is true of those things that are supposed to justify q, r, and so forth. What we are left with then (at best) is an infnite chain of conditional justifers, each one situated so as to justify its target proposition if the preceding link in the chain is justifed. But a series of conditional justifers, no matter how far back they stretch, cannot actually justify anything. No more than a series of conditional premises could logically entail an unconditional conclusion.16 A conditional justifer must inherit the ability to actually justify from a preceding justifer that already possesses this ability, but the ability to actually justify cannot always and at all points be inherited. For that ability would not exist in the frst place and could not be passed on from one conditional justifer to another if there were not something that possessed it in and of itself, apart from any metajustifcation. Once we relinquish the demand for metajustifcation, however, it becomes unclear why seemings should be considered insufcient to establish an intentional truth-connection absent defeaters. Our discussion of metajustifcation reminds us that, given foundationalism, there must be some way of securing such a connection apart from having higher-order evidence vouching for the belief’s reliability. What we need, then, is a kind of mental state that can secure this connection all on its own. This would
Seemings and Truth
31
be a state that intrinsically indicates the truth of its content. That is, merely being in this sort of state must, in and of itself, give some indication to the subject that its content is true. Let’s call a state of this sort foundational evidence. What kind of mental state could serve as foundational evidence? Seemings ft the bill as well as any.17 This is because seemings present their content to the subject as true. Like utterances, thoughts have not only content but also force, or a mode in which they present that content. Akin to assertive utterances, seemings present their content to the subject with assertive force. This is accomplished within the phenomenal character of the seeming—it feels as though this content is representative of the way things actually are. I have compared this elsewhere to the feeling one has when one is directly aware of the correspondence between one’s thought and the state of afairs that thought describes. It is no wonder then that this phenomenal character is sometimes called the feel of truth. It is quite literally the feeling one has when directly aware of the truth of some proposition.18 In light of this, a seeming seems like precisely the sort of mental state that can intrinsically indicate the truth of its content and so serve as foundational evidence. Simply by having a seeming that p (and not by having any evidence or beliefs about that seeming) one feels as though one is directly aware of the truth of p. The feeling that one is aware of p’s truth must surely count as some indication that p is true from the subject’s frst-person point of view. This feeling does not guarantee that p actually is true, or even that belief in p is externally probable, but it does indicate the truth of p from the frst-person point of view such that, if the subject is to form an attitude towards p at all, belief is the best option.19 Indeed, the more pressing question becomes whether anything else can intrinsically indicate the truth of its content as seemings can. There is a strong case that the answer is no.20 In brief, if the mental state does not present its propositional content as true, then one must rely on background evidence to establish the connection between that content and truth. Or if the mental state lacks propositional content altogether, then one will again need background evidence in order to appreciate that whatever is given in that mental state matches the propositional content one hopes to justify. Finally, if that mental state has both propositional content and feels true, then that state is a seeming. But what to say about Norman? We are now in a position to appreciate a salient diference between the original case—where Norman fnds himself struck out of nowhere with a completely baseless belief—and the revised case—where Norman’s belief is based on what feels like a revelatory experience. From Norman’s own perspective, the latter belief clearly has something going for it that the former belief does not. In the revised case, it is no longer plausible to call Norman’s belief accidentally true when
32
Blake McAllister
it feels to him that he is directly aware of the correspondence between the content of that belief and the fact it describes. Norman’s belief is indicated to be true from his frst-person point of view. What of those who still intuit that Norman’s belief is not justifed and, so, not indicated to be true in the manner required for justifcation? Firstly, the fact that a seeming intrinsically indicates the truth of its content does not guarantee that this content has sufcient indication to be justifably believed. The indication provided may only be strong enough to justify some positive level of confdence. Kevin McCain and Luca Moretti, for instance, say that “mere seemings” (ones that lack presentational or sensory phenomenology) only provide “an infnitesimal amount of positive evidence.”21 If we go in this direction, then we must retreat from PC, which guarantees some level of justifcation for believing in the absence of defeaters, to a version of phenomenal conservatism which merely guarantees justifcation for some positive doxastic attitude or other.22 My preferred formulation is: PC* If it seems to S that p is true, then (because of that and to that extent) S has foundational evidence for p. Applying something like this principle to Norman, McCain, and Moretti write: An internalist can, and should, admit that Norman’s epistemic position has improved, by an infnitesimal amount, because of his mere seeming. After all, it seems to him that the President is in New York, and he does not have any defeaters for this—it does not strike him as an odd seeming to have, or something he could not possibly know, or whatever. (McCain and Moretti 2021, 72) This is hard to argue with. Are we to maintain that Norman’s belief has absolutely nothing going for it from the frst-person point of view, even though its content feels evident? A second response, compatible with the frst, is to note that, despite Bonjour’s stipulations to the contrary, it is exceedingly hard to envision Norman as actually lacking any evidence against the reliability of his clairvoyant seeming. For one, Norman, if he is an ordinary adult, will have copious evidence about which sorts of faculties humans do and do not possess, and so he is bound to have defeaters (in this case, undercutting defeaters) that draw his clairvoyant seemings into question.23 For another, anyone with a basic knowledge of the United States will know that the chances of the President being in New York on any particular day are very low. So even if Norman’s clairvoyant seeming did provide some moderate
Seemings and Truth
33
evidence for this conclusion, it likely wouldn’t overcome the background evidence against it. It is perhaps our tacit recognition of these defeaters that makes us resistant to granting Norman justifcation in the revised case. What we intuit is that Norman’s belief is not ultima facie justifed, but that is of course compatible with its being prima facie justifed as PC maintains. Indeed, to truly imagine Norman lacking defeaters, we must imagine him as being in something of an infant-like state, wholly ignorant of what the world is like or what basic modes of access to that world humans like himself enjoy. Envision him as a kind of epistemological Adam, created whole cloth in adulthood, who is stretching out his mind for the frst time to see what the world is like. To begin, imagine that it seems to Norman that there are physical objects all around him of various shapes and sizes. What a remarkable ability to just tell that there are objects present and what they are like. Of course, what might seem magical to him were he to refect on it is familiar to us as ordinary perception. Do Norman’s perceptual seemings justify him in believing (or taking some other positive doxastic attitude)? Do they provide the requisite intentional connection to truth? Surely yes, or else we are all in trouble. Norman’s situation is essentially where we started in infancy, and we are only now able to build a case for the reliability of our perceptual faculties because we were then justifed in believing on the basis of those initial perceptual seemings.24 Now change the example slightly. Instead of it seeming to Norman that there are objects in his immediate environment, it seems that there are objects in some distant environment (although he would have no sense of what was “distant” or not since that notion is defned in terms of what we can ordinarily perceive). The situation is exactly like before. Remember, Norman has no idea whether he has sensory organs any more than he knows whether he has clairvoyant faculties. So, if Norman’s perceptual seemings sufciently indicate the truth of their content despite lacking evidence about the reliability of the faculties that give rise to them, then so do his clairvoyant seemings. Unlike Norman, you and I have a host of reasons for favoring perceptual seemings over clairvoyant ones. To give one example, we know through a posteriori observation that humans have reliable perceptual faculties but not clairvoyant ones. To give another example, a priori refection may assure us that the rich phenomenology of perceptions (which clairvoyant seemings presumably lack) makes them especially trustworthy in some regard. All well and good, but we must judge the situation from Norman’s point of view, not ours. And Norman, by stipulation, has no evidence for or against the reliability of the seeming (not even a priori evidence). Thus, such considerations cannot be taken into account by Norman. For him, the two cases of perception and clairvoyance really are alike. The upshot is that when we get an accurate sense of what it would really be like for Norman to lack any defeaters, it becomes quite plausible that
34
Blake McAllister
Norman’s clairvoyant seemings do provide some level of justifcation (for a positive level of confdence if not for full-fedged belief). At the very least, this is just as plausible a stance to take on the matter as the contrary, in which case proponents of the truth objection have failed to make their case. More generally, I think there is wisdom in not placing too much weight on our intuitions about such extreme cases, given that they are so difcult to imagine.25 What is it like to lack all evidence about the reliability of one’s seemings and the truth of one’s beliefs? It’s hard to say with any confdence. While I do not think our intuitions about such cases are useless,26 they should perhaps not stand alone as the centerpiece of one’s argument. And this applies just as much to those who would leverage such intuitions against phenomenal conservatism as those who (like me) would appeal to them in making the case for phenomenal conservatism. 2.3.
Two Dilemmas for PC
Lurking behind our discussion of PC and the intentional truth-connection are two perennial objections to foundationalism and internalism, respectively. On the one hand, there is Bonjour’s dilemma (a more precise version of Sellars’ dilemma), which asks how a state can justify without itself requiring justifcation.27 On the other hand, there is Bergmann’s dilemma, which asks how a state can make belief non-accidental without justifably conceiving of the state as relevant to belief and so setting of a vicious regress.28 In parting, I’d like to gesture, if only briefy, at the solutions to these problems implicit in our discussion above. To begin, seemings are not susceptible to normative evaluation in the way Bonjour’s dilemma supposes.29 The subject does not take any afrmative stance towards a proposition simply by virtue of its seeming to be true (as in belief). Nevertheless, a seeming presents its content in an assertive mode, thereby indicating its truth. Bonjour will demand a metajustifcation for why this seeming’s “assertion” should be trusted.30 However, as noted above, it is eminently plausible to think that the felt truth of a proposition counts at least somewhat in its favor even absent such verifcation. The felt truth of a proposition also makes its truth non-accidental from the subject’s perspective, and this is true even if the subject does not refect on this fact—merely having the seeming is enough.31 Of course, for a belief to be non-accidental, it must also be properly based on that seeming.32 The threat of regress arises only if we think that properly basing belief in p on a seeming that p requires justifably conceiving of that seeming as relevant to the belief. But this overintellectualizes the belief-forming process. The phenomenology of a seeming can directly incline us to believe its content. That is, the triggering condition for our rational belief-forming dispositions can be p’s feeling true, rather than S’s refecting on the fact that p feels
Seemings and Truth
35
true. Thus, the belief that p can be rationally based on the seeming that p without believing (at all) that the latter is relevant to the former, but a belief that is rationally based on the felt truth of its content could hardly be called “accidental” from the subject’s perspective.33 In sum, seemings are positioned to deftly split the horns of both dilemmas threatening internalism and foundationalism. Is there another internalist principle of foundational justifcation that can say the same? Notes 1 The distinction between efective and intentional means is from Kvanvig 1998, 434f. 2 Huemer 2007, 30. Defeaters should be broadly construed here to include background evidence (including a priori evidence) that gives that content a low prior probability. See the discussion in McAllister 2023, ch. 2. 3 See McAllister 2023, ch. 8, for a list of some historical and contemporary proponents. 4 Though see DePoe 2020, 82–83. 5 See Bergmann 2013a on the externalist way. 6 I argue in McAllister 2023 that many classical foundationalists should be understood as proponents of a restricted version of phenomenal conservatism, though I am sure many would eschew that label. 7 Huemer 2001 is the prime example, though this would also include the frst round of defenders following in Huemer’s footsteps. 8 For some of my own reasons for endorsing internalism, see McAllister 2023 and McAllister 2023, ch. 1. 9 A belief system can be rational without such a belief, but it will be vulnerable to defeat upon refection. 10 See McAllister 2022. 11 Kvanvig 1998. 12 As noted above, such proponents will think that in fact justifcation is an efective means to truth. They will simply deny that this is guaranteed by the nature of justifcation itself. 13 See also Reiland 2015, 524–525. 14 See Bonjour 1978, 5–6. 15 META does not require that S believe that S’s belief that p is externally likely, but only that S have justifcation for believing this. 16 . . . & s ⊃ r & r ⊃ q & q ⊃ p //∴ p is not valid. 17 The main competitor to seemings is perhaps beliefs. Epistemic conservatives maintain that a belief has some measure of justifcation simply because it is held. I argue that we should prefer phenomenal conservatism over epistemic conservatism in McAllister 2021 and McAllister 2023, ch. 7. 18 See McAllister 2023, ch. 4. I also argue there for the superiority of an experiential view of seemings over the alternatives and defend this experiential view against objections. This builds on the discussion in McAllister 2018. 19 For a more extended argument, see McAllister 2023, ch. 8. 20 For a complete defense, see McAllister 2023, ch. 6. 21 McCain and Moretti 2021, 71. I use this as a helpful example. I argue in McAllister 2023, ch. 9, that many mere seemings provide stronger evidence than this.
36
Blake McAllister
22 Such a principle can still allow that some seemings justify belief. It simply does not claim that this is true of all seemings. 23 See Moon 2018, section 4.2. In Moon’s terminology, Norman is past “the age of accountability” (265). 24 Alston 1991, ch. 2–3. See also Moon 2018, section 5. 25 Coren 2021. 26 I rely on them in building a case for phenomenal conservatism in McAllister 2019 and, more recently, McAllister 2023, ch. 8. 27 Bonjour 1978. 28 Bergmann 2006, ch. 1–2. 29 Huemer 2001, 97–98. 30 Bonjour 2004, 358. 31 C.f., Rogers and Matheson 2011, 60f. 32 This corresponds to the well-known distinction between propositional and doxastic justifcation. Bergmann 2013b, 171f, seems to think it an objection to PC that a seeming that p, without proper basing, does not guarantee that belief in p is non-accidental, but PC (a principle of propositional justifcation only) makes no such claim. See Gage 2016, 53–54 and Moretti and Piazza 2015, section 3. 33 For a full explanation of how seemings resolve Bergmann’s dilemma, see McAllister forthcoming.
References Alston, William P. Perceiving God: The Epistemology of Religious Experience. Ithaca, NY: Cornell University Press, 1991. Bergmann, Michael. Justifcation Without Awareness: A Defense of Epistemic Externalism. Oxford: Oxford University Press, 2006. ———. “Externalist Justifcation and the Role of Seemings.” Philosophical Studies 166, no. 1 (2013a): 163–184. ———. “Phenomenal Conservatism and the Dilemma for Internalism.” In Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism, edited by Chris Tucker, 154–178. Oxford: Oxford University Press, 2013b. BonJour, Laurence. “Can Empirical Knowledge Have a Foundation?” American Philosophical Quarterly 15, no. 1 (1978): 1–14. ———. “Externalist Theories of Empirical Knowledge.” Midwest Studies in Philosophy 5, no. 1 (1980): 53–73. ———. The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press, 1985. ———. “In Search of Direct Realism.” Philosophy and Phenomenological Research 69, no. 2 (2004): 349–367. Coren, Daniel. “Epistemic Conservatism and Bare Beliefs.” Synthese 198 (2021): 743–756. DePoe, John. “A Classical Evidentialist Response to PC.” In Debating Christian Religious Epistemology: An Introduction to Five Views on the Knowledge of God, edited by John DePoe and Tyler Dalton McNabb, 82–85. London: Bloomsbury, 2020. Foley, Richard. Working Without a Net: A Study of Egocentric Epistemology. Oxford and New York: Oxford University Press, 1993.
Seemings and Truth
37
Gage, Logan. “Phenomenal Conservatism and the Subject’s Perspective Objection.” Acta Analytica 31, no. 1 (2016): 43–58. Hasan, Ali. “Phenomenal Conservatism, Classical Foundationalism, and Internalist Justifcation.” Philosophical Studies 162 (2013): 119–141. Huemer, Michael. Skepticism and the Veil of Perception. Lanham, MD: Rowman & Littlefeld, 2001. ———. “Compassionate Phenomenal Conservatism.” Philosophy and Phenomenological Research 74, no. 1 (2007): 30–55.Kvanvig, Jonathan L. “Why Should Inquiring Minds Want to Know?: ‘Meno’ Problems and Epistemological Axiology.” The Monist 81, no. 3 (1998): 426–451. Lyons, Jack C. “Should Reliabilists Be Worried About Demon Worlds?” Philosophy and Phenomenological Research 86, no. 1 (2013): 1–40. McAllister, Blake. “Seemings as Sui Generis.” Synthese 195, no. 7 (2018): 3079–3096. ———. “A Return to Common Sense: Restorationism and Common Sense Epistemology.” In Restoration & Philosophy, edited by J. Caleb Clanton, 35–78. Knoxville, TN: University of Tennessee Press, 2019. ———. “From One Conservative to Another: A Critique of Epistemic Conservatism.” Southwest Philosophy Review 37, no. 2 (2021): 167–186. ———. “Justifcation Without Excuses: A Defense of Classical Deontologism.” American Philosophical Quarterly 59, no. 4 (2022): 353–366. ———. “Rescuing a Traditional Argument for Internalism.” Synthese 201, no. 4 (2023): 1–22. ———. Seemings and the Foundations of Justifcation: A Defense of Phenomenal Conservatism. Abington, UK: Routledge, 2023. ———. “How Seemings Resolve Bergmann’s Dilemma for Internalism.” Acta Analytica (forthcoming). McCain, Kevin and Luca Moretti. Appearance and Explanation: Phenomenal Explanationism in Epistemology. Oxford: Oxford University Press, 2021. Moon, Andrew. “How to Use Cognitive Faculties You Never Knew You Had.” Pacifc Philosophical Quarterly 99, no. S1 (2018): 251–275. Moretti, Luca and Tommaso Piazza. “Phenomenal Conservatism and Bergmann’s Dilemma.” Erkenntnis 80, no. 6 (2015): 1271–1290. Reiland, Indrek. “Experience, Seemings, and Evidence.” Pacifc Philosophical Quarterly 96, no. 4 (2015): 510–534. Rogers, Jason and Jonathan Matheson. “Bergmann’s Dilemma: Exit Strategies for Internalists.” Philosophical Studies 152, no. 1 (2011): 55–80.
3
Nonsubjectivism About How Things Seem Matthew McGrath
3.1.
Introduction
On a phone call with my friend, I listen to complaints and hear irritation in his voice about small matters. “You’re in a bad mood today, Seth,” I say. Challenged, I refect that I could mention things he’s said, the way he said them, etc. Worrying that I couldn’t articulate a strong case in this way, instead I say, “you seem to be in a bad mood.” In saying this, I think of myself as giving a reason to think he’s in a bad mood, one that somehow “gets at” the “real” reasons, viz. those hard-to-articulate but observable facts about what he said, his tone of voice, etc.—facts that are not about me, let alone my mind, but about him—facts that others could have observed if they were listening in and knew my friend well enough to pick up on them. Cases like this give us: Datum: in a range of familiar cases, we cite it seems/seemed that p in defense of p, and in doing so, we see ourselves as “getting at” support for p provided by a relevant body of “worldly” facts—that is, facts not about us, let alone our minds, but about the objects and events relevant to p. Other things being equal, in giving an account of its seeming that p, we should want to vindicate this datum. We should want our account to accommodate and explain how in such cases it seems that p, if true, very often is a reason to think that p, at least pro tanto and defeasibly. We should also want to explain how in citing it seems that p in such cases, we are somehow “getting at” worldly facts that support p. Let’s now remind ourselves of the standard view of how things seem. The standard view, rarely articulated but often presupposed, is what I’ll call simple subjectivism. On this view, its seeming that p is a matter of its seeming to relevant subject(s) that p. This view might seem unavoidable. DOI: 10.4324/9781003299349-5
Nonsubjectivism About How Things Seem
39
How could it simply seem that p? Mustn’t there be someone to whom it seems that p? Its seeming that p is like x’s being to the left. It needs saturation: it seems to S that p; x is to the left of y. The simple subjectivist grants that we often talk about how things seem without explicitly mentioning a subject. This is a convenient shorthand. When we assertively utter, “it seems that p,” what we say is true if it seems to the relevant subject(s) that p. As a default, the relevant subject is the speaker, but only as a default. John Hawthorne and Maria Lasonen-Aarnio, for instance, claim that a defendant’s remark in a courtroom, “although it seems that I am guilty, I am innocent,” asserts something about how things seem to the jury or audience, not to the speaker.1 Now consider Datum. According to simple subjectivism, when I claim to Seth, “you seem to be in a bad mood,” I’m citing a fact about myself, viz that it seemed to me that this is so—that I have a “seeming” to this efect.2 Is this fact a reason to think he’s in a bad mood? Unsupplemented with any claim about the reliability of my seemings, we might be skeptical. But even if it’s a reason, does it “get at” support from worldly facts? My seeming is caused by worldly facts, but this makes the relevant sort of “getting-at” quite indirect. When I say, “you seem to be in a bad mood,” I feel I’m saying something about those facts and not about myself. I don’t intend these considerations to come close to refuting simple subjectivism. But they should make us wonder whether there are defensible alternatives to it, ones that vindicate Datum more straightforwardly. Simple subjectivism is a species of subjectivism. The core idea of subjectivism is that its seeming to S that p (for appropriate subject S) is explanatorily prior to its seeming that p. The simple subjectivist insists that when it seems that p there must be a subject to serve as a “witness” for its so seeming—an instance of its seeming that p to a subject S. The sophisticated subjectivist, by contrast, gives an account of its seeming that p in terms of counterfactual or dispositional claims about things seeming certain ways to subjects under certain conditions. This allows for unwitnessed cases of its seeming that p. Sophisticated subjectivism might seem to perform better on Datum. In citing its seeming that p we wouldn’t be citing a reason that concerns ourselves in the frst instance but something closer to the worldly facts. We should also entertain the possibility of giving up subjectivism. This is the option I explore in this chapter. On the nonsubjectivist view I present, it seems that p if (roughly) the relevant body of facts supports p. This view, in essence, takes Datum and builds an account around it. We can see how, in cases like the phone call or the detective, it seems that p “gets at” support from a relevant body of worldly facts on this sort of nonsubjectivist view. The truth of it seems that p in these cases trivially requires that such support obtains. And in cases in which the body of facts is germane to
40
Matthew McGrath
whether p, it’s plausible that the fact that this body of facts supports p can itself lend support to p. Thus, this sort of nonsubjectivism will vindicate Datum. I say this view is nonsubjectivist. It does not explain it seems that p in terms of actual or possible facts of the form it seems to S that p. Still, it retains elements that might not qualify it as “objectivist,” including—but, as we’ll see, not limited to—the notion of a perspective, to which I appeal in explaining the relevant body of facts.3 Many epistemologists take seemings to comprise one of the fundamental sources of epistemic justifcation for our beliefs.4 I do not argue directly against such views here. However, I suspect that these views lose some of their appeal once we are clearer on the nature of the reasons we cite when we cite it seems that p in defense of p. For, if I am right, we are not citing facts about our “seemings.” Here is the plan for the chapter. In Section 3.1, I argue against subjectivism, both simple and sophisticated, without drawing directly on Datum. In Sections 3.2 and 3.3, I develop and defend a nonsubjectivist view. 3.2. Against Subjectivism My case against subjectivism proceeds by way of an argument against subjectivist semantics about talking about how things seem. If this semantics is mistaken, then barring a special philosophical use of “seems”, subjectivism itself is mistaken, too. And I think we can put aside a special philosophical use here. In cases such as the phone call case, “seems” is not used in a special philosophical way. The semantical accounts will concern bare “seem” sentences in English, that is, sentences of the form “it seems that p,” as well as ones that result from raising the grammatical subject of the embedded sentence such as “NN seems to be F” and “NN seems F,” and tensed variants.5 I employ contextualist semantics in formulating these accounts to provide fexibility about the relevant subject(s), as mentioned above.6 Let’s start with simple subjectivist semantics, the semantical theory that pairs naturally with simple subjectivism. According to this theory, an utterance of a bare “seems” sentence is true if it seems to the contextually relevant subject(s) that the target proposition is true. Setting aside any context-sensitivity associated with “p”, a sentence of the form “it seems that p” will be true in a context if it seems that p to the relevant subject(s) at the time of the context. I’ll raise problems for this theory before turning to sophisticated subjectivist semantics in Section 1.4.
Nonsubjectivism About How Things Seem 3.2.1.
41
Disputes
You and a friend are together at a barbecue. You both pay attention to the host’s manner, his conversation, his expressions, etc., as the barbeque proceeds. As the two of you are leaving the barbeque, you say to your friend, “the host seemed arrogant.” Your friend disagrees: “No, he didn’t. He seemed reserved, not arrogant.” Each of you might cite certain features of his manner to back up your claim or counter the other’s. At a certain point, if the dispute isn’t resolved, one or both of you might simply agree to disagree. At this point, you might retreat to explicitly autobiographical claims: “well, he seemed arrogant to me.” And this is a retreat, not a restatement. Or suppose Sherlock Holmes and Dr. Watson are at an early stage of their investigation. Still, they might argue over whether it seems, at this point in their investigation, that Moriarty is the killer. Neither says outright that he is the killer, but Watson says, “as of now, it seems that Moriarty is the killer,” while Holmes disagrees, “no, it doesn’t.” Each cites evidence in their joint possession to support their case. Such disputes are not about whether the relevant proposition seemed to be the case to one or both conversational parties or to anyone else. If they were, they would be decided once it became clear how things seem or seemed to the relevant people. But this doesn’t decide them. Nor are they disagreements merely over whether the target proposition is true. You and your friend are disputing how the host seemed, not, or at least not directly, whether he is arrogant.7 The existence of such disputes is evidence against simple subjectivist semantics. 3.2.2.
Error
Even in the barbeque case, you and your friend each think the other is in error about how the host seemed. One of you really could be in error; perhaps your friend is a poor judge of the signs of arrogance and just is wrong about whether the host seemed arrogant. But let’s consider an even clearer case of error. At the house of an acquaintance, Celia, I’m looking over a large record collection (Celia is otherwise occupied). Seeing a lot of recordings of Prokofev, Rachmaninof, and Stravinsky but no Bach, Mozart, Haydn, and none from any other 18th-century composers, I betray my ignorance of classical music by saying: (C) It seems that Celia likes 18th-century classical music.
42
Matthew McGrath
What I say is false. The only records in the room are of compositions by late 19th-century and later composers. It does seem to me that Celia likes 18th-century music. We can further suppose that the people with me are just as ignorant of classical music as I am. To them, too, it seems that Cecilia likes 18th-century classical music. Add, lastly, that this is how it seems to Celia herself. She knows that she does like this music, and she may be refecting on her fondness of it at the same time as we are having our conversation. (It turns out that she keeps her record collection of 18th-century composers in a diferent part of the house.) The problem with simple subjectivist semantics is that it will incorrectly classify my statement C as true.8 Such errors are found in non-perceptual cases as well. Dr. Watson may be wrong in his claim, “Moriarty seems to be the killer.” Suppose he focuses on one element of the case that points to Moriarty, neglecting several other elements known to him and Holmes that together undermine the force of the Moriarty-pointing elements. Watson’s statement is false. It would be false even if it seemed to Holmes, too, in a moment of clouded judgment, that Moriarty was the killer. 3.3.3.
Its Seeming That p Without Its Seeming to Anyone That p
Finally, we can test simple subjectivist semantics by determining if there are true statements about how things seemed at times when there were no subjects to whom they seemed that way. A prosecutor shows security camera video footage from a break-in. The person in the video didn’t act in ways one would expect him to act if he knew he was on flm. The prosecutor then says, “the whole time he was in the shop the defendant seemed unaware of the security camera.” This statement is in the past tense. The relevant past time is not the time of the showing of the video but the time of the break-in. At that time, there was no one to whom it seemed that he was unaware of the camera, and yet the seems statement is true.9 Think, second, of how natural it is to refer to how things seem when narrating true stories about past events. Before The Cretaceous—Paleogene extinction event, dinosaurs had dominated the earth for millions of years. They had no serious rivals and were well adapted to their environments. It seemed that dinosaurs were well-suited to dominate the earth for many years to come. And then the asteroid hit! It would not be a good objection to the highlighted statement that it must be wrong because no one had such a seeming at the time.
Nonsubjectivism About How Things Seem
43
There are further incorrect predictions. Consider this dialogue: A: B:
I wonder how the accused, John, seemed the morning before the crime. Did he seem calm? agitated? It would be good to fnd out. No one encountered him. So, he couldn’t have seemed any way at all.
B’s remark is absurd but is correct under simple subjectivist semantics, at least assuming, as is plausible, that John himself is not a relevant “S.” 3.3.4.
Sophisticated Subjectivism
A natural fallback position is sophisticated subjectivism. This pairs with sophisticated subjectivist semantics, according to which bare seem statements are true at a time if, at that time, it would seem to the right sorts of subjects in the right circumstances that p. What counts as the right sort of subject might vary with context, but often it would be a “normal” or “generic” subject. On this semantics, we can arguably fnd a point of disagreement in the barbeque case and an error in the record collection case, and we do not have to deny the possibility of things seeming ways at a time when they didn’t seem those ways to anyone at that time. Consider again: (D) It seemed that dinosaurs were well-suited to continue to dominate the earth for many years to come. When uttered as in the earlier speech. It is not sufcient for the truth of D that it would now seem to normal people, upon having the relevant information, that dinosaurs would have continued to dominate the earth. D is in the past tense. What we would need is for it to be true back then that the relevant sorts of subjects with the relevant sorts of information would have been disposed to have the relevant seeming. This in itself is peculiar. (Could there have been subjects with the relevant concepts back then at all?) Other statements create even more serious problems. Imagine I’m telling an episode of the history of an English village: The Saxon hoard under the Medieval town church was completely unknown for centuries. It was very well hidden indeed. It seemed that no one would ever have any idea of its existence. But then, randomly, some local teens decided to build an underground hiding place beneath the church and made the discovery. How could the underlined statement—call it “F”—be true on the sophisticated semantics? We would need to consider whether, at the relevant time(s),
44
Matthew McGrath
an appropriate subject given the relevant information, would be disposed to have it seem to them that no one would ever have any idea of the existence of the Saxon hoard under the kirk. But could a rational subject have such a self-falsifying seeming? In general, although it may have seemed that no one would ever have any idea of the fact that p, it can’t be true that a rational subject would be disposed to have a seeming to this efect. Thus, sophisticated subjectivism gets the truth-conditions for D and F wrong.10 There are ways to attenuate subjectivism. Perhaps what counts is not whether the right sort of subject would have a seeming that p, but whether they would have a spectator-like seeming: in that situation, p. Such a view could potentially handle many of the cases we’ve considered, including the dinosaur and Saxon hoard cases—although it would still have to put subjects back at the time of the dinosaurs, etc. Perhaps this can be made to work. Rather than pursuing this line further, I want to try a diferent tack. 3.3.
A Nonsubjectivist Account
Go back to the phone call case. Suppose my statement, “you seem to be in a bad mood” is true. Let’s consider how a sophisticated subjectivist would try to secure this result. This will point us toward a form of nonsubjectivism. What sort of information will the relevant generic subject have? It will have to be information about what Seth said and his manner on the phone call. This information corresponds to a perspective occupied by me during the phone call. We can imagine Seth’s partner at their ofce has also truly said at around the same time, “Seth’s in a good mood today,” thinking about his smile and the cheerful emails he’s sent her today. For his partner’s statement to be true, we’d need the generic subject to have a diferent body of information, corresponding to a diferent perspective. The sophisticated subjectivist needs to appeal to bodies of information in perspectives. She then adds claims about which seemings generic subjects with such information would have. I propose to appeal directly to the information—to the set of facts in the perspective—and to do without generic subjects and their seemings, in favor of what those facts support. I think there is something phenomenologically attractive about going this way. When I think “it seems that p,” I feel I am thinking about what the relevant facts support, rather than how things would seem to some generic subject, let alone a pure spectator looking down on our world. To start, consider a simple nonsubjectivist semantics, according to which “it seems that p,” is true in context C if the facts in the C-relevant perspective support p. This account accommodates many cases we have considered. But it could be usefully elaborated a bit further. Do the facts about which recordings Celia has in her music library support the idea that she
Nonsubjectivism About How Things Seem
45
likes 19th- and 20th-century Russian classical music? Yes, but arguably only given certain background information, for example, about a person’s music collections generally refecting their musical tastes. This gives us: “It seems that p” is true in context C if the body of facts in the C-relevant perspective, together with the C-relevant body of background information, supports p sufciently.11 3.4.
More Details
In this section, I fll in some further details in response to some natural questions: • • • •
What is a perspective? What is it to occupy a perspective? Why must the “information” in a perspective consist of facts rather than propositions, which could be true or false? Must the background information also consist only of facts? Can the facts in the relevant perspective ever be about the psychological states of the speaker? How does context determine truth-conditionally relevant perspectives?
3.4.1.
Perspectives and Occupation
We can model a perspective as a pair of a body of facts and a time. A subject occupies a perspective either by having knowledge of or by having the right sort of access (e.g., perceptual) to the information at the time. In the barbeque case, the perspective includes, in its informational component, facts that were observable for guests at the event. Its time is the time of the event. You and your friend occupy it insofar as this information at the time is perceptually accessible to you. In the case of Holmes and Watson discussing their murder case, the relevant perspective consists of a body of focal information about the crime. They occupy the perspective, not by virtue of being in a position to access it perceptually, but by knowing it. And we can have intermediary cases in which a subject occupies a perspective in part by knowing some of the contained information and in part by being positioned to access other parts of it. This notion of perspective is an abstraction from the notion of a subject’s epistemic position. A subject at a time might know certain facts and have perceptual or other forms of access to other facts. We can then take these facts—or a subset of them—together with time as a perspective. Some perspectives are not only unoccupied but also unoccupiable. Consider a perspective whose informational component includes the fact that
46
Matthew McGrath
no thinking subjects exist (at the time). A thinking subject couldn’t occupy this perspective. But the facts in the perspective can still support certain propositions.12 3.4.2.
Facts vs. Propositions
Suppose Roger is sure of the truth of the (false) claim that the US 2020 election was rigged for Biden; this is something he takes himself to know. Suppose he expresses this belief aloud and adds, “thus, it seems that Biden was not legitimately elected.” What Roger says is false. It’s false even if his interlocutors add his claim about a “rigged” election to the common ground of the conversation. If false propositions could enter into the information of a perspective, the content of my false belief would enter here, in which case Roger’s seems statement would be true, which it isn’t. Again, if any old false but believed “information” could enter into the background information, it would enter in for our original Celia case, and so “it seems Celia likes 18th-century classical music” would be true, which it isn’t. Still, there are cases that I fnd difcult to accommodate without allowing for some falsehoods to enter background information. Consider Simpson’s paradox. To use Titelbaum’s (2022) illustration of it, suppose I’m told player A’s success rate is higher than player B’s for 2-point basketball shots as well as for 3-point shots. I might then say, “so, it seems that A’s overall success rate has to be higher, too.” We could secure the truth of my statement if we allowed as part of the background the false assumption that A and B took 2- and 3-point shots at the same rate. If we don’t allow this, it is harder to see how it could be true. One option, then, is to allow for certain false assumptions to fgure in the background “information.” The crucial thing, I think, is to capture the intuitive idea that it seems that p is true if the relevant facts provide support for p. The support can be defeasible. When I claim, “it seems A and B must have the same overall success rate,” I do so because I think there is this support. The hard question is whether there really is such support or not. Obviously, if there is, it is defeated. But is it there at all? I am not sure. But we can say this: if there is such support in this case, then we may need to allow certain reasonableby-default but false assumptions into the background. This doesn’t mean opening the foodgates. The relevant facts in the Celia case do not support the claim that she likes 18th-century classical music. I am tempted to say the following: A probability pro could reply to me, “no, it doesn’t seem A has to have the same overall success rate as B,” and then explain how this requires an assumption that is not stated as true in the example. His claim is plausibly true. In response to this, I might retreat instead to “it seemed to me.” That is certainly true.
Nonsubjectivism About How Things Seem
47
However, I will leave this issue open. How we resolve it, I think, hinges on how we understand a body of facts supporting a proposition. If there can be support in some cases when that support depends on a false assumption, then we should let false assumptions of the relevant sorts into the background. If, as I’m inclined to think, there can’t be such support, we should insist on the background consisting only of truths.13 (The next section will be relevant to these issues as well.) 3.4.3.
Perspectives and Mental States
Can facts about mental states be elements of a perspective? Yes. For instance, when we say we seem to remember that p, the relevant facts are presumably psychological, such as a feeling as if one remembers or a memory image. The same goes for “I seem to see.” However, often psychological elements are not included in the contextually relevant perspective. When I look at the darkening sky and hear the thunder, my statement “it seems it’s going to rain” is in virtue of facts about the observable signs of rain, not in virtue of facts about my experience or psychological states. And that is why it is correct to add counterfactuals denying dependence on one’s psychological phenomena “and so even if I wasn’t here it would still seem it’s going to rain.” The same goes for the barbecue case and the recollection collection case. Nevertheless, it is possible to introduce psychological elements into a perspective. In thinking about radical skeptical hypotheses, I might reason like this: “I have visual experiences, auditory experiences of being before a fre. I have experiences as of heat. So, it seems a fre is in front of me.” Here, the fnal statement is plausibly true and remains true even if we imagine a brain in a vat doing the reasoning. My preceding remarks help determine the relevant perspective as one that includes my psychology. One can do this for the Simpson’s paradox case, too: “I have a strong intuition that A and B must have the same overall success rate. It seems it must be true.” 3.4.4.
Contextual Determination of Relevant Perspectives
How is the truth-conditionally relevant perspective determined in context? The usual sorts of answers are these: salience (i.e., by which perspectives are salient); relevance (by which perspectives are such that their support of p would matter for the purposes of the conversation); accommodation (i.e., by which perspectives make the speaker’s remark come out true); and normative factors (i.e., by which perspective the speaker or relevant subject should or is expected to occupy). Two comments here. First, these factors can pull in diferent directions. Suppose a member of a Congressional committee neglected to acquaint
48
Matthew McGrath
himself with the information in the fles that members were charged to examine. At the hearing, he relies on only a small subset of this information and says, “it seems there was no wrongdoing by the President.” Suppose the total information in the fles overall clearly supports the claim that there was wrongdoing by the president. Then relevance and normative factors outweigh accommodation: what the Congressman says is false. The relevant perspective includes all the information in the fles, even though the speaker doesn’t possess that information. He doesn’t occupy the perspective, but he should. Second, again, the contextually relevant perspective needn’t include the speaker’s total perspective, nor does it even need to include the total perspective determined by the common ground of the conversation.14 This is what makes possible cases in which one truly says, “although it seems that p, we know that not-p.” It is an interesting question under what conditions such statements can be true. As Brogaard (2018) puts it, some seemings “go away” in the presence of counterevidence, while others don’t. To put it in terms of my account, while it may be true to say “it seems that Jane is not at the party,” after being told by partygoers of this, it is extremely hard to say it truly again when Jane appears right before us. By contrast, we can truly say, “it seems the stick is bent,” despite it being common knowledge that it is straight. A question I won’t try to answer here is whether, simply by appealing to the four factors above—salience, relevance, accommodation, and normativity—we can explain why some perspectives can become contextually relevant easily while others cannot at all or cannot at least without great efort.15 3.5.
Objections
I have space to discuss only a few objections. 3.5.1.
Too Weak?
One objection is that my account gives the wrong result in cases of complicated but cogent arguments. In such a case, if a person hasn’t grasped any argument for p from that information, the person might say, “it doesn’t seem that p.” The objection is that this can be true, even though the information in the perspective does support p. I am unsure of the force of the objection. Suppose Dr. Watson says after such an investigation, “it doesn’t seem Moriarty did it.” Suppose the connection between the information and Moriarty’s guilt is so complicated that only a brilliant detective like Holmes can piece it together. Still, if Holmes is in the conversation, he can reply, “no, Watson, it does seem Moriarty did it,” and then proceed to outline the reasoning. What Holmes says can be true.
Nonsubjectivism About How Things Seem
49
However, if we do want to amend my account to cope with “hard-todiscern” support relations, we can do so as follows: We can say that “it seems that p” is true in C if facts in the C-relevant perspective, relative to the C-relevant background, provide strong enough support that is also suitably accessible. What counts as suitably accessible might itself vary with context. 3.5.2.
Too Strong?
We have already discussed the most pressing objection that my account is too strong (see the discussion of Simpson’s paradox in Section 3.2). But there is a second sort of objection worth discussing as well. Don’t we regularly say, “it seems that p” of-the-cuf, with little serious thought? If we were thinking about what the facts of the relevant perspective support, presumably we wouldn’t be so cavalier. We do occasionally give quick assent to “it seems that p.” Sometimes we’re just wrong in what we say. We can be corrected, “no, it doesn’t seem that way,” followed by an explanation of how the relevant information doesn’t support p. However, it matters a great deal which body of information is the relevant one. Sometimes the relevant body is very limited. And it can sometimes be a simple matter to determine whether a small sliver of information provides support. As the body of information grows, of-thecuf answers may become less reasonable. 3.5.3.
Sloppiness over “to S”
Often, we are happy to exchange “seems to me” for “seems.” If there were such a big diference between these, as there is in my view, why would we be so cavalier about it? There are big diferences in meaning between “I think that p” and “p,” and yet in conversation we’re often sloppy about which we utter. The diference in meaning nevertheless shows up in several ways. If one says “p,” one sticks one’s neck out: if p is false, the statement was incorrect. We can argue about whether p is the case, but not (or not as easily) about whether you believe it. The same goes for “it seems that p” at least as compared to “it seems to me that p.” The diferences in meaning are real, even if in many conversations we don’t in fact bother with the diferences. Another point is relevant as well. We often want to focus on the full perspective that we ourselves occupy (perhaps this is the default). We take ourselves to know certain information, and we’ll inquire about whether that information supports p. In this sort of case, the key diference between “it seems to me that p” and “it seems that p” is that the latter endorses
50
Matthew McGrath
the existence of a supportive relation, whereas the former hedges. If there is no support, “it seems that p” is false, but “it seems to me that p” could still be true.16 3.6.
Conclusion
If what I have argued here is correct, facts about how things seem are not “subjectivist.” They are not, and are not explained by, facts about how things do or would seem to actual or possible subjects. Rather, they concern what the facts of the relevant perspective sufciently support. How this afects epistemology is a question I leave for another day. But I suspect it does afect it. In many cases in which it is hard to fnd reasons for our beliefs, we have a reason in “that’s how it seems.” This reason isn’t about our minds in the typical case, but about the world. The reason is that some portion of the world backs up what we believe.17 Notes 1 This example is given in an early draft of their (2021). 2 A word about terminology: in discussing how things seem to a subject, I will use the jargon of “seemings.” It seems to S that p if S has a “seeming” with the content that p. I take these seemings to be mental states, but I make no assumptions about exactly what sort they are, for example, beliefs, inclinations to believe, sui generic presentational states. I also don’t assume that all such seemings are of the same sort. 3 I draw on Conee (2013) and especially Hawthorne and Lasonen-Aarnio (2021), though without the subjectivism implicit in their work. See also Cappelen’s (2012, ch. 2) and Tooley’s (2013). 4 Philosophers who see seemings, or some subset of seemings, as a source of justifcation include Bergmann (2013), Brogaard (2013), Chudnof (2013), Cullison (2013), Huemer (2001), Pryor (2000), and Tucker (2010). 5 One clarifcation here. In some cases, “to S” is elided. Suppose I ask you, “how did Jack seem to you?” You answer, “he seemed happy.” This is presumably a case of elision. The sentence uttered includes “to me.” What is uttered, then, is not a bare “seem” sentence. 6 Some philosophers of language might accept a close relative of subjectivist semantics, but for assertoric content rather than semantic content. I leave such technical matters to the side here while noting that, as far as I can see, my main arguments will apply equally well against such views. 7 In some cases, we’re not trying to get at how things are at all, only at how they seem. If you’re trying out for a certain acting role, such as the role of an arrogant person, the question of whether you really are arrogant won’t arise at all—only whether you seemed arrogant during your tryout. Two casting agents might disagree over whether you seemed arrogant in your tryout. 8 For readability, I will often allow myself below to speak loosely of “statements” about how things seem, using this term in some cases for bare “seem” sentences in context and in others for the propositions expressed by such sentences in context.
Nonsubjectivism About How Things Seem
51
9 Vary the case a bit. Imagine there was a security ofcer remotely monitoring the footage in real time. The ofcer truly said at the time of the break-in, “he seems unaware of the camera.” When the prosecutor later says, “he seemed unaware of the camera,” she says the same thing; they both state a fact about how things seemed. So far, so good for the simple subjectivist; after all, there was a contemporaneous seeming at the time (the security ofcer’s seeming). But what the prosecutor says in this example is presumably the same as what the prosecutor says in the original example, in which a contemporaneous seeming is lacking. One could insist that what the prosecutor says in the original example is false, but it would be exceedingly odd for someone to protest against the prosecutor, even in making a picky point, “that isn’t right, because he didn’t seem that way to anyone at the time of the break-in.” Think, also, about how the detectives might talk about what can be learned from the video as they’re about ready to view it for the frst time. “Was the thief aware of the camera? We should be able to tell if he seemed aware of it or not. Let’s roll the tape!” What is to be found out is naturally presumed to be something already true, not something that will become true as it’s found out. 10 For a yet more fanciful version of the problem, imagine a time at which it was highly unlikely, given what was happening then, that life would ever arise in the universe. At that time, it seemed life would never arise. And since seemings, as mental states, require life, it was also true that it seemed then that there would never be a case of its seeming to anyone that something was the case. Presumably, we do not want to explain the truth-conditions for it seemed at the time that seemings would never be enjoyed in terms of dispositions to enjoy seemings to the efect that no one would ever have any seemings. 11 A few words on the relationship between this account and that of Conee (2013) and Hawthorne and Lasonen-Aarnio (2021). Conee (2013) considers only what is “conveyed” by claims about how things seem to a subject. But I follow him in appealing to the notion of support from evidence. My account owes much to Hawthorne’s and Lasonen-Aarnio’s. I follow them (in their early draft) in distinguishing bare seemings statements from those that specify a subject of the seeming. I follow them as well in claiming that which body of information is relevant for the truth of a claim about how things seem varies with context. However, as I read them, they take the truth-conditions to be subjectivist. They write: An alternative idea is that “seems” claims say roughly that a proposition has positive epistemic status from a certain subject’s perspective (perhaps simpliciter, perhaps relative to a salient subset of the subject’s evidence), while falling short of ascribing outright belief. (Here it is useful to compare seemsconstructions to various other so-called hedge constructions such as “I have the impression that,” “It strikes me that,” and so on.) (2021, 2n6) For them, it is still the relevant subject’s evidence that counts, rather than a body of facts from a (possibly unoccupied) perspective. In my view, the issue of “approximating” outright belief is not relevant to bare seems statements, though it is relevant to seems to S statements: the latter are of a kind with ascriptions of outright belief but fall short of it, whereas the former ascribe no mental state at all. 12 As those in the consciousness debate will know, some philosophers employ the notion of a perspective or point of view in arguing for theses about the
52
13 14 15
16
17
Matthew McGrath subjectivity of experience. But as John Biro (2006) has pointed out, there is little in such notions by themselves to back this up. The clearest notion of a point of view, according to Biro, is that of a location of a sort that is occupiable by a subject, where the location may be straightforwardly spatial but may also be an element in a diferent sort of “space”, such as “attitudinal space.” Applied to attitudinal space—or, for my purposes, “epistemic space”—the notion of a point of view or perspective is content-based rather than owner-based. It is individuated by reference to contents rather than by occupants. Indeed, as Biro points out, using the content-based notion, we can truly say that there are unoccupied points of view. This may reduce the force of it seems that p as a reason to think p in some situations, but it is consistent with there being signifcant force in many situations, especially when the relevant background information is known to be true. Hawthorne and Lasonen-Aarnio (2021) make this point. See Note 11. I have proposed a full-on nonsubjectivism about how things seem, relying on a nonsubjectivist semantics for bare “seems” statements. It would be interesting to explore a weaker position about how things seem based on the weaker semantical view that bare “seems” statements sometimes have nonsubjectivist truth-conditions. Such a view is worth exploring, but I lack the space to do so here. Drawing epistemological conclusions from such a view might be a more subtle afair. Thanks to Kevin McCain for his comments on this issue. In future work, I hope to give an account of the relationship between how things seem and how things seem to a subject. The idea I plan to pursue is that the truth-conditions for “it seems to S that p” subjectivize the truth-conditions of “it seems that p.” The relevant perspectives and the background are composed not of genuine information but of putative information; the support condition becomes S’s fnding the putative information to support p, relative to the putative background. Here, fnding x to be F doesn’t entail that x is F. Warm thanks go to Jeremy Fantl, Allan Hazlett, Dave Liebesman, Maria Lasonen-Aarnio, Kevin McCain, Ernest Sosa, Scott Stapleford, and Chris Willard-Kyle for their written comments, as well as to members of the epistemology groups at Rutgers and at Washington University in St. Louis and an audience at the 2022 Seemings conference at the University of St. Thomas in Fredricton, New Brunswick. Special thanks to John Hawthorne for many vigorous discussions of these issues in the fall of 2018.
References Bealer, George (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 Littlefeld), pp. 201–239. Bengson, John (2015). “The Intellectual Given,” Mind 124 (495): 707–760. Bergmann, Michael (2013). “Externalist Justifcation and the Role of Seemings,” Philosophical Studies 166 (1): 163–184. Biro, John (2006). “A Point of View on Points of View,” Philosophical Psychology 19 (1): 3–12. Brogaard, Berit (2013). “A Sensible Dogmatism,” in C. Tucker (2013): 270–292. ——— (2018). “Phenomenal Dogmatism, Seeming Evidentialism and Inferential Justifcation,” in K. McCain (ed.), Believing in Accord with the Evidence: New Essays on Evidentialism (Cham, Switzerland: Springer).
Nonsubjectivism About How Things Seem
53
Cappelen, Herman (2012). Philosophy Without Intuitions (Oxford: Oxford University Press). Conee, Earl (2013). “Seeming Evidence,” in C. Tucker (2013): 52–70. Conee, Earl and Richard Feldman (2004). Evidentialism (Oxford: Oxford University Press). Chudnof, Elijah (2013). Intuition (Oxford: Oxford University Press). Cullison, Andrew (2013). “Seemings and Semantics,” in C. Tucker (2013): 33–51. Hawthorne, John and Maria Lasonen-Aarnio (2021). “Not So Phenomenal!” Philosophical Review 130 (1): 1–43 Huemer, Michael (2001). Skepticism and the Veil of Perception (Lanham, MD: Roman and Littlefeld). Jackson, Frank and Philip Pettit (1990). “Program Explanations: A General Perspective,” Analysis 50 (2): 107–117. McAlister, Blake (2018). “Seemings as Sui Generis,” Synthese 195 (7): 3079–3096. Pryor, James (2000). “The Skeptic and the Dogmatist,” Noûs 34 (4): 517–549. Titelbaum, Michael (2022). Fundamentals of Bayesian Epistemology Volume 1 (Oxford: Oxford University Press). Tooley, Michael (2013). “Phenomenal Conservatism and the Principle of Credulity,” in C. Tucker (2013): 306–327. Tucker, Christopher (2010). “Why Open-Minded People Should Endorse Dogmatism,” Philosophical Perspectives 24 (1): 529–545. ———, ed. (2013). Seemings and Justifcation (Oxford: Oxford University Press).
4
Against the Phenomenal View of Evidence Disagreement and Shared Evidence Elizabeth Jackson
4.1.
Introduction
Several philosophers defend the view that seemings play a central epistemological role. Michael Huemer and other phenomenal conservatives argue that seemings confer epistemic justifcation: “If it seems to S that p, then, in the absence of defeaters, S thereby has at least some degree of justifcation for believing that p” (Huemer 2007: 30).1 Here, I’m concerned with a variation of this view: the view that seemings are evidence. Most authors who connect seemings with epistemic justifcation afrm this, as they maintain that seemings confer justifcation because of their evidential role. And many phenomenal conservatives argue directly that seemings are evidence. For example, Chris Tucker (2011: 52) argues that “if it seems to a subject that P, then the subject thereby possesses evidence which supports P.” McCain and Moretti (2021) also defend the view that seemings are evidence; on their view, diferent types of seemings provide more or stronger evidence than others (depending, among other things, on explanatory ft). Others who argue that seemings are evidence include Pust (2000), Yandell (1993), and McAlister (2016, 2021). These authors don’t endorse this phenomenal view of evidence for no reason. As Kelly (2008) points out, the phenomenal view straightforwardly explains our access to evidence and provides a response to skepticism. On the other hand, it also makes evidence easy to come by—perhaps too easy, if every seeming has evidential weight.2 The worry I raise in this chapter is related to this concern, but focuses on disagreement. More precisely, I’ll argue that the phenomenal view of evidence makes it extremely difcult, if not impossible, for people who disagree to share evidence. Consider three possible connections between seemings and evidence: (a) If it seems to S that p, S has evidence for p. (b) If S has evidence for p, it seems to S that p. (c) It seems to S that p if S has evidence for p. DOI: 10.4324/9781003299349-6
Against the Phenomenal View of Evidence
55
The problem I raise in this chapter will be a problem for those who endorse (a) or (c)—I’m challenging the idea that seemings are sufcient for having evidence, but not the idea that they are necessary. Phenomenal conservatives largely accept (a) and (c); Tucker, for example, explicitly endorses (a), along with a number of those cited above. Insofar as (a) and (c) are widely endorsed by phenomenal conservatives, this may be a problem for the view more generally. However, I’ll target proponents of (a) and (c) specifcally here; this may not be a problem for those who endorse (b) only, or perhaps those who connect seemings and justifcation without appealing to evidence. The chapter proceeds as follows: In Section 4.2, I motivate a key claim: that S1 and S2 may disagree about whether p and share the evidence that bears on p. I show how this claim falls naturally out of several literatures in epistemology, including the permissivism and disagreement literatures. In Section 4.3, I explain why this key claim, along with other plausible premises, creates problems for the phenomenal view of evidence—specifcally, for (a) and (c). I also respond to objections. I conclude in Section 4.4. 4.2.
The Key Claim: Disagreeing People Can Share Evidence
In this section, I explain and motivate a key claim that conficts with the phenomenal view of evidence: Key claim: Possibly, S1 and S2 disagree about whether p and share all the evidence that bears on p. A few clarifcations about this claim. First, by “disagree” I simply mean that S1 believes p and S2 disbelieves p (or believes not-p). While there may be other ways of disagreeing (involving withholding belief or credence), this chapter focuses on this basic, uncontroversial way of disagreeing. “Share evidence” is a little harder to defne, as it hinges on controversial issues, some of which are at stake in this chapter (e.g., What is evidence? What does it mean for evidence to bear on a proposition? What does it mean to have evidence?). I don’t want to be overly committal, and I do want to invoke a pre-theoretical notion of evidence. That said, note two things the second conjunct of the key claim does not mean. First, S1 and S2 need not share all of their evidence full-stop; they simply need to share the evidence that bears on some proposition. Second, S1 and S2 do not need to be epistemic peers in the sense often invoked in the disagreement literature. For arguments that epistemic peers are rare, see King (2012) and Matheson (2014).
56
Elizabeth Jackson
Why should we think that the key claim holds? I’ll discuss at least two reasons. The frst involves the large (and still growing) literature on the epistemology of disagreement: how should we respond when smart, educated people disagree with us? Steadfasters argue that we can continue to hold onto our beliefs in the face of disagreement, whereas conciliationists argue that we should alter our beliefs in some way. However, suppose that when two people disagree, they never share evidence. This throws a wrench in the debate and makes its key questions much less interesting. For one thing, if all diferences in beliefs vary with diferences in relevant evidence, it’s hard to see what would motivate smart, educated, disagreeing people to conciliate. Furthermore, if disagreement with shared evidence is impossible, this wouldn’t merely mean that most people we encounter every day don’t share our evidence; this means that, even in idealized cases, disagreeing people couldn’t share evidence. This is a hard pill to swallow. Even those who think evidence is rarely shared in real life still acknowledge evidence is shared by disagreeing parties in idealized cases (see, e.g., Matheson 2014). The second consideration in favor of the key claim comes from the permissivism literature. The permissivism literature concerns whether a body of evidence can rationally permit more than one attitude toward a proposition. Those asking this question hold the evidence fxed, and ask if diferent responses to that evidence could be rational. This presupposes that disagreeing people—who take diferent attitudes to a proposition—can share evidence. Furthermore, most of this literature has focused on arguments for and against interpersonal permissivism: the view that two agents with the same evidence could both be rational, even if taking diferent attitudes. While there’s been a more recent interest in intrapersonal permissivism (concerning a single agent and her evidence), the majority of the literature on interpersonal permissivism would concern an impossible case if the key claim were false. Why care if two disagreeing persons with the same evidence can both be rational if disagreeing persons can’t share evidence in the frst place? Considering possible disagreement is not helpful here either, because if the key claim is false, in a possible world where S1 shares S2’s evidence, S1 would agree with S2.3 Kopec and Titelbaum put the point this way: in a discussion about why the conception of evidence is central to the permissivism debate, if we were to use a very mentalistic notion of evidence that includes every thought crossing through an agent’s head, we get a thesis that seems trivially true. As soon as one agent judges that P while the other judges that not P, the two agents would have diferent “evidence.” (Kopec and Titelbaum 2016: 191)
Against the Phenomenal View of Evidence
57
Thus, both the disagreement and the permissivism literature seem to presuppose the key claim: that disagreeing persons can share evidence. Furthermore, recent debates in epistemology either don’t make sense or are based on a false presupposition if the key claim is false. Now, I’ll argue that this key claim conficts with the phenomenal view of evidence. 4.3.
A Problem for the Phenomenal View of Evidence
We can now see the problem with the phenomenal view of evidence when considering the case of our disagreeing persons, S1 and S2. We can combine their situation with the phenomenal view of evidence as follows: (1) S1 believes p. (2) If S1 believes p, it seems to S1 that A. (3) If it seems to S1 that A, then S1 has evidence that A. (1*) S2 believes not-p. (2*) If S2 believes not-p, it seems to S2 that B. (3*) If it seems to S2 that B, then S2 has evidence that B (where B≠A). Given this set up, I’ll argue that defenders of the phenomenal view of evidence are committed to: (4) S1 does not have S2’s evidence that B, and S2 does not have S1’s evidence that A. And (4) conficts with the key claim. Let’s discuss each of these in turn. (1) and (1*) are true by stipulation. (2) and (2*) are perhaps the most controversial, but there’s good reason to think they are true: (2) and (2*) essentially state that people who disagree have diferent seemings. This does not mean that seemings are beliefs, or that if it seems to S that p, then S believes p. Things can seem true to us but we don’t believe them (e.g., because we have a defeater); it also could be appropriate to have contradictory seemings, but it’s never okay to have contradictory beliefs. To see why defenders of the phenomenal view of evidence are committed to (2) and (2*), we’ll consider two diferent views of the relationship between beliefs and seemings. The frst view is defended by Michael Huemer (2007: 40), who argues that (almost all) diferences in beliefs are directly due to diferences in seemings: when we form beliefs, with a few exceptions not relevant here, our beliefs are based on the way things seem to us. Indeed, I think that the way things appear to oneself is normally the only (proximately) causally
58
Elizabeth Jackson
relevant factor in one’s belief-formation. In other words, in normal contexts, including that of the present discussion of epistemic justifcation, one would not form diferent beliefs unless things appeared diferent to oneself in some way (belief content supervenes on appearances, in normal circumstances). Furthermore, in normal conditions, the way appearances determine beliefs is by inclining one towards believing what appears to oneself to be so, as opposed, say, to our being inclined to believe the things that seem false. According to Huemer, in the large majority of cases, if you believe p, it seems to you that p. The “exceptions” that Huemer mentions are “selfdeception and leaps of faith” and “[perhaps] severe disorders” (2007: n14). So on Huemer’s view, almost always, if S believes that p, it seems to S that p. In this case, S1 and S2 would have diferent seemings: it would seem to S1 that p, and it would seem to S2 that not-p. One might wonder how Huemer’s view fts with inferential beliefs. Consider Scott, who believes that Caesar either had epilepsy or sufered from mini-strokes (call this proposition C) because of what he’s read. Scott genuinely believes C, but it doesn’t seem to Scott that C. Scott’s belief is based on other beliefs and not based directly on a seeming that C.4 Huemer would reply that while Scott may not have the unconditional seeming that C, C seems true to Scott given E (Scott’s evidence, i.e., what Scott read). So Scott still has a seeming that C, conditional on E (see Huemer 2013: 338). Presumably, those who disagree with Scott would either not have the conditional seeming, or they would not have E, Scott’s evidence. Either way, it follows that those who disagree with Scott would not share Scott’s evidence. But not all defenders of the phenomenal view of evidence share Huemer’s view; some would argue that Scott doesn’t even have the conditional seeming that C. What’s the alternative? On another view, Scott’s belief that C isn’t based on a seeming that C, but is based on beliefs about what he’s read. Defenders of this second view don’t maintain this chain of beliefs goes on forever, though: at some point, things bottom out in a seeming. As Blake McAlister (2021: 4) says, “all of our justifed moral beliefs (and any other belief, for that matter) will ultimately be based on seemings.” So, while Scott’s belief that C may not be based directly on a seeming that C, it is based on other belief(s), which are ultimately based on seemings.5 According to this view, S1’s belief that p may not be based solely on a seeming that p. Instead, it could be based on a belief (or chain of beliefs). However, this chain would ultimately bottom out in a seeming. The same for S2: S2’s belief that not-p may also be based on belief(s) that bottom out in a seeming. But given that S1 and S2 disagree about p, their bottom-level seemings would almost certainly be diferent. This is a common occurrence
Against the Phenomenal View of Evidence
59
in philosophical disagreements that come down to a “clash of intuitions.” If their seemings were the same, it’s not clear that S1 and S2 would disagree. Therefore, regardless of the specifc view taken, S1 and S2 have diferent seemings. Either Huemer is right and S1 and S2’s disagreement is directly based on seemings. Or, alternatively, if S1 and S2’s beliefs are based on other beliefs, they would still have diferent seemings (although these seemings would have diferent content than the clashing beliefs). Finally, (3) and (3*) are statements of the phenomenal view of evidence, the target claim of this chapter (i.e., (a) and/or (c)). (3) is an instance of (a) and one direction of (c), and (3*) is (3) with variables changed. There is one thing to note about the interpretation of (3) and (3*), though: seemings must confer evidence in some special way and not merely in the generic “evidence is cheap” sense. This is because one could accept (3) and (3*) but also argue that S1 and S2 do share evidence: both S1 and S2 have the same evidence, and this evidence supports (or includes) both A and B. While this move is possible, I doubt the phenomenal conservative will want to take this position. Most phenomenal conservatives maintain that seemings have a justifcatory force: absent defeaters, seemings justify, and in fact, seemings are the primary vehicle of justifcation. And again, while perhaps not all seemings are created equal—there could be cases of weak or conficting seemings that don’t justify beliefs—the phenomenal conservative would likely not want to maintain that these are the only cases in which disagreeing parties can share evidence. Then, S1 has special evidence that A, and S2 has special evidence that B. This special evidence explains why they disagree and also explains the justifcatory role that seemings play, given phenomenal conservatism. From this, it follows that: (4) S1 does not have S2’s evidence that B, and S2 does not have S1’s evidence that A. But (4) conficts with the key claim. I suggest that (3) and (3*), and thus (a) and (c), are the claims to reject. The defender of the phenomenal view might object to the inference to (4), by appealing to the (plausible) principle that evidence of evidence of evidence. Given this principle, assuming S1 and S2 are aware of the disagreement, S1 may have evidence that B and S2 may have evidence that A (because they have evidence of evidence). So S1 and S2 might share evidence in any case.6 In response, this isn’t sufcient for S2 to share S1’s special evidence for A (and vice versa). My being aware that you disagree with me—even knowing you are smart and educated—isn’t the same as my sharing your seeming. And simply because we both have a piece of evidence for a proposition’s truth doesn’t mean we share evidence. Consider an analogy: a distant testimony that p is diferent than a frsthand
60
Elizabeth Jackson
experience that p, even though both are evidence for p. We wouldn’t say that someone with only distant testimony shares evidence with someone with frsthand experience (for one thing, diferent credences—and perhaps even diferent beliefs—would be justifed for them). So mere awareness of a disagreement (even with a peer) isn’t enough to secure shared evidence. Furthermore, two disagreeing people could share evidence without being aware of the disagreement. I don’t have to know that you disagree with me in order to share your evidence; it would be odd if shared evidence hinged on our awareness of disagreement. Note one fnal point. Suppose at least one of the parties in the disagreement is irrational, perhaps due to a fawed inference from seemings to beliefs or from beliefs to other beliefs. In this case, the phenomenal view of evidence might allow for disagreement and shared evidence. Specifcally, S1 and S2 might both have the same evidence (say, a seeming that B), but S1 irrationally believes p, even though the seeming that B supports believing not-p. Then, S1 and S2 disagree but share evidence. Note, however, that it would be surprising if the only cases where disagreeing parties could share evidence would be when one party made a faulty inference. This seems to fy in the face of the permissivism and disagreement literatures, which assume we can hold the evidence fxed and ask questions about whether reasonable disagreement can occur. These exceptional cases also do not align with the cases of ideal disagreement, in which the possibility of shared evidence is widely recognized. So even if it’s not impossible to share evidence in the face of disagreement on the phenomenal view, if it can only occur via irrationality, the problem for the phenomenal view of evidence remains. 4.4. Conclusion I’ve advanced an objection to the phenomenal view of evidence: the view that, if it seems to S that p, S has evidence for p. First, I’ve motivated the claim that two disagreeing parties can share evidence. Then, I’ve shown how this claim conficts with the phenomenal view of evidence. I’ve considered several ways out for the phenomenal view of evidence and argued that none of these are especially promising. While this, of course, is one consideration among many in the complex debate regarding phenomenal conservativism, I maintain that phenomenal conservatives owe us an explanation for the possibility of disagreement and shared evidence. Acknowledgments Thanks to Kevin McCain and Scott Stapleford for their very valuable written comments and to the members of the Board-Certifed Epistemologists group for their helpful discussion of the above material.
Against the Phenomenal View of Evidence
61
Notes 1 See Audi (1993), Chisholm (1989), Swinburne (2001), Pollock and Cruz (1999), Pryor (2000, 2004), and Huemer (2007, 2013, 2014) for defenses of the claim that seemings justify; see Hawthorne and Lasonen-Aarnio (2021) and Kelly (2008) for critical discussion. 2 See Markie (2005), Littlejohn (2011), and Tooley (2013). 3 Thanks to Kevin McCain. 4 Thanks to Scott Stapleford for this case. 5 Thanks to Kevin McCain and Scott Stapleford for helpful discussion. 6 Thanks to Kevin McCain for raising this objection.
References Audi, R. (1993). The Structure of Justifcation. Cambridge: CUP. Chisholm, R. (1989). Theory of Knowledge. Englewood Clifs, NJ: Prentice-Hall. Hawthorne, John & Maria Lasonen-Aarnio. (2021). “Not So Phenomenal!” Philosophical Review 130(1): 1–43. Huemer, Michael. (2007). “Compassionate Phenomenal Conservatism.” Philosophy and Phenomenological Research 74(1): 30–55. Huemer, Michael. (2013). “Phenomenal Conservatism Über Alles.” In Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism (C. Tucker, ed.), pp. 328–350. Oxford: OUP. Huemer, Michael. (2014). “Phenomenal Conservatism.” Internet Encyclopedia of Philosophy. www.iep.utm.edu/. Kelly, Thomas. (2008). “Evidence: Fundamental Concepts and the Phenomenal Conception.” Philosophy Compass 3(5): 933–955. King, Nathan. (2012). “Disagreement: What’s the Problem? Or A Good Peer is Hard to Find.” Philosophy and Phenomenological Research 85(2): 249–272. Kopec, M. & M. Titelbaum. (2016). “The Uniqueness Thesis.” Philosophy Compass 11(4): 189–200. Littlejohn, Clayton. (2011). “Defeating Phenomenal Conservatism.” Analytic Philosophy 52: 35–48. Markie, Peter. (2005). “The Mystery of Direct Perceptual Justifcation.” Philosophical Studies 126: 347–373. Matheson, Jonathan. (2014). “Disagreement: Idealized and Everyday.” In The Ethics of Belief: Individual and Social (J. Matheson and R. Vitz, eds.), pp. 315–330. Oxford: OUP. McAlister, Blake. (2016). Common Sense Epistemology: A Defense of Seemings as Evidence. Baylor University. Waco, TX: ProQuest. McAlister, Blake. (2021).” Understanding Moral Disagreement: A Christian Perspectivalist Approach.” Religions 318: 1–14. McCain, Kevin, and Luca Moretti. (2021). Appearance and Explanation: Phenomenal Explanationism in Epistemology. Oxford: OUP. Pollock, J. L. & J. Cruz. (1999). Contemporary Theories of Knowledge, 2nd ed. Lanham, MD: Rowman & Littlefeld. Pryor, J. (2000). “The Skeptic and the Dogmatist.” Noûs 34: 517–549. Pryor, J. (2004). “What’s Wrong with Moore’s Argument?” Philosophical Issues 14: 349–378.
62
Elizabeth Jackson
Pust, Joel. (2000). Intuitions as Evidence. New York: Routledge. Swinburne, R. (2001). Epistemic Justifcation. Oxford: Oxford University Press. Tooley, Michael. (2013). “Michael Huemer and the Principle of Phenomenal Conservatism.” In Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism (C. Tucker, ed.), pp. 306–327. Oxford: OUP. Tucker, Chris. (2011). “Phenomenal Conservatism and Evidentialism in Religious Epistemology.” In Kelly James Clark & Raymond J. VanArragon (eds.), Evidence and Religious Belief, pp. 52–73. Oxford: Oxford University Press. Yandell, Keith. (1993). The Epistemology of Religious Experience. Cambridge: CUP.
5
Appearances and the Problem of Stored Beliefs Kevin McCain and Scott Stapleford
Here’s something that should be obvious: We have many beliefs stored in memory. Here’s something else that should be obvious: At least some stored beliefs are justifed even when not currently in view. For instance, (assuming fairly typical experience) it is exceedingly plausible that not only did you believe that Thomas Jeferson was once President of the United States before you read this sentence, but also that your belief was justifed. These seemingly unremarkable facts are thought to be a source of trouble: Internalist theories of epistemic justifcation purportedly can’t explain them. There are diferent ways of pushing this objection. Two of the most discussed are the problem of forgotten evidence and the problem of stored beliefs.1 The frst problem arises because in many cases you will have forgotten your original evidence (say, for believing that Jeferson was once President of the United States). The second problem arises because it’s not clear what your current evidence might be (for believing that Jeferson was President) when you’re not at the moment considering it. Our focus is the second challenge.2 Proponents of appearance-based accounts of epistemic justifcation have options in responding to the challenge of stored beliefs.3 An extreme approach simply denies that there are stored beliefs.4 Of course, there’s no problem of explaining how stored beliefs can be justifed if there are no stored beliefs. Another approach, which we fnd especially unappealing, is to insist that justifcation doesn’t supervene on the evidence one has at a time. Rather, a stored belief simply retains whatever justifcation it had when it was originally formed.5 The cost of this move is too high because it seems to relinquish internalism. After all, two people could be mental duplicates now and diverge in justifcation. This could occur if the one person formed some belief on the basis of good evidence in the past and the other person formed the same belief without sufcient evidence, and both have forgotten their original bases. A less costly response—though one with a bad reputation—is to embrace doxastic conservatism. Stored beliefs are justifed on this view (at least to some degree) simply by being had.6 DOI: 10.4324/9781003299349-7
64 Kevin McCain and Scott Stapleford These responses have their defenders and warrant consideration. But we leave them out of account and suggest an alternative solution, one that we fnd more promising. To the problem of stored beliefs, we propose stored appearances. From where we stand, stored appearances are just the thing to justify stored beliefs.7 Our aim in this chapter is to mount a preliminary defense of stored/dispositional appearances8 and to argue that they provide a plausible solution to the problem of stored beliefs. The case will be made as follows: First, we present the problem of stored beliefs in more detail (Section 5.1). Next, we explain what appearances are, why things of that sort can be stored, and how the possibility of storing appearances solves the problem of stored beliefs (Section 5.2). Finally, we respond to likely objections to the stored appearances solution to the problem of stored beliefs (Sections 5.3 and 5.4). The frst objection rests on considerations pertaining to basing, a notion we therefore discuss. Ultimately, we argue that the objection from basing fails to establish that dispositional appearances are ill suited to justifying stored beliefs. The second objection relies upon the intuition that there are justifed stored beliefs for which there are no accompanying stored appearances. This objection fails, so we say, because the clearer it becomes that a stored belief has no accompanying stored appearances, the less clear it is that the belief is justifed; and the clearer it is that a stored belief is justifed, the clearer it becomes that there are stored appearances accompanying it. 5.1. The Problem of Stored Beliefs The problem of stored beliefs is thought to arise because it seems intuitive, on the one hand, that at least some of our stored beliefs are justifed and yet it isn’t clear, on the other hand, how an internalist theory can accommodate this obvious fact. To better appreciate the problem, it will help to refect on what it means for a belief to be stored in memory. Consider this common distinction. Beliefs can be occurrent (part of a subject’s current awareness) or non-occurrent (not part of the subject’s current awareness). It’s not the mere storing of beliefs that causes the problem. It’s storage plus non-occurrence. If being stored does not rule out being occurrent, then a stored occurrent belief can be justifed in any of the various ways that other occurrent beliefs might be justifed. Bring your phone number to mind. Right now, while you think about it, your belief that your phone number is whatever it happens to be is part of your current awareness. But that belief is also lodged in memory: It is one of your stored beliefs. That’s why you are currently remembering your phone number rather than forming a new belief about what it is. Evidence that justifes this belief is presumably before your mind now.9 The problem of stored beliefs arises for beliefs that are stored and non-occurrent.
Appearances and the Problem of Stored Beliefs
65
We can bring this problem into focus by envisioning a situation where it seems undeniable that someone has a justifed belief that is both stored and non-occurrent and yet the justifcation is hard to explain. The following example should excite the right intuition: Sleeping Sally
Sally is at her friend’s house for the frst time. While waiting for her friend to get ready, Sally encounters her cat, whom she’s never met before. The cat walks lazily across the room and Sally sees that it’s black. Once the cat shufes of to the kitchen, Sally gets cozy on the couch, where she slips into a dreamless sleep.10 Sleeping Sally puts a spotlight on the problem. Sally’s belief that her friend’s cat is black seems to be adequately justifed. It’s relevantly similar to countless everyday beliefs that we also take to be justifed: The lights are on, the room is cold, the phone is ringing, and that sort of thing. But it may also seem that Sally’s belief lacks evidence since she’s currently in a dreamless sleep: In the absence of conscious experiences, she’s not having any appearances.11 This is the problem of stored beliefs—accounting for the intuitive justifcation that one’s stored beliefs possess when one lacks occurrent appearances to support them. 5.2.
Solving the Problem of Stored Beliefs
To understand how an appearance-based theory can solve the problem of stored beliefs, we’ll frst need to get clear on the nature of appearances. Appearances are experiences with propositional content, mindto-world direction of ft, and a distinctive phenomenology.12 These are key features of appearances, according to McCain and Moretti (2021), and there is much to say about each of them. But brief comments will sufce to show the promise of our solution to the problem of stored beliefs. First, then, appearances have propositional content. This helps explain the prevalence of common expressions to the efect that it appears that p, or seems that p. An appearance in the relevant sense is always an experience that something is such or so. Second, appearances have mind-to-world direction of ft. They represent the world as being a certain way by virtue of their propositional content. It is precisely because they are representations of the world that appearances can be more or less accurate. This contrasts with mental states having world-to-mind direction of ft, desires being the stock example. Though a desire may be appropriate or inappropriate in some context, it cannot be
66 Kevin McCain and Scott Stapleford accurate or inaccurate because it represents how one would like the world to be rather than how the world is. Third, appearances are characterized by a distinctive phenomenology— what is often called “forcefulness.” The phenomenology of an appearance is such that it presents its propositional content as true. In other words, appearances “have the feel of truth, the feel of a state whose content reveals how things really are” (Tollhurst 1998: 298–299). Or, again, appearances represent their content “in such a way that it ‘feels as if’ we could tell that those propositions are true—and that we’re perceiving them to be true— just by virtue of having them so represented” (Pryor 2000: 547). To make the idea concrete, consider the diference between imagining that you are looking at a squirrel (i.e., calling up a mental image of a squirrel in your imagination) and having a visual appearance of a squirrel while walking in the park. It could be that the mental images formed in these cases are indistinguishable. However, there is a forcefulness accompanying the visual appearance of a squirrel encountered in the park which does not attend the image of a squirrel you merely conjure.13 Now that we have a handle on appearances—they are experiences that forcefully represent propositional content with a mind-to-world direction of ft—let’s turn for a moment to beliefs. Before laying eyes on the cat, Sally did not believe that her friend’s cat is black. (She had no belief as to its color.) But she did have a disposition to believe that her friend’s cat is black. All it took to give Sally the belief was for her to be placed in the right perceptual circumstances—the circumstances of seeing the cat. Once she actually looked at the cat, Sally occurrently believed that her friend’s cat is black. Later, when the belief has been put into storage—when it’s locked away in her memory—Sally dispositionally believes that her friend’s cat is black. Sally has the belief that her friend’s cat is black and is disposed to bring it to mind—given the right sorts of triggers—or to use the belief in her reasoning.14 This three-way division can be usefully applied to appearances. As Sally can have the appearance that her friend’s cat is black, she can plainly have the disposition to have this appearance. After all, Sally’s having the appearance is just the manifestation of that disposition. It should be equally obvious that Sally can have an occurrent appearance that her friend’s cat is black. As Sleeping Sally is described, we can assume that Sally’s vision is functioning properly and that she is in favorable perceptual conditions for seeing that the cat is black. Rather less obvious is whether Sally can have a dispositional appearance that her friend’s cat is black when the appearance is no longer occurrent. That’s the real heart of the matter: Can Sally have a stored appearance? We don’t see why not. Appearances are experiences, and experiences can be stored in memory for later recall. They belong to what psychologists refer to as “episodic memory.” In the case
Appearances and the Problem of Stored Beliefs
67
we’re imagining here, Sally forms the occurrent belief that her friend’s cat is black. She then stores this belief in memory in the usual sort of way (there’s nothing weird going on in the example). Later, when the belief is tucked away in memory, Sally has the dispositional belief that her friend’s cat is black. An analogous story about appearances seems to us equally plausible. When Sally frst beholds the cat, she as an occurrent appearance that it’s black. She then stores the appearance in memory such that it could be retrieved: She could later recall the appearance as an episodic memory. At some point beyond the present, when she’s not currently looking at the cat, Sally has a dispositional appearance that her friend’s cat is black. This is just to say that she can recall her experience of seeing the cat and have an appearance that it is black. Putting some thought to defeat might further motivate the idea that there can be dispositional appearances. Let’s modify Sleeping Sally so that the dispositional appearance can plausibly serve as a defeater. In this variation, Sally still has the appearance of her friend’s cat being black, but for some reason we needn’t specify she fails to form the corresponding belief. Now suppose that after the cat wanders of to the kitchen, another friend, Sadie, also stops by for a visit. She asks about the cat and mentions in passing that it’s brown. Does Sally have a defeater for Sadie’s assertion? It certainly seems that way. If Sally were now to believe that her friend’s cat is brown, the belief would not be justifed. For an appearance that the cat is black can easily be brought to mind in the form of an episodic memory. Given that Sally’s episodic memory in this case is arguably a stored appearance, it seems that not only can she have a dispositional appearance, but that dispositional appearances can impact the justifcation of her beliefs. This is our solution to the problem of stored beliefs. In cases like Sleeping Sally, the stored belief is justifed because there is a corresponding stored appearance conferring justifcation. Sally’s occurrent belief that her friend’s cat is black is justifed by her occurrent appearance that it’s black. When she stores the belief and the appearance in memory, they both become nonoccurrent. But the stored belief remains justifed because it is supported by her stored appearance.15 5.3.
A Worry About Basing
Resistance to our claim that an appearance-based theory of justifcation can solve the problem of stored beliefs may be grounded in a worry about basing. Andrew Moon (2012) expresses this worry. In pressing the problem of stored beliefs as a challenge to internalist theories (whether appearance based or not), he lets nothing ride on the existence of dispositional appearances. Even if such things exist, says Moon, stored beliefs are not based on them. If presented with Sleeping Sally, Moon would insist that Sally’s belief
68
Kevin McCain and Scott Stapleford
is not based on a stored appearance that the cat is black—even allowing for the sake of argument that Sally has such an appearance. The problem of stored beliefs thus remains. At best, internalist theories can account for Sally’s propositional justifcation that the cat is black, according to Moon. (They can explain how Sally has justifcation for so believing.) What they can’t seem to make sense of is the intuition that Sally’s belief is doxastically justifed. (They fail to render the verdict that Sally’s belief that her friend’s cat is black is actually justifed.)16 Let’s try to clarify this objection before responding. “Basing” picks out the relation that one’s belief must bear to the source of one’s propositional justifcation in order to be doxastically justifed. In the context of raising his objection, Moon assumes that basing is some sort of causal relation. Granting this assumption,17 the worry comes down to this: Should we think that a stored belief can be causally sustained by a stored appearance? For Moon the answer is “no.” He’s presumably not thinking that appearances just aren’t the sort of thing that can causally sustain beliefs. For Sally’s occurrent belief that the cat is black is uncontroversially caused by her occurrent appearance that the cat is black. If there’s a basing problem here, it must be something else. Two ways of shaping Moon’s worry come to mind. In storing the appearance that justifes Sally’s belief, the causal basis of that belief is altered (it changes from an occurrent to a non-occurrent appearance). Maybe that’s a problem: The causal basis of a belief can’t change. Or perhaps—on the second way of taking Moon’s worry—there is something special about stored beliefs and appearances that interferes with basing. That could also be a problem. We’ll consider these possibilities in turn. On the frst way of framing Moon’s worry, then, storing the appearance that supports a belief changes its causal basis. And that supposedly can’t happen. But a good reason for thinking that it can happen is that it does happen quite often. It’s a fairly routine occurrence for the causal basis of some belief to change. Consider another variation on Sleeping Sally. After Sally observes the cat, her friend comes down and announces that her own cat is currently at the vet. The cat that walked by is her roommate’s. But Sally’s friend mentions that her own cat is black as well. On the heels of this short conversation, Sally’s belief that her friend’s cat is black—the one currently away at the vet—is no longer causally sustained by the appearance from a moment earlier. But thanks to her friend’s description, she still maintains the belief. That’s the crucial point: She still maintains the belief, though one causal basis has been swapped out for another. If there’s any doubt that the causal basis has changed, try to imagine the following. Say we have some sort of neural manipulation technology that allows us to wipe specifc memories from a person’s mind while leaving other memories intact. In Sally’s case,
Appearances and the Problem of Stored Beliefs
69
we wipe the memory of the cat’s appearance while leaving in place the memory of the testimony as to its color. If the facts were so arranged, Sally would carry on believing that her friend’s cat is black. Now alter the circumstances again: Suppose we wipe from Sally’s mind the testimony as to the color of the cat. And suppose we leave the testimony that the cat Sally saw does not belong to her friend. It seems equally obvious that Sally would no longer believe that her friend’s cat is black. Her belief about the cat’s color thus counterfactually depends upon her memory of the relevant testimony and yet it doesn’t depend in this way (in the latest variation of the case) on her appearance that the cat is black. This suggests that Sally’s belief is now causally based on the testimony of her friend but not on her earlier appearance. So its causal basis has changed.18 A possible counter is to admit the general point that the causal basis of a belief can change but then deny that this happens in a case like Sally’s. When she falls asleep on the couch, we might say, the causal basis of her belief stays the same. So the stored belief that her friend’s cat is black is not based on the stored appearance. Such a claim could certainly be made, but we see no reason to accept it. The naïve metaphor of memory being something like an empty container into which beliefs and appearances are dropped “as is” has some initial attraction. But empirical research doesn’t bear this out. When a belief is transferred from working memory (where it is occurrent) to more long-term storage (where it is non-occurrent), a signifcant amount of cognitive processing takes place. There is thus very good reason, from the standpoint of cognitive psychology, to think that a belief’s causal basis can change as it comes to be stored in memory.19 On the second way of framing Moon’s worry about basing, there is some feature of stored appearances or stored beliefs that prevents the former from causing the latter. That an occurrent belief can be caused by an occurrent appearance is very easy to accept. Why does Sally believe that her friend’s cat is black while she is looking at it? Because it appears that way. Is there any good reason to doubt that something similar can be said about stored beliefs and stored appearances? Not from our perspective. It’s arguably true that one stored belief can causally sustain another belief. For example, if you have the justifed stored beliefs that p, p entails q, and q, it’s reasonable to think that the stored belief that q is causally sustained by the other two beliefs, even if only in part. Were you to lose your belief that p, after all, you might stop believing that q (assuming that q had no independent support). We see no reason to deny that a stored appearance can causally sustain a stored belief in a similar way. If Sally were to lose her stored appearance that her friend’s cat is black, it seems unlikely that she would continue to believe that it’s black. We can easily imagine cases where someone like Sally would lose the stored appearance but retain the stored belief. But this is hardly a problem. In such cases, the person either
70 Kevin McCain and Scott Stapleford has a new basis which justifes the stored belief or doesn’t. If she does, then the stored belief is justifed—being held, as it is, on a new justifying basis. If she doesn’t, then the stored belief is unjustifed. Appearance-based theories are fne either way. Such theories needn’t claim that being based on a stored appearance is the only way for a stored belief to be justifed. They’re committed to claiming that being based on a stored appearance is one way for a stored belief to be justifed. And so the worry about basing comes to nothing. It’s perfectly possible for stored appearances and stored beliefs to causally interact in ways that mirror the interactions of their occurrent counterparts. 5.4.
A Worry About Absent Appearances
We’ve argued that stored beliefs are justifed by stored appearances. And we’ve responded to a worry about basing. In both cases, the discussion revolved around a key example: Sleeping Sally. The example was used for illustration, but one might accuse us of picking cherries: We chose a thoroughly perceptual example (about seeing a cat and believing that it has a certain color). And though it’s plausible that stored perceptual beliefs are justifed by stored appearances, there are many stored beliefs that aren’t perceptual. Many of these stored beliefs lack accompanying stored appearances—at least according to the objection.20 To see how this might be so, consider a particular stored belief—say, Sally’s belief that her birthday is January 1. If Sally is like most people, she can’t recall any particular experience of learning that her birthday is January 1. It’s also very unlikely that Sally has any appearance stored along with her belief that her birthday is January 1 comparable to the stored appearance of the cat. Nevertheless, it seems that Sally’s belief that her birthday is January 1 is justifed, even while she sleeps (even when the belief is stored). But this cuts against our claim that stored beliefs are justifed by stored appearances. So what are we to say? Two things. First, if push comes to shove, we could soften our claim that stored appearances solve the problem of stored beliefs. Stored appearances would only be part of the solution. Anyone who wants to maintain an appearance-based theory of epistemic justifcation could pair the stored appearances response with some other promising idea. For example, one might appeal to something like Steup’s (2001) notion of global epistemic self-trust. Basically, the centrality of appearances to epistemic justifcation could be preserved by taking on Steup’s plausible suggestion that we have an appearance that we tend to be reliable believers. Such an appearance could provide justifcation for believing that we are reliable believers, and this could provide justifcation for thinking that our stored beliefs are true. In Sally’s case, her appearance of being a reliable believer would provide
Appearances and the Problem of Stored Beliefs
71
her with direct justifcation for thinking that she is a reliable believer and indirect justifcation for her stored belief that her birthday is January 1. Though Steup’s notion of global epistemic self-trust is defensible— and though it pairs well with an appearance-based theory of epistemic justifcation—neither it nor any other supplemental notion is forced upon us. Thinking further about Sally and her belief that her birthday is January 1 should make this clear. When Sally frst acquired the belief, it was in all likelihood on the basis of testimony. Initially, there would have been a stored appearance to go along with the stored belief that her birthday is January 1—namely, the appearance that so and so told her as much. No problem for us so far. What could be a problem is the possible circumstance that Sally’s stored belief remains justifed years later even when the stored appearance (of being told that her birthday is January 1) has long since departed. Now we must admit that if Sally’s stored belief is justifed and there is no stored appearance that justifes it, then our response is on the ropes. But we deny the antecedent: In cases where it is clear that the stored belief is justifed, it is also clear that there are stored appearances. As McCain and Moretti (2021) argue with respect to instances of forgotten evidence, it is plausible that even when the initial justifying appearance is forgotten, stored beliefs enjoy a signifcant amount of support. For example, even if Sally lacks the sort of appearance of general reliability that Steup describes, she likely has information about her track record concerning basic facts about her past. This information consists of stored appearances or justifed stored beliefs (such as memories of being right about this kind of thing in the past). Additionally, Sally has plenty of evidence by way of meta-cognition. When she recalls that her birthday is January 1, it appears to her as something she knows, it appears familiar, it appears easy to recall, and it appears that she recalls it quickly. Allowing that Sally has this evidence when she is currently recalling that her birthday is January 1, one might yet worry that she doesn’t have these appearances to justify the belief when it is stored. Nevertheless, it’s reasonable to think she had these appearances the last time she recalled the date of her birthday. And there’s no reason to doubt that such appearances were stored at the time. Hence, even though her initial stored appearance has been forgotten—and thus no longer justifes her stored belief—she has more recent stored appearances (from the last time she recalled her birthday) that do justify the belief. Now, one might respond by simply stipulating that Sally doesn’t have any of these stored appearances. In other words, one might insist that Sally has no appearance with respect to her track record on things like dates of birthdays and that on the most recent occasion of recalling that her birthday is January 1, she either had no appearances from meta-cognition or else failed to store the ones that she did. Though one could indeed make
72
Kevin McCain and Scott Stapleford
such stipulations, and thereby guarantee that Sally lacks the sort of justifying stored appearances we described, doing so makes it far less clear that Sally’s stored belief about the date of her birthday is actually justifed. After all, if it doesn’t appear to Sally that she tends to be reliable about that sort of thing, it doesn’t appear to her that her birthday is something she knows, it doesn’t appear to be a familiar fact, and it doesn’t appear to be something she recalls quickly and easily, then it’s not the least bit evident that her stored belief is justifed. To be sure, either Sally’s belief that her birthday is January 1 is accompanied by stored appearances that justify it or it isn’t. If it is, there is no problem for the stored appearances response. If it isn’t, then her belief doesn’t seem justifed, and there’s still no problem for the stored appearances response. Either way, fans of appearance-based theories of epistemic justifcation have nothing to fear in this objection and should lose no sleep over the problem of stored beliefs. Notes 1 Smithies (2019) helpfully identifes two additional challenges for internalist theories of epistemic justifcation arising from issues pertaining to memory—what he calls the “problem of stored defeaters” and the “problem of background beliefs.” For responses to these challenges on behalf of an appearance-based account of justifcation, see McCain and Moretti (2021). 2 For responses to the frst, see McCain (2015a) and McCain and Moretti (2021). 3 In much of the relevant literature, “seeming” is used interchangeably with “appearance.” In fact, “seeming” is the more common term. However, following McCain and Moretti (2021), we prefer to use the term “appearance” so as to remain neutral on important debates concerning the unity/disunity of experience. 4 Frise (2018) argues for this response conditionally. In particular, he argues that if representationalism is correct about the nature of belief, then there are no stored beliefs. Frise’s conditional formulation of the “no stored beliefs” response is fairly plausible. Though we are sympathetic to a dispositionalist account of belief, we’re not committing to it here. We can deal with the problem of stored beliefs without taking any view on the metaphysics of belief. 5 See Huemer (1999). 6 See Smithies (2019). While we don’t endorse this response, one of us did in the past. And we both think there’s more to be said for the view than is often supposed in the literature. Various objections to doxastic conservatism are dealt with in McCain (2008, 2020, 2021). 7 See Madison (2014) and McCain and Moretti (2021). Huemer (2013) also suggests that dispositional appearances may exist, though he doesn’t fully commit. 8 We use “stored appearance” and “dispositional appearance” interchangeably. 9 Plausibly, when you are currently recalling your phone number, you have a number of meta-cognitive appearances. For example, it strikes you that your phone number is something you know, you recall the number easily, and so on. See McCain and Moretti (2021) for discussion. 10 Sleeping Sally is modeled on an example from Moon (2012) that raises the problem of stored beliefs as a challenge to evidentialism. The key diference is
Appearances and the Problem of Stored Beliefs
11 12
13 14 15
16
17 18
19 20
73
that in Moon’s example, the belief in question concerns a basic truth of logic, whereas ours is straightforwardly empirical. This feature of Moon’s example brings in unnecessary complications about a priori justifcation, which we would rather avoid. See Moon (2012) for a careful discussion of how this sort of case provides an argument against internalist theories of justifcation. See Madison (2014) and McCain (2014, 2015b) for responses. Of course, there are other (and, we think, mistaken) views on the nature of appearances. Leaving competing views aside, we assume the account that we take to be correct. See McCain and Moretti (2021) for discussion of several alternatives as well as reasons for preferring the experience account of appearances. For fuller discussion of this and other features of appearances, see McCain and Moretti (2021). See Audi (1994) for a very helpful discussion of these distinctions. It is worth noting that our explanation of how Sally’s stored appearance can defeat her justifcation also solves what Smithies (2019) calls the “problem of stored defeaters.” We bypass that problem here because it strikes us as essentially the same as the problem of stored beliefs and, as expected, it admits of the same solution. Acknowledging this sort of move from Moon, Madison (2014) essentially concedes the point. The internalist has a response, according to Madison, when it comes to propositional justifcation but not when it comes to doxastic justifcation. We think this concession is given too readily. A causal understanding of the basing relation is defended in McCain (2012) and (2014). One might quibble that we have the wrong account of causation. Of course we disagree with that. But we can’t get into a debate about causation here. We’ll simply note that if one insists that the sort of counterfactual dependence we’re assuming isn’t equivalent to causation, then we’ll concede that our refections on Sally don’t make clear that the causal basis of beliefs can change. But the concession wouldn’t matter. The sort of counterfactual dependence we’ve described is well suited to flling the role of basing as understood by those defending causal accounts of the basing relation (see McCain 2012, 2014). So even if we don’t call it causal, it’s fair to say that the bases of one’s beliefs—in the sense relevant to epistemic basing—can change over time. See Frise (2018) for a helpful discussion of the cognitive psychology literature on this issue. Thanks to Matthias Steup for suggesting this objection.
References Audi, Robert. 1994. Dispositional belief and dispositions to believe. Nous 28: 419–434. Frise, Matthew. 2018. Eliminating the problem of stored beliefs. American Philosophical Quarterly 55: 63–79. Huemer, Michael. 1999. The problem of memory knowledge. Pacifc Philosophical Quarterly 80: 346–357. Huemer, Michael. 2013. Phenomenal conservatism über alles. In Chris Tucker (ed.) Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, pp. 328–350.
74
Kevin McCain and Scott Stapleford
Madison, B.J.C. 2014. Epistemic internalism, justifcation, and memory. Logos & Episteme 5: 33–62. McCain, Kevin. 2008. The virtues of epistemic conservatism. Synthese 164: 185–200. McCain, Kevin. 2012. The interventionist account of causation and the basing relation. Philosophical Studies 159: 357–382. McCain, Kevin. 2014. Evidentialism and Epistemic Justifcation. New York: Routledge. McCain, Kevin. 2015a. Is forgotten evidence a problem of evidentialism? Southern Journal of Philosophy 53: 471–480. McCain, Kevin. 2015b. No knowledge without evidence. Journal of Philosophical Research 40: 369–376. McCain, Kevin. 2020. Epistemic conservatism and the basing relation. In Patrick Bondy and J. Adam Carter (eds) Well-Founded Belief: New Essays on the Epistemic Basing Relation. New York: Routledge, pp. 201–214. McCain, Kevin. 2021. Epistemic conservatism: A non-evidentialist epistemology? In Luca Moretti and Nikolaj Pedersen (eds) Non-Evidentialist Epistemology. Leiden: Brill, pp. 152–169. McCain, Kevin and Luca Moretti. 2021. Appearance and Explanation: Phenomenal Explanationism in Epistemology. Oxford: Oxford University Press. Moon, Andrew. 2012. Knowing without evidence. Mind 121: 309–331. Pryor, James. 2000. The skeptic and the dogmatist. Nous 34: 517–549. Smithies, Declan. 2019. The Epistemic Role of Consciousness. Oxford: Oxford University Press. Steup, Matthias. 2001. Epistemic duty, evidence, and internality. In Matthis Steup (ed.) Knowledge, Truth, and Duty. Oxford: Oxford University Press, pp. 134–148. Tollhurst, William. 1998. Seemings. American Philosophical Quarterly 35: 293–302.
6
Emotions as Evidence for Evaluations Earl Conee and Richard Feldman
6.1.
Clarifying the Issue
There’s no question that our emotions can provide us with evidence. For example, experiencing an emotional reaction gives us evidence that we are having that emotion, at least when it is strong and distinctive. A harder question is whether emotions provide evidence for associated evaluations. For example, is fear evidence that the feared thing is dangerous, and is enthusiasm evidence that its object is good? The general question we will address in this chapter is whether any emotions are evidence for any evaluations. We will discuss the classic example of purported evidence of danger from fear. But we will concentrate on moral evaluations, where the interest is highest in fnding supporting evidence and understanding it. It could be that there are important diferences among emotions—some are evidence for related evaluations and others are not. We do not assume that if emotions are in some cases evidence for evaluations, there must be a fully general principle according to which all emotions are evidence for evaluations. We aim to remain neutral about what exactly emotions are. Some theorists associate them with bodily feelings. Others with cognitive attitudes. Others with some combination of feelings and attitudes.1 We think that the considerations advanced here apply regardless of which account of emotions proves best. We intend to base our case on examples that are uncontroversial instances of emotional states. It is reasonably clear that often, when we have an emotion, we become justifed in believing some evaluative proposition. This leaves open what, exactly, provides that justifcation. It could be that in some sense and in some cases the emotion itself provides “direct,” or, as we will say, “independent,” evidence for the evaluative proposition. We understand the relevant independence of evidential support for an evaluative proposition as evidence that is entirely sufcient on its own to give some epistemic reason to believe the evaluative proposition, even when it is not supplemented by DOI: 10.4324/9781003299349-8
76
Earl Conee and Richard Feldman
any contingent background information or any evidence that is part of the emotion’s basis. In other words, ‘independent’ evidence for a proposition is evidence that is a reason on its own to believe the proposition, not just something that can combine with other evidence to form a reason to believe the proposition. In contrast, “indirect” or “dependent” evidence, as we understand it, supports a proposition but only in conjunction with other evidence and thus not independently. A perceptual example will illustrate the diference between independent and dependent support as we think of it. Suppose that we touch a surface and it feels coarse in texture. This feeling may be independent evidence for the proposition that the surface is coarse (if any perceptual appearance is ever sufcient on its own as an epistemic reason for an external world proposition). The feeling may serve by itself as a (defeasible) reason to think that the surface is coarse. In contrast, feeling coarse is at best dependent evidence for the proposition that it is a surface of some sandpaper. Our having epistemic reason to believe that it is sandpaper that feels coarse requires our also having some background reason to think that what feels that way has enough likelihood of being sandpaper. Almost anything can be dependent evidence for anything else; it takes only the right background evidence to link that evidence to what it is dependent evidence for. Independent evidence, on the other hand, is more limited, and it is a philosophically more interesting topic. It will be our focus in this chapter. One important point about the independent/dependent evidence distinction is that sometimes evidence that might be thought to be independent support for a proposition turns out to be dependent evidence at best. When one is aware of this, it justifes making a kind of reality check: Since the evidence is at most dependent evidence for p, it supports p only if it is working in combination with other evidence. Sometimes we mistakenly think we have the rest of a combination like that. In such cases, what we have instead is a mere suspicion of p’s truth, or an activated bias in favor of p, an overgeneralization. We may think that we have, in inchoate form, the rest of a good reason when in fact we do not have any supplement that gives us genuine support for p. It is important for us to appreciate the force of our reasons to try to fesh out the rest of the combination. Success shows us that the candidate is indeed dependent on evidence in support of p; failure at least casts doubt on that and sometimes refutes it. We acknowledge that certain emotive responses seem to be pretty well hard-wired into us, for example, fear of snakes. But this leaves it open whether the fear is independent evidence for any fearfully induced evaluation, for example, that snakes are dangerous. Even if our fear of snakes is best explained by their dangerousness, the fear is not independent evidence for this evaluation. We will return to this example.
Emotions as Evidence for Evaluations
77
Our question is not whether any independent evidence provided by emotions is sufcient to provide knowledge-level justifcation for associated evaluations. If they do provide evidence, it might not be that strong and might be defeated by other evidence. The question is whether emotions provide any independent evidence at all for evaluations.2 The evaluations that arise in the case of emotional responses are in some ways like the beliefs that arise in perception, but we will argue that even if perception can in some cases provide independent evidence for propositions about perceived objects, emotions never provide independent evidence for evaluations. Our argument for this conclusion will derive entirely from the examination of cases that, we think, provide it with strong inductive support. A very diferent argument for a conclusion somewhat like ours might be based on the alleged unreliability of emotions. The argument relies on the frequently sensible advice not to let our emotions shape our beliefs or the related idea that, due to their unreliability, rational belief requires setting emotions aside. When these denials of evidence from emotions are taken to be completely general, they would imply that emotions are not any kind of evidence for evaluations. But in their general form, these are bad ideas. We can have evidence that our emotional evaluations of a particular sort have tended to be borne out. For example, we might learn that our feeling a spontaneous, unspecifc concern for someone has turned out to be a reliable sign that the person is in some sort of trouble. Under such circumstances, we get dependent evidence from a new instance of that sort of emotion for the corresponding evaluation. This evidence can help to make justifed the belief that the person is in trouble. But there is, nevertheless, some merit to the advice to avoid reliance on emotions as evidence—there are numerous examples in which emotional responses confict with rational responses. For example, we sometimes learn that our own enthusiasms are usually feeting and untrustworthy. In such cases, it would not be rational to be guided by our emotions. However, none of these considerations about emotions and rational belief resolve our question about independent evidence. Even if emotions sometimes contribute to dependent evidence for an evaluation, they could at times be independent evidence. And even if there is reason to distrust them in many cases, it still could be that the reason to distrust them outweighs some independent evidence that they provide for the associated evaluations. In other words, even if an emotion-based evaluation is not rational, it might be that the emotion itself is independent evidence that is defeated (because outweighed) by what we have learned about our responses. No general point about the unreliability of emotions resolves our question about whether emotions ever provide independent evidence for evaluations.
78
Earl Conee and Richard Feldman
In “Against Emotional Dogmatism,” Berit Brogaard and Elijah Chudnof present two general arguments for a conclusion that is very similar to ours.3 They assert general requirements based on independent evidence that they go on to claim no emotion meets. We appreciate the philosophical advantages of theoretical arguments like these. If successful, such arguments would be fully general, efcient, and efective. Unfortunately, we believe that the arguments by Brogaard and Chudnof are unsuccessful, and we see no other satisfactory general argument. We will discuss their arguments critically in Section 6.4. 6.2.
An Example-Based Argument Against the “Independent Evidence” View
We contend that no emotion provides independent evidence for any evaluations. As we stated in Section 6.1, the view under discussion is somewhat analogous to views about perception according to which in some cases perception is independent evidence for propositions about perceived objects. For example, the independent evidence view about emotions might hold that when we are frightened by something, it seems dangerous and that the emotion itself—the fear of the thing—is independent evidence that it is dangerous. This kind of view is sometimes formulated as a view about “seemings.” It is said that in perception an object might “perceptually seem” to be blue and this seeming is independent evidence that it is blue. The idea that emotions provide independent evidence can be formulated analogously. Eilidh Harrison discusses a view that holds that some emotive states are what we are calling independent evidence for evaluations. Harrison uses the phrase “it emotionally seems to S that e” to formulate those emotive states.4 As a point of initial resistance to the claim that some emotive evidence parallels perceptual independent evidence, we note that the emotional language that is closest to an ascription of a perceptual seeming has no clear meaning. It is easy to understand sensory claims like this: “Perceptually, the wall seems to be blue.” In contrast, it is not at all clear what it would mean to say: “Emotionally, the snake seems to be dangerous.” At best, this has the unintended meaning that the snake’s emotions seem to pose a danger. The “seeming” that can make it true that a snake “seems dangerous” is not an emotion toward the snake. It is something apparently threatening about the snake’s perceptual appearance, such as its hissing or coiling. Our fear of a snake is a response to the apparent danger. It is not a way that the danger appears. This diference in how perceptions of objects and emotions toward objects can be described in the language of appearance indicates that the relations
Emotions as Evidence for Evaluations
79
are importantly diferent. The fear, and more generally, an emotion, is not a way that its object appears to the subject. An emotional attitude toward its propositional content does not give whatever sort of epistemic reason an appearance gives for the truth of what appears to be so. It can be correctly said that objects “seem” to have evaluative properties when we experience some emotions. For example, upon receiving a gift, we might feel gratitude and say that giving the gift seems kind and thoughtful. Whether “seems” here expresses anything other than belief, or perhaps an inclination to believe, is unclear. If the “seems” claim reports only a belief or an inclination to believe, then our question is not even addressed by the claim. Our question is whether an associated emotion ever provides independent evidence for that belief. If the “seems” claim implies that we have some evidence for the evaluation, then the epistemic question is more nearly addressed, but it remains unanswered. An implication of evidence leaves it open whether or not it is the associated emotion that is providing evidence. It also remains open that the emotion is providing only dependent evidence. The ready use of this sort of “seems” talk does not support any particular answer to our question about independent evidence. The point about the absence of “emotional seemings” leaves it open that emotions give independent evidence for evaluations in some other way. The most extreme version of the independent evidence thesis might be that for every emotion, there is some associated evaluation for which the emotion provides independent evidence. But such a thesis is highly implausible since some emotions are not always linked with any particular object of evaluation or any particular evaluation. For example, we might experience free-foating anxiety without making any particular judgment. The epistemic thesis that is disputed here is the more limited claim that some emotions are independent evidence for associated evaluations. Perhaps the examples that are most favorable for the independent evidence thesis are cases in which the object associated with the emotion and the evaluation it leads to are clearly identifed and generally associated with one another. We will consider some examples of this sort. We will use the frst two to argue that adding or subtracting the emotion makes no diference to the evidence for the evaluation. Case 1
N is an ordinary person who is currently in an ordinary psychological condition. N sees O going to considerable trouble to help a stranger, P, who has been struggling to climb a staircase. P is appreciative and O assures P that O is happy to be of assistance. N notes O’s exercise of kindness, consideration and generosity and the gratefully received
80
Earl Conee and Richard Feldman
benefts to P. N feels admiration for O’s conduct and evaluates O’s action as a good one. At this point the story takes an unusual turn. As N continues to think about what O did, a special sort of tranquilizer that N has taken goes into efect. N becomes emotionally numb. N continues to think that O has done a good deed, though N no longer has any emotional reaction to it. The drug also induces a highly selective amnesia that prevents recall of recent emotive reactions. As a result, N does not recall feeling admiration toward O’s conduct. In this case, it seems clear that N’s epistemic justifcation for the evaluation does not change. N remains aware of the commendable things that N noted about the deed. N does not lose any epistemic reason to think that O’s deed is a good one by losing N’s positive feelings toward it and lacking any recollection of having had them.5 In the following example, the relevant things go in the opposite direction. Case 2
T has been tranquilized. The drug has rendered T emotionally numb, but otherwise unimpaired. In this condition T learns of a case of harassment. T learns that U has been cruelly mocking the meek and innocent V for weeks, U enjoys repeatedly upsetting V deeply by the continual mocking, and it is for the sake of this enjoyment that U has done it. On the basis of the clear negative moral bearing of these features of U’s treatment of V, T evaluates the treatment as morally wrong. Even in the absence of an emotion, T is capable of such an evaluation, and it would be justifed. So far, proponents of independent emotional evidence can agree, since they can acknowledge that there are also non-emotional sorts of independent evidence. Case 2 continues. Now the tranquilizer wears of and T resumes having conscious emotive reactions. T thinks again about the harassment case. T is now morally indignant and outraged by U’s treatment of V. T judges again that the treatment was wrong. The best candidates for being independent emotional evidence for the associated evaluations are emotive attitudes that seem quite appropriate. T’s emotive responses to U’s harassment of V do seem quite appropriate. So, if emotive responses ever give independent evidence for evaluations, then T’s indignation and outrage are independent evidence of the wrongness of U’s conduct. But we think that it is intuitively clear that T’s newly felt
Emotions as Evidence for Evaluations
81
emotions are not new independent evidence that T has of U’s wrongdoing. They could well give T-dependent inductive evidence for that conclusion. A typical person can recall such emotions being associated with justifed evaluations. If T has such background evidence, the emotion does provide additional dependent evidence. But absent any such history, as far as T can tell, such feelings are new and they provide no new evidence for the negative evaluation. T’s independent evidence continues to be the reasons that T already had. These are reasons consisting in T’s information concerning the negative moral bearing of the motives, intentions, and efects on those involved. In thinking about Case 2, we might be misled by noticing a change in a sort of justifcation that T has. After the tranquilizer wears of, T has the emotive reactions of indignation and outrage. These are appropriate ways to feel about U’s behavior. We can plausibly describe these emotions themselves as “justifed.” But in this case, the justifed reactions derive their appropriateness partly from what justifes the evaluation—the features of U’s conduct that T is aware of that make it clearly wrong. This is a kind of justifcation of the emotions, rather than a contribution by the emotions to the epistemic justifcation of the evaluation.6 Examples like Cases 1 and 2 of losing or gaining emotional reactions argue that emotional attitudes do not give independent evidence for their propositional contents. The subject’s evidence and strength of epistemic justifcation for the moral evaluations stay the same, despite the comings and goings of the emotions. It helps to support this conclusion to identify what sort of independent evidence subjects do have when emotive attitudes have epistemically justifed contents. We turn now to that question. In our view, what justifes the contents of emotive attitudes, when they are justifed, is mainly two types of evidence. One of them is the subject’s independent non-emotive evidence for the evaluation. The subject has evidence for facts that make true the evaluation, anything from signs of threatening features of snakes to manifestations of meritorious features of conduct. Again, it is this sort of evidence that also typically induces the emotional reaction. The other type of evidence is inductive. Here are some further illustrative examples. Case 3
A, B, and C are philosophy colleagues participating in a departmental discussion group. Work by philosopher X comes up in the course of the discussion. A is irascible and notoriously has a low opinion of the work of X. B, a new colleague, is unaware of these things about A. B mildly commends a view by X that is disputable but defensible. A interrupts before B is quite fnished, sneering, “You can’t be serious! That is the
82
Earl Conee and Richard Feldman
sort of rubbish that is typical of X. Can you really not have seen Y’s classic refutation?” Not having even heard of Y, B is stunned into silence and clearly upset by A’s comment. Witnessing all of this, C immediately and strongly feels disapproval toward A’s behavior, evaluating it as A’s having wronged B. In Case 3, the judgment that A has wronged B is as closely connected as possible with C’s disapproving emotion. This evaluative judgment is epistemically justifed. C’s spontaneous disapproval is an emotional state that is a candidate for being independent evidence that C has for his wrongness judgment. This seems to be a prime example of the sort of emotive attitude to which the independent evidence view is supposed to apply. Again, in our view, the disapproving attitude is not independent evidence for the wrongness evaluation. It is at most dependent on evidence. C’s disapproval does have an evidential basis: C has noticed the interruption, the condescension, and the professional insult in A’s utterance. Such features of an utterance are evidence of its wrongness. They are good evidence for A’s having wronged B even in the absence of any emotive response by A. C’s disapproval may add dependent inductive evidence, if C is aware that C’s evaluations that have been based on C’s spontaneous disapprovals have been largely confrmed. But we think that Case 3 illustrates the usual sort of epistemic justifcation for the evaluations associated with emotions. The subject has non-emotive evidence for the evaluation. The emotion arises partly on the basis of that evidence and does not add independent evidence for the evaluation. To supplement our support for this point, a little later we will describe examples (Cases 5 and 6) of an emotion associated with an evaluation where there is no other justifcation for the evaluation. It will be clear that the emotion does not add evidence for the evaluation. Adam Pelser discusses another case in which he claims the emotion provides independent evidence for an evaluation.7 Case 4
A father is grieving over the death of his daughter in a car crash. The father was alienated from the daughter prior to her death. His discussions with similarly estranged fathers had given him testimonial evidence against his daughter’s being valuable and her death being a great loss. Upon experiencing the grief, though, the father comes to believe that he has undergone an irrevocable loss of someone of great value (p. 118). Pelser claims that the father “need not have non-emotion-based epistemic justifcation for believing her death to be an irrevocable loss of someone of great value” (p. 118). He says that this evaluative belief—that she is of
Emotions as Evidence for Evaluations
83
great value—is “consequent to the grief” and that the grief, that is, the emotion, not other factors, justifes it. We will take up one moral belief that seems to us to be a central part of the father’s broader judgment of his having undergone a great and irrevocable loss. It is the belief that the daughter was a good person. We will assume that this belief arises as a response to his grief and the belief becomes justifed at that time. If so, then in our view, something in what the father has come to think about the daughter is prompting this evaluation. The father may well have been moved by fnding in himself the emotion of grief to recall good qualities of the daughter and good things about some of the ways that they related to one another. Perhaps, for instance, the father is brought to recall occasions when the daughter manifested spontaneous consideration and generosity on encountering others’ misfortunes. Such recollections would be evidence for the father’s judgment that the daughter had been a good person. They are evidence that he might have recovered, whether or not the emotion induces him to bring them to mind. Absent any such non-emotive support, however, the father’s evaluation of his daughter would not be evidentially justifed. (Of course, the father’s grief could serve as dependent evidence for the belief that the daughter was a good person if the grief is part of an inductive argument linking feelings of grief to that sort of evaluation.) Unpleasant as it may be to consider, grief can give rise to unreasonably favorable evaluations of a person who has died. We deny that the favorable evaluations in such circumstances are evidentially supported by the feeling of grief alone. It might instead direct attention to some good features of the deceased that were outweighed by others that are not then considered. When the favorable evaluations are justifed, supporting background information is always the independent evidence that does the justifying. Here is another example where it might seem more difcult to deny that an emotion is serving on its own as evidence for an evaluation.8 Case 5
S has just had a difcult conversation with Q, involving a variety of sensitive topics and some degree of awkwardness, but upon its conclusion S judges that S handled it as well as possible. As a little time goes by, however, S experiences a nagging, unsettling feeling that it wasn’t actually handled so well, and S fnds himself feeling a not fully articulable sense of something like guilt, or perhaps it is regret, or even remorse. Where exactly is that coming from? It’s not clear: S has not yet found any identifable holes in S’s earlier construal of the conversation, according to which S did nothing wrong; yet something isn’t sitting right, and S begins to suspect that S may have behaved badly, wronging Q. At the
84
Earl Conee and Richard Feldman
time, all the evidence that S has from the conversation itself still seems to S to point to S’s having done well: S can’t clearly see any hole in that construal. But S’s sense of guilt, or some similar emotion, keys S into the fact that S likely behaved worse than S thinks. To assess the evidential role of S’s guilt-like emotion toward the evaluation that S wronged Q in some way during their conversation, it is crucial to consider something that is not explicit in this description of Case 5. What are S’s memories, if any, of S’s having similar feelings in the past?9 Assuming that S is a typical morally sensitive adult, S will have experienced guilt-like feelings in other cases. It is likely that in most of those cases, some guiltmaking features were apparent to S. It was a needless harm or the like. This gave S evidence that the guilt-like feeling was directed toward something S did, or neglected to do, that was wrong. In Case 5, S has yet to notice anything similar. But S is also likely to have some evidence that wrong-making features are not always easy to discern and articulate. If S has in memory a background of that sort, then the guilt-like feeling is inductive evidence of S’s wrongdoing in the conversation. The emotion is dependent evidence in that it is just part of the inductive argument that constitutes the evidence. To the extent that nothing wrong-making shows up on S’s continued refection, this evidence is weakened because its relevant similarity to discerned wrongdoing decreases. But some inductive evidence remains. Even if S has in memory no history about a guilt-like feeling in particular, having memories of borne out misgivings, inchoate concerns, or suspicions of any sort makes for some inductive evidence that S has in this case too. The guilt-like feeling on its own need not be evidence of wrongdoing for S to have this sort of evidence. If S has the typical background memories just described, then there is no reason to think that S’s guilt-like feeling is independent evidence for his conclusion that he engaged in some conversational wrongdoing, We can fll out the case to make S atypical. For example, we can consider the possibility that S’s guilt-like feeling is a new sort of feeling for S. So, S has no remembered history of how often wrongdoing is a source of such a feeling. Nevertheless, S is led by it to suspect conversational wrongdoing. Yet on refection S has found nothing apparently wrong about S’s conversational conduct. In such a case we do not see that S has any evidence of actual wrongdoing. Instead, the feeling would be a mystery to S. It would be like having for the frst time a completely unspecifc sense of having accomplished something special at home yesterday, with no other evidence for any such accomplishment. These would be puzzling stray impressions. They would not be evidence for the corresponding thoughts. We have been discussing cases in which any appearance of having independent emotional evidence can be explained by other evidence or
Emotions as Evidence for Evaluations
85
dependent emotional evidence. Michael Brady has ofered a diferent sort of objection to the thought that emotions are independent evidence for evaluations. Brady has objected on the grounds that emotions often serve as reasons to look for evidence rather than as evidence themselves.10 Pelser, however, responds that we sometimes do not treat perceptual evidence as by itself sufcient to justify belief but instead as reason to get more evidence.11 That, he correctly notes, does not show that perceptions are never independent evidence themselves. So, similarly, the fact that sometimes emotions prompt us to look for more evidence does not show that they are not evidence themselves. We agree. But when we look for independent evidence from emotions, we cannot fnd any. Another example of emotion without other evidence available will be additionally helpful to show that the emotion is not independent support. Unlike Case 5, it is an example in which we who think about it are not inclined to assume any background evidence that makes the emotion contribute to inductive support for the evaluation. Case 6
W is induced by a drug that she is unaware of having taken to feel revulsion at the thought of eating a strawberry. The revulsion inclines W to regard the strawberry eating as morally wrong. W wonders why she feels as she does. She seeks other evidence of moral wrongness. She fnds no such evidence. We believe it to be clear that W has no epistemic reason to attribute wrongness to the eating of the strawberry. Instead, W has reason to fnd her revulsion puzzling and inappropriate. Of course, if she has reason to think that such feelings arise in her only in the presence of wrong actions, then she does have inductive evidence for some such conclusion. That possibility provides no support for the independent evidence view. Further, any such inductive support may well be defeated by her evidence that she is aware of nothing else supporting her moral conclusion. Finally, we return to the fear of snakes and the associated evaluation of them as dangerous. Considerations against independent emotional evidence similar to those from the other examples apply here. If our belief about a perceived snake’s dangerousness is justifed, then background evidence about the harm potentially caused by snakes with the perceived snake’s appearance provides the justifcation.12 Note that if we had no reason to think that snakes ever harmed anything, then our fearing them would not seem to be a good basis to think that snakes are dangerous. At best the fear would be dependent evidence for some negative evaluation, and perhaps also reason to keep looking for potential harm.
86
Earl Conee and Richard Feldman
It might be thought that our fear of some things is a “natural” rather than a learned response. That may make it understandable, perhaps excusable, that we believe the resulting negative evaluations. But this does not suggest that they are epistemically justifed, any more than any prejudicial beliefs we may fnd ourselves naturally inclined to form. It should be clear that we are not arguing that the evaluations made in the presence of emotions are not justifed. We will further discuss how we think the justifcation works in the next section. 6.3.
Emotions, Perceptions, and Evidence
The view that perceptions can be independent evidence for propositions about the sensory qualities of perceived objects has been a topic of considerable discussion. A version of this view, as noted earlier, holds that in perception objects can perceptually seem to have certain properties, and these perceptual seemings provide defeasible independent evidence for propositions to the efect that the objects do have those properties. This is thought to be an appealing solution to puzzles about perceptual justifcation and skeptical worries about how external world propositions can be justifed. We have our doubts about the merits of this view about perceptual seemings, but we will not go into them here. We do acknowledge that the view proposes a solution to a genuinely perplexing question about how and why external world beliefs are justifed. Our preferred solution relies on justifcation being conferred on external world propositions that are parts of best explanations of perceptual experiences. But we will assume for the present discussion that perceptual seemings are sometimes independent evidence for their contents. This does not give reason to think that emotive attitudes are also independent evidence for associated evaluations. As we have noted, evaluations such as that snakes are dangerous do not “emotionally seem” to be true. That does not make good sense. Furthermore, in the perceptual case the appearance is how an environmental fact appears to us. But an emotion is not how an evaluation appears to us. Another example helpfully illustrates this. Case 7
R feels appalled at someone’s casually disclosing a friend’s highly confdential communication. It seems to R that the disclosure is wrong. When we are in the position of R in such a case and we consider how the disclosure seems wrong to us, we have no inclination to cite simply our emotion: “It seems wrong by appalling me.” Instead, its appearance
Emotions as Evidence for Evaluations
87
of being wrong would consist of something that we might put this way: “What makes it seem to be wrong is that it so drastically betrays the trust that a friend deserves to have. That is what appalls me about it.” Both the seeming wrongness and the emotion have their source in a non-emotive basis for wrongness. One further point about perceptual justifcation will clarify our view. Perceptual experiences often serve as dependent evidence for beliefs about the world. Indeed, this is the typical case. We learn to classify things, so that just seeing them enables us to know what they are. But our background evidence is making a diference; the perception itself is not independent evidence even if the belief is formed immediately and without conscious inference. Consider, for example, our ability to classify birds, trees, or people. We do need, and have, evidence about the appearance of these things and of our ability to classify them. The bare perceptual experience by itself does not sufce, even if the classifcation of objects becomes automatic and even if we are not always able to articulate the identifying characteristics of the things we are able to discern.13 As our discussion of the examples in the previous section shows, we think that in the case of evaluative responses associated with emotions, something similar occurs. We can learn that our emotions are reasonably reliable indicators of the characteristics of the things we react to. But we can also learn that they are not, for example, we can fnd that often we are frightened when there is not really any danger. As in all of the examples considered here, if the evaluation is justifed, then it is justifed by other evidence present in the situation, often evidence that gives rise to the emotion as well as the evaluation. There are cases in which we witness or hear about some behavior and immediately react with a negative emotion toward it, but we are unable to say what features provoke that reaction. A defender of the independent evidence view might contend that in such a case, the emotional reaction provides independent evidence for the evaluation. This might be defended on the grounds that the evaluation is justifed, yet there is no other obvious candidate for its evidential support. We dispute this assessment of the case. Such a reaction can occur in examples like this: Case 8
Z learns of a way of dressing that is common in another culture. To Z it seems strange. Z immediately feels disapproval toward the practice. Z thinks that it is a bad way to dress. On refection though, Z cannot identify anything bad about it.
88
Earl Conee and Richard Feldman
A negative emotion like Z’s appears to be a product of bias. We think it is clear that the emotion itself is not evidence for Z that it is a bad practice. If Z’s evaluation is justifed, this is either because Z has reason to trust Z’s reactions or because the reported features of the practice justify the reaction even though Z is unable to identify them. Perhaps, for instance, Z has a vague sense that the gaps in the clothing may result in its not providing enough warmth for what Z supposes to be the culture’s climate. In that case the justifcation works something like one’s ability to recognize things without being able to describe their identifying features. The reported features, not the emotion, do the justifying. Thus, we do not claim that emotions play no role in the justifcation of associated evaluations. In some cases, emotions may prompt the search for evidence that justifes the evaluation. Additionally, as we have acknowledged several times, the presence of an emotion may serve as part of an inductive argument for an evaluation. We note that awareness of an emotion in another person may also be part of an inductive argument for an evaluative conclusion. We can learn something about the world by noting the occurrence of emotional responses, both our own and those of others. But to admit these evidential roles for emotions is not to endorse them as independent evidence for evaluations. 6.4. Other Arguments for Our Conclusion We have argued for our view by appealing to what we think is a proper assessment of a variety of cases. In contrast to this approach, Berit Brogaard and Elijah Chudnof argue for a conclusion similar to ours on the basis of two general arguments that, if successful, would provide convincing theoretical support for our view.14 We will turn next to an assessment of those arguments. Brogaard and Chudnof argue that emotions provide what we are calling independent evidence for evaluations only if both of two things are true: (1) emotions have “presentational phenomenology with respect to the proposition attributing that evaluative property to that object or event” (p. 69) and (2) emotions “represent objects and events as having evaluative properties in an appropriately evidence insensitive way” (p. 74). They claim that emotions do not have the properties identifed in (1) and (2) and thus do not provide independent evidence for evaluations. We will explain what they think these conditions mean, why they think that they are not satisfed, and why we think that their arguments do not succeed. Here’s how Brogaard and Chudnof characterize presentational phenomenology: In general an experience has presentational phenomenology with respect to the proposition that p just in case it both makes it seem as if p and
Emotions as Evidence for Evaluations
89
makes it seem as if it makes you aware of a portion of the world making it true that—i.e. a truthmaker for—p (p. 68) One helpful example that they consider concerns a car parked next to a fre hydrant. They say the experience of a parking attendant seeing the car so parked may make it seem to him that the car is illegally parked, but the experience lacks presentational phenomenology related to illegality. This is because the truthmaker for the illegality proposition includes facts about the relevant laws and the experience does not seem to make him aware of such facts. In contrast, the perceptual experience does seem to make the attendant aware of the truthmaker for a proposition about the position of the car, namely the relevant fact about the position of the car and the parking meter. As applied to emotions, their view is that the emotional experience may make its subject seem to be aware of various facts about feelings and the object to which the emotion is a reaction, but the emotion does not seem to make the subject aware of the truthmakers for the related evaluations. For example, fear does not seem to make one aware of the fact that the object feared is actually a threat. Brogaard and Chudnof’s argument turns on their claims that something is independent evidence only if it is phenomenologically presented, which in turn requires that one has a seeming awareness of its truthmaker. We don’t think that giving a seeming awareness of a truthmaker is needed for something to be independent evidence. Consider memory. Coming to have a recollection has the same seemingly self-sufcient justifying role as does having visual perception. Our recovery of a recollection of a factual proposition that we had learned about the distant past, say, that ancient Carthage was in Africa, is as plausible an instance of independent evidence for its truth as is a visual perception of the color of an object. (We leave it open that fnally neither one is independent evidence.) Yet recollecting the proposition as true does not seem to make us aware of the truth-making presence of ancient Carthage in Africa. Thus, the seeming awareness of truthmakers seems not to be a necessary condition for independent evidence. It’s possible that an alternative analysis of phenomenological presentation would avoid these problems. Since we lack such an account, and we lack any independent grasp of the idea, we don’t see how the argument is capable of establishing its conclusion.15 The concept of evidential insensitivity used in Brogaard and Chudnof’s second argument is illustrated by the familiar Müller-Lyer illusion: two lines of equal length with arrowheads at both ends, one pair pointing outwards and the other inwards, appear to us to be of unequal length. This appearance persists, even when we learn that they are of equal length. That is, the illusion persists—the line with the outward
90
Earl Conee and Richard Feldman
arrows still seems longer, despite our evidence that it is not. In contrast, their example of evidence that is sensitive goes like this: Our hearing an early forecast of an approaching severe hurricane makes it seem to us that we ought to depart from the area, but upon hearing a later forecast of an unthreatening storm path, it no longer seems to us that we should depart (p. 72). Brogaard and Chudnof claim that what we are calling independent evidence must be evidentially insensitive. They claim that our emotional responses do not represent objects to have evaluative properties in evidentially insensitive ways. This is because people can adjust their evaluations in light of new evidence so that the object no longer seems to have the evaluative property it initially seemed to have. For example, a remark that initially seems to be cruel may no longer seem so when one learns that it is part of an ongoing joke shared by two people. Brogaard and Chudnof acknowledge that the seeming truth of an emotional evaluation does not always continue upon acquiring decisively negative evidence concerning the evaluation. That is, in some cases, emotions are evidentially insensitive. However, they claim that this is due to the irrationality of the emotion. Their example goes like this: Someone continues to fear air travel and it continues to seem dangerous to the person, both initially and even after being well informed about the high level of air travel safety. They claim that this is an irrational fear because it does not ft its object (p. 73). And, they claim, this irrationality undermines the claim that the emotion provides independent evidence for the evaluation. They conclude that emotionally responsive evaluations are never appropriately evidentially insensitive and that this precludes them from being independent evidence for evaluations (p. 74). We do not see why having appropriate evidential insensitivity is a good test of being independent evidence. In the cases in question, what may or may not have the sensitivity are emotions that are associated with evaluations that seem true. The emotions are candidates for being independent evidence for the evaluations. If the subject’s gaining new evidence that bears negatively on the evaluation makes a justifed evaluation no longer seem true to the subject, then the emotion is not appropriately evidentially insensitive. Thus, this sensitivity is a capacity of new negative evidence to terminate a mental state of seeming true that was generated by the emotion.16 We do not see how the presence or absence of that psychological efect of new evidence tells us anything about whether the emotion is independent evidence for the evaluation. It may be that sometimes a subject receives new negative evidence about the content of an emotive attitude and this causes the subject’s evaluation of that content to stop seeming true. But independent evidence from the emotion for the evaluation may remain, though it is counterbalanced or outweighed by the new evidence.
Emotions as Evidence for Evaluations
91
Thus, it may be that independent evidence is not always appropriately evidentially insensitive. These considerations are not conclusive. They do lead us to doubt that the causal capacity of appropriate evidential insensitivity is required for independent evidence. Since a premise asserting such a requirement is part of Brogaard and Chudnof’s second argument against emotions being independent evidence for evaluations, we doubt that the argument succeeds. 6.5. Conclusion We have not attempted to make a general argument that emotions are not independent evidence for evaluations. We have not ofered any necessary condition based on independent evidence that we contend emotions never meet. Instead, we have argued from examples. Concerning the examples, we have contended that any appearance of emotional independent evidence for an evaluation is better explained by the existence of other, non-independent evidence. We have tried to illustrate this with optimal candidates for emotional independent evidence. Concerning most of the examples, we have observed that in the most similar situations where emotion is absent, the subject’s independent evidence for the evaluations remains the same. We have also observed that in examples where an emotion associated with an evaluation is not accompanied by any other evidence for the evaluation, the evaluation lacks evidential support. Thus, we have made an inductive case for the thesis that emotions never provide independent evidence for any evaluation. We think that the emotions in our examples are representative of the optimal candidates for being this independent evidence. We have contended that the emotions in these examples provide only dependent evidence. If we are right, then the examples provide strong support for our thesis. Notes 1 See Scarantino, Andrea and Ronald de Sousa, “Emotion”, in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta (Summer 2021 ed.), URL = https://plato.stanford.edu/archives/sum2021/entries/emotion/. 2 We are assuming a non-subjective “cognitivist” understanding of evaluations in this paper. Thus, the question we address is whether emotions provide evidence for the truth of the evaluations, where the evaluations are not reports or expressions of our emotive attitudes or reactions. 3 Brogaard, Berit and Elijah Chudnof, “Against Emotional Dogmatism”, Philosophical Issues 26 (2016): 59–77. 4 “The Prospects of Emotional Dogmatism”, Philosophical Studies 178 (2021): 2535–2555. 5 Our claim here is that N’s evidence does not change at all. A slightly more modest claim would be that his evidence does not get any weaker. The more
92
6 7 8 9
10 11 12
13
14 15 16
Earl Conee and Richard Feldman modest claim allows that the emotion does provide evidence, but that it does not combine with the other evidence to more strongly support the conclusion. In other words, the evidence is not “additive.” We agree that evidence does not always need to be additive, but we do not agree that new evidence in a case like this would not be additive. It seems clear to us that if the emotion itself were evidence, it would strengthen the case for the evaluation. We thank Bill FitzPatrick for pressing us on this point. That the emotion is justifed is especially clear in accounts of emotions in which the evaluative reaction is taken to be part of the emotion. “Emotion, Evaluative Perception, and Epistemic Justifcation”, in Emotion and Value, ed. Sabine Roeser and Cain Todd, Oxford: Oxford University Press, 2014, pp. 106–122. This is a slight re-wording of an example that was ofered to us by Bill FitzPatrick in correspondence. We are grateful for it and for his other helpful comments on an earlier draft. Plausibly, the basis of S’s evaluation of his conduct could include information he has about the feelings others have in such cases and his similarity to those people. That is, he might have reason to think that generally, when people have feelings like his, they have behaved poorly. Brady, M., “Emotions, Perception, and Reasons,” in Morality and the Emotions, ed. C. Bagnoli, Oxford: Oxford University Press, 2012, pp. 135–149. “Emotion, Evaluative Perception, and Epistemic Justifcation,” Section 4.2. We favor the background evidence view of the justifcation. That view is not crucial to a denial that the fear is independent evidence when the belief is justifed. Moon, Andrew, “How to Use Cognitive Faculties You Never Knew You Had”, Pacifc Philosophical Quarterly 99 (April 2018): 251–275, discusses a version of the snake case and defends the view that the belief that it’s dangerous is justifed, using a proper functionalist account. See p. 263f. We realize that the view described here is not universally accepted. For discussion, see Lyons, Jack, “Epistemological Problems of Perception”, in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta (Spring 2017 ed.), URL = https://plato.stanford.edu/archives/spr2017/entries/perception-episprob/. See section 2.1.5. “Against Emotional Dogmatism.” For much more about presentational phenomenology, see Chapter 1 of Chudnof, Elijah, Intuition, Oxford: Oxford University Press, 2013. We are perplexed by the general nature of this purported mental state of seeming true, but for present purposes what matters is only that it is supposed to be some sort of mental state.
References Brady, Michael. “Emotions, Perception, and Reasons”, in Morality and the Emotions, ed. C. Bagnoli, Oxford: Oxford University Press, 2012, pp. 135–149. Brogaard, Berit and Elijah Chudnof. “Against Emotional Dogmatism”, Philosophical Issues 26 (2016): 59–77. Chudnof, Elijah. Intuition, Oxford: Oxford University Press, 2013. Harrison, Elidah. “The Prospects of Emotional Dogmatism”, Philosophical Studies 178 (2021): 2535–2555.
Emotions as Evidence for Evaluations
93
Lyons, Jack, “Epistemological Problems of Perception”, in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta (Spring 2017 ed.), URL = https://plato. stanford.edu/archives/spr2017/entries/perception-episprob/. Moon, Andrew. “How to Use Cognitive Faculties You Never Knew You Had”, Pacifc Philosophical Quarterly 99 (April 2018): 251–275. Pelser, Adam. “Emotion, Evaluative Perception, and Epistemic Justifcation”, in Emotion and Value, ed. Sabine Roeser and Cain Todd, Oxford: Oxford University Press, 2014, pp. 106–122. Scarantino, Andrea and Ronald de Sousa, “Emotion”, in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta (Summer 2021 ed.), URL = https://plato. stanford.edu/archives/sum2021/entries/emotion/.
7
How to Be Irrational Michael Huemer
7.1.
The Puzzle of Internally Unjustifed Belief
How is it that we often hold unjustifed beliefs? On some theories of justifcation, this is easy to understand. For instance, perhaps a belief is unjustifed whenever the belief-forming method is in fact unreliable, regardless of whether it appears to us to be reliable.1 On this account, our holding unjustifed beliefs might be perfectly understandable, as our unjustifed beliefs might be indistinguishable, from our own point of view, from justifed beliefs. But matters are more puzzling for internalist conceptions of justifcation. How and why would one form a belief that is unjustifed from one’s own point of view? The question is particularly puzzling for Phenomenal Conservatives, myself included, who hold that justifcation derives from how things seem to the subject. As I have elsewhere articulated the thesis: PC
If it seems to S that P, then, in the absence of defeaters, S thereby has at least some justifcation for believing P.2
Do people sometimes form beliefs that don’t even seem true to them? One might try appealing to defeaters to explain unjustifed beliefs: Perhaps when we believe irrationally, the belief typically does seem true to us, but there are defeaters that we are ignoring. But why do we ignore these defeaters? If, as I have also claimed, defeaters must be propositions that one has justifcation for believing, and this justifcation must also derive from how things seem to oneself, then it is on its face puzzling that we would ignore our defeaters. If the appearance that supports a defeater has enough force to defeat the justifcation for P, then why would it not have enough force to defeat our psychological inclination to believe P? None of this is obviously impossible, but it calls for explanation. In what follows, I ofer two related explanations. First, we sometimes have nontruth-oriented desires that move us to believe a proposition more than the DOI: 10.4324/9781003299349-9
How to Be Irrational
95
evidence warrants. Second, these same desires prompt us to violate epistemic duties regarding how we conduct inquiry. 7.2. Forms of Irrationality 7.2.1.
Credence and Belief
I assume there is a distinction between two kinds of doxastic attitudes: credences and outright beliefs. Credence is a gradable attitude, the degree of confdence that one has in a proposition. This can be roughly understood in terms of fair betting odds: if you have an 80% credence in A, then you should regard a bet on A at 4 to 1 odds as fair. (This is not meant as an analysis of “credence,” but it nevertheless serves to give a sense of what we’re talking about.) By contrast, outright belief is a qualitative attitude toward a proposition. This attitude is incompatible with having a low credence, but it is not simply having a high credence. When one outright believes a proposition, one ends inquiry into that proposition’s truth with the conclusion that it is true; one subsequently uses the proposition itself, rather than an estimate of the proposition’s probability, in further reasoning and in planning action.3 When we ask about justifcation for belief, then, we may have in mind either justifcation for a credence or justifcation for an outright belief. The justifed credence in a proposition is a matter of how likely that proposition is given one’s current epistemic situation; one’s credence should be proportional to that probability. Justifed outright belief, however, depends on more than just the probability of the proposition (unless the proposition is 100% certain, in which case outright belief is always justifed). However likely a proposition is (short of 100%), there is always a nontrivial question of whether such a probability is high enough, in the context, to warrant outright belief. The threshold probability required will difer for diferent propositions. Outright belief is a cognitive attitude, so justifcation for outright belief is a kind of epistemic justifcation; nevertheless, pragmatic and even moral factors can afect this justifcation.4 Since adopting an outright belief involves ending inquiry, the justifcation for outright belief may depend upon the expected fruitfulness of further inquiry, the expected costs of inquiry, and the importance of avoiding error on the matter in question. Similarly, since outright belief results in using the proposition itself, rather than an estimate of its probability, in further reasoning, the potential costs and benefts of this can infuence whether outright belief is justifed. For instance, in lottery cases, where one knows that exactly one of a large number of lottery tickets will win, one may not outright believe, of
96
Michael Huemer
a randomly chosen ticket, that it is a loser. In this context, the threshold for outright belief is 100% (or something very close to it), since adopting a lower threshold results in outright believing of every ticket that it is a loser, ignoring the small probability that it is a winner (again, the discarding of this probabilistic information is just part of what it is to have an outright belief). One could then deduce that the prize will not be given, which we know is not true. Practical reasoning would also be disrupted; for example, a person running a lottery could not decide how to appropriately price each ticket if he outright believes of each that it will lose. In other contexts, a proposition with the same probability may be suitable for outright belief. For instance, if I seemingly remember clearly where I parked my car, I may outright believe that the car is in that location, even if the probability of its being there is no higher than the probability that my lottery ticket is a loser. This is because in the car context, a nonextreme threshold for outright belief, permitting belief with less than 100% probability, would not be expected to generate inferences that contradict background knowledge, nor to generate unreasonable results in practical reasoning. 7.2.2.
Three Forms of Irrationality
Taking the above into account, there are three ways of having an epistemically irrational doxastic attitude.5 First, one may have a credence in a proposition that is not proportioned to the proposition’s degree of justifcation. For example, A is 70% probable on your evidence, but you believe it with 90% confdence, or vice versa. Second, one may have an outright belief in a proposition whose probability is below the appropriate threshold for outright belief (or withhold outright belief when the proposition’s probability is above the threshold at which belief is rationally required). For example, if the minimum threshold for rational belief in A is 95%, your evidence renders A only 80% probable, but you nevertheless outright believe A. Third, a belief may be unjustifed in virtue of the subject’s violation of epistemic obligations regarding the conduct of inquiry. For example, you were obligated to listen to the objections to A before drawing a conclusion, but you refused to do so and adopted a belief anyway. This third kind of irrationality calls for further explanation. In my view, a subject’s violation of his epistemic duties may sufce to impugn the justifcation for his outright beliefs, though not for other doxastic attitudes such as credences or suspense of judgment. Regardless of how well or badly you conduct an inquiry into A, there will be a probability that A has on your actual evidence. If your credence matches that probability, then it is epistemically justifed. Likewise, if the evidence that you currently have fails
How to Be Irrational
97
to adequately justify A, and as a result you withhold judgment about A, then your suspense of judgment is justifed, regardless of how shoddy your investigation may have been and regardless of how much evidence a better investigation would have turned up. Our epistemic obligations regarding the conduct of inquiry are conditional: one must conduct a responsible inquiry into A if one is to form an outright belief about it. In other words, one is often epistemically obligated not to form an outright belief without conducting a responsible inquiry, since only the completion of such an inquiry justifes the attitude of having closed inquiry into A. This is not to deny that one might have an unconditional obligation of some kind to inquire and form a defnite belief. For instance, if a prosecutor receives a criminal case, she may be unconditionally obligated to try to fnd out whether the suspect is guilty; her duty is not discharged by merely refusing to form an outright belief about the case. But the unconditional obligation to investigate would be a moral or professional obligation, not an epistemic obligation. Epistemology cannot tell a person when to gather evidence; it can only tell a person what she may believe, or what credences she may have, given her epistemic situation. Thus, the prosecutor’s epistemic obligations would be to proportion her credences to the evidence and to refrain from believing the defendant guilty as long as she has not conducted a responsible investigation. What constitutes a responsible inquiry? I won’t try to give an exhaustive list of the obligations of responsible inquiry, but I have in mind such norms as the following: •
• •
If one is to form a belief about A, one should gather further evidence about A as long as one can do so at low cost and the evidence can be expected to make a signifcant diference to the probability of A. For example, if I can easily fnd out for sure what time it is by simply looking at my phone, then I should not form a belief about the time using only my intuitive sense. Before adopting an outright belief in a controversial thesis, one should listen attentively and charitably to the arguments against it. Before accepting an argument about a controversial subject, one should attempt to think of objections to that argument, as well as arguments for the opposing view.
Under normal conditions, the violation of such duties renders one unjustifed in forming an outright belief. One way to think of this kind of irrationality is that it is a subset of the second form of irrationality, namely, that of forming an outright belief when the probability of the target proposition is below the necessary threshold for justifed belief. In other words, even if A has a very high probability on your evidence,
98
Michael Huemer
that probability is still insufcient if you failed to conduct a responsible inquiry that you were required to conduct. This way of describing things is somewhat misleading, however, because it might seem to suggest that, if one were to conduct further inquiry, this inquiry would render outright belief justifed only if it further raised the probability of the target proposition. And that need not be so. It may be that, after conducting responsible inquiry, the target proposition would be left with the same probability, yet outright belief would now be justifed, simply because one had discharged one’s obligation to do sufcient investigation. It is for this reason that I have listed the violation of duties of inquiry as a third form of epistemic irrationality. All of that was to explain the sense in which a doxastic attitude might be irrational. Let us turn to explaining why we often form such irrational attitudes. 7.3. 7.3.1.
The Doxastic Inclination Theory Inclinations to Believe
Appearances, or “seemings,” are closely tied to inclinations to believe. Nearly everyone agrees that when it seems to one that P, one is naturally inclined to believe that P. Indeed, one account holds that seemings just are inclinations to believe something.6 If all inclinations to believe were appearances, or derived from appearances, then it would be particularly puzzling that a person might believe something in the absence of a corresponding appearance. But not all inclinations to believe are connected to appearances, as we see in the following example: Accused Child: Your son is accused of an infamous crime, say, deadnaming Caitlin Jenner. You are inclined to believe your son innocent because you love your son and do not want to believe he is capable of such a heinous act. Surely such a case is possible. Notice that that would be quite diferent from the case where you are inclined to believe your son innocent because he just seems to be innocent. So seemings are only one possible ground among others for an inclination to believe. This suggests a simple account of how we come to have unjustifed doxastic attitudes: They are caused by non-appearance-based doxastic inclinations. That is, we form unjustifed beliefs when we are inclined to believe something for any reason other than that the belief’s content seems correct. These reasons could include the emotional satisfaction we derive from
How to Be Irrational
99
holding a certain belief, the belief that it is morally virtuous to hold a certain belief, the desire to ft in with others who hold a particular belief, and so on. These will in general be non-epistemic reasons for belief, including prudential, moral, and instrumental reasons. This basic account applies to all three kinds of epistemic irrationality. In the Accused Child case, the parent may attach a higher credence to [My son is innocent] than the evidence warrants, based partly on the parent’s desire to believe the son innocent. The parent may outright believe the son innocent when the probability of innocence is insufcient to justify this attitude. As evidence of the son’s guilt accumulates, the parent might give up this belief but still refuse to adopt the opposite belief, [My son is guilty]; the parent might then continue to withhold judgment after the probability of guilt was high enough to rationally require belief. Finally, the parent might refuse to look at evidence of the son’s guilt, focus exclusively on arguments for his innocence, and refuse to think about possible objections to those arguments. 7.3.2.
The Doxastic Involuntarist Objection
The above picture portrays beliefs as similar to actions in that they may be motivated by non-epistemic, practical reasons. This does not entail but at least suggests doxastic voluntarism, the view that we can control our beliefs in roughly the same sense that we can control our actions. If we can control our beliefs, then it makes sense that we would often believe what we (for whatever reason) want to believe. However, many philosophers reject doxastic voluntarism, citing hypothetical scenarios such as the following: Girafe Prize: I announce that I have a $1 million prize that I will award to you if and only if you sincerely believe that you are a girafe. First you must be hooked up to my infallible polygraph device. Then I will ask you, “Are you a girafe?” If you answer “Yes” and the machine says you are not lying, you get the money. Could you win the prize? Most people answer “no,” thus demonstrating that we cannot in general control our beliefs and that the mere desire to believe a proposition is impotent to produce belief.7 But perhaps the example does not quite show that, since believing that one is a girafe could have serious negative consequences, perhaps even outweighing the value of $1 million. In addition, those of us who aren’t girafes have decisive epistemic reasons against the belief that we are girafes, so perhaps the example shows only that one cannot believe something for practical reasons when one has conclusive evidence against it.
100
Michael Huemer
That leaves open that one may be able to believe a proposition for practical reasons in other cases. So imagine a variant of the case: Even Number Prize: As above, except that the prize is for believing that the number of atoms in the universe is even. Here, it seems that you would have overall most practical reason, if able, to adopt the belief in question. Furthermore, there are no evidential reasons against the required belief, which has exactly a 50% probability on your evidence. Thus, if people can ever adopt a belief for practical reasons, it seems that Even Number Prize is the sort of case in which they should be expected to do so. Yet it still seems that most people could not, in fact, win the prize in this case.8 You could of course say, “Yes, the number of atoms is even” and perhaps put on a good performance, but you would only be pretending, and my infallible polygraph would detect the lie. All of which suggests that the simple account of Section 7.3.1 must be mistaken: A normal person cannot believe a proposition simply because he has a non-epistemic reason for wanting to have that belief. 7.3.3.
Doxastic Semi-voluntarism
On its own, Even Number Prize would seem to show that practical reasons are impotent to afect belief. But we know that is not so, because of such cases as Accused Child. We all know that in a case such as that, a parent very likely would believe his son innocent, even in the face of strong evidence of guilt. So the puzzle is to say when and why we can believe for non-epistemic reasons. How is Accused Child diferent from Even Number Prize? I suggest that the problem with Even Number Prize is that the evidential situation is too simple and the practical motivation for belief too explicit. There is zero evidence either for or against the number of atoms’ being even. If you were to adopt the belief that the number is even, there is no story you could tell yourself about how the belief could be considered rational. We know from cases such as Girafe Prize that one cannot choose to adopt an obviously false belief. Cases such as Even Number Prize show us, in addition, that one cannot choose to adopt an obviously unjustifed belief. But that is not to say that one cannot choose to adopt an unjustifed belief. One may be able to do so when the belief’s irrationality is more disguised. Nor does it imply that practical reasons cannot infuence belief. They may be able to do so when the evidential situation is more complex and ambiguous. That is how I think most unjustifed beliefs
How to Be Irrational
101
come about: They occur when we have non-epistemic reasons for belief and the evidential situation is somehow unclear, such that it is less than obvious to the believer that her belief is unjustifed. On this account, non-epistemic reasons have some direct infuence on beliefs, but this infuence is limited—the epistemic reasons must not be too strong or too clear, if we are to believe for non-epistemic reasons. We can only stray so far from epistemic rationality (though individuals may vary in how far they can stray). Compare the view that, while human beings have free will, our will is only so strong—most of us cannot, for example, simply choose to hold our hand in a fre for an extended period of time. Thus, the frst important element in typical cases of unjustifed believing is that the subject should have information that appears to support the desired belief (else the belief will be too obviously unjustifed), alongside information that casts doubt on it (else the belief will actually be justifed). This will make it possible for the subject to attribute his belief to the positive evidence. It helps if the evidence for and against P is of diferent kinds, such that it is unclear how such information should be weighed. In the Accused Child case, the parent can tell himself, plausibly, that he has a great deal of evidence regarding his son’s character, drawn from all his years observing the child. That evidence will normally be very complex and contain many bits that support the child’s having a virtuous character. It will be unclear how this evidence should be weighed against the more direct evidence regarding the specifc crime, for example, witness testimonies. Although the evidence of innocence is instrumental for the parent’s belief, the belief is not simply caused by that evidence. We may assume that the evidence is not strong enough to cause a belief by itself—if the parent had comparable evidence regarding an alleged crime by some other child, the parent would not deem that child innocent. The parent’s desire is able to cause belief in the context of an ambiguous evidential situation; alternately, one might say that the ambiguous evidence is able to cause belief in the context of a desire to believe. And the reason for this is that the subject is able to avoid recognizing the epistemic irrationality of the belief given the unclear evidential situation. 7.3.4.
Mistaking Desires for Appearances
Another key element in most cases of unjustifed belief is the emotional reaction that one has to a particular thought. When we entertain some propositions, we have a positive or negative afective response. Suppose, for example, that you actively dislike a particular rival social group—say, people who hold political views opposite to your own on some issue of great import. (I leave it to you to fll in the details so that you would have
102
Michael Huemer
the stipulated attitude.) Someone tells you a disparaging claim about that group—say, that they tend on average to be more dishonest and less benevolent than your own group. This is meant as a factual claim, not a pure insult, to which some of your experience is relevant. Your informant also gives some superfcial evidence for the claim, for example, that some survey came up with that result. When you start to think about the claim, you have a positive reaction. Rehearsing the claim in your head feels good, in something like the way that punching something feels good when you are angry. The disparaging claim fts with your current attitudes, and you can feel that ft. You are thus favorably inclined toward the claim. I intend this as an example of an unjustifed belief. The evidence for the claim (let us assume) is not sufcient for outright belief and would not have induced you to form a belief in the absence of your antipathy toward the rival group. Nevertheless, you are able to form the belief because you do not notice that it is unjustifed. If the issue arises, you will deny that you adopted that belief for the pleasure of insulting the rival group. Instead, you will say, and to some extent believe, that the disparaging claim just seems true, based on all your experience as well as the evidence given by your interlocutor. You will be able to say this, in part, because the attraction that you feel toward the disparaging claim is similar to an appearance. It is merely a diferent source of inclination to believe. You confuse the felt appropriateness of asserting the claim with its seeming true. It is important for my account that the attraction to the disparaging claim is not an appearance. Careful introspection would reveal that it derives from antipathy, rather than from a truth-oriented cognitive faculty. But the same motive that leads you to adopt the belief also discourages you from such introspection. You don’t carefully examine the claim or your own thought processes, because you fnd the claim congenial. A similar phenomenon happens in reverse when we hear claims that we fnd repugnant. Thus, if you hear a disparaging claim about your own social group, you will sense the clash between the claim and your current attitudes. You feel resistance, and that resistance infuences you to reject the claim if there is any way of doing so. The felt resistance can be confused with an appearance, such that you fail to notice that you are forming beliefs in an epistemically irrational manner. This points up another way in which the case of the Even Number Prize difers from Accused Child or other cases in which we commonly adopt beliefs for non-epistemic reasons. This is that in Even Number Prize, there would be no feeling of ft or appropriateness when one thinks about the proposition [The number of atoms in the universe is even]. There would be no such feeling because you do not (if you are normal) have any preexisting attitudes that the belief could be said to ft with. By contrast, in
How to Be Irrational
103
the Accused Child case, the parent’s pre-existing attitudes of love, loyalty, and trust in the child ft, in an obvious sense, with the parent’s believing the child innocent. What makes a belief ft with an attitude? Usually, this turns on evaluative claims that the attitude would match.9 If you feel afection for someone, that fts with the belief that the person is worthy of such afection. By extension, the descriptive claims that would make such evaluative beliefs true also ft with the attitude. Thus, if Carole feels afection for Ted, that fts with Carole’s belief that Ted is not the sort of person who would hurt other people for fun. If Jon feels afraid of fying, that fts with his belief that airplanes have frequent accidents. 7.4. Corruption of Doxastic Practices 7.4.1.
The Selection of Information Sources
Though it is controversial that practical goals can directly infuence credences or outright beliefs, it is uncontroversial that they can infuence credences or beliefs indirectly, by afecting how we conduct inquiry. Regardless of whether we can directly choose beliefs, we can certainly choose how we gather evidence. We can, for example, decide to listen only to news sources that share our political orientation. This efectively guarantees that we will receive mainly information that fts with that orientation. A disinterested truth-seeker would prioritize the information sources that are most likely to tell him something he does not already know, including sources with views that contrast with his own. Most human beings with strong political opinions, however, favor reinforcement of their existing opinions: conservatives read and watch conservative sources; progressives read and watch progressive sources. Often, these sources have entirely diferent news stories.10 A conservative source, for example, might carry a story about a crime committed by an immigrant, or a story about a law-abiding citizen heroically using his gun to stop a crime. A progressive source would likely overlook those stories, instead favoring a story about police abusing a black suspect, or a crime committed using a gun. Each type of news source presents evidence supporting the beliefs of their viewers, thus helping them to maintain and strengthen the beliefs that they prefer to hold. Such evidence is anecdotal and cherry picked, and thus of limited actual probative value. But as a matter of psychological fact, such evidence tends to strongly infuence people’s opinions. Since we are aware of this fact, the selection of news sources presents us with a way of intentionally, indirectly controlling what we believe. The service goes beyond the mere presentation of objective, factual information. By prudently selecting information sources, we can usually
104
Michael Huemer
count on being supplied with arguments congenial to our worldview without cluttering our minds with opposing arguments. If some evidence is ambiguous and capable of being interpreted in light of diferent ideological commitments, we can usually count on our news source to supply us with a congenial interpretation, which we might not have been able to think of on our own. The technique for controlling our beliefs goes beyond selection of news sources. Most people with strong, controversial beliefs of any kind prefer to associate with others who share those beliefs, rarely discussing the topics of those beliefs with people who hold other views. For instance, most religious individuals prefer spending time with those who share their faith, while avoiding discussion of religion with those who hold other faiths or no faith. Most Christians would not spend time reading a book defending atheism. Most people who read philosophical work tend to prefer authors with whom they generally agree. (Caveat: The norms of the academic world force academics to read opposing authors in order to publish their own work. The practical goal of publication thus tends to override the preference for material that we agree with. But when reading for pleasure, even academics tend to prefer works by likeminded authors.) 7.4.2.
Direction of Attention
Just as we can choose where we get external information from, we can choose where we direct our attention within the range of things we are aware of. If we know of a variety of arguments and pieces of evidence on a given subject, we can choose to dwell on the arguments and evidence that are most pleasant to contemplate, which would be the arguments and evidence congenial to our preferred beliefs. This is likely to increase the psychological infuence of those arguments and evidence, relative to the arguments and evidence to which we devote less attention. We may also decide what sorts of ideas to attempt to think of: we can try to think of arguments in favor of a particular view, and we can refrain from trying to think of objections to those arguments. In most cases, if we try to think of reasons for a controversial view, we will think of some. In most cases, if we don’t try to think of objections to an idea, then we will not think of objections (though if an idea’s faws are sufciently obvious, or we are sufciently intelligent, we may be unable to avoid seeing some objections). All of this is to say that we can, at will, violate the norms of responsible, truth-oriented inquiry. We can put on a sort of show trial for ourselves aimed at vindicating a pre-selected conclusion.
How to Be Irrational 7.4.3.
105
The Role of Doxastic Semi-voluntarism
The above techniques are useful to us because doxastic semi-voluntarism is true; that is, non-epistemic desires have some direct infuence on belief but this infuence is limited by the strength of our epistemic reasons. Let us pause to see why this is so. First, note that the technique of selecting evidence sources in a biased manner would have no expected payof for a perfectly epistemically rational being in whom non-epistemic desires had no direct infuence on belief. The perfectly rational being could of course deliberately select evidence sources favorable to his preferred belief, but upon doing so, his mind would automatically conditionalize on the fact that he had selected evidence sources in that way. Thus, the conclusion superfcially supported by the evidence would have to be discounted by the likelihood that counterevidence would have been turned up had the agent consulted other sources. The agent’s ex ante expectation about how strong the counter-evidence would be would perfectly match how strongly his mind would discount the desired conclusion. For example, let’s say I want to believe that gun control laws are benefcial. I currently have a 0.8 credence in this proposition, which I would like to go up, or at least not to go down. Now I must decide whether to read John Lott’s book, More Guns, Less Crime, which defends the utility of private gun ownership.11 Before reading it, I know that it will present evidence against gun control. I have some expectation about how strong this evidence would be. Without worrying about exactly how strength of evidence should be measured, let’s suppose that the expected value of the strength of that evidence, based on my current information (before reading the book) is S. Then, if I am perfectly rational, my current credences have already taken into account the fact that there is evidence against gun control with an expected strength of S. Therefore, if the actual evidence in the book turns out to be of strength S, then my credence in the utility of gun control will remain unchanged. If it turns out to be stronger than S, then my credence will unfortunately go down. But if it turns out that Lott makes a weaker case than I was expecting, then my credence in the utility of gun control will go up. The possibility in which my credence goes up exactly counter-balances the possibility in which it goes down, so that there is zero net expected impact on my credence from reading the book. A similar observation applies to the technique of strategically directing one’s attention and energy. If I try to think of arguments for gun control and don’t try to think of any objections, I will, as a rational being, have to conditionalize on the fact that I did that. The possibility that the objections I would think of would be weaker than I currently expect would exactly counterbalance the possibility that they would be stronger; thus,
106
Michael Huemer
strategically directing my attention would have zero net expected impact on my credence. But we all know that this is not true—strategic selection of sources and direction of attention do have a predictable efect on our doxastic attitudes. This is because, apart from our choices about where to get information and how to direct our attention, our belief formation is not perfectly epistemically rational to begin with; it is to some extent directly infuenced by our preferences. Thus, if we wish to believe P, we can to some extent overlook the biases in our selection of information sources. We can neglect to conditionalize on, or at least underweight, the likely existence of unknown counter-evidence. By contrast, if we desire not to believe P, and we have somehow run into evidence that strongly supports P, we can choose to take into account the likely existence of unknown counter-evidence and use this as grounds for remaining skeptical. The techniques of manipulating evidence and attention would also be of no use to a being with perfect voluntary control over her beliefs. Such a being could simply believe whatever she wanted by directly deciding to do so. There would be no need to select evidence sources, as she could simply decide not to be infuenced by any evidence that would support beliefs she does not want to hold. Nor would there be any need to strategically direct her attention, as she could, upon thinking of an objection to her view, simply decide not to accept that objection, however forceful it might be. Thus, the sort of corruption of inquiry that I have described is best suited to semi-rational beings such as ourselves, with imperfect rationality and imperfect doxastic control. 7.5.
Conclusion
There are two main demands of epistemic rationality: to conduct inquiry in a truth-directed way and to base our beliefs on what seems true to us. But human beings have goals other than truth and rationality, and these goals are often furthered or hampered by what we believe. Thus, we sometimes violate one of the requirements of epistemic rationality, motivated by nonepistemic inclinations. But we cannot believe anything we wish; we cannot adopt beliefs or credences that are too obviously false or unjustifed. Hence, we tend to adopt unjustifed credences or beliefs when the evidential situation is somewhat unclear and when we have doxastic inclinations that can be confused with appearances. When we sense a kind of ft between our emotional attitudes and a particular proposition, we may confuse this with the proposition’s seeming correct. In addition, we often conduct inquiry in a biased manner, seeking to avoid gathering evidence against our preferred beliefs, directing our attention toward information that favors our preferred beliefs, and
How to Be Irrational
107
refraining from thinking of objections. This is the explanation for most, perhaps all, internally unjustifed beliefs. These processes are rarely if ever fully conscious, and we do not normally expressly decide to adopt irrational beliefs. We simply fnd ourselves naturally doing so under the infuence of our doxastic inclinations, just as a person naturally follows his inclinations in his ordinary behavior unless he makes a specifc efort to resist them. Rationality is thus not merely the absence of irrationality; rationality requires specifc eforts to avoid engaging our natural self-deceptive mechanisms. The importance of rationality lies in the fact that it leads to beliefs that are more likely to be correct. In many cases, individuals and society as a whole are seriously harmed by incorrect beliefs, partly due to the temptation to believe what we want to believe rather than what seems true. The above analysis of how unjustifed beliefs come about is meant as an aid to rationality: the conscientious thinker should take note of how irrational beliefs commonly arise in order to be on guard against them. One should periodically ask oneself such questions as whether a proposition we are inclined to endorse really seems true or only emotionally comfortable, and whether we have conducted a fair and thorough inquiry before adopting a belief. We should make eforts to gather evidence and ideas from a variety of sources and to entertain objections to the ideas to which we are attracted. In this way, though we may have less emotionally satisfying beliefs in the short run, our beliefs are likely to become more accurate and rational in the long run. Notes 1 2 3 4 5 6 7 8
Goldman 1979. Huemer 2007, 30. On outright belief, see Owens 2000; Stafel 2019; Williamson forthcoming. On moral encroachment, see Case 2019. Leaving aside problems with proper basing. Sosa 1998, 258–259; Swinburne 2001, 141–142. Alston (1988, 263) gives a similar example. When I raise these sorts of examples in class, a few students think they could win the Even Number Prize or even the Girafe Prize. I worry for those students’ sanity. But even if some unusual people can form such beliefs, I wish to examine how ordinary people who lack capacities for radical doublethink may acquire unjustifed beliefs. 9 On the notion of ftting attitudes, see Brentano 2009; Ewing 1948. 10 Keegan 2016. 11 Lott 2010.
References Alston, William. 1988. “The Deontological Conception of Epistemic Justifcation,” Philosophical Perspectives 2, Epistemology: 257–299.
108
Michael Huemer
Brentano, Franz. [1889] 2009. The Origin of Our Knowledge of Right and Wrong, ed. Oskar Kraus and Roderick Chisholm, tr. Roderick Chisholm and Elizabeth Schneewind (London: Routledge & Kegan Paul). Case, Spencer. 2019. “From Epistemic to Moral Realism,” Journal of Moral Philosophy 16:541–562. Ewing, A. C. 1948. The Defnition of Good (London: Routledge). Goldman, Alvin. 1979. “What Is Justifed Belief?”, pp. 1–25 in George Pappas (ed.), Justifcation and Knowledge (Boston: D. Reidel). Huemer, Michael. 2007. “Compassionate Phenomenal Conservatism,” Philosophy and Phenomenological Research 74:30–55. Keegan, Jon. 2016. “Blue Feed, Red Feed” (web page), Wall Street Journal, https:// graphics.wsj.com/blue-feed-red-feed, accessed July 26, 2022. Lott, John. 2010. More Guns, Less Crime: Understanding Crime and Gun Control Laws, Third Edition (Chicago, IL: University of Chicago Press). Owens, David. 2000. Reason Without Freedom: The Problem of Epistemic Normativity (London: Routledge). Sosa, Ernest. 1998. “Minimal Intuition,” pp. 257–270 in Michael DePaul and William Ramsey (eds.), Rethinking Intuition (Lanham, MD: Rowman & Littlefeld). Stafel, Julia. 2019. “How Do Beliefs Simplify Reasoning?”, Noûs 53:937–962. Swinburne, Richard. 2001. Epistemic Justifcation (Oxford: Oxford University Press). Williamson, Timothy. Forthcoming. “Knowledge, Credence, and the Strength of Belief” in Amy Flowerree and Baron Reed (eds.), Expansive Epistemology: Norms, Action, and the Social World (London: Routledge).
Part 2
Seemings in Inference and Inquiry
8
Dogmatism, Seemings, and Non-deductive Inferential Justifcation Berit Brogaard and Dimitria Electra Gatzia
8.1.
Introduction
Dogmatism—understood as a general thesis—is the view that an experience or seeming that p can provide defeasible (or prima facie) immediate justifcation for believing p in virtue of its phenomenology (e.g., Pryor, 2013). Dogmatism, as popularized by Jim Pryor (2000), is a thesis about perceptual justifcation and presupposes a particular view of perception. As we shall see, phenomenal seemings play a crucial role in dogmatism of this kind. However, dogmatism is not by defnition a thesis about perceptual justifcation but can be—and has been—defended for other kinds of justifcation, including memorial, introspective, and a priori justifcation (Huemer, 2001; Brogaard, 2013; Chudnof, 2013). Furthermore, while dogmatism about perceptual justifcation has appealed primarily to advocates of representational theories of perceptual experience, dogmatism is consistent with other views of perceptual experience. Naïve realists and disjunctivists who hold that perceptual experience is a perceptual relation of direct acquaintance with an external-world fact or object can also endorse dogmatism. Indeed, we will argue that they ought to do so. Otherwise, it is not clear that they can coherently maintain that relationalism about perceptual experience has an epistemic advantage compared to representational views. After arguing that advocates of relational approaches to perceptual experience ought to be dogmatists, we look more closely at the alleged advantage of relational views over representational views. We then show that even if we grant that relationalism has this advantage, only dogmatism that takes perceptual experience to be representational can be extended to account for non-deductive inferential justifcation. As an account of the latter is required to avoid succumbing to skepticism, relationalism, we argue, does not have the epistemic advantage its defenders claim it has.
DOI: 10.4324/9781003299349-11
112
Berit Brogaard and Dimitria Electra Gatzia
8.2. Seemings and Two Forms of Phenomenal Dogmatism Dogmatism—understood as a general thesis—is the view that an experience or seeming that p can confer at least some degree of prima facie immediate justifcation on the belief that p by virtue of its phenomenology.1 The terms “immediate” and “prima facie” are crucial to this view of justifcation. To say that justifcation for believing a proposition p is immediate is to say that it’s not even partly constituted by justifcation for believing another proposition q. For example, if your perceptual experience as of a dog being S-shaped, tall, slim, short-haired, and long-tailed provides justifcation for believing that the dog is a Greyhound only together with justifed background assumptions about what Greyhounds look like, such as the justifed background assumption that Greyhounds are S-shaped, then your justifcation for your belief is mediate rather than immediate. This is because your perceptual experience serves as justifcation for believing that the dog is a Greyhound only together with background assumptions that themselves provide justifcation for other propositions, for instance, the proposition that Greyhounds are S-shaped. It should be noted that while immediate justifcation is a form of non-inferential justifcation, the converse is not true. That is, inferential justifcation can be a form of immediate justifcation. For example, intuitions can provide immediate justifcation for the proposition that p entails p or q. The question of whether there is immediate justifcation can be formulated either propositionally or doxastically (Brogaard and Gatzia, 2020). Phenomenal dogmatism is a claim about propositional justifcation, not doxastic justifcation (Turri, 2010). Where doxastic justifcation is something beliefs possess, propositional justifcation is something subjects have for believing a proposition. Propositional justifcation is also sometimes referred to as a “warrant” or “evidence.” To say that your justifcation is prima facie is to say that it is defeasible, that is, you have evidence that can weaken or wholly undermine your justifcation. A defeater is evidence that you come to possess, which directly or indirectly calls into question your justifcation for believing a proposition. A rebutting defeater undermines by directly providing justifcation for an opposing proposition. For instance, if Junior has a perceptual experience as of the ashtray being square but Lily tells him that it’s round, then what Lily said directly provides justifcation for the proposition that the ashtray is round, which opposes the content of Junior’s seeming that it is square. An undercutting defeater undermines by undercutting the support otherwise provided by one’s justifcation. For instance, if Junior has an experience as of the ashtray being square but his doctor tells him that his shape vision is defective, then the doctor’s testimony undercuts his experience’s justifcatory status. Defeaters can be misleading, which is to say that they are
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
113
inaccurate unbeknownst to the subject yet still defeat the subject’s justifcation to the same extent as non-misleading defeaters. For example, suppose Junior has an experience as of the ashtray being square, but Lily tells him that it’s round. Unbeknownst to Junior, Lily’s testimony is inaccurate. In this case, Lily’s testimony provides Junior with a misleading defeater of his experience’s justifcatory status. Since defeaters can be misleading yet still defeat a subject’s justifcation to the same extent as non-misleading defeaters, all justifcation is defeasible. Dogmatism, as advanced by Jim Pryor (2000), is a thesis about perceptual justifcation and which presupposes a particular view of perception. However, the scope of dogmatism is not limited to perceptual justifcation but can be—and has been—applied to other kinds of justifcation, including memorial, introspective, and a priori justifcation (Chudnof, 2013, 2014; Brogaard, 2013). Since Pryor (2000) gave currency to the theory, dogmatism about perceptual justifcation has been taken to involve a representational account of perceptual experience. Representational theories hold that perceptual experience fundamentally is a matter of representing how things phenomenally seem to be. Proponents of representational views take the phenomenology of perceptual experience to consist exclusively or mostly of representational phenomenal properties, such as the property of representing something as being oval (e.g., Chalmers, 2004; Siegel, 2012; Brogaard, 2018). Thus understood, representational theories are compatible with the view that some of the phenomenal properties of perceptual experience are non-representational. For instance, it may be said that while appearance properties such as colors, extension, or texture are refected in the experience’s representational character, experiential properties such as imprecision or salience are refected only in the experience’s non-representational character (Block, 2015). Call dogmatism about perceptual justifcation that takes perceptual experience to be representational “representational phenomenal dogmatism.” On a representational view, perceptual experience makes the external world phenomenally seem a certain way to a perceiver in virtue of its phenomenology, and it is in virtue of how its phenomenology makes the external world seem that it provides justifcation for believing a proposition. Accordingly, we can render representational phenomenal dogmatism about perceptual justifcation in terms of phenomenal seemings as follows: Representational Phenomenal dogmatism (Perceptual Justifcation)
If S’s perceptual experience makes it phenomenally seem to S that p, then S thereby has at least some degree of prima facie immediate (propositional) justifcation for believing that p.
114
Berit Brogaard and Dimitria Electra Gatzia
Figure 8.1 Müller-Lyer Illusion. While both lines have the same length, the line with the arrows pointed inward (top) phenomenally seems longer than the line with the arrows pointed outward (bottom).
Phenomenal seemings difer from epistemic seemings in that the former normally are evidence-insensitive, whereas the latter are not. An evidenceinsensitive seeming persists even when you have counterevidence suggesting that the seeming is inaccurate (Chisholm, 1957; Brogaard, 2018). In the Müller-Lyer illusion (Figure 8.1), the two lines phenomenally seem to you to have diferent lengths even if you know that they have the same length. As your seeming persists in spite of you possessing counterevidence suggesting that it is inaccurate, it is evidence-insensitive and thus phenomenal. Like the content of degrees of belief, the content of epistemic seemings is merely probable. On a Bayesian model, the subjective probability, or credence, of a hypothesis H is the product of the likelihood of H, given your new evidence, and the probability of your prior beliefs about the world (the “prior”).2 In the case of epistemic seemings, their content is the hypothesis, and the seeming attributes a high subjective probability of that content. Say that you are listening to an episode of the Podcast The Rewatchables, and the host Bill Simmons says that his favorite movie of all time is E.T. The Extra-Terrestrial. Given Bill Simmons’ testimony, you take it to be subjectively probable that his favorite movie is E.T. As your credence is higher than 0.5 but lower than 1, it comes to epistemically seem to you that his favorite movie is E.T. As the content of your seeming is merely subjectively
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
115
probable, it easily yields to defeating evidence. Suppose at the end of the roundtable discussion, Bill Simmons says “Actually, when I said I couldn’t stop rewatching E.T., I was just kidding. My favorite movie of all time is, of course, Michael Mann’s Heat.” Assuming you are rational, you will update on your new evidence, and it will no longer epistemically seem to you that Bill Simmons’ favorite movie of all times is E.T. Instead, it will come to epistemically seem to you that his favorite movie is Heat. Although dogmatism about perceptual justifcation has appealed primarily to advocates of representational accounts of perceptual experience, the dogmatist thesis is consistent with other accounts of experience. Naïve realists, disjunctivists, and reliabilists, for instance, can also adopt dogmatism about perceptual justifcation as long as they hold that perceptual justifcation is immediate and defeasible (Pryor, 2013; Brogaard, 2013, 2021). Most forms of naïve realism and disjunctivism are relational views, viz. views that hold that perceptual experience fundamentally is a matter of being directly acquainted with an external-world fact or object (Brewer, 2011; Allen, 2016). Relationalists do not normally use the term “dogmatism” to characterize their views of the epistemic role of perceptual experience. However, one of the main motivations for relationalism is that this sort of view of experience is required to provide an adequate account of perceptual justifcation (McDowell, 1982; Fumerton, 1995; Fish, 2009). The gist of their argument is this: on non-relational representational views, perceptual experience directly acquaints us only with an intermediary between us and the external world (Brewer, 2011; Allen, 2016). But if we are only directly acquainted with such intermediaries, then it’s hard to see how we can have epistemic access to the external world, which leaves the representational views vulnerable to skepticism about perceptual justifcation. Relationalists take this to suggest that only experience that directly acquaints us with an external-world fact p can provide justifcation for believing that p.3 Note, however, that for this argument to be compelling, relationalists cannot say that perceptual experience only provides mediate justifcation for propositions. Suppose otherwise. Then advocates of representational views could say that the experience provides mediate justifcation for propositions together with justifed background assumptions guaranteeing that the right sort of relation obtains between us and the external world. This way advocates of representational views would avoid the relationalists’ skeptical argument made against them. This move by the representational theorists would even the score. The relational views would thus lose their epistemic advantage over representational views. So, it seems that relationalists must hold that experience provides immediate justifcation for believing a proposition. Moreover, because defeaters can be misleading, all justifcation is defeasible. So, even though relationalists are not in the
116
Berit Brogaard and Dimitria Electra Gatzia
habit of referring to their view of perceptual justifcation as a form of dogmatism, they have compelling motivation for their account of perceptual justifcation only if they commit to a form of dogmatism, which is committed to the possibility of immediate justifcation arising from perceptual experiences. Relationalists take perceptual experience to provide perceptual justifcation for a proposition by virtue of the experience’s relational phenomenology, which they argue puts us in direct contact with an external-world fact. As phenomenal seemings have propositional content and thus do not put us in direct contact with the external world, relationalists must reject that experience provides justifcation for propositions in virtue of how it makes things phenomenally seem to us. Call dogmatism about perceptual justifcation that takes experience to be relational “relational phenomenal dogmatism.” This view can be glossed as follows: Relational Phenomenal dogmatism (Perceptual Justifcation)
If S stands in a perceptual relation of direct acquaintance to p, S thereby has at least some degree of prima facie immediate (propositional) justifcation for believing that p. In the next section, we look more closely at how the two forms of phenomenal dogmatism propose to solve the standard skeptical challenge about perceptual justifcation. 8.3.
The Relationalist Argument Against the Representational View of Perceptual Experience
Defenders of representational phenomenal dogmatism argue that their view rebuts a standard skeptical challenge to internalist views of justifcation (e.g., Pryor, 2000, 2004). For example, your experience as of having hands provides at least some degree of prima facie immediate justifcation for believing that you have hands. As skeptical alternatives, such as the hypothesis that you are a handless brain in a vat (BIV), are merely possible, they do not defeat your experience’s justifcatory status. Whether the Moorean reasoning from the belief that you have hands to the belief that you are not a handless BIV is admissible is a further question.4 Even if the Moorean reasoning is no good, this doesn’t by itself show that your experience as of having hands cannot provide at least some degree of prima facie immediate justifcation for believing you have hands. Relationalists about perceptual experience, however, argue that their view has an epistemic advantage compared to non-relational views of experience (e.g., McDowell, 1982; Fumerton, 1995; Fish, 2009). If the
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
117
relationalist argument is sound, it follows that representational theories cannot fend of the skeptical challenge to perceptual justifcation. The relationalist argument runs as follows: on non-relational accounts, perceptual experience directly acquaints us only with an intermediary interposing between us and the external world. Unlike sense-datum theories, representational theorists do not take the intermediary to be a sense-datum, but rather a propositional content that represents the world as being a certain way (Siegel, 2012; Brogaard, 2018). Representational views thus entail that perceptual experience does not provide direct conscious access to the external world. But if experience doesn’t do that, then it can make it phenomenally seem that things are a certain way when they are not. Given a representational view, if your perceptual experience were to make things phenomenally seem a certain way to you, then it would also make things seem that way had a skeptical alternative been true. So, your experience does not provide epistemic evidence against skeptical alternatives. But in that case, representational views succumb to skepticism after all, or so the argument goes. For illustrative purposes, let’s consider a particular instance of this argument: Premise 1: S’s perceptual experience as of having hands makes it phenomenally seem to S that S has hands. Premise 2: If S had been a handless BIV, S’s perceptual experience would also have made it phenomenally seem to S that S has hands. Intermediary Conclusion: A perceptual experience that makes it phenomenally seem to S that S has hands provides no evidence for the proposition that S is not a handless BIV. Premise 3: A perceptual experience as of p can provide some justifcation for p, only if it provides some evidence against skeptical alternatives that also make it phenomenally seem that p. Conclusion: S’s perceptual experience as of having hands does not provide any justifcation for the proposition that S has hands. Relationalists take the vulnerability of non-relational views to skepticism to suggest that only experience that directly perceptually acquaints us with an external-world fact p can provide perceptual justifcation for believing that p. According to relationalists, their view fares better than representational theories because experience directly phenomenally acquaints us with an external-world fact. As our experience directly acquaints us with the external-world fact that we have hands, but our experience does not directly acquaint us with this fact in the skeptical scenario where we are handless BIVs, the relational view can explain the epistemic asymmetry between the non-skeptical and skeptical scenarios, thus blocking the skeptical argument against the evidential role of perceptual experience.
118
Berit Brogaard and Dimitria Electra Gatzia
In the next section, we argue that relationalists face even greater difculties than those they attribute to representational theories. In particular, we argue that relationalism cannot be extended to account for ampliative inferential justifcation. 8.4.
Relationalism and Ampliative Inferential Justifcation
Ampliative arguments are non-deductive, which is to say that their conclusion contains information not contained in the premises. Consider the following ampliative arguments: (1) Every time Sara gets solicitation emails, she deletes them without reading them. So, Sara doesn’t like reading solicitation emails. (2) All 10,000 times we fipped this coin, it came up heads. So, this coin is rigged. (3) Wes was planning to surprise Sylvia with an engagement ring. Sylvia is wearing an engagement ring today. So, Wes and Sylvia must have gotten engaged. (4) The German professor told Lily that the German word “Krankenschwester” means nurse. So, the German word “Krankenschwester” means nurse. (5) 99.99% of students at the French Art Academy are French. Rose is a student at the French Art Academy. So, Rose is French. (6) For the past 10 years, Eddy’s car has never been stolen from the condo garage. So, Eddy’s car will not be stolen from the condo garage today. (7) All the 1,000 black candies I tasted from this urn are licorice-favored. So, the next black candy from the urn I taste will also be licorice-favored.
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
119
(8) There is a 99.99% risk of getting injuries from an electrical shock. Zari will get an electrical shock this afternoon. So, Zari will get injured this afternoon. There are two kinds of ampliative arguments: abductive and inductive.5 (1)—(4) are abductive, or, as they are also known, inferences to the best explanation (see, e.g., Douven, 2021). They are the kinds of arguments we use to explain ordinary phenomena such as people’s behaviors in light of their mental states. We also rely on abductive arguments to arrive at empirical hypotheses that can then be empirically tested more thoroughly. (5)—(8) are inductive arguments. What makes an argument inductive is not merely the fact that it’s based on empirical or statistical data. Rather, the diference between abductive and inductive arguments is that the former implicitly or explicitly appeal to explanation, whereas the latter merely appeal to empirical or statistical data. Unlike deductive arguments, ampliative arguments are not said to be valid or invalid, but rather strong, moderately strong, or weak, depending on how much the premises, if true, would increase the likelihood that the conclusion is true. The fact that the premises of a valid deductive argument entail the conclusion corresponds to an a priori necessary truth of the form “P ⇒ C.” Likewise, the fact that the premises of a reasonably strong ampliative argument support the conclusion corresponds to a probable, contingent a posteriori truth of the form “P(A|B) > 0.5,” which roughly means that A makes B probable. For want of a better term, call necessary a priori truths and probable, contingent a posteriori truths of this kind “inferential truths,” or “inferential facts.” When “P ⇒ C” is a priori, it is plausible that we can stand in an experiential relation of direct acquaintance with the fact that P ⇒ C. So, relational phenomenal dogmatists can maintain that we have a priori inferential justifcation for a priori inferential facts by virtue of standing in an experiential relation of direct acquaintance to those facts. Relationalists evidently cannot take the immediate justifcation for a priori inferential facts to be perceptual in nature. However, they could take immediate justifcation for a priori inferential facts to be intuition experiences.6 Intuition experiences, as we shall use the term, are distinct from beliefs and inclinations to believe (Huemer, 2005; Chudnof, 2013, 2014). They are sui generis experiences that in principle can be construed as having either a representational phenomenology or a direct relational phenomenology just like perceptual experience, although they are not themselves perceptual. The most natural option for relationalists is to take intuition experiences to consist in experiential relations of direct acquaintance between the subject and an inferential fact. If they adopt
120 Berit Brogaard and Dimitria Electra Gatzia this view of intuition experiences, dogmatism about intuitive inferential justifcation can be formulated as follows: Relational Phenomenal dogmatism (Intuitive Inferential Justifcation)
If S stands in an intuitive experiential relation of direct acquaintance to p, S thereby has at least some degree of prima facie immediate (propositional) justifcation for believing that p. Now, consider the following simple deductive argument: (9) The tray is square. So, the tray is either square or round. Let’s say that Hector has a perceptual experience of the tray being square, and an intuition experience of the fact that necessarily, if the tray is square, then the tray is either square or round. Assuming a relational view of both perceptual experience and intuition experience, Hector’s perceptual experience provides prima facie immediate justifcation for the premise that the tray is square. Moreover, Hector’s intuition experience provides prima facie immediate justifcation for the deductive inferential fact that necessarily, if the tray is square, then the tray is either square or round. As we can reasonably assume that prima facie immediate justifcation is closed across deductive entailment supported by intuition experiences, it follows that Hector has prima facie immediate justifcation for believing that the tray is either square or round. However, although it is intuitively plausible that we can be immediately aware of deductive inferential facts, it is rather implausible to think that we can immediately “see” or “grasp” ampliative inferential facts (Ramsey, 1992; Fumerton, 1995: 218). After all, unlike deductive inferential facts, ampliative inferential facts are a posteriori rather than a priori. A relationalist may reply that it is possible that ampliative inferential truths are contingent a priori truths (e.g., Fumerton, 1995: ch. 7).7 If that is indeed the case, then we can plausibly be immediately aware of these truths. By way of reply, it is doubtful that ampliative inferential truths are a priori rather than a posteriori. First, ampliative inferential truth are very diferent from paradigm cases of contingent a priori truths, such as “Phosphorus is visible in the morning sky” and “Jack the Ripper murdered and mutilated female prostitutes in London in 1888.” “Phosphorus” and “Jack the Ripper” are singular terms introduced by defnite descriptions that conceptually guarantee that the predicates “λx (murdered and mutilated
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
121
female prostitutes in London in 1888(x))” and “λx(visible in the morning sky(x))” are true of their referents, viz. Venus and the Whitechappel Murderer, respectively. Ampliative inferential truths also difer from paradigm cases of deeply contingent a priori truths. A deeply contingent proposition p does not by itself guarantee that there is a fact that confrms p in the actual world. In other words, it is conceivable that while p is true in some possible world, p is false in the actual world. As confrming p would require empirical investigation, deeply contingent truths do not seem to be a priori (Evans, 1985). Even so, John Hawthorne (2002: 11–12) has argued that there could be deeply contingent a priori truths. Suppose that you have not had any experiences yet, but that you anticipate a variety of experiential life histories, H1, H2, . . ., Hn, and you conceive of various theories T1, T2, . . ., Tn that describe possible structures of the actual world. Suppose further that you possess an innate body of true principles you can apply to determine whether a theory about the actual world is the best explanation of an experiential life history. Now, whether you have justifcation to believe that if Tn is true, then He is true depends on whether Tn is the best explanation of He, which is something you can determine on the basis of your innate principles. For example, if T1 best explains H8, then you have justifcation for believing T1 → H8. But the material conditional T1 → H8 is a deeply contingent a priori truth. Although you are able to conceive of any such material conditional as false, this does not undermine the justifcation you have for believing the conditional. So, you can come to have justifcation for believing a deeply contingent truth on purely a priori grounds. While Hawthorne makes a compelling case for the possibility of deeply contingent a priori truths, his argument doesn’t support the thesis that we can come to have justifcation for all ampliative inferential truths on purely a priori grounds. To assess whether an ampliative inferential proposition is intuitively plausible, we need to draw on our prior experiences, including our past experiences of co-occurrences of events, testimonial evidence, and explanatory principles we have committed to memory. For example, to assess the intuitive plausibility of “If the sky is completely dark, then it will rain,” we may be able to draw on our past experiences of co-occurrences of a dark sky and subsequent rain. As assessing the intuitive plausibility of ampliative inferential propositions normally requires drawing on prior experience, we cannot come to have justifcation for their truth on purely a priori grounds. But if we cannot come to have justifcation in this way, then we must reject the suggestion that we can stand in intuitive experiential relations of direct acquaintances to ampliative inferential facts. At this point, relationalists may bite the bullet when it comes to intuition experience and insist that even though they hold that perceptual experience does not have a representational structure, this does not prevent them
122
Berit Brogaard and Dimitria Electra Gatzia
from taking intuition experiences to have this structure. In other words, relationalists could in principle agree with the representational theorists about intuition experience but not about perceptual experience. Representational phenomenal dogmatism about intuitive inferential justifcation can be glossed as follows: Representational Phenomenal Dogmatism (Intuitive Inferential Justifcation)
If S has an intuition experience that makes it phenomenally seem to S that a deductive or ampliative proposition p is true, then S has at least some degree of defeasible immediate justifcation for believing that p. If, however, relationalists regard intuition experiences as representational experiential states, then they become vulnerable to the same sort of skeptical challenge that they claim representational views of perceptual experience succumb to. On a representational view of intuition experience, the representational content is an intermediary interposing between us and the inferential fact. A representational view of intuition experience thus entails that intuition experience does not provide direct conscious access to inferential facts by virtue of its phenomenology. But if intuition experience doesn’t do this, then an intuition experience can make it phenomenally seem that an inferential fact obtains when it doesn’t. But given a representational view of intuition experience, if an intuition experience makes things phenomenally seem a certain way to us, it would also have made things seem the same way had a skeptical alternative been true. So, given a representational view of intuition experience, intuition experience does not provide evidence against skeptical alternatives. But if experience doesn’t provide such evidence, then representational theories of intuition experience succumb to skepticism. For illustrative purposes, consider the following argument. Let F be the inferential fact that because Rosalinda’s dog Fido never bit any of her 1000 past visitors, Fido will not bite her next visitor, which should have a high credence. Premise 1: S’s intuition experience makes it phenomenally seem to S that F has a high likelihood of being true. Premise 2: If S had come into existence 4 seconds ago complete with all her current experiences and memories, then S’s intuition experience would also make it phenomenally seem to her that F has a high likelihood of being true. Intermediary conclusion: An intuition experience that makes it phenomenally seem to S that F has a high likelihood of being true provides no evidence against the skeptical alternative that S came into existence 4 seconds ago complete with all her current experiences and memories.
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
123
Premise 3: An intuition experience as of p can provide some inferential justifcation for p, only if it provides some evidence against skeptical alternatives that also make it phenomenally seem that p. Conclusion: S’s intuition experience as of F having a high probability of being true does not provide any inferential justifcation for the proposition that F has a high probability of being true. So, if relationalists take this route of rendering intuition experiences to be representational, then they cannot also push their perceptual account’s alleged virtue, viz. that it has an edge over a representational view when it comes to the skeptical challenge to perceptual justifcation. This then would raise the question of what motivates the relational view in the frst place. So, relationalists would face a dilemma. They can take intuition experience to be either relational or representational. If they accept the frst horn of the dilemma, and thus hold that intuition experiences are experiential relations of direct acquaintance between a subject and an inferential fact, then they cannot account for ampliative inferential justifcation. If they accept the second horn of the dilemma and thus hold that, unlike perceptual experience, intuition experience is representational rather than relational, then they can no longer coherently maintain that their view of perceptual experience has an epistemic edge over a representational view. These considerations thus give us reason to reject the relational views of perceptual experience and the correlated epistemic thesis of relational phenomenal dogmatism. In the next section, we argue that the representational views of experience don’t run into the same kind of trouble as the relational views of experience. 8.5.
Intuition Seemings and Representational Phenomenal Dogmatism
For proponents of the representational view of perceptual experience, the most natural account of intuition experiences is one that takes them to be structurally analogous to perceptual experiences. If intuition experiences are representational states, then representational phenomenal dogmatism is true not only for perceptual justifcation but also for intuitive inferential justifcation. Here is, once again, representational phenomenal dogmatism about intuitive inferential justifcation: Representational Phenomenal Dogmatism (Intuitive Inferential Justifcation)
If S has an intuition experience that makes it phenomenally seem to S that a deductive or ampliative proposition p is true, then S thereby has
124
Berit Brogaard and Dimitria Electra Gatzia
at least some degree of defeasible immediate justifcation for believing that p. Because the representational view treats intuition experiences analogously to perceptual experience, it avoids the challenges facing the relational view when it comes to ampliative inferential justifcation. This is because, unlike the relational view, the representational view does not take intuition experiences to involve an experiential relation of direct acquaintance. The representational view is, therefore, not subject to the objection that we cannot be directly phenomenally acquainted with merely probable truths. Some have argued that intuitions can only provide justifcation for a priori truths (e.g., Bealer, 1998). Thus, it may be thought that proponents of the representational view should defend the claim that intuition experiences can provide immediate justifcation for merely probable a posteriori truths. By way of reply, the hypothesis that our human psychology makes us capable of undergoing intuition experiences that make it phenomenally seem to us that certain merely probable propositions are true is an empirical claim. So, it would need to be empirically justifed. But, in fact, we do have empirical evidence to justify this claim. For example, it will likely phenomenally seem to competent speakers on intuitive grounds that the following ampliative arguments are cogent: (10) Only 1 in 10,000 students at the German Film Academy are non-German. Bella is a student at the German Film Academy. So, Bella is German. (11) Carl has been on time every day for the past 10 years. So, Carl will be on time today. As it will likely phenomenally seem to competent speakers on intuitive grounds that ampliative arguments like (10) and (11) are cogent, there is evidence to suggest that our intuition experiences can provide immediate justifcation for merely probable a posteriori truths. Now, consider the following simple ampliative argument (12): (12) The sky is completely dark. So, it will rain soon. (12) arguably is a reasonably strong abductive argument. Let’s say that Oscar has a perceptual experience that makes it phenomenally seem to him
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
125
that the sky is completely dark, and an intuition experience that makes it phenomenally seem to him that it is probable that if the sky is completely dark, then it will rain soon. Assuming a representational view of both perceptual experience and intuition experience, Oscar’s perceptual seeming provides prima facie immediate justifcation for believing that the sky is completely dark. Moreover, his intuition seeming provides prima facie immediate justifcation for the ampliative transition from the premise to the conclusion. For one-premise ampliative arguments, we can safely assume that prima facie immediate justifcation is closed across ampliative inference for which we have prima facie immediate justifcation. So, Oscar has immediate justifcation for believing it will rain soon. The question here arises, however, whether we can also safely assume that immediate justifcation is closed across ampliative inference in multipremise ampliative arguments. Consider the multi-premise ampliative argument below. (13) Premise 1: On this occasion, overwatering my succulent led to root rot. Premise 2: On this occasion, overwatering my succulent led to root rot. Conclusion: Overwatering succulents leads to root rot. (13) is a reasonably strong inductive argument. But one may worry that the representational phenomenal dogmatist cannot safely assume that immediate justifcation is closed across an ampliative transition in multipremise ampliative arguments. Representational phenomenal dogmatism, the objector may remind us, holds that seemings are immediate justifers. But “it seems” fails to agglomerate with conjunction. For example, it may seem to Oscar that lottery ticket 1 will win, that lottery ticket 2 will win, and that lottery ticket 3 will win. Yet it may not seem to Oscar that all three tickets will win. By way of reply, “it seems” fails to agglomerate in the envisaged case because it ascribes epistemic seemings to Oscar. As we have seen, epistemic seemings assign a probability higher than 0.5 but lower than 1 to their contents. But suppose that it epistemically seems to Oscar that ticket 1 will win, because he implicitly assigns a 0.6 probability to the proposition that ticket 1 will win, that it epistemically seems to him that ticket 2 will win, because he implicitly assigns a 0.6 probability to the proposition that ticket 2 will win, and that it epistemically seems to him that ticket 3 will win, because he implicitly assigns a 0.6 probability to the proposition ticket 3 will win. In that case, Oscar rationally ought to assign probability 0.216 to “All three tickets will win.” So, if he is rational, it will not epistemically seem to him that all three tickets will win. Unlike epistemic seemings, phenomenal seemings do agglomerate with conjunction. This is because phenomenal seemings do not modify their
126
Berit Brogaard and Dimitria Electra Gatzia
contents’ apparent certainty, which is the feature that grounds their evidence-insensitivity. For example, if it phenomenally seems to Beatrice that car 1 is red, it phenomenally seems to her that car 2 is red, and it phenomenally seems to her that car 3 is red, then it will phenomenally seem to her that all three cars are red. As phenomenal seemings agglomerate with conjunction, we can safely assume that immediate justifcation is closed across ampliative inference in multi-premise ampliative arguments. So, if you have immediate perceptual justifcation for each of the premises in (13), and you have immediate intuitive justifcation for thinking the premises make the conclusion likely, then you have immediate justifcation for thinking that the conclusion is true. Given that representational views can be extended to accommodate ampliative inferential justifcation, representational views fare better than relationalism. 8.6.
Conclusion
In this chapter, we have explored two forms of phenomenal dogmatism about perceptual experience. One is the standard view that presupposes that experience has representational character. The other is one that presupposes relationalism, which holds that perceptual experience is a perceptual relation of direct acquaintance between us and the external world. Phenomenal seemings, we have argued, play a crucial role in representational phenomenal dogmatism but not in relational phenomenal dogmatism. The main aim of this chapter has been to show that even if we grant that relationalism has an alleged epistemic advantage over representational views when it comes to perceptual justifcation, only the form of dogmatism that takes experience to be representational can be extended to account for non-deductive inferential justifcation. As an account of the latter is required to avoid succumbing to skepticism, relationalism, we have argued, does not have the claimed epistemic virtue.8 Notes 1 For thinkers sympathetic to something like dogmatism, see, for example, Pollock and Cruz (1999), Pryor (2000, 2004, 2012, 2013), Sillins (2008, 2013), Tucker (2010a), Chudnof (2012, 2013, 2014), Brogaard (2013, 2018, 2021, In Press), Ghijsen (2015), Moretti (2015), Fuqua (2017), Pace (2017). 2 Subjective probability, or credence, refers to the degree of belief, which is specifed by a real number in the [0,1] interval, where 0 indicates certainty that a proposition is false and 1 indicates certainty that it is true. 3 Not all relationalists hold that we are directly acquainted with facts; see for example, Brewer (2005). 4 For arguments that Moorean arguments are question-begging, see, for example, Wright (2002, 2007), Davies (1998), McLaughlin (2000), Dretske (2005). For
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
5 6 7 8
127
arguments that they are not, see, for example, Pryor (2004, 2012, 2013), Tucker (2010b). Some argue that all good inductive arguments are in fact instances of abductive arguments. See, for example, McCain (2016). Addressing this question here, however, would take us too far afeld. We borrow the term “intuition experience” from Chudnof (2013). Sympathizers with the view that we can be directly aware of the fact that the truth of the premises of an ampliative argument makes the truth of the conclusion probable include, for example, Keynes (1921), Bonjour (1998), Beebe (2009), Hasan (2017). We are grateful to Kevin McCain and Scott Stapleford for comments on a prior version of this chapter.
References Allen, K. (2016). A Naïve Realist Theory of Colour. Oxford: Oxford University Press. Bealer, G. (1998). Intuition and the Autonomy of Philosophy. In M. R. DePaul & W. M. Ramsey (eds.), Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry. Lanham, MD: Rowman and Littlefeld, 201–240. Beebe, J. (2009). The Abductivist Reply to Skepticism. Philosophy and Phenomenological Research, 79(3): 605–636. Block, N. (2015). The Puzzle of Perceptual Precision. In T. Metzinger & J. M. Windt (eds.), Open Mind, vol. 5. Frankfurt am Main: MIND Group, 1–52. Bonjour, L. (1998). In Defense of Pure Reason. Cambridge: Cambridge University Press. BonJour, L. (2005). In Defense of the A Priori. In M. Steup & E. Sosa (eds.), Contemporary Debates in Epistemology. Malden: Blackwell, 98–105. Brewer, B. (2005). Do Sense Experiential States Have Conceptual Content? In M. Steup & E. Sosa (eds.), Contemporary Debates in Epistemology. Malden: Blackwell, 217–230. Brewer, B. (2011). Perception and Its Objects. Oxford: Oxford University Press. Brogaard, B. (2013). Phenomenal Seemings and Sensible Dogmatism. In C. Tucker (ed.), Seemings and Justifcation. Oxford: Oxford University Press, 270–289. Brogaard, B. (2018). Seeing and Saying. Oxford: Oxford University Press. Brogaard, B. (2021). Dogmatism and Ampliative Inference. Veritas, 66(1): e42186. Brogaard, B. (In Press). The Rational Roles of Experience. In R. Rosenhagen (ed.), Reformed Empiricism and Its Prospects. New York: Synthese Library. Brogaard, B., & Gatzia, D. E. (2020). Introduction. In B. Brogaard and D. E. Gatzia (eds.), The Epistemology of Non-visual Perception. New York: Oxford University Press. Chalmers, D. J. (2004). The Representational Character of Experience. In B. Leiter (ed.), The Future for Philosophy. Oxford: Oxford University Press, 153–181. Chisholm, R. (1957). Perceiving. Ithaca: Cornell University Press. Chudnof, E. (2012). Presentational Phenomenology. In S. Miguens & G. Preyer (eds.), Consciousness and Subjectivity. Frankfurt am Main: Ontos Verlag, 51–72. Chudnof, E. (2013). Intuition. Oxford: Oxford University Press. Chudnof, E. (2014). The Rational Roles of Intuition. In A. R. Booth & D. P. Rowbottom (eds.), Intuitions. Oxford: Oxford University Press, 9–36.
128
Berit Brogaard and Dimitria Electra Gatzia
Davies, M. (1998). Externalism, Architecturalism, and Epistemic Warrant. In C. Wright, M. Smith, & C. Macdonald (eds.), Knowing Our Own Minds: Essays in Self-Knowledge. Oxford: Oxford University Press, 321–361. Douven, I. (2021). Abduction. In E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Summer 2021 ed.), URL = . Dretske, F. (2005). The Case Against Closure. In M. Steup & E. Sosa (eds.), Contemporary Debates in Epistemology. Malden, MA: Blackwell, 13–25. Evans, G. (1985). Reference and Contingency. In Collected Papers. Oxford: Oxford University Press, 178–213. Fish, W. (2009). Perception, Hallucination, and Illusion. New York: Oxford University Press. Fumerton, R. A. (1995). Metaepistemology and Skepticism. London: Rowman & Littlefeld Publishers, Inc. Fuqua, F. (2017). Dogmatism Without Mooreanism. American Philosophical Quarterly, 54(2): 195–211. Ghijsen, H. (2015). Grounding Perceptual Dogmatism: What Are Perceptual Seemings? Southern Journal of Philosophy, 53(2): 196–215. Hasan, A. (2017). Defense of Rationalism About Abductive Inference. In T. Poston & K. McCain (eds.), Best Explanations: New Essays on Inference to the Best Explanation. Oxford: Oxford University Press, 150–170. Hawthorne, J. (2002). Deeply Contingent a Priori Knowledge. Philosophy and Phenomenological Research, 65(2): 247–269. Huemer, M. (2001). Skepticism and the Veil of Perception. Lanham, MD: Rowman and Littlefeld. Huemer, M. (2005). Ethical Intuitionism. London: Palgrave Macmillan. Keynes, J. (1921). A Treatise on Probability. London: Macmillan. McCain, K. (2016). The Nature of Scientifc Knowledge: An Explanatory Approach. Berlin & Heidelberg: Springer. McDowell, J. (1982). Criteria, Defeasibility, and Knowledge. Proceedings of the British Academy, 68: 455–479. McLaughlin, B. (2000). Skepticism, Externalism, and Self-Knowledge. Aristotelian Society, 74: 93–118. Moretti, L. (2015). In Defense of Dogmatism. Philosophical Studies, 172(1): 261–282. Pace, M. (2017). Experiences, Seemings, and Perceptual Justifcation. Australasian Journal of Philosophy, 95(2): 226–241. Pollock, J. & Cruz, J. (1999). Contemporary Theories of Knowledge (2nd ed.). Lanham: Rowman & Littlefeld. Pryor, J. (2000). The Skeptic and the Dogmatist. Noûs, 34: 517–549. Pryor, J. (2004). What’s Wrong with Moore’s Argument? Philosophical Issue, 14: 349–378. Pryor, J. (2012). When Warrant Transmits. In A. Coliva (ed.), Wittgenstein, Epistemology and Mind: Themes from the Philosophy of Crispin Wright. Oxford: Oxford University Press, 269–303.
Dogmatism, Seemings, & Non-deductive Inferential Justifcation
129
Pryor, J. (2013). Problems for Credulism. In C. Tucker (ed.), Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism. Oxford: Oxford University Press, 89–132. Ramsey, W. (1992). Prototypes and Conceptual Analysis. Topoi, 11: 59–70. Siegel, S. (2012). Cognitive Penetrability and Perceptual Justifcation. Noûs, 46(2): 201–222. Sillins, N. (2008). Basic Justifcation and the Moorean Response to the Skeptic. Oxford Studies in Epistemology, 2: 108–140. Sillins, N. (2013). The Signifcance of High-Level Content. Philosophical Studies, 162(1): 13–33. Tucker, C. (2010a). Why Open-Minded People Should Endorse Dogmatism. Philosophical Perspectives, 24: 529–545. Tucker, C. (2010b). When Transmission Fails. Philosophical Review, 119: 497–529. Turri, J. (2010). On the Relationship between Propositional and Doxastic Justifcation. Philosophy and Phenomenological Research, 80(2): 312–326. Wright, C. (2002). (Anti-)Skeptics Simple and Subtle: G. E. Moore and John McDowell. Philosophy and Phenomenological Research, 65: 330–348. Wright, C. (2007). Perils of Dogmatism. In S. Nuccetelli (ed.), Themes from G. E. Moore: New Essays in Epistemology and Ethics. Oxford: Oxford University Press, 25–48.
9
Inference Without the Taking Condition Declan Smithies
9.1.
The Taking Condition
What is involved in making an inference? This chapter argues against what Paul Boghossian calls the Taking Condition: The Taking Condition: Inferring necessarily involves the thinker taking his premises to support his conclusion and drawing his conclusion because of that fact. (2014: 5) I won’t argue that the Taking Condition is incoherent—that nothing can coherently play the role that takings are supposed to play in inference. Instead, I’ll argue that it cannot plausibly explain all the inferential knowledge that we ordinarily take ourselves to have. Moreover, I’ll argue that we don’t need it to understand the nature of inference. It’s worth noting from the outset that the Taking Condition doesn’t explain the nature of inference in more basic terms. Instead, it presupposes the concept of drawing a conclusion without any further analysis. Moreover, it doesn’t explain what it takes to draw a conclusion because you take it to be supported by your premises. Even so, the Taking Condition imposes a substantial and disputable requirement on making an inference; namely, that you must take your premises to support your conclusion. What exactly does this mean? We cannot interpret this as a mere placeholder for whatever relation holds between your attitudes toward the premises and the conclusion when you make an inference. Otherwise, taking some premises to support a conclusion means nothing other than inferring the conclusion from those premises. I will ignore this defationary interpretation since it is entirely vacuous. Instead, my goal is to argue against a representational interpretation of the Taking Condition, which says that the thinker—and not just her sub-personal systems—must represent that the premises of an inference support its conclusion. DOI: 10.4324/9781003299349-12
Inference Without the Taking Condition
131
On the view I defend, you can infer a conclusion from some premises without representing that the premises support the conclusion. Perhaps you must treat your premises as if they support your conclusion, but only in the trivial sense that you infer the conclusion from the premises. What I deny is that your inference from premises to conclusion must be mediated by some personal-level representation that those premises support the conclusion. 9.2.
Regress Problems
What kind of representational state is involved in taking your premises to support a conclusion? On a doxastic construal of the Taking Condition, making an inference requires drawing a conclusion because you believe it is supported by your premises. As Boghossian argues, however, the doxastic construal generates what he calls an ingress regress. If inferential knowledge is possible, then some inferences can be justifed. And yet the belief that some premises support a conclusion cannot play any role in justifying inference unless the belief is justifed. So how is it justifed? If it can be justifed only by inference, then we face an infnite regress: you cannot be justifed in making an inference unless you’re antecedently justifed in making an infnite series of inferences beforehand. And this makes it impossible for inference to be justifed at all. To avoid the ingress regress, Boghossian (2018: 62) endorses an intuitional construal of the Taking Condition, according to which inference involves drawing a conclusion from some premises because you intuit that the premises support the conclusion. An intuition is a conscious representational state—a seeming, appearance, or presentation—that is distinct from belief and that can justify belief without standing in need of any justifcation. Because intuition—unlike belief—needs no justifcation, the ingress regress never gets started. In earlier work, Boghossian (2014: 9) argues that the intuitional view faces a regress problem of its own, which he calls the egress regress. The problem is that any representation that some premises support a conclusion—whether doxastic or not—must be combined with a representation of the premises before it can be used in inferring the conclusion from those premises. But this transition has the form of an inference: (1) P supports Q. (2) P. (3) Therefore, Q. Hence, inference seems required to bring the intuition that some premises support a conclusion to bear in the act of inferring the conclusion from
132
Declan Smithies
those premises. Once again, this generates an infnite regress in which making an inference requires having already made an infnite series of inferences beforehand. More recently, however, Boghossian suggests that we can block the egress regress by appealing to mundane examples in which representational states with conditional contents are used in rationally guiding action without any need for inference: For example: the tennis player’s intention to respond with a backhand if the ball comes to his left, and with a forehand if it comes to his right. That general intention can rationally control the tennis player’s behavior without his having to conduct personal-level inferences in order to act upon it. (2018: 63) However, Boghossian’s use of this example is questionable. When the tennis player prepares to hit a backhand, he does so because he believes that the ball is coming to his left. If he didn’t believe this, then he wouldn’t have formed the intention to hit a backhand. His unconditional belief that the ball is coming to his left combines with his conditional intention to hit a backhand when the ball comes to his left to generate the unconditional intention to hit a backhand now. And this causal transition has the form of an inference: (1) If the ball comes to my left, then I’ll hit a backhand. (2) The ball is now coming to my left. (3) So, I’ll hit a backhand now. Moreover, this inferential transition operates on personal-level representations—namely, beliefs and intentions—rather than representations in some sub-personal system. So, pace Boghossian, the tennis player needs to make a personal-level inference in acting on his conditional intention. Otherwise, we cannot explain why his action is rational in the context of the game. Of course, he cannot make the inference consciously by thinking through the premises in sequence: that would take too much time. As I argue in Section 9.3, however, not all personal-level inferences are made consciously. Inference requires only the right kind of causal dependence between personal-level representations of premises and conclusion. Perhaps a representational state is properly attributed to the person, rather than one of his sub-personal systems, only if its contents are accessible to consciousness (Smithies 2019: ch. 4). Even so, it doesn’t follow that their contents must be consciously accessed in the act of making an inference.
Inference Without the Taking Condition
133
Perhaps Boghossian will deny that the tennis player draws an inference because he violates the Taking Condition: he doesn’t take his premises to support his conclusion. And yet it remains to be seen whether this response can be motivated on independent grounds. I have not given any principled argument that it’s incoherent to posit personal-level representational states with conditional contents that can be used in non-inferentially guiding inference. Even so, we’ve not yet seen any plausible precedent for this controversial idea. If we can, we should try to make sense of inference without it. 9.3.
Over-intellectualization Problems
Another problem with the Taking Condition is that it imposes overly demanding intellectual requirements on making an inference. As a result, it struggles to accommodate the possibility of unrefective reasoning in nonhuman animals, human children, and even human adults. Adult human life involves participation in a social practice that is sometimes called “the game of giving and asking for reasons.” We challenge each other to articulate reasons for our beliefs and actions, and we evaluate whether those reasons are any good. In responding to such challenges, we give arguments in which we cite considerations that we take to support our beliefs and actions. In evaluating these arguments, we explicitly consider whether the cited premises support their conclusions. This is what Tyler Burge (1996) calls critical reasoning. As Burge notes, however, not all reasoning is critical reasoning. We often make inferences without evaluating them frst. Such inferences can extend our knowledge without any need to represent the support relation between premises and conclusion. As I’ll explain, these unrefective inferences can be fully conscious, partially conscious, or fully unconscious. First, some unrefective reasoning is fully conscious in the sense that we consciously represent the premises and conclusion of an argument, although we don’t consciously represent the support relation between them. Suppose you learn that the weather forecast predicts rain tomorrow and this prompts you to wonder whether your friend will cancel the bike ride that you planned together. Conscious thought might be needed to fgure out how your friend will react to the weather forecast. But once you decide that they will cancel if it rains, and you already know that it will rain, you don’t need to consider whether these premises support the conclusion that the ride will be cancelled. The conclusion will be evident without any further consideration at all. Second, some reasoning is partially conscious in the sense that you consciously represent some premises of an argument but not all of them. If you already know from previous experience that your friend will cancel the
134
Declan Smithies
bike ride if it rains, then no further conscious thought is needed to fgure out how they will react to the weather forecast. You can deduce that they will cancel the ride without frst needing to recall your background knowledge that they will cancel if it rains. You can draw on this background knowledge in making the inference without accessing its content in conscious thought. Indeed, there are principled reasons why you cannot bring all your background knowledge into consciousness in the act of making an inference. Attention limits the capacity of conscious thought: you cannot think consciously about too many diferent things at the same time. But the background knowledge that is relevant to making an inference can overfow these attentional limits on conscious thought. Third, some reasoning is fully unconscious in the sense that the conclusion is drawn from the premises without any conscious representation at all. You might go to bed wondering how your friend will react to the weather forecast and wake up already knowing that they will cancel. In that case, your knowledge is based on unconscious inference from background knowledge about your friend. This background knowledge may be consciously accessible in the sense that you can access it on demand when you wake up and ask yourself how you know they will cancel. But none of this background knowledge needs to be consciously accessed in the process of drawing the inference. If reasoning can be partially conscious, then it can be wholly unconscious for the same reason—your knowledge can fgure in inference without conscious access. If Boghossian denies that these are genuine inferences, then how can he explain your knowledge that the ride will be cancelled? One option is to deny that you have knowledge, rather than mere reliable belief (Boghossian 2019: 122). But why make this concession to skepticism? Boghossian sometimes invokes an internalist condition for knowledge, which requires that your reasons for belief must be in principle knowable by refection alone: On internalist ways of thinking . . . you have most reason to believe something only if it is possible for you to fgure out, by refection alone, that you have reason to believe it. And that in turn requires that, in the ideal case, all the factors that are relevant to assessing your reasoning should be open to refective view. (2016: 49) And yet a weak version of the internalist requirement is satisfed so long as your conclusion is inferred from known premises that are accessible to consciousness upon demand. You don’t need to actually refect on and consciously access your reasons for belief in the act of making an inference. This stronger version of the internalist requirement is too demanding
Inference Without the Taking Condition
135
because it excludes the possibility of unrefective knowledge altogether (Smithies 2019: 264–268). A second option is to avoid this concession to skepticism by explaining your knowledge in some other way. Thus, Boghossian suggests that perhaps your knowledge is based on intuition, rather than inference: The most natural description of the case where p suddenly strikes you as true is that you suddenly have the intuition that p is true. No doubt there is a causal explanation for why you suddenly have that intuition. And no doubt that causal explanation has something to do with your prior experiences with p. But all of that is a far cry from saying that you inferred to p from premises that you are not aware of. (2019: 122) This suggestion fails to explain how your knowledge that the ride will be cancelled depends on your knowledge about the weather forecast together with your knowledge of how your friend will react. Boghossian claims that the dependence is causal, rather than epistemic, because intuitions are caused rather than justifed by background beliefs. What this ignores, however, is that your background beliefs must constitute knowledge to serve as a causal basis for knowing that the ride will be cancelled. Hence, your knowledge that the ride will be cancelled depends epistemically, and not just causally, on your background knowledge. We cannot plausibly explain this while denying that it’s a genuine case of inference. A third option is to accommodate these examples of inferential knowledge by positing tacit representations to satisfy the Taking Condition. Boghossian (2018: 66) claims that unrefective reasoning is guided by some tacit representation of support relations that is not consciously accessed in the act of making the inference. But what are these tacit representations? They cannot be sub-personal representations, since the Taking Condition requires that support relations are represented by thinkers themselves, rather than their sub-personal systems. And they are not conscious experiences either. So, presumably, they are dispositional beliefs. I’m willing to concede that refective adults typically have such dispositional beliefs when they make inferences. After all, it is irrational to make an inference while disbelieving or suspending belief about whether the premises support the conclusion: this kind of Moorean incoherence is an instance of epistemic akrasia (Hlobil 2014; cf. Smithies 2019: ch. 9). While not strictly impossible, this extreme form of irrationality is rather unusual. We’re normally disposed to defend our beliefs in response to challenges by citing the premises from which we infer them. In other words, we tend to believe at least in a dispositional sense that the premises of our inferences support their conclusions.
136
Declan Smithies
Even so, the Taking Condition requires that our reasoning must be guided by these dispositional beliefs about the support relation. This is what happens in critical reasoning. In some cases, however, the direction of explanation is the other way around. For instance, we sometimes use suppositional reasoning as a strategy for evaluating arguments: we assume that their premises are true, and we evaluate the conclusion in light of these assumptions. In such cases, we may believe that some premises support a conclusion only because we’re disposed to infer the conclusion from the premises in the context of suppositional reasoning. Moreover, there are independent reasons to doubt that the Taking Condition can be salvaged by invoking dispositional beliefs. As we’ve seen, the doxastic construal of the Taking Condition generates an infnite regress of justifcation. To block the regress, Boghossian appeals to intuitions that can play a role in justifying inference without needing justifcation themselves. Intuitions are conscious experiences with the phenomenal character of seemings, appearances, or presentations. Dispositional beliefs have no such phenomenal character. Perhaps they are associated with dispositions to experience the phenomenal character of intuition. And yet the mere disposition to experience an intuition—unlike the experience of having one— plays no role in justifying belief. An additional problem is that you need the concept of support to have a belief or intuition that some premises support a conclusion. Plausibly, however, unrefective creatures can make inferences without any conceptual capacity to represent support relations. Consider a child who infers that her father is home when she hears the front door open. Presumably, she needs no logical or epistemic concepts—such as deductive entailment or probabilistic support—to make the inference. Boghossian (2018: 67) speculates that children may acquire a generic concept of support expressible by the word “so” around 3–4 years old as part of coming to understand the distinction between appearance and reality. And yet toddlers with no more than a few simple words can make the inference in question when they respond to the door opening by saying, “Dada!” And even nonhuman animals, such as domestic dogs, can display inferential knowledge of who is at the door through various forms of non-linguistic behavior, such as grabbing a ball to play catch (see Andrews 2020: ch. 4). Boghossian (2018: 61) seems forced to deny that these are genuine cases of inference. But it’s implausible to deny that toddlers or pets can know who is at the door. After all, it’s extremely natural to ascribe this knowledge to explain why they act as they do. And it’s not clear how else they can acquire this knowledge except by inference. To deny that they have inferential knowledge is to make an implausible concession to skepticism. So Boghossian needs strong reasons to deny that animals and children can acquire knowledge through inference.
Inference Without the Taking Condition
137
His main argument is that you can be held responsible for your reasoning. When you make an inference, your belief in the conclusion is based on the reasons provided by the premises from which you inferred it. We can evaluate the quality of the inference depending on whether it is based on good or bad reasons. Moreover, we hold people responsible for the quality of their reasoning: we regard good reasoning as praiseworthy and bad reasoning as blameworthy. Interestingly, however, we don’t hold animals or toddlers responsible for their beliefs. Does this show that they don’t engage in inference at all? No: the argument proves too much. After all, animals and children are agents who act for reasons, whether good or bad. Even so, we don’t hold them responsible for their actions by subjecting them to reactive attitudes, such as praise and blame. Arguably, responsibility requires the capacity for refection (Smithies 2019: 280–282). You cannot be held responsible for your beliefs and actions unless you’re capable of regulating them in light of refection on your reasons for belief and action. We don’t hold animals and toddlers responsible for their beliefs and actions because they lack these refective capacities. But that doesn’t show that they cannot believe and act for good or bad reasons at all. In particular, it doesn’t show that they cannot make inferences by believing things for reasons provided by their other beliefs. I don’t claim to know where inference frst makes its appearance on the phylogenetic or ontogenetic chain. That is an empirical issue that goes beyond the scope of this chapter. Even so, it seems overwhelmingly likely that this occurs before the capacity to represent support relations. Presumably, you cannot have the concept of support without the capacity to think thoughts of the form, “p supports q.” But attention imposes limits on the capacity of thought: there are only so many thoughts we can entertain at once. For some creatures, these attentional limits may be severe enough that they can only think one atomic thought at a time. I see no principled reason to deny that such creatures can move inferentially from one thought to another without representing support relations between them. In any case, the burden is on proponents of the Taking Condition to explain why this is impossible. 9.4. Generalization Problems As we’ve just seen, our actions are based on reasons in much the same way as our beliefs. Just as we believe things for reasons—whether good or bad—so too we act and react for reasons. In other words, the basing relation extends beyond the epistemic domain of belief and knowledge into the practical domain of action and reaction. Inference is just one species of the basing relation. When you make an inference, your belief in the conclusion is based on the reason provided by
138
Declan Smithies
the premises from which you infer it. But not all beliefs are inferentially based on other premises that you believe. Otherwise, we face an infnite regress in which every justifed belief is inferentially based on reasons provided by some other justifed belief and so on without end. Where does the regress come to an end? The usual foundationalist answer is that justifed beliefs can be non-inferentially based on reasons provided by perception, introspection, or intuition. On this view, beliefs can be based on reasons either with or without inference. What this means is that an account of inference must cohere with a more general account of the basing relation. After all, inference is just one species of the genus. So, if we endorse the Taking Condition, then we face a choice: either we must extend it to other species of basing or we must give special reasons for restricting it to inference. Unfortunately, however, neither option seems attractive. We cannot plausibly extend the Taking Condition to all instances of the basing relation. We sometimes act, react, and believe things for reasons without representing them as reasons that support our response. Here are some examples: • • •
You can choose something (e.g., chocolate ice-cream) for the reason that you like it without representing that the fact that you like it is a reason to choose it. You can feel sad for the reason that you’ve lost something you value (e.g., a toy) without representing that the loss is a reason to feel sad. You can believe something (e.g., that the sky is blue) for the reason that it looks that way without representing that the visual appearance is a reason to believe it.
Young children—and non-human animals too—can make choices, feel emotions, and believe things for reasons without representing their reasons as reasons for responding as they do. Even mature adults often fnd themselves in this predicament. Although we have the conceptual capacities required for articulating our reasons for belief and action in general, we’re not always capable of exercising these capacities accurately on any given occasion. Sometimes, we act for one reason while believing we act for another. As we know from social psychology, examples of selfdeception, confabulation, and post-hoc rationalization are all too common (Wilson 2004). Boghossian (2019: 122–123) resists extending the Taking Condition to all instances of the basing relation. He allows that you can acquire perceptual knowledge based on perceptual appearance without taking the appearance to support the belief. But why should the requirements for inferential basing diverge from the requirements for non-inferential basing? According to Boghossian, the Taking Condition doesn’t apply when
Inference Without the Taking Condition
139
there is a match in content between the justifying state and the belief that it justifes. He continues: The inference from p to q is not like that. A belief that p, which is the input into the inferential process, is not a seeming that q. And while the transition to believing that q may be familiar and well-supported, it is not simply like acquiescing in something that is already the proto-belief that q. (2019: 123) While there is indeed a structural diference between inferential and noninferential justifcation, I don’t see how this motivates any restriction on the Taking Condition. The Taking Condition constrains inference whether justifed or unjustifed: to make an inference, you must take your premises to support believing your conclusion. So why doesn’t the non-inferential transition from perception to belief require taking your perception to support your belief? This question isn’t settled by noting the structural diference between inferential and non-inferential justifcation. We need some other reason to restrict the Taking Condition. 9.5.
Reasoning as Mental Action
A more promising answer is that the Taking Condition applies only to instances of the basing condition that involve agency. Agency is not typically involved in forming beliefs that endorse the contents of perception: this is something we do automatically and involuntarily. In contrast, Boghossian maintains that reasoning is a form of mental action: Reasoning is something we do, not just something that happens to us. And it is something we do, not just something that is done by subpersonal bits of us. And it is something that we do with an aim—that of fguring out what follows or is supported by other things one believes. It’s hard to see how to respect these features of reasoning without something like the Taking Condition. (2014: 5) Moreover, the Taking Condition is supposed to explain why reasoning is a form of mental action. The general idea is that all actions are based on some desire, goal, or aim to achieve an end together with some belief-like representation that the action is a means to this end. Here is Boghossian’s succinct summary of this idea: The agent has an aim; she has a view about a way of accomplishing that aim; and she performs an action as a result of that combination. (2018: 61–62)
140 Declan Smithies Reasoning fts this mold when it is based on the aim of fguring out what follows from some premises together with a representation that inferring a given conclusion will achieve the aim because it follows from those premises. In that case, the Taking Condition is satisfed. Jonathan Way and Daniel Whiting (2016: 325) complain that Boghossian’s explanation is viciously regressive. According to Boghossian, every action is based on the aim of achieving some end together with a representation that the action is a means to this end. But this seems tantamount to saying that every action is based on practical reasoning of the following form: (1) I shall do A. (2) Doing B is a means to doing A. (3) So, I shall do B. On Boghossian’s view, however, all reasoning—including practical reasoning—is mental action. This yields the problematic result that every action requires some prior mental action, and all reasoning requires some prior reasoning. Something has clearly gone awry. To avoid the regress, Boghossian must deny that all action is based on practical reasoning. This seems like a sensible thing to deny. Presumably, a toddler can choose chocolate ice-cream for the reason that they prefer it without reasoning from the premise that they prefer it together with the further premise that choosing it will satisfy their preference. If he denies this, however, then he needs to explain how an action can be non-inferentially based on an aim, goal, or end. We already encountered this challenge in Section 9.2. Even if this challenge can be met, however, another regress problem looms nearby. According to Boghossian, all action is based—whether inferentially or otherwise—on some end together with some representation of the means to that end. Now the question arises whether all forms of basing involve agency. If so, then we face an infnite regress: every action is based on some prior act of basing. Hence, Boghossian must deny that all basing involves agency. But if agency is not required for basing in general, then why should it be required for inferential basing in particular? Ultimately, I’m not persuaded that Boghossian provides any compelling answer to this question. His opening gambit is that reasoning is something that we can be said to do, rather than something that happens to us. As a linguistic point, this seems entirely correct. At the same time, however, it cannot bear much theoretical weight. We often describe events that are not actions in the active voice as things we do, rather than things that merely happen to us. For instance, we describe people as falling asleep, sweating, and vomiting,
Inference Without the Taking Condition
141
as well as perceiving things and believing that things are thus-and-so. Although these are all things we can be said to do, the Taking Condition does not apply to them. So, we cannot rely on this linguistic point in making the case that inference is distinguished from non-inferential basing by its agential character. A more promising consideration is that reasoning involves the experience of agency. When you concentrate on solving a puzzle, for example, your experience has an agential dimension: it feels as if you’re actively trying to fgure out the solution. As we saw in Section 9.3, however, not all reasoning is like this. Much of the time, we draw inferences efortlessly without any need to actively focus our attention. In such cases, there is no salient experience of agency. Our reasoning is often guided by background beliefs that are not manifest in consciousness at all. And we sometimes update our beliefs through reasoning that is wholly unconscious—as when we wake up from a deep sleep to fnd that we have already fgured out the solution to a problem. Reasoning need not involve any experience at all, let alone an experience of agency. Reasoning is one of many psychological phenomena that can be experienced either actively or passively. One such example is imagination: you can actively try to visualize a scene, or you can simply fnd yourself experiencing visual imagery in a dream or a daydream. Another is attention: you can focus your attention on following a moving target, or your attention can be captured when a stationary target suddenly moves. A third example is judgment: you can reach a verdict by actively assessing the evidence for and against a hypothesis, as in jury deliberations, but you can also experience a sudden realization as if out of the blue without any experience of agency. If all these other psychological processes can be experienced both actively and passively, then why not reasoning too? I contend that reasoning is no diferent from perceptual belief-formation in this respect. We typically acquire perceptual knowledge automatically with no need to actively focus our attention on the question of what to believe. In some cases, however, we actively deliberate about whether to believe that things are how they perceptually appear. Similarly, our reasoning can be actively guided by focusing attention on whether one thing is a reason to believe another. But much of our reasoning occurs automatically without any need for active control. Boghossian’s main argument is that we cannot make sense of our practice of holding people responsible for their reasoning unless we suppose that reasoning is a mental action: For it to make sense to hold you responsible for your inferences, inferring has to be something you do, and not just something that happens to you. It has to be a mental action of yours, something you have control
142 Declan Smithies over, and which you could have done diferently, had you thought it desirable to do so. (2018: 60) As Mark Richard (2019) notes, however, we hold people responsible not only for actions over which they have direct control but also for the consequences of their actions over which they have only indirect control. We can blame someone for ill health that results from bad decisions, for example, and we can praise them when they turn things around. So, we don’t need to assume that reasoning is a mental action to explain why we hold people responsible. We just need to assume that people have indirect control over their reasoning by performing other mental actions, such as considering questions, paying attention, or gathering evidence. This objection doesn’t strike at the heart of Boghossian’s argument. His main point is that we hold people responsible for the reasons on which their inferences are based. Although we hold people responsible for their health, we don’t hold them responsible in the same way, since the state of your health is not based on reasons at all. A more charitable version of Boghossian’s argument is that we need to assume that inference is a mental action to explain the distinctive way in which we hold people responsible for basing their inferences on good reasons. The real problem with Boghossian’s argument is that it proves too much. After all, we hold each other responsible for basing our beliefs on good reasons even when they are not based on inference. For instance, we credit people for basing their beliefs on perceptual appearances and we blame them when they fail to do so without good reason. If his argument succeeds at all, then it shows that all basing—whether inferential or non-inferential—is agential. As we’ve seen, however, this claim generates a vicious regress. I conclude that Boghossian cannot solve the generalization problem by appealing to the idea that inference is a mental action. 9.6.
Inference and Association
One of Boghossian’s main arguments for the Taking Condition is that we need it to explain the distinction between inference and association. Suppose you believe one thing because you believe another, although you don’t take the one thing to support the other. What distinguishes this transition from a mere process of association that is sensitive to the contents of your beliefs? Here is Boghossian’s example: A habitual depressive’s judging “I am having so much fun” may routinely cause and explain his judging “Yet there is so much sufering in
Inference Without the Taking Condition
143
the world,” as directly as you please, without this being a case in which he is inferring the latter thought from the earlier one. (2014: 4) The depressive doesn’t make an inference because his judgment that the world is full of sufering isn’t based on his judgment that he is having fun. Rather, one judgment causes the other by a process of associative thinking. But then what more is needed for inference beyond the causation of one judgment by another? The premises of an inference need not support its conclusion, since inference is not always justifed. Instead, Boghossian claims, the thinker must take their premises to support their conclusion. The depressive doesn’t draw an inference because he doesn’t take his judgment that he is having fun to support his judgment that the world is full of sufering. In this way, Boghossian uses the Taking Condition to explain the distinction between inference and association. Boghossian’s example is somewhat under-described. On the most natural interpretation, the depressive already believes that the world is full of sufering, since his depression darkens his outlook on the world. The realization that he is having fun today merely serves as a trigger that activates this standing belief in conscious judgment. The same judgment might have been activated in some other way, such as considering whether the world is full of sufering. And this is enough to explain why the transition doesn’t count as an inference. The one judgment merely activates the other, rather than serving as its epistemic basis. Hence, we can explain why the transition doesn’t count as an inference without endorsing the Taking Condition. Of course, Boghossian might deny that this version of the case is the one he has in mind. He could stipulate that the realization that you’re having fun doesn’t merely trigger the activation of a pre-existing belief, but rather causes the formation of a new belief. This version of the case is not so realistic—it has no obvious connection with the psychology of depression—but it is perfectly coherent all the same. The problem is that the example is not fleshed out in enough detail to make it clear why this should count as association, rather than inference. To be sure, the inference in question would be bizarrely irrational. And yet Boghossian imposes no limits on how irrational one’s inferences can be. Rather than assuming from the outset that there are rationality constraints on inference, he requires that any such constraints should emerge as consequences of a correct theory of inference (Boghossian 2014: 4). I’m not disputing that we can construct examples in which the causal transition from one belief to another is the wrong kind to constitute
144
Declan Smithies
inference. Indeed, examples of this kind are familiar from the literature on deviant causal chains. Here is one from John Turri: Through some random quirk—the result of a neural assembly malfunctioning—Wilt’s belief that the lettuce has wilted is the proximate mental cause of his belief that the Patriots will win twelve games this season. But it certainly seems false that Wilt’s belief that the lettuce has wilted is his reason for believing that the Patriots will win twelve games this season. (2011: 389) Such examples show that inference is not just believing one thing because you believe another. To count as an inference, one belief must cause another in the right kind of way. The problem is that it’s hard to specify what counts as right kind of causal relation to constitute inference. This is the familiar problem of deviant causal chains. The problem of distinguishing between inference and association is just one aspect of this more general problem. Boghossian takes his challenge to be distinctive because it arises when the causal relation between beliefs is direct, rather than mediated by an intervening causal chain of events. As Turri’s example shows, however, mental causation can be deviant even when it is direct. As Boghossian (2014: 5, n. 2) admits, the Taking Condition cannot solve the problem of deviant causal chains. It says that making an inference requires drawing a conclusion from some premises because you take the premises to support your conclusion, but it doesn’t specify which kind of causal relation is required for inference. Nothing precludes the representation of support relations from fguring in deviant causal chains. Suppose your belief in some conclusion is caused in some deviant way by your taking your premises to support the conclusion. This doesn’t count as an inference because your belief in the conclusion isn’t caused in the right way. And yet the Taking Condition doesn’t specify which kind of causal relation is required. For much the same reason, the Taking Condition cannot explain the distinction between inference and association. If beliefs can fgure in processes of content-sensitive association, then so can the representational state of taking some premises to support a conclusion. Suppose our depressive thinker mistakenly takes the premise that he is having fun to support the conclusion that the world is full of sufering. And suppose this causes him by a process of association to form some arbitrary belief—say, that he is the King of Spain. This is surely not an inference. If so, then the same is true when the same inputs cause him to believe by a process of association that the world is full of sufering. In this example, the Taking Condition is satisfed, since the thinker believes that Q because he believes that P and he
Inference Without the Taking Condition
145
takes P to support Q. And yet this is a case of association, rather than inference. Hence, the Taking Condition cannot explain the distinction between inference and association after all. To explain the distinction in full generality, we need to specify which kind of causal relation is necessary and sufcient for inference. In other words, we need a solution to the problem of deviant causal chains. Unfortunately, however, there is no consensus about how to solve this problem or even whether it can be solved at all. Perhaps we must take the concept of inference as primitive rather than try to analyze it in more basic terms. As we’ve seen, Boghossian’s Taking Condition presupposes the concept of drawing an inference because you take the premises to support the conclusion. He ofers no analysis of what counts as the right kind of causal relation for inference. But if he can take this as primitive, then his opponents are entitled to do the same. The disputed issue is whether the relata of this causal relation must include representation of not only premises and conclusion but also the support relation between them. We’ve not yet seen any compelling motivation for this claim. 9.7. Conclusions This chapter makes no attempt to explain what it is to make an inference in more basic terms. Indeed, I doubt that any reductive explanation of inference is possible. In any case, Boghossian’s Taking Condition does no such thing: it merely states a requirement for making an inference. I’ve argued that this requirement is too demanding because it cannot explain all of the knowledge that we ordinarily take ourselves to have. Moreover, I’ve explained why Boghossian’s main arguments for the Taking Condition are not persuasive. I conclude that the Taking Condition is false: you can make an inference without taking your premises to support your conclusion. References Andrews, K. 2020. The Animal Mind: An Introduction to the Philosophy of Animal Cognition. Routledge. Boghossian, P. 2014. What Is Inference? Philosophical Studies 169: 1–18. Boghossian, P. 2016. Reasoning and Refection: A Reply to Kornblith. Analysis 76: 41–54. Boghossian, P. 2018. Delimiting the Boundaries of Inference. Philosophical Issues 28: 55–69. Boghossian, P. 2019. Inference, Agency, and Responsibility. In Reasoning: New Essays on Theoretical and Practical Thinking, edited by M. Balcerak Jackson & B. Balcerak Jackson, 101–124. Oxford University Press. Burge, T. 1996. Our Entitlement to Self-Knowledge. Proceedings of the Aristotelian Society 96: 91–116.
146
Declan Smithies
Hlobil, U. 2014. Against Boghossian, Wright, and Broome on Inference. Philosophical Studies 167: 419–429. Richard, M. 2019. Is Reasoning a Form of Agency? In Reasoning: New Essays on Theoretical and Practical Thinking, edited by M. Balcerak Jackson & B. Balcerak Jackson, 91–100. Oxford University Press. Smithies, D. 2019. The Epistemic Role of Consciousness. Oxford University Press. Turri, J. 2011. Believing for a Reason. Erkenntnis 74: 383–397. Way, J. & Whiting, D. 2016. Against the Taking Condition. Philosophical Issues 26: 314–331. Wilson, T. 2004. Strangers to Ourselves: Discovering the Adaptive Unconscious. Harvard University Press.
10 Zetetic Seemings and Their Role in Inquiry Verena Wagner
10.1.
Introduction
Seemings are mental states or events with propositional content that have a specifc phenomenology often referred to as “felt truth.” In epistemology, seemings are mainly discussed as a possible justifcation for belief. Prominent ways of making seemings relevant for non-inferential justifcation are Jim Pryor’s (2000) Dogmatism and Michael Huemer’s (2007) principle of Phenomenal Conservatism (PC), both of which state that seemings provide at least some justifcation for believing the respective seeming’s propositional content. William Tolhurst (1998), Chris Tucker (2010), and others have defended similar accounts. [S]eemings provide psychological and epistemic support for belief. (Tolhurst 1998, 295) [W]hen it perceptually seems to you as if p is the case, you have a kind of justifcation for believing p that does not presuppose or rest on your justifcation for anything else. (Pryor 2000, 519) If it seems to S that p, then, in the absence of defeaters, S thereby has at least some degree of justifcation for believing that p. (Huemer 2007, 30) Necessarily, if it seems to S that P, then S thereby has prima facie (noninferential) justifcation for P. (Tucker 2010, 529) Various worries and objections have been raised against the view that seemings are mental states that can provide epistemic justifcation for beliefs. For the purposes of this chapter, I will focus on two objections here. The DOI: 10.4324/9781003299349-13
148
Verena Wagner
frst is about the nature or reality of seemings as distinct mental states, and about a peculiarity in how they manifest: seemings are hard to identify via introspection in cases of justifed belief but surprisingly easy to be spotted in cases of known illusions, fallacies, and false memories. The second problem concerns the justifcatory force of seemings in cases of benign cognitive penetration, as in expert cognition. If a seeming that p can be caused by a justifed belief in q, and q supports believing the target proposition p, then believing p would be justifed both by the belief in q and additionally by the seeming that p. Thus, the justifcation provided by the q-belief would be counted twice: once directly, and once via the seeming it has caused. This is called the Double Counting Objection (Tucker 2013a,13n; Tooley 2013, 322n; Huemer 2013, 339n). In this chapter, I want to address the nature and justifcatory force of seemings (as well as the two problems described above) from a diferent perspective, namely, that of inquiry. I want to show that even if seemings cannot provide epistemic justifcation for doxastic attitudes, they do not have to be analyzed as mere epiphenomena or even as non-existent. Seemings are real experiences, and they are important for belief formation in the sense that they play a crucial role in the sort of rational inquiry that typically leads to the formation of doxastic attitudes. In this contribution, I want to take seemings seriously as mental events that are experienced by subjects when asking questions and assessing possible answers. For this purpose, I will introduce the position of the Zetetic-Seemings Realist. Seemings, so this position suggests, provide reasons for engaging in certain zetetic, that is to say, inquiry-related tasks, and thus play a governing role with respect to the overall aim of answering the question that is currently on one’s research agenda. The realist position about zetetic seemings acknowledges that seemings are experiences of felt truth (or at least felt support for the truth) of a proposition. As such, zetetic seemings provide automatic feedback for inquiring subjects regarding their own evidential standing. This feedback is brought to consciousness during deliberation if there is a “zetetic push,” a push to fnd the correct answer for a question that is mentally entertained. Since a seeming that p might be caused by undue optimism or wishful thinking that does not provide evidential support for p, its mere occurrence cannot provide epistemic justifcation for making up one’s mind about the correct answer to the question at issue; however, experiences of felt support for the truth (or falsity) of a possible answer provide (in the absence of zetetic defeaters) justifcation for pursuing the seeming’s content in inquiry, for example, as a to-be-checked working hypothesis, as a new sub-question, or as an answer that can be provisionally ruled out. The feedback that seemings provide for inquiring subjects regarding their own evidential standing is helpful for achieving a positive interpretation of
Zetetic Seemings and Their Role in Inquiry
149
Jane Friedman’s Zetetic Instrumental Principle. By itself, ZIP does not give any guidance about when it is rational to engage in certain zetetic acts rather than others, for example, when to gather new evidence, when to retrieve and re-evaluate stored pieces of information, or which possible answer to pursue frst. Moreover, I will show that seemings, understood as psychological and justifcatory support for zetetic activities, can deal with both problems mentioned above. Their zetetic role can explain the peculiar occurrence pattern of seemings (via the absence or presence of a zetetic push), and it avoids the Double Counting Problem because cognitively penetrated seemings (of the benign sort) can justify zetetic activities without justifying beliefs. In Section 10.2, I will address the two problems concerning the nature (the peculiar occurrence pattern) and justifcatory force (Double Counting Problem) of seemings. In Section 10.3, I will introduce the zetetic turn and suggest that zetetic seemings are relevant for motivating and justifying zetetic activities and decisions. In Section 10.4, I will spell out the position of ZeteticSeemings Realism, which suggests a view on the nature of zetetic seemings that is a fusion of the Experience View and the rival Taking Evidence View. Further, the realist position regarding zetetic seemings will explain the justifcatory force of seemings and their role in justifying zetetic acts. 10.2. Epistemic Seemings: Two Problems 10.2.1.
A Peculiar Occurrence Pattern
Seeming-realists emphasize the phenomenal character of seemings, which, they say, sets them apart from other mental entities: when it seems to S that p is true, S experiences p as revealing or recommending or representing its truth. Thus, if seemings exist as distinct mental events or attitudes with this special phenomenal character, it should be easy to point at occurrences of seeming truth. Chris Tucker formulates a typical objection from the point of view of someone who fails to fnd any seemings when introspecting (he ascribes this view to Timothy Williamson 2007, 217): I understand what a belief is. I understand what an inclination to believe is. When I introspect, I can fnd beliefs and inclinations to believe. But I don’t have a grip of some sui generis propositional attitude thing you call a “seeming,” and I can’t fnd it when I introspect. (Tucker 2013a, 5) Admittedly it is not easy to “fnd” seemings via introspection, particularly when we look for them as entities involved in cases of justifed belief. This is odd because that is where we would expect them to be if seemings have
150
Verena Wagner
the role of justifers for beliefs that dogmatists and phenomenal conservatives ascribe to them. If S believes that there is a cup of tea in front of her because she visually perceives a cup of tea in front of her, it is not introspectively obvious that there is a third mental state involved—a seeming that there is a cup of tea in front of her—in addition to the perception and the belief.1 Yet, contra (Tucker’s) Williamson, seemings are surprisingly easy to fnd via introspection in cases of known illusions where the corresponding beliefs would not be justifed: a half-submerged stick in a water glass seems bent to us even if we know that it is not bent. This efect is not restricted to perceptual seemings alone but can be found for memorial and for intellectual seemings as well. For example, S may unreservedly believe the conclusion of an argument because S believes in its validity and takes all its premises to be true. Nonetheless, S may lack any experience of felt truth with respect to the derived conclusion she has come to believe. The conclusion just does not reveal, recommend, or represent itself as true to S. On the other hand, there are established solutions to theoretical problems that remain counterintuitive even when fully believed. Take the Monty Hall Problem,2 which makes it seem that the odds are equally distributed between the two closed doors. Even if one knows that one is statistically better of switching to the door that one initially did not choose, it still strongly seems to one that it is not so. Similarly, S may recall her colleague’s room number without experiencing any “revelation of truth” about the recalled information. It’s just a fact she remembers, but it does not specifcally seem true to her (nor does it seem false). Yet, false memories may persist even in the presence of counterevidence and countervailing beliefs. Suppose that S has been caught in the tube with an invalid ticket that shows no sign of a stamp on it. Let’s further suppose S is shown the video tape that displays her walking by the stamping machine without using it. On this basis, S forms the belief that she didn’t validate her ticket. Yet, due to her false memory it still seems true to S that she has stamped her ticket. These cases show that we can fnd seemings, contra (Tucker’s) Williamson, but the context in which we fnd them is surely surprising. Michael Tooley (2013) does not reject the phenomenon of “felt truth” but he is unconvinced that such an occurrence of what he calls cognitive qualia presupposes the existence of a state or event other than occurrent belief. He thinks that there is “no ground for postulating the existence of a fundamentally diferent type of mental state—a seeming” (p. 313). He takes it that the phenomenon of felt truth or assent can be fully explained if we replace “the everyday concept of belief by the broader notion of degrees of belief or degrees of assent” (p. 313). So how would degrees of assent (or credences) help to explain the peculiar cases of felt assent like in the bentstick scenario? In those cases, Tooley could say, S assigns a high credence
Zetetic Seemings and Their Role in Inquiry
151
to the proposition that is backed by the evidence (the stick is not bent, it’s statistically better to switch doors, I did not stamp my ticket), but at the same time assigns a non-zero credence to its negation. The countervailing seemings might then be explained as cognitive qualia being produced by the latter credence. If this works, there is no distinct type of mental state required to account for the seemings. This suggestion fails to explain the data for two reasons. First, it is psychologically plausible and perfectly rational that one assigns a credence of 0 to the proposition that the stick is bent (p) and yet have the seeming that it is bent. A credence of 0 certainly cannot account for the cognitive qualia. Second, even if we accept that S is rational in assigning a high credence to p and a non-zero credence to not-p, the low credence involved here would not be able to account for the strength of the countervailing seeming. It’s not the case that S has merely a subtle seeming that the stick is bent; rather, it strongly seems to her that the stick is bent. Likewise, it strongly seems to S that it doesn’t make a diference which of the two doors she chooses, and it strongly seems to S that she validated her ticket before using the tube. If credences come with a feeling of assent, the degree of the credence assigned somehow needs to be refected in the degree of felt assent. Thus, low credences cannot explain why people experience a strong “revelation of truth” in these cases despite their countervailing beliefs. It is a peculiar feature of seeming-experiences that they are not reliably found via introspection in cases in which subjects have justifed beliefs but can be spotted easily in cases of countervailing beliefs. A theory of seemings should be able to explain this datum, especially if it is a theory that distinguishes between seeming states and other states (here perceptions, memories, and intuitions), as in Tucker (2013a). 10.2.2.
The Double Counting Objection
Another worry concerning views like PC accepts the existence of seemings as distinct mental states (at least for the sake of argument) but doubts the legitimacy of the epistemic justifcation that such states can allegedly provide in addition to other justifers. It is plausible that seeming-states may be caused by other mental states. As Tucker (2013a) points out, this is not always a bad thing. Trained ornithologists, for example, have seemings that are partly caused by their knowledge about birds. This is a benign form of cognitive penetration. Yet, this form of benign cognitive penetration creates a problem for the alleged justifying role of seemings. When it seems to the trained ornithologist that the bird over there is a hawk, and her seeming is caused by her expert knowledge about hawks, there are two justifers at work that are interdependent. One is directly provided by her justifed beliefs about hawks, the other one is provided by her seeming
152 Verena Wagner that the bird over there is a hawk, which causally depends on her expert knowledge. Thus, if seemings provide additional epistemic justifcation for belief no matter their causal history, the expert’s knowledge about birds is somehow counted twice. Tucker (2013a, 13) calls this an “illegitimate boost” of the overall justifcation for the respective belief. It is plausible to think that this problem not only pertains to seemings in the context of professional expertise but also is prevalent in many situations. Most people are well trained in identifying certain objects in their daily afairs, and it is plausible to think that their seemings are cognitively penetrated by their background knowledge and experiences as well. Thus, it must be asked whether there are any seemings that are not somehow cognitively penetrated (in the benign sense) and, if there are, how to tell them apart from those that are. How can we distinguish seemings that provide a legitimate boost from those that do not? Huemer (2013) and Tucker (2013a) suggest we must consider what the respective subject knows or believes about the (in)dependence of the two justifcations. According to Tucker, if S reasonably believes that the two are fully independent, then the provided boost is legitimate, but if S believes them to be fully dependent it is not. Tucker correctly points out that the problematic cases are those in between, where subjects do not have any evidence about the (in)dependence of the two justifcations. He may also be right that this problem is a general problem for any account of justifcation. However, the proposed solution that subjects must consider the interdependency of their seemings and other justifers as their potential causes does signifcant damage to the purported simplicity and intuitiveness of the project pursued by Dogmatism and Phenomenal Conservatism. Also, this move is rather surprising if one considers the objection usually levelled against higher-order accounts of seemings, which require refection on one’s own mental states, as “over-intellectualized, or over-introspectivized” (Huemer 2013, 334). As I will show later (Section 10.4.1), the subject’s consideration of the dependence relation between one’s seemings and their causes comes naturally for a zetetic interpretation of seemings. 10.3.
The Zetetic Turn
Epistemology has recently taken a zetetic turn, that is, a turn toward the study of inquiry (Friedman 2017, 2019, 2020; Stafel 2019; Thorstad 2021). I will follow Jane Friedman in taking inquiry to be a state of mind that is best characterized as mentally asking a question or as “having a question on one’s research agenda” (Friedman 2017, 308). A question can be characterized by the set of its complete, possible answers. Each complete, possible answer in the set closes the target question, but only one of them is true: this is the correct answer.
Zetetic Seemings and Their Role in Inquiry
153
The inquisitive mindset of asking a question is to be distinguished from the activity typically involved when a subject has the goal of resolving the question on her research agenda. To use a common example, a detective who investigates a crime is in the inquisitive state of mentally asking the question “Who committed the crime?” and performs certain investigative acts like questioning witnesses and searching the crime scene with the overall aim of answering the target question. Friedman stresses that one can be in the state of inquiry independently of performing any inquiry-related acts, and vice versa. In support of the latter, Friedman introduces Detective Morse, who knows or believes that he himself committed the crime he has been called in to solve. He performs the necessary acts for the purpose of covering up his involvement, but without being in the relevant inquisitive mindset. Thus, the state of inquiry must not be confused with the actions that are typically performed to make progress in resolving the question on one’s research agenda. Julia Stafel (2019) suggests an additional distinction between “inquiry” and “deliberation”: inquiry is the inquisitive state of mind that Friedman describes, while deliberation is a mental activity or process of evidence assessment and evaluation. I will introduce the notion of zetetic acts, which includes all kinds of activities that are pursued by the inquiring subject for the purpose of fguring out the answer to the question on her research agenda. Zetetic acts include the deliberative tasks of evidence assessment and evaluation that Stafel suggested, but it is broader insofar as it further includes all activities (mental or otherwise) that are performed with the aim of making progress with respect to the target question. The notion of zetetic acts comprises activities as diferent as the mental act of opening a relevant sub-question (Who has an alibi? Who has a motive?) and the physical act of picking up objects at the crime scene. These acts, as diferent as they may be, are both zetetic in my sense if they are done for the purpose of resolving the question on the subject’s research agenda (Who committed the crime?). Morse, who committed the crime himself, may act as if he inquires, but the actions he performs are not zetetic because they are not done for the aim of fguring out who committed the crime. Similarly, going for lunch in between questioning witnesses does not count as a zetetic act of the inquiring detective if she doesn’t do it with the aim of resolving the question. Subjects typically have various options for how to proceed and structure their zetetic activities for resolving the target question. Some of these options will be more sensible than others, given that the time, cognitive capacities, and resources of zetetic subjects are limited. On these grounds, it is possible to assess subjects as more or less rational when engaging in certain zetetic tasks. Thus, it is plausible that zetetic decisions and acts are assessable by zetetic norms that difer from epistemic norms. Friedman
154
Verena Wagner
suggests that inquiry is governed by the Zetetic Instrumental Principle (ZIP), which is an instance of the general norm of instrumental rationality applied to the end of inquiry: ZIP If one wants to fgure out Q?, then one ought to take the necessary means to fguring out Q?. (Friedman 2020, 503) Compliance with ZIP can be approached from a negative and a positive side. On the one hand, rational subjects should avoid certain things in order to remain focused on the to-be-resolved question Q?, and on the other hand, they should do certain things to reach their end of resolving Q? efciently. Friedman is concerned with what subjects should avoid in order to comply with ZIP, because her aim is to point to a tension between zetetic and standard epistemic norms. To stay focused on resolving Q? within a given time frame, one must avoid all kinds of distractions, also those that come in the form of Q?-irrelevant evidence one already possesses and Q?irrelevant information one could easily acquire. Friedman argues that complying with ZIP means that “we should sometimes not follow our evidence, and we should sometimes not come to know things we are in a position to know” (2020, 503). She concludes that compliance with ZIP by avoiding Q?-irrelevant evidence is in tension with traditional epistemic norms, which generally permit the formation of beliefs based on whatever evidence one possesses or can easily obtain. In this chapter, I will not pursue the alleged tension between epistemic and zetetic norms that Friedman suggests. Thus, I will not be concerned with how epistemic and zetetic norms interact or how potential tensions can be resolved.3 Rather, I will be concerned with a positive approach to pursuing one’s inquiry. It is one thing to remain focused on the tobe-resolved question Q? by avoiding various distractions, as described by Friedman, but there are still many ways to actively pursue the aim of fnding the correct answer to Q?. Being in the inquisitive mindset does not suggest any route by itself. Subjects who comply with ZIP by avoiding various distractions can still do better or worse zetetically: how to start, which possible answer to pursue frst, which information to assess as potential evidence, what information to double-check, when to pause and think again, etc. Seemings, so I will argue, can be crucial for answering some of these questions. For this purpose, I will introduce the notion of zetetic seemings and the fgure of a realist about zetetic seemings (the “zeemings realist”), who accepts that zetetic norms are distinct from epistemic norms but who focuses on the positive side of advancing inquiry. In the next section, I will introduce the realist position about the nature and justifcatory force of seemings.
Zetetic Seemings and Their Role in Inquiry 10.4. 10.4.1.
155
Zetetic-Seemings Realism The Nature of Zetetic Seemings
The realist position about the nature of zetetic seemings suggested here is a modifed version of the so-called Taking Evidence View (TEV) about epistemic seemings which is combined with the most striking feature of the rival Experience View about seemings. TEV is attributed to Earl Conee (2013) and Michael Tooley (2013). Both analyze epistemic seemings as involving higher-order mental states of taking oneself to have evidence for some proposition p. Their views difer in various respects, but they both take a seeming that p as involving the combination of two psychological facts: (1) a mental state the subject is aware of, for example, believing or recalling that q, and (2) an inclination to believe (or a belief) that the state in (1) evidentially supports the target proposition p. Given these two facts, the subject is under the impression that she has evidence in support of the truth of p. The here suggested modifed view adopts the idea of taking oneself to have evidence in support of p but is diferent from TEV in at least two respects. First, according to the zeemings realist, the subject does not have to be aware of the mental state in (1). Second, the subject does not have to believe that the mental state is evidentially relevant for p but may rather merely experience herself as having evidence. The connection between the mental state that is potentially relevant for supporting p and the higherorder state that tracks this relevance is of a causal nature and gives rise to a specifc experience of felt support. In accordance with TEV, we may well imagine a case in which S’s impression of having evidence for p is caused by S’s recalling that q, in combination with S’s believing that the truth of q is evidentially relevant for the truth of p. Yet, contra TEV, S may also be under this impression even if the described causal connection is not fully transparent to S and even if S cannot access the potentially justifying mental state. One may be caused by some (not yet occurrent) mental state (e.g., a stored belief that q) to experience oneself as having evidence for p even if one does not immediately know which mental state caused this impression and thus without knowing whether that state really qualifes as evidence in the frst place. Often, subjects just have the raw feeling of having evidence for or against the truth of some proposition and cannot point to the respective mental state that has caused this feeling. While this may sound mysterious, we must consider that seemings do not occur out of the blue but in the context of inquiry. If a subject has a question on her research agenda, the mind is prone to consider and evaluate possible answers. This happens automatically as a psychological response to questions that are open to thought. This psychological feedback mechanism is what I call the “zetetic push.” Seemings are suggestions
156
Verena Wagner
for correct answers to the question currently entertained that are pushed into consciousness. In this sense, a possible answer reveals (or recommends or represents) itself as the correct answer to the target question. The zetetic perspective also helps with avoiding typical objections brought against TEV. Huemer (2013) charges higher-order theories of seemings that involve beliefs about other mental states to be psychologically “over-intellectualized, or over-introspectivized” (p. 334). These accounts, says Huemer, cannot explain the immediacy of seemings and the passive character of the subject to whom the proposition presents itself as true. The modifed version of TEV suggested here does not view refection on one’s own mental states to be required for experiencing seemings. What is psychologically required, is to have a question open in one’s mind which is not overly complicated and does not require higher-order refection. Subjects don’t have to think or ask about their mental states—they simply have a question on their research agenda, and as a result of the zetetic push, possible answers recommend themselves, caused by potentially relevant mental states. If everything goes well, this feedback is caused by mental states that really provide evidence for the truth of the possible answer, but this is not necessarily the case. To illustrate the modifed version of TEV, let me introduce detective Zet: Detective Zet
Detective Zet is called to a crime scene to start with her new case. Upon arrival, she searches the scene and talks to witnesses. Zet cannot determine what it is, but one witness, Will, just seems guilty to her. Thus, Zet decides to treat Will as if he were the prime suspect: Zet puts Will (but none of the other witnesses) under police surveillance and searches for potential links between Will and the victim. Zet is following a hunch. Her seeming that Will is guilty is triggered by the inquisitive mode of asking who committed the crime. By having this question on her research agenda, her mind pushes possible answers into consciousness that are somehow connected to what she believes, experiences, has experienced, remembers, or wishes to be true. Zet experiences herself as having evidence in support of Will’s being guilty even if she does not yet know what it is that makes it seem so to her. When Zet refects on her impression, she may become aware of what it was that led her to think that Will is guilty. She may realize retrospectively that Will was nervously picking his right ear and that her previous experience with guilty suspects behaving similarly caused her hunch. This provides us with enough material to explain the peculiarity of seeming-occurrences. As experiences, seemings are feeting mental events
Zetetic Seemings and Their Role in Inquiry
157
that are diferent from standing attitudes like belief. They don’t show in introspection for cases where there is no question to be settled. Seemings are elusive experiences that show up if there is a zetetic push. Such a push is present in those cases where subjects focus on a confict within their body of evidence (e.g., visual information of a bent stick vs the stick’s actual straightness) which typically prompts re-inquiry and double-checking. Introspection into unchallenged beliefs does not prompt inquiry and, thus, no possible answer propositions are brought up to consciousness. Think of the unstamped-ticket case: It is plausible to think that S lacked any seeming about having stamped the ticket before she was challenged by the ticket inspector. Only due to the challenge does her false memory prompt an experience of having evidence of having validated the ticket. The same process would occur if S correctly remembered having stamped the ticket in a situation where someone, let’s suppose, secretly switched her validated ticket against an unstamped one. In response to the planted evidence, S might give up her true belief and question her sanity, but she would still experience herself as having evidence for having validated the ticket. Psychologically, it is the same situation as in the false memory case. It doesn’t matter whether it is the seeming or the countervailing belief that is false. What matters for the occurrence of seemings is the zetetic push that is here provided by challenging evidence. Thus, it is not the falseness of the memorial seeming (or the falseness of the bent-stick perception) that accounts for the peculiar occurrence pattern of seemings. Seemings are not rationally assessable. They just give us prompts and hints about our own evidential standing. In the next section, I will turn to the justifcatory force of seemings. 10.4.2.
The Justifcatory Force of Zetetic Seemings
Subjects who comply with ZIP by avoiding various distractions can still do better or worse in what they do (and not only avoid) zetetically. Some zetetic acts will be more rational than others because inquiring subjects have limited time, limited cognitive capacities, and limited resources. Zetetic decisions need to be made: how to start, which possible answer path to pursue frst, which information to assess as potential evidence, what information to double-check, etc. The zeemings realist suggests that seemings provide zetetic reasons for pursuing inquiry into their respective content proposition. For this purpose, the zeemings realist suggests the following zetetic version of PC: ZPC If it seems to S that p, and p is somehow relevant for S’s fguring out the currently asked question Q?, then, in the absence of zetetic defeaters, S has zetetic justifcation for pursuing p in her inquiry into Q?.
158
Verena Wagner
A proposition p is somehow relevant for S’s fguring out the currently asked question Q? if p is or entails a possible answer to Q? that partially or completely closes Q?, or so S believes. Zetetic defeaters are reasons that speak against pursuing a particular zetetic path, for example, information that pursuing this path would be too costly or too time-consuming, or is unlikely to deliver useful results. Epistemic reasons that are sufcient for adopting the belief that p can also be zetetic defeaters against pursuing p further in inquiry (provided that the adoption of a p-belief closes the relevant question). A subject can pursue a proposition in inquiry in multiple ways: one can inquire whether p is true, one can treat p as a to-be-checked working hypothesis, or one can temporarily rule out not-p in one’s inquiry. Zetetic justifcation is practical justifcation with respect to zetetic tasks and activities. Let me apply ZPC to the example of Detective Zet. Zet’s zetetic decisions to treat witness Will as a suspect, put him under surveillance, and check out his connection to the victim are motivated by her seeming that Will is guilty. If there are no defeaters, Zet’s zetetic acts are also rationally supported by her seeming. Zet would have a defeater if there were a more promising path to pursue, for example, if there were another suspect, who was identifed by several witnesses, or if Zet knew that she tends toward suspecting innocent witnesses. As I introduced the case in the previous section, Will’s behavior together with Zet’s expertise in witness and suspect behavior caused Zet’s experiencing herself as having evidence for Will’s guilt even if Zet was not immediately aware of the specifc body language she automatically processed as she was talking to Will. This makes Zet’s seeming a paradigm case of expert cognition and a case of benign cognitive penetration. In this context, it is easy to see that zetetic seemings are not challenged by the Double Counting Problem. Zetetic seemings are caused by other mental states and are thus cognitively penetrated by defnition. However, there is no double counting of their justifcatory force because seemings alone do not provide epistemic justifcation. A seeming that p merely provides justifcation for zetetic acts, such as treating p as a working premise, inquiring into p’s truth, or inquiring into the seeming’s causes. While Zet is justifed, based on her seeming, to pursue the hypothesis that Will committed the crime, Zet is not justifed to believe that it is so. This is plausible because Zet’s seeming could be caused by mental states that are either not justifers for belief at all or are of insufcient strength for the adoption of the respective belief. For example, Will might have been nervous for a reason that is completely unrelated to the case, maybe because he was driving his car without a license and does not want the police to fnd out. Only by pursuing the content of her seemings in inquiry
Zetetic Seemings and Their Role in Inquiry
159
will Zet be able to fnd out whether the cause of her seeming is actual evidence for believing its content or not. This fts well with what Audi (2013) has in mind when he writes about fnding the seeming’s “basic source”: [S]uppose I refect on how to justify a proposition I believe which I want others to believe. If I can say to myself only that it seems true to me and can fnd no story associating it with a basic source, I will very likely wonder if I have failed and would be widely taken to have failed. (Audi 2013, 193) Zetetic seemings do not provide us with epistemic justifcation for believing their contents but they hint at potential evidence that might need to be assessed. Yet, there are exceptions. If a subject has evidence that her seemings (in a certain feld) are mostly reliable, this constitutes both an epistemic justifcation for the seeming’s content as well as a zetetic defeater. In this case, S has no reason to pursue p in inquiry but a reason to adopt the corresponding belief directly. If it seems to the ornithologist that the bird over there is a hawk, and if she has a good track record in identifying local birds at frst sight, then she is justifed to believe that the bird over there is a hawk. The same goes for trained gold diggers, chess champions, experienced fre fghters as well as non-professional experts who have evidence that their seemings are reliable in a certain feld. What these subjects have in common is that their seemings are reliably caused by mental states that directly justify belief in the content proposition. These subjects have evidence that their seemings are reliable shortcuts to relevant and sufcient evidence that does not have to be confrmed by inquiry. This is true even if the causally responsible mental states that also provide the justifcation are not accessible to the subject. Even if the expert frefghter cannot explain what caused her seeming that the burning house would collapse any minute, she is justifed in so believing if she has a good track record in identifying imminent dangers arising from fres. 10.5.
Conclusion
My zetetic account of seemings takes the phenomenology and experiential character of seemings seriously and ofers a unifed view of seemings in light of their role as experiences of having evidence that can monitor and guide inquiry. It can explain the peculiar occurrence pattern of seemings by the presence or absence of a “zetetic push” and avoid the Double Counting Problem by restricting the justifcatory force of seemings to the zetetic realm. My account can also explain under which conditions expert seemings can provide epistemic justifcation.
160
Verena Wagner
Acknowledgments I am grateful to the audiences of the 2023 Zurich research colloquium on Normativity and the 2022 workshop Seemings: New Arguments, New Angles in Fredericton, New Brunswick, for feedback on earlier versions of this chapter. My special thanks go to the editors of this volume, especially Kevin McCain and Scott Stapleford, who provided me with helpful comments and questions. I thank the Baden-Württemberg Stiftung for fnancial support and Christopher von Bülow for proofreading. Notes 1 This may not be a problem for Huemer and Tolhurst, who take perceptual seemings to be perceptions (see also McCain and Moretti 2021). However, many other seemings-realists distinguish between perception and perceptual seemings (see Tucker 2013a, 7). 2 The Monty Hall Problem is a puzzle about subjective probabilities. There are three doors, behind two of which a goat is waiting, whereas there is a prize behind the other one. The player doesn’t know where the prize is and picks one door. Instead of opening the chosen door, the gamemaster opens one of the other two and shows a goat. Now, the player is asked whether she wants to remain at her chosen door or switch to the other door. 3 See Friedman’s (2020) Unity View and, alternatively, Thorstad’s (2021) Focal Point View.
References Audi, Robert (2013). Doxastic Innocence: Phenomenal Conservatism and Grounds of Justifcation. In Chris Tucker (ed.), Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism, pp. 181–201. New York: OUP. Conee, Earl (2013). Seeming Evidence. In Chris Tucker (ed.), Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism, pp. 52–68. New York: OUP. Friedman, Jane (2017). Why Suspend Judging? Noûs 51 (2):302–326. Friedman, Jane (2019). Inquiry and Belief. Noûs 53 (2):296–315. Friedman, Jane (2020). The Epistemic and the Zetetic. Philosophical Review 129 (4):501–536. Huemer, Michael (2007). Compassionate Phenomenal Conservatism. Philosophy and Phenomenological Research 74 (1):30–55. Huemer, Michael (2013). Phenomenal Conservatism Über Alles. In Chris Tucker (ed.), Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism, pp. 328–350. New York: OUP. McCain, Kevin, and Luca Moretti. (2021). Appearance and Explanation: Phenomenal Explanationism in Epistemology. Oxford: OUP. Pryor, James (2000). The Skeptic and the Dogmatist. Noûs 34 (4):517–549. Stafel, Julia (2019). Credences and Suspended Judgments as Transitional Attitudes. Philosophical Issues 29 (1):281–294.
Zetetic Seemings and Their Role in Inquiry
161
Thorstad, David (2021). Inquiry and the Epistemic. Philosophical Studies 178 (9):2913–2928. Tolhurst, William (1998). Seemings. American Philosophical Quarterly 35 (3):293–302. Tooley, Michael (2013). Michael Huemer and the Principle of Phenomenal Conservatism. In Chris Tucker (ed.), Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism, pp. 306–327. New York: OUP. Tucker, Chris (2010). Why Open-Minded People Should Endorse Dogmatism. Philosophical Perspectives 24 (1):529–545. Tucker, Chris (2013a). Seemings and Justifcation: An Introduction. In Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism, pp. 1–29. New York: OUP. Tucker, Chris (ed.) (2013b). Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism. New York: OUP. Williamson, Timothy (2007). The Philosophy of Philosophy. Malden, MA: Wiley-Blackwell.
11 Intuition in Philosophical Inquiry John Bengson
What role, if any, do “seemings” play in philosophy? Clarifying the relevant type of philosophical inquiry will enable us to pinpoint what epistemic contributions “seemings” could make thereto. It will also reveal what turns on the question—nothing less, I propose, than the very possibility of philosophy. However, the answer I develop appeals not to seemings but rather to intuitions conceived as a particular type of presentation—a “consciousness of seizing upon” how the world is, as Husserl put it.1 After distinguishing presentations from seemings, I prise apart two kinds of presentational phenomenology, eventually arguing that the “contentful” sort is poised to do the needed epistemic work. 11.1.
The Structure of Philosophical Inquiry
Philosophy can be undertaken in a practical mode, as a means to some social, political, or ethical end. It’s also possible to philosophize in an aesthetic mode; perhaps this is how to think of those who engage in philosophical activity just because they find it elegant, arresting, or awe-inspiring. Still other modes are possible. A pluralistic commitment to the legitimacy of these different options is fully compatible with viewing philosophical inquiry as often rightly undertaken in a theoretical mode. This is a species of a genus, theoretical inquiry: refection aimed at the provision of a theory. While often a messy business in practice, such inquiry can be modeled as a process structured around two stages.2 The frst centers on the gathering of relevant considerations, or data, that must somehow be handled when addressing the questions that open inquiry. The second involves constructing a theory that adequately addresses those questions. The two stages are intimately connected insofar as the outputs of the former are the inputs to the latter. Together, they represent the transition from the opening of inquiry, concerning a set of questions, to its closing. DOI: 10.4324/9781003299349-14
Intuition in Philosophical Inquiry
163
In what does such “closing” consist? In theoretical inquiry, undertaken for its own sake, a ftting end is making an intellectual improvement, in the form of an epistemic achievement such as justifed belief, knowledge, or understanding. These constitute success or excellence at theoretical inquiry, and so are among its proper goals. Of course, not every epistemic achievement is an ultimate proper goal: one that resolves theoretical inquiry in a fully successful way, reaching the point at which, to put it simply, there is no more work to be done. For example, although justifed belief or ordinary knowledge regarding an answer to a philosophical question—for example, What is knowledge? Is existence a property? Does consciousness afect behavior?—may leave inquirers feeling satisfed, neither state by itself guarantees the sort of comprehensive and systematic illumination needed to complete the job. After all, both can be gained on the basis of rote memorization, which pairs with obliviousness to the answer’s rationale, explanation, or broader ft in the web of the world; and both may incorporate muddle of various sorts, as when one only hazily grasps a key concept in the answer known. In either case, inquiry isn’t fnished. Suppose, however, that inquirers were to achieve understanding of their target, in the sense of fully grasping a theory that is accurate (correct), reasonbased (positively supported), robust (answers a multitude of questions about the domain), illuminating (explanatory), orderly (systematic), and coherent (internally and externally). Call such a theory “successful.” If a philosophical theory qualifes as such, then it handles the data in a manner that fosters understanding with respect to the questions—regarding topics such as knowledge, existence, consciousness, truth, beauty, God, and morality—giving voice to genuine philosophical perplexity. This is no small feat. Plausibly, fully grasping a theory that is successful in this sense would ensure the sort of comprehensive and systematic illumination that renders it ftting for inquirers to call it a day. Their curiosity, wonder, or puzzlement would be appropriately relieved thereby.3 11.2.
The Possibility of Philosophy
That is, of course, just a sketch of the basic contours of philosophical inquiry, its two-stage structure and epistemic télos. Despite being schematic
164
John Bengson
in certain respects, the sketch is both informative and fecund. Among other things, it enables us to sharpen a longstanding Big Question about philosophy itself: How is philosophy possible? For the sketch positions us to view this question as querying either i. the character and source of philosophical data (at stage one), or ii. the prospect of a method of theorizing (at stage two) capable of turning out successful philosophical theories. In efect, we can see the Big Question as asking whether we have the tools for the job: do we have intellectual states by whose exercise philosophers can, however fallibly, gather data and develop theories that adequately handle those data? Appeals to intuition in metaphilosophy are naturally read as proposing an afrmative answer to this question. To reach for intuition in this context is to advance the hypothesis that it is suited to shoulder the epistemic burdens at each stage of philosophical inquiry (data collection and theorizing).4 Eventually I’ll spell out my preferred way of thinking about intuition. For now, let me briefy comment on what this hypothesis says about the two stages. First, data collection. Very roughly, philosophical data are pre-theoretical claims that inquirers have good reason to believe. For example, that Gettier’s Smith has a justifed true belief but lacks knowledge, or that an alien creature could undergo mental activity while lacking an organ possessing the physical features characteristic of the human brain, or that it is wrong to harvest the organs of an innocent person to save the lives of fve others, or that it is unjust to privilege the contributions of men over the perfectly identical contributions of women—these are candidate data in epistemology, philosophy of mind, normative ethics, and feminist philosophy, respectively. These candidates have been gathered through refection on thought experiments. General claims regarding the supervenience of the aesthetic on the non-aesthetic, transitivity of identity, essence of singleton sets and their members, or moral status of persons are also widely treated as data. Such data do not magically appear in ready-made lists, but must be collected. Sometimes this is done via perception, introspection, linguistic judgment, testimony, or statistical analysis of surveys. While such empirical sources plausibly yield data describing facets of the actual world (e.g., data registering discrimination based on race, gender, or sexuality), it’s far from obvious that they alone can supply the sort of data required to anchor successful inquiry across the philosophical landscape. After all, many important philosophical topics have two features encouraging recognition of data with a non-empirical source: they are at least partly modal5 or normative in character, and are often not about anyone’s immediate environment, current psychological states, use of language, or lay opinions. Insofar as we
Intuition in Philosophical Inquiry
165
have intuitions about what is possible or necessary, essential or non-essential, right or wrong, good or bad, these suggest themselves as resources when collecting the sort of data philosophers need. The hypothesis stated earlier embraces this suggestion. Second, theorizing. At the heart of this second stage of inquiry is a method whose function is to take the data as inputs and to yield a successful theory as an output. At a minimum, such a method will instruct theorists to handle the data—that is, accommodate and explain them—through a series of claims each of which is, in turn, given an adequate defense.6 Many familiar defenses, such as various iterations of the Consequence Argument for incompatibilism about free will, or prominent ethical justifcations of vegetarianism, take the form of arguments. These are regularly built out of modal or normative premises regarded as true, perhaps obviously so, or supported by further arguments whose premises have this feature. Other defenses may be more direct—as per (a common interpretation of) traditional defenses of classical logic, according to which they treat its core principles as self-evident rather than the conclusions of arguments. Whatever its form, to be adequate a defense must provide positive epistemic support for a given claim: a reason for belief. Such a reason needn’t be dialectically persuasive, capable of rationally convincing an opponent. Nor must it be indefeasible, or even conclusive. Still, it needs to be good, being not only undefeated but also strong, in the sense that it would not be easily defeated by competing considerations, and its possession implies the possession of at least some evidence. The hypothesis articulated earlier efectively states that intuition fts the bill. I’ve suggested that if intuition is to help answer our question about the possibility of philosophy, it should slot in to inquiry’s two-stage structure in the right ways. The “should” indicates that the hypothesis that it does so must be earned, not presumed. One way to make progress towards this end would be to develop and refne this hypothesis, showing that it satisfes the criteria of a good theory by handling data about the epistemology of our two stages in a way that promotes understanding of the enterprise (philosophical inquiry) they compose. This is my primary goal in what follows. 11.3.
Data Regarding the Epistemology of Philosophy
An adequate theory of a given domain must handle the data regarding that domain. When it comes to the epistemology of philosophical inquiry, these data include observations about the epistemic practices of philosophers engaged in philosophical discussion and debate at the two stages of inquiry. These practices are many and varied. While any attempt at a comprehensive list is bound to be controversial, it is safe to say that any plausible efort will cite practices such as refection on thought experiments and critical scrutiny of general claims or principles. I will focus on this pair,
166
John Bengson
though my remarks plausibly extend to many other epistemic practices, such as assessment of relevant inferences (i.e., determining what follows from what) or conceptual ethics. Refection on thought experiments and general principles facilitates both the collection of data at inquiry’s frst stage and the defense of theoretical claims at its second. (Section 11.2 cited a few examples.) Our responses to hypothetical scenarios and candidate principles are not invariably based on reasoning through multi-premise arguments. Often the verdicts strike us as immediately clear or obvious. Nor are they usually the products merely of vision, audition, olfaction, gustation, or somatosensation. Yet, in such cases, we fnd ourselves in possession of knowledge, or at least justifed belief. That isn’t to say we never proceed cautiously; we sometimes withhold judgment. But in many cases, we take a stand, and regard our responses as suitable bases for a very high level of confdence. Having called attention to a few key epistemic practices in philosophy, highlighting the central role played by our responses to thought experiments and general principles, we are now ready to take the step of identifying some of the data that an adequate theory of these practices must handle. A few have already been mentioned; a bit more thought reveals several others. It will prove helpful to compile a list: D1. When engaging in these practices, it often strikes us that this, that, or the other thing. D2. These responses are conscious states or events. D3. These responses occur straightaway, in the absence of any consciously mediated transition from the presumed truth of one set of claims to the truth of the verdict that strikes us as true. D4. These responses may prompt us to endorse the verdict, though we may instead withhold judgment (e.g., from caution, or because we lose the temptation to afrm a given response). D5. These responses are sometimes reactions to highly unusual or fantastical scenarios, far removed from any actual situations that we’ve faced, or to principles whose scope of application encompasses situations we could not encounter. D6. These responses sometimes confict with our antecedent views, and so to what we are antecedently inclined to think about the topic. D7. These responses are regularly taken to justify the beliefs they prompt, provide evidence for those beliefs’ contents, and put their subjects in a position to know those contents. D8. In many cases, we regard it as appropriate to possess a very high level of confdence in these beliefs. D9. We are often unable to identify a set of considerations, independent of our responses, that provide a correspondingly strong rationale for these beliefs.
Intuition in Philosophical Inquiry
167
D10. While these beliefs are sometimes achieved with little efort, they are also sometimes the result of substantial refection or analysis. This list is not meant to be exhaustive. Nevertheless, it identifes a number of important data that an adequate theory of the relevant epistemic practices is called to accommodate and explain.7 To get a feel for how to proceed from here, consider an inferentialist theory of these practices: when collecting data and assembling theories in the course of philosophical inquiry, we acquire justifed belief or knowledge of philosophically signifcant claims by performing inferences from a stock of warranted background beliefs, where inference is a certain kind of normatively evaluable mental transition.8 This theory sticks to components— beliefs and inferences—that are familiar. However, it is an open question whether they enable the theory to handle all the data enumerated earlier (and other pertinent data). It is not simply that inferentialism is silent about the responses described in the frst six data, D1—D6, which it thereby fails to either accommodate or explain. Its central thesis is also difcult to square with several of these data: the idea that we infer our verdicts from our background beliefs appears to point in the opposite direction from the observations that those verdicts are not consciously mediated (per D3) and are sometimes novel (per D5) or revolutionary (per D6). Inferentialism also looks ill-suited to handle the observations about justifcation, evidence, and knowledge in the remaining data. For example, inductive and abductive inferences do not typically support the high level of confdence described in D8. The point is not decisive. But it becomes concerning when combined with D7, which tells us that our conscious responses are regularly treated as adequate epistemic bases, and D9, regarding our inability to identify independent considerations that might provide the indicated support. Of course, it is open to inferentialists to supplement their theory with additional claims that handle the data. Alternatively, inferentialists may provide strong reasons to think that some or all of the putative data I’ve enumerated are not really data or needn’t be accommodated and explained.9 While I welcome eforts to take these steps, it seems to me doubtful that either strategy will succeed. I am more optimistic about an intuition-based approach, to which we turn next. 11.4.
An Intuition-Based Theory
While there are many ways to develop an intuition-based theory of the relevant epistemic practices, my preferred version combines four main theses. The frst tells us what an intuition is: Quasi-Perceptualism: Intuitions are a specifc type of presentational state, a species of contentful state that does not merely represent the world as
168
John Bengson
being some way, but also presents it as being that way. (The diagram below portrays these taxonomical divisions.) Such a state has several characteristic features: it is conscious, non-factive, gradable (it can be more or less clear, vivid, hazy, or obscure), baseless (it is not consciously formed, by a subject, on the basis of any other mental states), fundamentally non-voluntary (it happens to one), compelling (it inclines one to belief), and rationalizing (its content shows up as to-be-believed).10 Contentful
Merely contentful
Representational
(e.g., hopes, wishes)
Merely representational
Presentational
(e.g., beliefs, acceptances, hypotheses)
(e.g.,
intuitions,
perceptual experiences)
The next three theses apply to all presentational states. The frst focuses on justifcation: Presentationalism: A thinker is justifed in believing that p on the basis of a presentational state—such as intuition—with content p because in having that state, it is presented to the thinker that p. Such justifcation is both defeasible and gradable: it is greater to the extent that the corresponding presentation is clear and vivid, and it is lesser to the extent that the presentation is hazy and obscure.11 Another addresses evidence: Content Is Evidence: The content of a presentational state—such as intuition—belongs to a thinker’s body of evidence because it is the content of a presentational state. Such evidential status is once again both defeasible and gradable.12 The last confronts the possibility of knowledge: Naïve Realism: When a presentational state—such as intuition—with content p is non-accidentally correct, hence able to serve as a source of knowledge that p, this is so because the state is constituted by the (possibly mind-independent) fact that p.13
Intuition in Philosophical Inquiry
169
Elsewhere I endeavor to clarify and substantiate each of these theses, along with several other claims about the nature and epistemic status of intuition—for example, concerning intuition’s relation to epistemically fecund social practices, place in disagreement, and role as a sense-maker.14 What I wish to emphasize here is that these four theses are well-positioned to accommodate and explain all the data enumerated earlier. To appreciate this, notice that the frst thesis, Quasi-Perceptualism, not only renders it likely that the initial six data are true; it also helps to explain why those data hold. For suppose that participants in the relevant epistemic practices have intuitions that are presentational in the sense described by the Quasi-Perceptualist thesis. This would make sense of the fact that subjects experience conscious strikings (per D1 and D2) that, while occurring without conscious mediation (per D3), could but need not prompt beliefs (per D4), perhaps even beliefs that are novel (per D5) or revolutionary (per D6).15 The remaining three theses—Presentationalism, Content is Evidence, and Naïve Realism—accommodate and explain the data about justifcation, evidence, and knowledge in D7 and D8. Together, the four theses help to make sense of D9, regarding our inability to identify independent considerations that provide the requisite support for our verdicts. And when the four theses are supplemented with a plausible claim about the etiology of presentational states, they also handle D10. The claim is this: Efort: While some presentational states—including some intuitions— require very little preparation, others may occur only in the wake of considerable mental efort, including sustained attention to and careful analysis of various questions, claims, and scenarios. While my use of this etiological claim to accommodate and explain a datum about philosophical inquiry may appear somewhat novel, the claim itself is old. A severe version is endorsed by Plato, who required ten years of advanced mathematical training, plus more experience and education besides, to properly “see those very things one cannot see except with the mind.”16 In the mid-20th century, A. C. Ewing emphasized in Reason and Intuition the comparatively prosaic point that Even where intuition is not backed by any explicit process of inference, . . . intuition is not a quasi-miraculous fash of insight standing by itself and not essentially connected with any thought-process at all. Intuition presupposes at least a partial analysis of the situation or a selecting of certain aspects of it, a process which presumably takes some time and may be more or less gradual, and it is certainly afected deeply by our previous experience and thought. What we see immediately may be the
170
John Bengson
result of a careful survey or long experience of the whole situation or the whole system involved.17 Though the labor at issue is often framed in individualistic terms, its socialhistorical dimensions are at least equally important. As Frege observed in the Grundlagen, Often it is only after immense intellectual efort, which may have continued over centuries, that humanity at last succeeds in . . . stripping of the irrelevant accretions which veil [a topic] from the eyes of the mind.18 While individual thinkers can give it their all, success is often contingent on the best eforts of those whom they follow. In what do such eforts consist? As with any psychological state, one may have an intuition in better or worse cognitive conditions. We can think of the good conditions as determined by characteristic features of the state, specifed by reference to its success. For example, a characteristic feature of perception is that an important range of its instances require normal lighting conditions for their success; consequently, such lighting will be among the good conditions in the case of perception. Similarly, insofar as a characteristic feature of intuition is that the success of an important range of its instances requires • • • • • • • •
mastery of relevant concepts, attentiveness to germane distinctions, alertness to pertinent instances of vagueness, ambiguity, under-specifcation, context-sensitivity, pragmatic efects, and the like, sensitivity to the potentially distorting infuence of afect or cognitive bias, familiarity with relevant facts (e.g., non-moral ones when addressing moral questions), acquaintance with relevant qualities (e.g., colors, felt pain), intellectually virtuous refection (e.g., calm, careful, attentive, openminded, sober), and the intention to uncover the truth,
such will be among the good conditions in the case of intuition. Such conditions play several important theoretical roles, described by the following thesis: Good Conditions: A presentational state—such as intuition—is regulated by good conditions for its operation, specifed by reference to the state’s success. Satisfaction of all these conditions implies that it
Intuition in Philosophical Inquiry
171
is constituted by the relevant fact (per Naïve Realism). Deviance from them is a potential defeater or attenuator. Awareness of them positions thinkers to correct, detect, and avoid errors or omissions. The efort to which Efort refers is essentially a matter of working—individually and collectively, over time—to realize the good conditions. Those who join are on track to intuit responsibly.19 11.5.
Presentations vs. Seemings
Quasi-perceptualism, which afrms that intuitions are presentational states, seeks to improve on the idea that intuitions are seemings. The latter classifcation is not particularly helpful, as “seeming” can be said in many ways (as J. L. Austin, Roderick Chisholm, Frank Jackson, and many others have emphasized).20 To give just one example, “It seems to S that p” is true merely if one is inclined to believe that p—as when one truly utters “It seems that the formula is indeed a theorem” after slogging through a highly complicated proof. One might insist that something else, or more specifc, is meant by “seeming.” This is George Bealer’s strategy: When you have an intuition that A, it seems to you that A. Here “seems” is understood, not in its use as a cautionary or “hedging” term [e.g., to designate an inclination to believe], but in its use as a term for a genuine kind of conscious episode.21 Bealer goes on to tell us what the relevant “genuine kind of conscious episode” is not: it is neither a belief, nor guess, nor hypothesis, nor hunch, and so forth. We are also given a few examples. But little is said about it itself. The relevant type of seeming—whatever it is—remains an unexplicated primitive. For instance, Bealer and other proponents of the seemings theory of intuition (such as Joel Pust and Michael Huemer) do not identify any of the type’s characteristic features, or systematically locate it in a broader taxonomy of mental state types. (By way of contrast, Quasi-Perceptualism does both of these things.) Consequently, the seemings theory is unilluminating: it is unable to handle relevant data. Although it afrms the facts about subjects’ responses in D1 and D2, it fails to explain them. Nor does it make sense of why those responses have the features specifed in D3, D4, D5, and D6. While an unexplicated notion may ofer a neutral starting point for subsequent investigation, it fails to deliver theoretical understanding of its target.22 To be clear, my intention is not to deny a connection between intuitions and the language of “seeming.” I recognize the legitimacy of uttering “It
172
John Bengson
seems that p” when one has an intuition. (Alternatives include: “It strikes me that p,” “It’s obvious that p,” “It’s clear that p,” “Plainly, p,” “p is evident,” or simply “p.”) But the legitimacy of this way of talking does not support the seemings theory. It’s simply to allow that the language of “seeming” is diverse and inclusive. Some proponents of the seemings theory might interpret this D&I view of “seeming”-talk to license viewing Quasi-Perceptualism as a friendly elaboration of the seemings theory: some seemings are just presentational states. I’m not completely hostile to this perspective. However, it’s worth noting that there appear to be specifc diferences between presentations and conscious seemings. Two stand out. First, although many mental states that merit the label “seeming” represent the world as being the way it would be were their contents true, they do not always present those contents as true. That is, seemings needn’t be presentational states. The point applies even to seemings that qualify as instances of a “genuine kind of conscious episode.” To fx attention, consider Ernest Sosa’s observation that when one feels attracted to assent to a proposition p, one is in a conscious state that can be truly expressed by uttering “It seems to me that p.”23 (If inclinations to believe can be truly reported with this form of words, as noted earlier, I see no reason to deny that felt attractions to assent can be as well.) But one may feel attracted to assenting to some claim without it being presented as true. A susceptible listener responding to a charismatic speaker who professes that nature demonstrates God’s greatness might feel attracted to assenting to this proposition even if the listener does not then have the intuition that it is true; the listener does not possess a mental state that presents it as being so. Second, it seems fair to say that whatever a conscious seeming is, it is explicit in the sense that its content is available, at the moment it occurs, as the content of a conscious thought fully articulable by its subject. In other words, if one enjoys a genuine conscious episode in which it seems to one that p, then one is able at the time to formulate explicitly—out loud or in one’s head—the way things seem: that p.24 In contrast, presentations are sometimes inexplicit: one need not be able at the time to formulate explicitly—out loud or in one’s head—the way things are presented as being. Just as one might have a visual experience but not at that moment be able to articulate fully what was presented in one’s visual consciousness, one might have an intuition but not at that moment be able to articulate fully what was presented in one’s intellectual consciousness. The implication is that what is presented may not be one and the same as what seems true. An example may help to bring this out. A moral philosopher might have an intuition that presents a given action as negatively valenced in a particular way. Unfortunately, she struggles to articulate the way; at the moment, it eludes her. What seems true is that the action is morally problematic—this
Intuition in Philosophical Inquiry
173
content is, right then and there, available for her to think and say. But she suspects that her intuition’s content is more specifc. So she sets herself the task of pinpointing the negatively valenced normative feature that her intuition ascribes to the action. Impermissibility? No, too strong. Disvalue? No, too weak. Later that evening, when perusing Richard Price’s Review of the Principal Questions in Morals, she lands on it: the conduct is unftting. This, she realizes, is what she intuited. The basic point should be familiar: it is one thing to experience or intuit something, and another thing entirely to be able to formulate it explicitly. It is a virtue of Quasi-Perceptualism, which understands intuitions as presentations (which are sometimes inexplicit) rather than seemings (which are explicit), that it clarifes and explains this datum.25 11.6.
Presentational Phenomenology
I’ve distinguished presentations from seemings. While I’ve identifed several characteristic features of presentational states, those features do not (individually or jointly) amount to an analysis of presentationality. This is a phenomenological property, concerning what it is like for a subject to be in such a state. What is presentational phenomenology? There are in fact several candidates for the label “presentational phenomenology.” Susanna Siegel’s treatment of visual experience highlights one species of the genus: It seems manifest to introspection that visual phenomenology presents spatial properties (such as being nearby or in front of the perceiver), color properties (or properties closely related to colors), and shape and luminance properties.26 Focusing also on visual experience, Scott Sturgeon suggests that the same goes for bearers of such features: What it’s like to enjoy visual experience is for it to be as if objects and their features are directly before the mind.27 Let’s use the expression “objectual presentation” to cover this species of presentational phenomenology, understanding “object” broadly to include such things as properties and relations, as well as individuals. Jim Pryor identifes another species when describing the peculiar “phenomenal force” or way our experiences have of presenting propositions to us. Our experiences represent propositions in such a way that it “feels as if” we could tell that those propositions are
174
John Bengson
true—and that we’re perceiving them to be true—just by virtue of having them so represented.28 Pryor’s suggestion is not that experiences present contents such as propositions to us in just the way that they present color or shape properties. Some mental states present contents in a diferent manner: what is presented, or present to mind, is its being the case that p, where p is a content such as a proposition. The content is presented as true (holding, obtaining, the case, so, etc.). Siegel provides a helpful illustration: Suppose you see a cube, and it looks red and cubical. Here your experience presents it as being the case that there is a red cube before you. Contrast a hope[:] in hoping that there is a red cube in front of you, it need not be presented to you as being the case that there is a red cube in front of you.29 Although the perceptual experience and hope both somehow relate you to the content there is a red cube before you, only the former presents this content as being the case. Quasi-Perceptualism asserts that intuitions are like perceptual experiences in just this way: they, too, present their contents as true. For instance, an intuition with the content a person persists after bicycling in the rain presents this content as being the case. Call this second species of presentational phenomenology “contentful presentationality.”30 I have identifed two species of a genus, presentational phenomenology, clarifying which is ascribed to intuitions by the theory presented in Section 11.4. Are there any other species? According to Elijah Chudnof, who takes both intuition and perceptual experience to possess presentational phenomenology, What it is for an experience of yours to have presentational phenomenology with respect to p is for it to both make it seem that p is true and make it seem as if this experience makes you aware of a truth-maker for p,31 where a truth-maker may be a property, relation, or individual—some “chunk of reality that makes it true that p.”32 This proposal is evocative. But I do not think it unearths a third species of presentational phenomenology. In fact, I doubt that it provides a theoretically adequate description of any type of presentational phenomenology. My skepticism is motivated by concerns about each conjunct. The frst conjunct understands presentational phenomenology in terms of seemings. Per Section 11.5, this renders it not only theoretically unilluminating (since the notion of seeming remains an unexplicated primitive)
Intuition in Philosophical Inquiry
175
but also mistaken (since conscious seemings are not always presentational but are always explicit, whereas presentations are always presentational but potentially inexplicit).33 As for the second conjunct, I harbor two reservations. First, the demand that subjects have in mind putative truth-makers excludes many ordinary moral intuitions. That it’s right to protect your children from lethal danger, wrong to recreationally slaughter other people, ftting to repay kindness with kindness (but not unkindness with unkindness)—intuiters often feel in the dark about the “chunk of reality,” if any, that makes the contents of these intuitions true.34 Likewise, familiar intuitions about Gettier cases, causal preemption scenarios, deviant causal chains, normative supervenience theses, the law of noncontradiction, or the transitivity of identity famously leave wide open what, if anything, makes their contents true (even putatively). Second, the second conjunct’s demand for seemingawareness35 overreaches. Suppose you have the intuition that anything green is a determinate shade of green, or that infnite magnitudes are larger than fnite ones, or that it’s blameworthy to violate a moral requirement absent excuse. Does it—indeed, must it—thereby seem to you as if you are aware of any relevant objects, such as green, infnity, or blameworthiness, that might plausibly serve as these contents’ truth-makers? It may be that your intuition involves thinking about or cognitively attending to such things. But it is far from clear that it involves seeming to be aware of them in a manner relevantly similar to the way that perceptual experience makes it seem as if you are aware of colors and shapes, which goes well beyond mere thinking or cognitive attention. One possibility is that neither conjunct is to be read strictly or literally, and Chudnof’s proposal simply aspires to conjoin the two species of presentational phenomenology I’ve identifed. This is a natural way to interpret Chudnof’s comment (issued in response to my distinction between the two species) that, as he sees it, “perceptual experiences both present propositions as being true to us, and present objects and features to us.”36 However, Chudnof immediately remarks, “I am using ‘presentational phenomenology’ to pick out the second property”; this is followed by the suggestion that experiences “have the frst because they have the second.”37 I doubt the remark, which is difcult to square with Chudnof’s demand for a seeming that p (recall the frst conjunct of his proposal, discussed in the previous three paragraphs); moreover, the idea that the second property captures a notion of presentational phenomenology applicable to intuition is questionable for (inter alia) the reasons ofered in the preceding paragraph. Regarding the suggestion: importantly, it commends an alternative to the theory articulated in Section 11.4, which gives pride of place to contentful presentationality. Let me now succinctly raise two objections to Chudnof’s suggestion that we instead view its objectual sibling as occupying the privileged position.
176
John Bengson
The frst is a variant on Frank Jackson’s “many property problem.38 Consider the following sequence. Your initial experience presents it as being the case that a red sphere and a green cube are on a surface before you. The two objects then change colors. Your subsequent experience presents it as being the case that a green sphere and a red cube are on a surface before you. Here is a claim about this sequence: in the two experiences, the same objects and properties are present to mind, though diferent contents are presented as being the case. If this claim is correct, as seems plausible, it tells against the suggestion that contentful presentational phenomenology can be fully explained in terms of objectual presentational phenomenology. The second objection is epistemic. I’ll approach it indirectly, by frst commenting on the interaction between the main elements of my position. Though neutral with respect to objectual presentationality, QuasiPerceptualism asserts that intuition displays contentful presentationality. Presentationalism, Content Is Evidence, and Naïve Realism put this assertion to epistemic work. In doing so, they explain the distinct epistemic profles of contentfully presentational states, such as perceptual experience and intuition, on one hand, and imagistic states, including states of imagination, whether sensory or intellectual, on the other. This might be thought unachievable by a set of epistemic theses focused on phenomenology. But Quasi-Perceptualism’s attention to contentful presentationality positions them to manage the feat. To illustrate, notice that an imaginer who deliberately forms a vivid mental image of a multi-colored balloon and imagines that it is hovering before her is relevantly diferent from a perceptual experiencer to whom it is presented that there is a multi-colored balloon hovering before her. The two subjects may enjoy identical visual images (involving redness, blueness, yellowness, circularity, etc.), perhaps thereby seeming to undergo the same objectual presentations (of the corresponding qualities); but the imaginer lacks the relevant contentful presentation—it is not presented to her as being the case that there is such a balloon—and so lacks, according to Presentationalism, the perceptual experiencer’s justifcation for corresponding belief. A similar contrast shows up in the intellectual case. A subject capable of rich geometrical imagination might achieve a vivid mental image of two triangles which agree in two sides and the enclosed angle; unable to discern what follows, but instructed by a teacher to ponder various options, she entertains the idea that any two such triangles are congruent. Contrast an intuiter to whom it is actually presented that any two such triangles are congruent: as we say, she just “sees” that this is so, when she refects on it. The imaginer and the intuiter may enjoy identical imagery (involving lines, angles, etc.), perhaps thereby seeming to undergo the same objectual presentations (of lines, angles, etc.); but the former lacks the relevant contentful presentation—it is not presented to her that any two such triangles are congruent—and so, according to
Intuition in Philosophical Inquiry
177
Presentationalism, lacks the latter’s justifcation for corresponding belief. An approach narrowly focused on objectual presentational phenomenology would arguably need to do gymnastics to make sense of the epistemic diference. (This is the second problem, promised earlier, facing such an approach.) By contrast, the epistemic diference is smoothly handled by an intuition-based theory that privileges contentful presentational phenomenology, taking it to be the epistemic star.39 11.7.
Conclusion
We began with a question about the role of “seemings” in philosophical inquiry. This question is connected to a larger one concerning the possibility of philosophy, which (on my sharpening) queries whether we have the intellectual capacities to succeed at each of the two stages of inquiry: data collection and theorizing. The question targets i. the character and source of philosophical data (at stage one), and ii. the prospect of a method of theorizing (at stage two) capable of turning out successful philosophical theories. Attending to some of the epistemic practices in which philosophers partake at both stages enabled us to uncover a series of relevant data. If an intuition-based theory highlighting presentations (rather than seemings) that possess contentful presentational phenomenology (rather than the objectual sort) handles these data in an illuminating manner, as I’ve proposed, then we’ve taken steps to earn the answer, initially foated as a hypothesis, that intuition so-construed is poised to help. I say “taken steps” because much work remains. After all, skeptics about intuition are legion; they cannot responsibly be ignored. Though I’ve ofered resources to respond to some of their objections, the foregoing leaves many of their questions unanswered. Likewise for a range of others querying (say) the explanatory basis of the theses I’ve proposed. This is hardly surprising. Here, as elsewhere in philosophy, progress does not lead us near the end but only farther from the beginning.40 Notes 1 Husserl (1913/1982, 9). 2 This two-stage model is not intended to describe the temporal phases of instances of theoretical inquiry as they occur on the ground but rather to represent the basic structure of such inquiry: its core components and their confguration. 3 This section briefy summarized ideas I’ve developed with co-authors in Philosophical Methodology: From Data to Theory (Bengson et al. 2022; chapter one elaborates and defends our claims about the structure of philosophical inquiry,
178
4 5 6
7
8
9 10 11 12
13 14 15
John Bengson
while chapters two and three furnish an account of philosophical data). Except for a couple of footnotes below that reference this joint work, subsequent sections of this chapter express my own views. Though they do not employ this particular framing device, I submit that it ofers a charitable reconstruction of the positions endorsed by Bealer (1996 and 1998), Bonjour (1998), and Pust (2000), among others. I use “modal” to cover a range of intensional as well as hyper-intensional phenomena: necessity, possibility, counterfactual dependence, grounding, essence, structure, and so forth. Additional criteria charge theorists with the tasks of explaining those claims and integrating them with our best picture of the world (e.g., the deliverances of our best science and mathematics); when, but only when, two or more theories do roughly equally well with respect to these other desiderata will the method instruct theorists to highlight theoretical virtues such as simplicity. Throughout, I focus on defense because this is where intuition is poised to make an epistemic contribution at inquiry’s second stage. For elaboration of the criteria I’ve listed, plus defense of the claim that their satisfaction delivers the features of understanding enumerated earlier, see Bengson et al.’s (2022, ch. 5) presentation of the Tri-Level Method. For example, additional data include that the responses sometimes lack the phenomenology of desire (e.g., the “hedonic aspect” described by Sinhababu 2017, section 2.2), may occur in the absence of a felt inclination to endorse their contents, resist certain sorts of normative assessments, and are often not about us— for instance, our linguistic practices or psychological states, whether actual or counterfactual—but instead concern non-linguistic, non-psychological matters. I lack the space to canvass the full range of core data here, though Section 11.6 presents the opportunity to discuss another datum (concerning inexplicitness). Such transitions may involve “the ofine application . . . of cognitive skills originally developed in perception,” as in Williamson’s (2013, 309) version of inferentialism. Other versions are suggested by Harman (2010) and Biggs and Wilson (2021). There are numerous ways to develop an inferentialist theory; my brief remarks in the text focus on a neutral version that can be feshed out in various ways. The option to reject a datum or forgo its accommodation and explanation protects against excessively conservative theorizing, while the demand for strong reasons discourages over-zealous revisionism (see Bengson et al. 2022, ch. 3). Bengson (2015a, sections 2–4). Bengson (2015a, section 5). Content Is Evidence is an alternative to the psychologistic view that intuition itself—the mental state—is evidence (see, e.g., Weatherson 2003; Alexander and Weinberg 2007; Goldman 2007; Alexander 2010; Mizrahi 2012, 2013; Boghossian 2020; Mofett 2023; for criticism, see Williamson 2007, ch. 7). At the same time, Content Is Evidence rejects the idea that contents are evidence independently of any connections to a thinker’s mental states, and so preserves the main motivation for psychologicism. Bengson (2015b). Bengson (2015c) and Bengson et al (2020 and forthcoming). Quasi-perceptualism also accommodates and explains the other data mentioned in Note 7, which are challenging for broadly doxastic theories of the responses described in D1—D6.
Intuition in Philosophical Inquiry
179
16 Republic 511a. Burnyeat’s (2000) discussion of Plato’s position is edifying. Descartes (1628/1985) articulated some of his own lofty demands in Rules for the Direction of the Mind, explaining how “we can make our employment of intuition . . . more skilful” (Rule IX). 17 Ewing (1941, 35). He writes a few pages later (Ibid., 39), “the more we have studied a subject and thought rationally about it, using our powers of inference, the more likely we are to be in a state of mind in which we shall intuit rightly. A person is not likely to have new, clear, true, and fruitful intuitions in regard to a subject if he has made no study of that subject.” See Koksvik (2013) and Chudnof (2019) for recent examples and helpful discussion of the mental efort involved in some cases of intuition. 18 Frege (1884/1950, vii). 19 The remarks in the text address concerns about intuiting being a free for all (cp. Weinberg 2007; Backes et al. forthcoming, section 6; cp. Wright 2010). I lack the space to address other skeptical worries here. Bengson (2020, sections 3–5) critically discusses several recent challenges to the intuition-based theory outlined here (cp. Bengson 2013a, 2014a, 2015a, 2015b). 20 Bengson (2014b, section 3.2) ofers a summary and relevant citations. 21 Bealer (1992, 101–102). Cp. Pust (2000, ch. 2) and Huemer (2007, section 1), among others. 22 Recall the features of theoretical understanding enumerated in Section 2.1. The criticism of the seemings theory advanced in the text can be extended to treatments of intuition that rely on unexplicated notions of the “feeling of rightness” (Thompson et al. 2011; Clavien and Fitzgerald 2017) or “pushiness” (Koksvick 2021), which do not furnish theoretical understanding. 23 Sosa (2007, ch. 3). Others have suggested that the same is true of temptations to believe: they are genuine kinds of conscious episodes, and they are truly reported by “It seems that p.” (Throughout I use double quotes where, strictly speaking, corner quotes are required.) 24 The point applies to a complex seeming, or collection of seemings, with multiple contents; if those contents all seem true, then all are articulable in the manner described. 25 Another virtue is normative. For example, distinguishing presentations from seemings has the potential to place Presentationalism outside the scope of objections challenging the epistemic import of seemings (e.g., some of the objections to phenomenal conservatism advanced by Markie 2006; Sosa 2007; Hawthorne and Lasonen-Aarnio 2021). 26 Siegel (2010, 52). 27 Sturgeon (2000, 24). Pautz (2007) uses the term “item” in a way similar to my broad use of “object.” 28 Pryor (2000, 547). 29 Siegel (2010, 48), though she does not distinguish contentful from objectual presentationality. Bengson (2013b, sections 2–3) draws this distinction before providing arguments helping to confrm it. 30 To appreciate that it is distinct from the frst species, suppose you willfully form a vivid mental image of a green leaf and imagine that it hangs before you. Imagination presents you with colors and shapes, but it does not present it as being the case, or true, that there is something hanging before you. This is a case of objectual presentationality without contentful presentationality. 31 Chudnof (2013, 38; cp. 2011, 2012, 2020).
180
John Bengson
32 Chudnof (2013, 18). I attempt to reconstruct the various elements of Chudnof’s characterization more carefully in Bengson (2014b). 33 In recent work, Chudnof (2020, 106) has replaced “seem” in the frst conjunct of his proposal with “represents.” This update may evade the criticism in the text, but at a cost, since representation does not imply contentful presentational phenomenology. This observation precipitates the epistemic objection described at the end of this section. 34 Chudnof acknowledges this point, yet seems oddly unconcerned by it (2013, 106–107). Elsewhere he suggests that “presentational phenomenology and its objects can be difcult to describe” (2013, 60). For sure. But this doesn’t quell the worry that we sometimes enjoy intuitions possessing presentational phenomenology with respect to a given moral proposition, but from the subject’s perspective the truth-maker of that proposition is missing. 35 In recent work Chudnof (2020, 106) has replaced “make it seem as if this experience makes” in his proposal’s second conjunct with “is felt as making.” As far as I can tell, this update only serves to intensify the objection I’m about to raise. 36 Chudnof (2011, 630n16). 37 Ibid. Siegel’s (2010, 48) treatment of the red cube example also suggests commitment to the idea that objectual presentationality sufces for contentful presentationality: “when such a property cluster (redness, cubicality, and being nearby) fgures in visual perceptual experience, the experience presents it as being the case that a red cube is nearby.” 38 Jackson’s (1975) argument targeted an adverbialist theory of perceptual experience. Since that is not my target here, replies to Jackson’s argument are not replies to the variant developed in the text. 39 In a rich and challenging discussion of presentationality, Deutsch (2019) allows that “perceptual experiences and intuitions share” the features listed in the Quasi-Perceptualist thesis plus the further feature (inexplicitness) spotlighted in Section 5, but he worries that those features and associated examples “do very little to convey what Bengson means by describing a [type of] mental state as ‘presentational’” (633). He concludes that presentation is a “mystery” (634), adding that we lack reason to deny that belief is also presentational (Section 2), in which case being presentational is not “content-justifying” (637–640). However, the features I’ve identifed distinguish presentation from belief: the latter type of state has many instances lacking one or more of them. More importantly, Deutsch’s critique fails to adequately engage the full range of elucidatory eforts in my original essay (2015a, sections 2–4), comprised of diverse illustrations, applications (e.g., to cases of intellectual illusion, hallucination, and blindsight), contrasts (e.g., between presentation and imagination, as well as inclination, attraction, and temptation), and intuitive glosses (e.g., in terms of one familiar use of “having the impression” as opposed to “being under the impression”—puzzlingly, Deutsch (637) dismisses this distinction “as a hair that is too fne to split”). The preceding discussion enriches that multifaceted treatment by directly examining the phenomenology of presentation. It thereby responds to Deutsch’s call to “give [the] terminology of ‘presentations’ a clear meaning” (645) while addressing his further, introspection-based worry that “there is nothing ‘it is like’ to have the Gettier intuition [or any other] philosophical intuition” (638), since “perceptual experiences are rich with phenomenal detail [which is “keyed to . . . mode”] and are phenomenally structured [which is “keyed to representational content”] as well,” whereas “[t]here simply is no phenomenal detail or structure to intuitions” (640). To the contrary,
Intuition in Philosophical Inquiry
181
intuitions enjoy both detail and structure: what it is like for a thinker to intuit Smith lacks knowledge is quite diferent from what it is like for her to wonder whether this is so; the former also difers from what it is like for her to have an intuition with another content, such as wanton torture is wrong or nothing can explain itself. At any rate, a state needn’t be phenomenally similar to perceptual experience in every respect in order to be like perceptual experience in enjoying contentful presentational phenomenology. It is this commonality that (pace Deutsch) renders both “content-justifying,” per Presentationalism. 40 I am grateful to participants in the NYU Mind & Language seminar on metaphilosophy, and especially to Crispin Wright for his illuminating comments on an early version of this paper. I also benefted from questions at a colloquium hosted by Ruhr-Universität Bochum, as well as input from Thomas Grundmann and the editors of this volume.
References Alexander, Joshua. 2010. “Is Experimental Philosophy Philosophically Signifcant?” Philosophical Psychology, 23: 377–389. Alexander, Joshua and Jonathan Weinberg. 2007. “Analytic Epistemology and Experimental Philosophy.” Philosophy Compass, 2: 56–80. Backes, Marvin, Matti Eklund, and Eliot Michaelson. Forthcoming. “Should Moral Intuitionism Go Social?” Noûs. Bealer, George. 1992. “The Incoherence of Empiricism.” Proceedings of the Aristotelian Society, 66: 99–138. Bealer, George. 1996. “A Priori Knowledge and the Scope of Philosophy.” Philosophical Studies, 81: 121–142. Bealer, George. 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. Rowman & Littlefeld. Bengson, John. 2013a. “Experimental Attacks on Intuitions and Answers.” Philosophy and Phenomenological Research, 86: 495–532. Bengson, John. 2013b. “Presentation and Content.” Noûs, 47: 795–807. Bengson, John. 2014a. “Review of Elijah Chudnof’s Intuition.” Notre Dame Philosophical Reviews. Bengson, John. 2014b. “How Philosophers Use Intuition and ‘Intuition’.” Philosophical Studies, 171: 555–576. Bengson, John. 2015a. “The Intellectual Given.” Mind, 495: 707–760. Bengson, John. 2015b. “Grasping the Third Realm.” Oxford Studies in Epistemology, 5: 1–38. Bengson, John. 2015c. “A Noetic Theory of Understanding and Intuition as SenseMaker.” Inquiry, 58: 633–668. Bengson, John. 2020. “The Myth of Quick and Easy Intuitions.” In S. Biggs and H. Geirsson, eds. The Routledge Handbook of Linguistic Reference. Routledge. Bengson, John, Terence Cuneo, and Russ Shafer-Landau. 2020. “Trusting Moral Intuitions.” Noûs, 54: 956–984. Bengson, John, Terence Cuneo, and Russ Shafer-Landau. 2022. Philosophical Methodology: From Data to Theory. Oxford University Press.
182
John Bengson
Bengson, John, Terence Cuneo, and Russ Shafer-Landau. Forthcoming. “Socially Conscious Moral Intuitionism.” Noûs. Biggs, Stephen and Jessica Wilson. 2021. “Abduction Versus Conceiving in Modal Epistemology.” Synthese, 198: 2045–2076. Boghossian, Paul. 2020. “Do We Have Reason to Doubt the Importance of the Distinction Between A Priori and A Posteriori Knowledge?” In Debating the A Priori. Oxford University Press. Bonjour, Laurence. 1998. In Defense of Pure Reason. Oxford University Press. Burnyeat, Myles. 2000. Plato on Why Mathematics Is Good for the Soul.” In T. Smiley, ed. Mathematics and Necessity: Essays in the History of Philosophy. Oxford University Press. Chudnof, Elijah. 2011. “What Intuitions Are Like.” Philosophy and Phenomenological Research, 82: 625–654. Chudnof, Elijah. 2012. “Presentational Phenomenology.” In S. Miguens and G. Preyer, eds. Consciousness and Subjectivity. Ontos Verlag. Chudnof, Elijah. 2013. Intuition. Oxford University Press. Chudnof, Elijah. 2019. “In Search of Intuition.” Australasian Journal of Philosophy, https://doi.org/10.1080/00048402.2019.1658121. Chudnof, Elijah. 2020. Forming Impressions: Expertise in Perception and Intuition. Oxford University Press. Clavien, Christine and C. Fitzgerald. 2017. “The Evolution of Moral Intuitions and Their Feeling of Rightness.” In R. Joyce, ed. The Routledge Handbook of Evolution and Philosophy. Routledge. Descartes, René. 1628/1985. Rules for the Direction of the Mind. Trans. J. Cottingham, R. Stoothof, and D. Murdoch. The Philosophical Writings of Descartes, Vol. 1. Cambridge University Press. Deutsch, Max. 2019. “Are Intuitions Quasi-Perceptual ‘Presentations’?” Metaphilosophy, 50: 631–648. Ewing, A.C. 1941. Reason and Intuition. London: Humphrey Milford Amen House, E.C. Frege, Gottlob. 1884/1950. The Foundations of Arithmetic. Trans. J. L. Austin. Northwestern University Press. Goldman, Alvin. 2007. “Philosophical Intuitions: Their Target, Their Source, and Their Epistemic Status.” Grazer Philosophische Studien, 74: 1–26. Harman, Gilbert. 2010. “Epistemology as Methodology.” In J. Dancy, E. Sosa, and M. Steup, eds. A Companion to Epistemology, 2nd edition. Wiley-Blackwell Hawthorne, John and Maria Lasonen-Aarnio. 2021. “Not So Phenomenal!” The Philosophical Review, 130: 1–44. Huemer, Michael. 2007. “Compassionate Phenomenal Conservatism.” Philosophy and Phenomenological Research, 74: 30–55. Husserl, Edmund. 1913/1982. Ideas Pertaining to a Pure Phenomenology and to Phenomenological Philosophy. Trans. F. Kersten. Kluwer. Jackson, Frank. 1975. “On the Adverbial Analysis of Visual Experience.” Metaphilosophy, 6: 127–135. Koksvik, Ole. 2013. “Intuition and Conscious Reasoning.” Philosophical Quarterly, 63: 709–715.
Intuition in Philosophical Inquiry
183
Koksvick, Ole. 2021. Intuition as Conscious Experience. Routledge. Markie, Peter. 2006. “Epistemically Appropriate Perceptual Belief.” Noûs, 40: 118–142. Mizrahi, Moti. 2012. “Intuition Mongering.” The Reasoner, 6: 169–170. Mizrahi, Moti. 2013. “More Intuition Mongering.” The Reasoner, 7: 5–6. Mofett, Marc A. 2023. “Intuitions as Evidence.” In M. Lasonen-Aarnio and C. Littlejohn, eds. The Routledge Handbook of the Philosophy of Evidence. Routledge. Pautz, Adam. 2007. “Intentionalism and Perceptual Presence.” Philosophical Perspectives, 21: 495-541. Pryor, Jim. 2000. “The Skeptic and the Dogmatist.” Noûs, 34: 517–549. Pust, Joel. 2000. Intuitions as Evidence. Garland Publishing. Siegel, Susanna. 2010. The Contents of Visual Experience. Oxford University Press. Sinhababu, Neil. 2017. Humean Nature: How Desires Explain Action, Thought, and Feeling. Oxford University Press. Sosa, Ernest. 2007. A Virtue Epistemology: Apt Belief and Refective Knowledge, Vol. 1. Oxford University Press. Sturgeon, Scott. 2000. Matters of Mind. Routledge. Thompson, V. A., J. A. Prowse Turner, and G. Pennycook. 2011. “Intuition, Reason, and Metacognition.” Cognitive Psychology, 63: 107–140. Weatherson, Brian. 2003. “What Good Are Counterexamples?” Philosophical Studies, 115: 1–31. Weinberg, Jonathan. 2007. “How to Challenge Intuitions Empirically Without Risking Scepticism.” Midwest Studies in Philosophy, 31: 318–343. Williamson, Timothy. 2007. The Philosophy of Philosophy. Blackwell. Williamson, Timothy. 2013. Modal Logic as Metaphysics. Oxford University Press. Wright, Jennifer Cole. 2010. “On Intuitional Stability: The Clear, the Strong, and the Paradigmatic.” Cognition, 115: 419–503.
Part 3
Seemings and Perception
12 Veridical Perceptual Seemings Elijah Chudnof
When I use the term “seemings” I mean to pick out a determinable of which various familiar experiences are determinate varieties. Examples include perceptual experiences of your environment, experiences of recalling events from your past, and experiences in which you intuit an abstract truth. In each case, there is some way you experience a subject matter as being, some way it seems to you. Here I focus on perceptual experiences.1 Readers of this volume will be familiar with dogmatism and phenomenal conservatism about perception. These are views about prima facie justifcation. Their core idea is that if you have a perceptual experience as of p, that is, if it perceptually seems to you as if p, then you thereby have some prima facie justifcation for believing that p. The core idea provokes diferent debates.2 Is there any merit to it at all? Supposing there is merit to the core idea, is it correct as stated or should it be revised or supplemented along various dimensions? Supposing all this is sorted out in favor of a specifc version of the core idea, what non-epistemic features ground the epistemic signifcance therein attributed to perceptual experiences? These debates are in the background of what I want to discuss here. My foreground topic is knowledge. Suppose you have an experience as of your environment being some way. Your experience might or might not be veridical. You might or might not take it at face value and thereby form a belief that your environment is how it seems to be. With respect to prima facie justifcation these matters can go either way. Suppose, however, your experience is veridical, and you do form a belief based on it by taking it at face value. What is the epistemic upshot? Everyone should agree that the belief you form is true. I’m interested in what else might result. Here are two positions. The Minimal View: If you believe that p based on a veridical perceptual experience as of p, then your belief is thereby prima facie justifed and true. Any additional epistemic status derives from additional factors.
DOI: 10.4324/9781003299349-16
188
Elijah Chudnof
The Maximal View: If you believe that p based on a veridical perceptual experience as of p, then your belief might thereby amount to knowledge. The belief might be knowledge just because it is based on the veridical perceptual experience. Between these extremes, there is plenty of room to develop one or another moderate view. Maybe you think belief based on veridical perceptual experience manifests a connection to reality that involves more than accuracy but less than knowledge. In this chapter, I set such possibilities aside and focus on the Maximal View. I fnd it attractive and want to explore what can be said in its defense. Here is the plan. The Maximal View is not popular. In Section 12.1, I explain why by describing the main challenges it faces. One prominent view about perceptual knowledge that is at least akin to the Maximal View is John McDowell’s view that a belief might amount to knowledge because it is based on a suitable perceptual state. In Section 12.2, I consider whether a line of thought suggested by McDowell’s view of perceptual knowledge contains resources to meet the challenges described in Section 12.1. I express some doubts that the challenges can be fully met this way. In Section 12.3, I explore the prospects of an alternative approach. The approach draws on some ideas from the phenomenological tradition. After promising but tentative results, I conclude in Section 12.4 with some reflections on the motivations for the Maximal View. 12.1.
Problems for the Maximal View
I know that a seen pepper is red. The Maximal View implies that my belief might amount to knowledge because it is based on my veridical perceptual experience as of a red pepper. Problems for the Maximal View derive from the fact that knowledge requires more than true belief based on good evidence. If my evidence is a veridical perceptual experience, then it might guarantee the truth of some beliefs based on it, but that is not enough for them to count as knowledge.3 One problem is environmental luck: The market is short on red peppers. They insert the few remaining red peppers in a row of mostly yellow peppers and install red lights where necessary to make it look like a full row of red peppers. I happen to look at one of the genuinely red peppers under normal illumination. What looks to me like a red pepper really is a red pepper. Though it happens to be true, my belief that the seen pepper is red could too easily have been false and does not amount to knowledge.
Veridical Perceptual Seemings
189
Alternatively, my environment might be normal, but I might have misleading evidence that it is not. This is the problem of epistemic defeat: The market is now well stocked with red peppers. I vividly recall being fooled by red lights installed over a row of yellow peppers in the past. I notice a shifty store manager with a bundle of red lights. Without good reason to discount the defeating evidence of memory and circumstance, when I see one of the red peppers, I form the belief that it is red anyway. My belief that the seen pepper is red is irrational and does not amount to knowledge. These cases strongly suggest that if I know that a seen pepper is red, then my veridical perceptual experience as of a red pepper is at best part of what explains this fact. There must be additional factors ruling out environmental luck and epistemic defeat. If so, then the Maximal View is false. This reading of the examples, however, depends on implicitly assigning circumscribed veridicality conditions to my perceptual experience as of a red pepper.4 One idea along these lines is that my experience is veridical just in case there is a red pepper nearby. This would be obviously mistaken about any actual experience. Any actual experience as of a red pepper would represent much else about the pepper and the situation in which it is seen. Let’s distinguish between full and partial veridicality. An experience is fully veridical just in case all its veridicality conditions are met, partly veridical if only some of its veridicality conditions are met. A more considered statement of the Maximal View should take this distinction into account as follows: If you believe that p based on a fully veridical perceptual experience as of p, then your belief might thereby amount to knowledge. The belief might be knowledge just because it is based on the fully veridical perceptual experience. This statement of the view still isolates one proposition—p—as expressing the content of experience informing the belief whose status is in question, but it also makes the epistemic signifcance it attributes to the experience depend on the satisfaction of veridicality conditions aside from those expressed by p. This dependence on full veridicality is relevant to assessing the challenges to the Maximal View. When I have a perceptual experience as of a red pepper, clearly it is possible for me to think to myself that I’m in a normal environment and that I have no defeating evidence. Considering full veridicality conditions, then, maybe I don’t need to think these things about my environment and evidence because my perceptual experience
190
Elijah Chudnof
already represents them. If so, then the examples do not show that the Maximal View is false since they are not examples of fully veridical perceptual experience. In the next section, I consider a view along these lines. 12.2. Perceptual Self-Consciousness John McDowell’s discussions of perceptual knowledge suggest a way to defend the Maximal View. McDowell addresses the problems of environmental luck and epistemic defeat while defending the idea that a belief might be knowledge because it is based on a suitable state of perceptual consciousness. To a frst approximation, if perceptual consciousness secures knowledge, then that is because it is also a form of self-consciousness, namely consciousness of oneself as enabled to know. Here is one way McDowell puts it: An experience in which the subject perceives that things are a certain way contains a potential for knowledge that the experience has that rational signifcance, even if the experience’s potential for grounding a judgment that things are that way, which would be knowledgeable, is not actualized (McDowell 2018, pg. 93). For it to make your belief that p based on it amount to knowledge, a perceptual state must constitute both perceiving a truth-maker for p and veridical consciousness of yourself as enabled to know that p. Veridical consciousness of yourself as enabled to know that p rules out epistemic defeat because it is a form of consciousness incompatible with the presence to mind of defeating considerations. Veridical consciousness of yourself as enabled to know that p also rules out environmental luck because its veridicality is also inconsistent with the presence in the environment of knowledge disabling circumstances. If my experience as of a red pepper puts me in a position to know that a seen pepper is red, for example, then I am in a perceptual state that constitutes perceiving a truth-maker for the proposition that the seen pepper is red and veridical consciousness of myself as enabled to know that the seen pepper is red. I couldn’t be in this perceptual state in either of the two cases discussed in the previous section. Suppose some perceptual states constitute both perceiving truth-makers and veridical consciousness of oneself as enabled to know. Then plausibly some fully veridical perceptual experiences are such perceptual states. It is not clear what else other than my fully veridical experience as of a red pepper might be the perceptual state that constitutes both perceiving the red pepper and veridical consciousness of myself as enabled to know that the seen pepper is red. Full satisfaction of veridicality conditions with respect
Veridical Perceptual Seemings
191
to my environment and my epistemic position is just what would make a perceptual experience into the sort of perceptual state McDowell describes. A natural idea for the Maximal View to adopt, then, is that if a veridical perceptual experience makes a belief based on it amount to knowledge, then that is because its full veridicality makes it into the sort of perceptual state McDowell describes. This would circumvent the problems of environmental luck and epistemic defeat. So, if there are such perceptual states, then Maximalists can recruit this fact in defense of their view. This is a big if. It is one thing for a perceptual state to constitute perceiving a situation in which a nearby pepper is red. That is the sort of job perception is for. It is another thing for a perceptual state to constitute veridical consciousness of oneself as enabled to know that a seen pepper is red. That job requires cooperation from one’s larger environment and one’s other mental states, such as thoughts and memories. So, how exactly are perceptual states themselves supposed to constitute both perceiving truthmakers and veridical consciousness of oneself as enabled to know propositions about them? The force of the question can be clarifed by considering an inadequate answer. It goes as follows. Setting skeptical worries aside, there are cases in which I do self-consciously know that a seen pepper is red. Call these good cases. In good cases, I perceive a truth-maker for the proposition that a particular pepper is red, and I am veridically conscious of myself as enabled to know that the seen pepper is red. So, there are situations in which one both perceives a truth-maker for p and is veridically conscious of oneself as enabled to know that p. Call the state of being in such a situation a good perceptual state. Good perceptual states are perceptual states that constitute perceiving a truth-maker for p and veridical consciousness of oneself as enabled to know that p. Fully veridical perceptual seemings are sufcient bases for knowledge because they are good perceptual states. There are a few reasons why this is an inadequate answer. First, perceptual state is a kind whose extension is not open to stipulation. Calling situations in which one both perceives a truth-maker for p and is veridically conscious of oneself as enabled to know that p “good perceptual states” does not show how they are perceptual states. Second, there should be some internal connection between perceiving a truth-maker for p and veridical consciousness of oneself as enabled to know that p. Their cooccurrence in some cases is not enough to show that the class of those cases forms a natural kind. On its own this is a probative consideration. But, and thirdly, the worry deepens when considered along with the fact that the perceptual states we are trying to understand are supposed to be bases for ordinary beliefs formed in response to perception, such as the belief that a seen pepper is red. Those bases should be available in the ordinary course of things. For all the view on ofers tells us about good perceptual states,
192
Elijah Chudnof
however, they are only recognizable by those who know their defnitions. Compare: normally sighted adults recognize instances of green; only the initiated recognize instances of grue. Fourth, and fnally, the conception of good perceptual state under consideration yields a coordinate notion of veridical perceptual seeming that obscures the distinction between the Minimal View and the Maximal View. If there are no constraints on what is included in the veridicality conditions of a perceptual seeming, then it isn’t clear what debates over whether factors in addition to their satisfaction are required for knowledge could be about. A potentially better answer to the question of how perceptual consciousness includes epistemically loaded self-consciousness derives from another idea of McDowell’s. In the paper already quoted, he writes: Knowledge of one’s ground for a perceptually knowledgeable judgment is an act of the same capacity for knowledge that is in act in the perceptual knowledge constituted by the judgment; and more generally, the potential for knowing that an experience has a rational signifcance by virtue of which it can ground perceptually knowledgeable judgments is contained in the same experience by virtue of its being, as such, a partial act of that same capacity. (McDowell 2018, pg. 95, italics in original) There is a capacity that is exercised in making perceptual judgments amounting to knowledge. Perceptual consciousness includes consciousness of oneself as enabled to know because it is “a partial act of that same capacity.” So, if a perceptual state constitutes both perceiving a truthmaker for p and veridical consciousness of oneself as enabled to know that p, then that is because it is the partial exercise of a capacity that, when completely exercised, results in making perceptual judgments amounting to knowledge. When I am in a perceptual state that constitutes both seeing a red pepper and veridical consciousness of myself as enabled to know that the seen pepper is red, for example, then that is because my perceptual state is the partial exercise of a capacity that, when completely exercised, results in my knowing by perception that the seen pepper is red. Talk of capacities suggests there is an explanation here that is missing from the inadequate view considered above. But this isn’t entirely clear to me. What is it for perceptual consciousness to be a partial act of a capacity for knowledge? Here are two options. Option 1: Being perceptually conscious is part of the process of making a perceptual judgment amounting to knowledge. This is a natural understanding, and it is true, but it doesn’t explain how perceptual consciousness intimates its own rational signifcance—comes to include epistemically loaded self-consciousness. One might reply that making a
Veridical Perceptual Seemings
193
perceptual judgment amounting to knowledge is, or at least sometimes is, a self-conscious act, so there must be a form of perceptual consciousness that includes epistemically loaded self-consciousness. This reply, however, returns to the inadequate view already rejected. It moves from the co-occurrence of perceiving truth-makers and veridical consciousness of being enabled to know in good cases to the existence of a form of perceptual consciousness that constitutes both. Option 2: The processes that determine the content of perceptual consciousness include partial exercise of a capacity for knowledge. This suggests the beginnings of a causal explanation. Perceptual consciousness results from exercising various capacities such as capacities for motion detection and shape constancy. Maybe it also results from partial exercise of a capacity for knowledge and this partial exercise of the capacity adds content about rational signifcance to content about motion and shape and the like. Perhaps an idea along these lines is workable, but to be empirically credible it would have to sharply distinguish how a capacity for knowledge determines perceptual content from how capacities for motion detection and shape constancy determine perceptual content. Let’s take stock. The main idea considered in this section is that a fully veridical perceptual experience as of p can be a sufcient basis for knowledge of p because it is a perceptual state that constitutes both perceiving a truth-maker for p and veridical consciousness of oneself as enabled to know that p. Though any ambition to give a non-circular account of knowledge is clearly of the table, this idea raises a non-optional explanatory question. Why does perceptual consciousness include epistemically loaded selfconsciousness? I’ve found it difcult to see through to an adequate answer in terms of the kinds of capacities involved in being perceptually conscious. The next section considers an approach that draws on some ideas in the phenomenological tradition. 12.3.
A Phenomenological Explanation
One might ask, from the position of a skeptic, for reasons to think some perceptual seemings include epistemically loaded self-consciousness. I’m concerned with an explanatory question. Suppose there are such perceptual seemings. How are they possible? What might explain their occurrence? Addressing this issue has potential to allay grounds for skepticism, but I set that project aside here. The phenomenological explanation I consider is that perceptual consciousness and epistemically loaded self-consciousness are tied together by their common grounding in consciousness itself. The nature of consciousness explains their unifcation in the sorts of perceptual seemings whose full veridicality requires being in a good case. For ideas about necessary
194
Elijah Chudnof
structures in consciousness one naturally looks to the phenomenological tradition.5,6 There is a diversity of opinion within that tradition, but the ideas I’ll draw on are largely common currency. The frst two occur with regularity in central texts by the likes of Husserl, Heidegger, MerleauPonty, and Sartre. I’ll quote from Aron Gurwitsch, who deserves wider recognition. Briefy put, the frst idea is that consciousness necessarily includes selfconsciousness. Here is Gurwitsch: When an object is given in experience, the experiencing subject is conscious of the object and has an awareness of this very consciousness of the object. Perceiving a material thing, listening to a musical note, thinking of a mathematical theorem, etc., we are not only conscious of the thing, the note, the theorem, etc., but are also aware of our perceiving, listening, thinking, etc. (Gurwitsch 2009, pg. 451). Classic and contemporary developments of this idea belabor not reading more into it than is intended (cf. Gallagher and Zahavi 2021). The main point I want to emphasize is that the awareness Gurwitsch and other phenomenologists have in mind is supposed to be constitutive of phenomenal consciousness. Part of what it is for a mental state to be phenomenally conscious is for its subject to be aware of that state. The relevant form of awareness should be available to any phenomenally conscious subject and operative in every phenomenally conscious state of that subject. So, it should not require higher cognitive capacities, nor should it consist in optional acts of refection. My seeing a red pepper includes my being aware of my seeing a red pepper just because it is phenomenally conscious, and not because I can or do think about it. That said, it could be that my awareness of my seeing a red pepper has a content that is enriched relative to the content of analogous states in phenomenally conscious lower animals. This point depends on attributing content to the relevant form of awareness, which brings us to the second idea. The second idea is that the self-consciousness partly constitutive of phenomenal consciousness imposes conditions on what consciousness is like. Perhaps the most general condition, taken up by every major writer in the phenomenological tradition, is temporality. Again, here is Gurwitsch: Experiencing an act, we are then aware of it prior to refection and even without grasping the act at all as a temporal phenomenon, as beginning, enduring and growing, and fading. Our awareness of the temporal development of an act is one and the same with our awareness of its
Veridical Perceptual Seemings
195
being experienced. Thus, in the case of an enduring act, we are aware of the fact that the act experienced now is the “same” as that experienced a moment ago. (Gurwitsch 2009, pg. 459). One way to think about the idea is this. Awareness of something is always awareness of it as being some way, and the awareness we’ve supposed to be constitutive of phenomenal consciousness is always awareness of it as having one or another temporal structure. So, it is not enough that experiences have temporal structure. The temporal structure must be felt, and the idea is that it is felt in the awareness of experience we’ve supposed to be constitutive of phenomenal consciousness. The frst idea discussed above is that my seeing a red pepper includes my being aware of my seeing a red pepper. The second idea introduced here is that my being aware of my seeing a red pepper is an awareness of my seeing as having a certain temporal structure: maybe it is felt as a new, or as continuing, or as feeting, etc. These are attributions of diferent temporal characteristics to my experience. Because they are attributions of characteristics, I take it they imply that my awareness of my seeing a red pepper has representational content. It is this content that might be enriched relative to the content of analogous states in phenomenally conscious lower animals. For example, maybe human experience is felt as temporally structured in ways not available to lower animals. Like the frst two ideas, the third and fnal idea I’ll draw on traces back to Husserl, and in this case, I prefer sticking to the source. Its later infuence on the tradition is relatively muted compared to that of the frst two ideas. The idea I have in mind is one Husserl discusses under the heading “modalization.” In addition to a felt temporal structure, states of consciousness are felt as having a “mode of validity.” I’ll quote, then comment: Consciousness, which gives its object in the fesh (originally), does not only have the mode of presentation in the fesh, which distinguishes it from presentifying consciousness and empty consciousness (both of which do not present in the same sense in the fesh); it also has a variable mode of being or a variable mode of validity. Original, normal perception has the primordial mode, “being valid simpliciter”; this is what we call straightforward naïve certainty. The appearing object is there in uncontested and unbroken certainty. What is uncontested points to possible contestations, or even to breaks, precisely to those we have just described, and by becoming bifurcated, they undergo a modifcation in their mode of validity. In doubt, both presentations in the fesh contending with one another have the same mode of validity, “questionable,”
196
Elijah Chudnof
and each presentation that is questionable is precisely in dispute and contested by the other. (Husserl 2001, pgs. 74–75) Consciousness which gives its object in the fesh is perceptual experience. Its “in the fesh” aspect distinguishes it from imagination (“presentifying” consciousness) and thought (“empty” consciousness). But two instances of in the fesh perceptual consciousness might difer from each other in their “validity.” In this passage Husserl contrasts perceptual consciousness that is felt with naïve certainty and perceptual consciousness that is felt with doubt. The doubt he has in mind arises by a confict in perception over time. Diference in felt validity is what Husserl means by modalization, which in addition to naïve certainty and doubtfulness includes possibility, probability, and certainty again but this time in a non-naïve form due to its resulting from a resolution of conficts. The key takeaway for present purposes is that these modes of validity are felt in the same way that temporal structure is felt. They are conditions self-consciousness might impose on consciousness. So, the frst phenomenological idea is that my seeing a red pepper includes my being aware of my seeing a red pepper. And the second is that my being aware of my seeing a red pepper is awareness of my seeing as having a temporal structure. The third and fnal idea is that my being aware of my seeing a red pepper is also awareness of my seeing as having a mode of validity. Why think that experiences are felt as having a mode of validity in addition to a temporal structure? Husserl presents examples and develops considerations that look like what are now called phenomenal contrast arguments (cf. Siegel 2007). Here is the kind of example he discusses. Consider the following two cases: Case 1. You seem to see a movie star, approach to get a selfe, fnd out it is a wax statue, and as you walk away seem to see a famous television actor. Case 2. You do see a movie star, succeed in getting a selfe, and as you walk away see a famous television actor. The visual experience as of a famous television actor in Case 1 has a certain temporal structure: it is felt as occurring after a visual experience as of a movie star. According to Husserl, and this seems plausible to me, the visual experience as of a television actor in Case 1 also has a mode of validity: it is felt as put in doubt by the previous visual experience as of a movie star. Husserl draws a further consequence, this time about the visual experience as of a television actor in Case 2. It also has a mode of validity: it is felt with naïve certainty. We might not have attended to this
Veridical Perceptual Seemings
197
aspect of the experience without frst contrasting it with the experience in Case 1, but once we do make the contrast, then, according to Husserl, we should recognize naïve certainty as a characteristic the experience has all along. The modes of validity Husserl discusses under the heading of certainty— whether naïve or not—are naturally taken to be forms of consciousness of oneself as enabled to know. With this in mind, let’s return to questions about explanation. Why do some fully veridical perceptual seemings constitute both perceiving truth-makers and veridical consciousness of oneself as enabled to know propositions about those truth-makers? The answer is that the relevant perceptual seemings are phenomenally conscious, and part of being phenomenally conscious is being felt as having one or another mode of validity. In the good cases in question, the mode of validity is certainty, and so is a mode in which the perceptual seeming includes consciousness of oneself as enabled to know. Since the perceptual seeming is fully veridical, its epistemically loaded veridicality conditions are met, and so it is the sort of perceptual seeming required by the Maximal View. The phenomenological answer to the explanatory question appears to improve on the inadequate answer rejected in the previous section. The inadequate answer begins with the observation that we sometimes do perceive a truth-maker for p and are veridically conscious of ourselves as enabled to know that p, and it defnes “good perceptual state” to mean being in a state in which those two conditions are met. Supposing there are such good perceptual states, then full veridicality of a perceptual seeming sufces for being in one of them. The problem with this answer is that it doesn’t show how it isn’t just yoking together two accidentally related conditions, one of which is not genuinely perceptual. The phenomenological answer suggests that the nature of the form of consciousness that occurs when one perceives a truth-maker for p in a good case must include consciousness of oneself as enabled to know p. There is supposed to be a necessary connection imposed by the nature of our kind of phenomenal consciousness. If phenomenal consciousness itself results from a partial act of the capacity for knowledge, then the phenomenological answer to the explanatory question might be a version of the answer that says perceptual experiences intimate their own rational signifcance by being partial acts of the capacity for knowledge. However that turns out, the phenomenological answer suggests a clear diference in the ways shape and motion on the one hand, and rational signifcance on the other, come to be perceptually represented. Perceptual experiences represent shape and motion relatively independently of how they relate to other experiences within a total phenomenal state or stream. Perceptual experiences represent their own rational
198
Elijah Chudnof
Figure 12.1 Kanizsa Triangle.
Figure 12.2 Context Efect.
signifcance largely because of how they relate to other experiences within a total phenomenal state or stream. The diference can be illustrated within perception: You see the top part of the triangular confguration as colored black. You also see it as a partly occluded disc. You see it as colored black independently of how you represent the other parts of the confguration. You see it as a partly occluded disc because of how you represent the other parts of the confguration. They are required for producing the illusory triangle which appears to partly occlude the disc. Without them, it would look like a disc with a wedge cut out. On the right, you see two shapes, each consisting of two approximate verticals connected by a horizontal. One looks like an H, the other looks like an A. You see them as two approximate verticals
Veridical Perceptual Seemings
199
connected by a horizonal independently of how you represent their neighbors. You see one as an H and one as an A because of how you represent their neighbors. If Husserl’s modes of validity are genuine aspects of perceptual states, then they are aspects that accrue to perceptual states via their participation in larger states, or streams, of consciousness. The foregoing suggests that the phenomenological answer to how perceptual seemings come to include epistemically loaded self-consciousness requires additional theoretical backing. For example, it depends on generalizing theories of gestalt efects that occur within perceptual experiences to theories of gestalt-like efects that occur between perceptual experiences and other experiences. Aron Gurwitsch pursues this project in his Field of Consciousness (reprinted in Gurwitsch 2009). The methodology of phenomenology, however, is out of step with the times, and further developments should seek integration with empirically grounded theories of consciousness. There are reasons for optimism on this front. First, empirical work on gestalt efects and holistic representations in perception at diferent levels of processing and across diferent sensory modalities is part of the psychological mainstream (cf. Wagemans 2015). Models developed there provide well worked out starting points for generalization. Second, theories that accord structural organization an explanatory role have already proved fruitful in some extra-perceptual domains, for example in the domain of insight problem solving (cf. Ohlsson 1984; Knoblich et al. 1999). Third, and fnally, a recent variety of higher order thought theory of consciousness called perceptual reality monitoring suggests something like Husserlian modalization plays a role in making sensory representations phenomenally conscious (cf. Lau 2022), giving reason to think the idea has some neurobiological plausibility. I leave further consideration of the matter to another occasion. 12.4.
Why Bother?
According to the Maximal View, a belief that p might amount to knowledge that p because it is based on a fully veridical perceptual experience as of p. Defending the Maximal View depends on assigning veridicality conditions to perceptual experiences that rule out environmental luck and epistemic defeat. Grounds for such an assignment can be found in ideas about perceptual phenomenology with some introspective plausibility, some historical precedent, and some empirical promise, but that lie well outside the mainstream of analytic epistemology. What motivation is there for pursuing this line of thought? The basic observation motivating the Maximal View is that ordinary knowledge in ordinary circumstances can consist in simply taking the
200
Elijah Chudnof
world as it presents itself for granted. You do not need to do anything special to know about your ordinary environment, you just need to take your perceptual experiences at face value. One might aim to accommodate the observation by saying that the basis for belief in such cases is fully veridical perception while denying that it is a belief’s being so based that makes it amount to knowledge. The basis is one factor, the conditions making the belief amount to knowledge include additional factors. This would imply that people generally don’t know how they know what they do know, that “I see a red pepper right there,” does not fully answer the question, “How do you know the store has red peppers?” This might be a defensible view, but it is just as dependent on theoretical commitments as the Maximal View. To my mind, pursuing the theoretical commitments of the Maximal View is a worthwhile endeavor. It demands fuller exploration of the phenomenology of seemings. This can shed light on their suitability to playing epistemic roles aside from the one articulated in the Maximal View. For example, even if you think they are just prima facie justifers, you will want some conception of what they are like that makes sense of this fact, and defationary conceptions on which they are inclinations to believe or sui generis somethings other than sensations fall fat. Notes 1 Some writers distinguish perceptual experiences from something else they call perceptual seemings. There are disagreements about what those are. For some options, see the papers in (Tucker 2013). I think drawing the distinction in the frst place is ill-advised. For critical discussion and additional references to the literature, see (Chudnof and DiDomenico 2015). 2 (McCain and Moretti 2021) is a recent book-length treatment that weighs in on the issues I mention and contains references to relevant literature. 3 One might challenge the idea that my basis for belief could be having a veridical perceptual experience. It could be having a perceptual experience, which experience might be veridical, but the identity of my basis does not include its veridicality since I am not appropriately sensitive to this aspect of the experience. I’m setting this issue aside. 4 It also depends on assuming that the ground of a fact must be a sufcient condition for that fact to be obtained. I explore ways of defending a version of the Maximal View that depend on rejecting this assumption in (Chudnof 2011, 2013). The aim here is to explore another option that is more in line with mainstream views about grounding (cf. Bliss and Trogdon 2021). 5 It is a nice question to ask what strength of necessity is involved in the claims I’ll be considering. They are not conceptually necessary. Their advocates may have taken them to be metaphysically necessary or grounded in the essences of the basic kinds fguring in them. I suspect meeting the explanatory questions raised by the Maximal View only requires that they hold with the strength of true generics about those kinds, but I set this issue aside here. 6 Toward the end of this section, I briefy consider whether the ideas drawn from the phenomenological tradition can be integrated with empirically grounded theories of consciousness. There is room for optimism.
Veridical Perceptual Seemings
201
References Bliss, R., & Trogdon, K. (2021). Metaphysical grounding. In The Stanford Encyclopedia of Philosophy (Winter 2021 Edition), Edward N. Zalta (ed.), URL = . Chudnof, E. (2011). What should a theory of knowledge do? Dialectica, 65(4), 561–579. Chudnof, E. (2013). Intuition. Oxford University Press. Chudnof, E., & Didomenico, D. (2015). The epistemic unity of perception. Pacifc Philosophical Quarterly, 96(4), 535–549. Gallagher, S., & Zahavi, D. (2021). Phenomenological approaches to self-consciousness. In The Stanford Encyclopedia of Philosophy (Spring 2021 Edition), Edward N. Zalta (ed.), URL = . Gurwitsch, A. (2009). The Collected Works of Aron Gurwitsch (1901–1973): Volume III: The Field of Consciousness: Theme, Thematic Field, and Margin. Springer Science & Business Media. Husserl, E. (2001). Analyses Concerning Passive and Active Synthesis Lectures on Transcendental Logic. Springer Verlag. Knoblich, G., Ohlsson, S., Haider, H., & Rhenius, D. (1999). Constraint relaxation and chunk decomposition in insight problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25(6), 1534. Lau, H. (2022). In Consciousness We Trust: The Cognitive Neuroscience of Subjective Experience. Oxford University Press. McCain, K., & Moretti, L. (2021). Appearance and Explanation: Phenomenal Explanationism in Epistemology. Oxford University Press. McDowell, J. (2018). Perceptual experience and empirical rationality. Analytic Philosophy, 59(1): 89–99. Ohlsson, S. (1984). Restructuring revisited: II. An information processing theory of restructuring and insight. Scandinavian Journal of Psychology, 25(2), 117–129. Siegel, S. (2007). How can we discover the contents of experience? The Southern Journal of Philosophy, 45(S1), 127–142. Tucker, C. (Ed.). (2013). Seemings and Justifcation: New Essays on Dogmatism and Phenomenal Conservatism. Oxford University Press. Wagemans, J. (Ed.). (2015). The Oxford Handbook of Perceptual Organization. Oxford University Press.
13 Perceptual Seemings and Perceptual Learning Harmen Ghijsen
13.1.
Introduction
Many philosophers have recently become more sensitive to the importance of perceptual learning in debates about perceptual justifcation (Brogaard & Gatzia, 2018; Chudnof, 2018b; Jenkin, 2022; McGrath, 2017, 2018; Pace, 2017; Vaassen, 2016) and cognitive penetration (Arstila, 2016; Cecchi, 2014; Connolly, 2014, 2019). Here I want to focus on how the phenomenon of perceptual learning, understood as long-term changes in perception due to practice or experience (Gibson, 1963; Connolly, 2019), can impact the debate about what I will call perceptual dogmatism, the view that if it perceptually seems to S that P, then S thereby has immediate prima facie perceptual justifcation for the belief that P.1 Many adherents of perceptual dogmatism have felt it necessary to distinguish between perceptual sensations and perceptual seemings to accommodate apparent problems for the view, such as (1) the problem of non-inferential perceptual identifcation (or the novice/expert problem), (2) the problem of the speckled hen, and (3) the problem of cognitive penetration.2 In contrast, I aim to demonstrate that at least some of these problems can be solved by paying closer attention to the phenomenon of perceptual learning, thus undercutting an important part of the motivation for a distinction between perceptual sensation and perceptual seeming.3 However, once we unpack what goes on in perceptual learning, two important questions about perceptual dogmatism resurface: (1) can it adequately explain the justifcatory force of perceptual phenomenology—what it is in virtue of which perceptual seemings are capable of generating immediate justifcation, and (2) can it really support the immediacy of perceptual justifcation? The plan for this chapter is as follows. In Section 13.2, I introduce perceptual dogmatism in more detail and explain the epistemic motivations for distinguishing between perceptual sensations and perceptual seemings. In Section 13.3, I elaborate on the notion of perceptual learning and argue that this notion undercuts some of the epistemic motivation for DOI: 10.4324/9781003299349-17
Perceptual Seemings and Perceptual Learning
203
distinguishing between perceptual sensations and perceptual seemings. In Section 13.4, I outline the two problems that resurface for perceptual dogmatism and discuss some responses. 13.2.
Perceptual Dogmatism and the Sensation-Seeming Distinction
According to perceptual dogmatism, if it perceptually seems to S that P, then S thereby has immediate prima facie perceptual justifcation for the belief that P. This view is in part motivated by the thought that it allows for having a specifc type of accessible reasons—this is the role of perceptual seemings—that provide a type of justifcation strong enough to support ordinary beliefs about the world without epistemically depending on one’s justifcation for other beliefs—this is the purported immediacy of the provided justifcation. Of course, this justifcation can still be defeated by further considerations (hence the prima facie clause), but if everything goes well, perceptual seemings will deliver ordinary perceptual beliefs (such as that there is a table in front of me) that are justifed merely by having the relevant seemings. Almost all perceptual dogmatists will agree that it’s in virtue of a specifc type of phenomenology that perceptual seemings are able to provide this immediate justifcation. This is nicely in line with the ambition to ground justifcation in features that are completely accessible to the subject: if phenomenology is what matters for immediate perceptual justifcation, then this justifcation-providing feature will be accessible to the subject of the experience. Although it’s disputed how we should characterize the relevant type of phenomenology exactly, the general idea is that there is something specifc about the way in which perceptual seemings present their contents to their subjects that grounds their specifc justifcatory power.4,5 Some perceptual dogmatists have argued that we should make further distinctions with regard to perceptual seemings in order to prevent counterexamples to their preferred view (e.g., Brogaard, 2013; Reiland, 2015; Tucker, 2010). In particular, a distinction between perceptual seemings and perceptual sensations is supposed to help with (1) the problem of noninferential perceptual identifcation, (2) the problem of the speckled hen, and (3) the problem of cognitive penetration. I’ll briefy go over these three problems in turn, and explain why the proposed distinction between sensations and seemings is supposed to alleviate some of the worries. Let’s start with the problem of non-inferential perceptual identifcation. In many cases of perception, it appears that we are able to “just see” (McGrath, 2017) that something is, say, an avocado, a pine tree, or a great spotted woodpecker, without having to infer this belief from other mental
204
Harmen Ghijsen
states. However, there is a clear diference between perceivers with regard to which objects they are able to identify in this non-inferential way: for instance, people unfamiliar with birds will not be able to identify a great spotted woodpecker at a glance. Yet it is not immediately clear in what sense the perceptual seeming of an expert birdwatcher difers from the perceptual seeming of a novice. Don’t they have the same perceptual representation of the bird in front of them? One type of response to this challenge maintains that we should clearly distinguish between perceptual sensations and perceptual seemings to explain the similarities and diferences between novice and expert. Although novice and expert both have the same perceptual sensation, the same visual image, of the bird in front of them, only the expert will also have a perceptual seeming that the bird is a great spotted woodpecker (Brogaard, 2013, pp. 283–284; Tucker, 2010, pp. 537–538; Reiland, 2015, p. 517). This will mean that only the expert has the right kind of mental state to provide them with immediate justifcation for the belief that the bird is a great spotted woodpecker. The novice will, at best, only have immediate justifcation for the belief that there is a bird in front of them because of a corresponding perceptual seeming. Similar to this response to the problem of non-inferential perceptual identifcation, the distinction between perceptual sensation and perceptual seeming is also thought to help with the problem of the speckled hen (Chisholm, 1942; BonJour & Sosa, 2003). Here, the challenge for perceptual dogmatism is to explain how perception can immediately justify us in believing that a three-speckled hen has three speckles, even though it cannot immediately justify us in believing that a 47-speckled hen has 47 speckles. Are the perceptual representations in these cases not relevantly similar? Again, the response on behalf of perceptual dogmatism can make use of the idea that there is a relevant similarity between perceiving the threespeckled hen and the 47-speckled hen, even though there also is a crucial diference. The similarity consists in the role of perceptual sensations: in both cases there is a perceptual sensation, a visual image, that represents the hen and its number of speckles. However, in the case of perceiving the 47-speckled hen, we are simply unable to have a perceptual seeming with the content that the hen has 47 speckles (Brogaard, 2013, pp. 282–283; Tucker, 2010, pp. 534–535). One might put this in terms of conceptualizing the content of the perceptual sensation: even though the perceptual sensation nonconceptually represents all 47 speckles, we are not able to conceptualize this content to form the relevant seeming due to our cognitive limitations. In the absence of this perceptual seeming, there is no reason to suppose that we can have immediate justifcation for the belief that the hen has 47 speckles—even if we have all of those speckles in view.
Perceptual Seemings and Perceptual Learning
205
That leaves us with the third problem for perceptual dogmatism: that of cognitive penetration. This problem challenges perceptual dogmatists to account for the idea that inappropriately caused perceptual seemings cannot provide immediate justifcation for corresponding perceptual beliefs. To illustrate this problem, just imagine that the above novice perceiver of the great spotted woodpecker suddenly acquires a perceptual seeming that the bird is a spotted woodpecker on the basis of wishful thinking.6 Such an inappropriately caused seeming, cognitively penetrated by one’s desires, appears unable to provide immediate justifcation, but how could a perceptual dogmatist explain this? According to Brogaard (2013, pp. 280–281), one way of dealing with cases like this is to claim that the relevant perceptual seemings are not properly grounded in the relevant perceptual sensations. The seemings are simply too insensitive to the contents of the sensations (as they would follow just as easily from sensations of pileated woodpeckers rather than great spotted woodpeckers) to be able to provide immediate justifcation.7 Thus, the distinction between sensations and seemings can also be wielded to locate where things go wrong in bad cases of cognitive penetration: the conceptualization step from perceptual sensation to perceptual seeming has to be performed properly in order for the seeming to be able to provide immediate justifcation. Although much more can be said about each of these three problems, they do seem to provide some epistemic motivation for introducing a distinction between perceptual sensations and perceptual seemings. Most proponents of this distinction also seem to agree that both sensations and seemings are necessary for providing immediate justifcation, as sensations can then be said to provide the qualitative awareness of the states of afairs that seemings present to be the case (e.g., Reiland, 2015; Berghofer, 2020). What’s more, by requiring the presence of both sensations and seemings for generating immediate perceptual justifcation, there will be no possibility of having “blind seemings” that provide immediate justifcation, such as those that might be present in cases of unconscious perception.8 Without a perceptual sensation to properly ground a perceptual seeming, immediate justifcation will simply not get of the ground. 13.3. Perceptual Learning and Its Consequences 13.3.1.
Perceptual Learning
So far we have been going along with the idea that the distinction between perceptual sensation and perceptual seeming can be useful in alleviating some of the worries for perceptual dogmatism, most notably those related to the problem of non-inferential perceptual identifcation, the problem of
206
Harmen Ghijsen
the speckled hen, and the problem of cognitive penetration. In this section, we will look more closely at the phenomenon of perceptual learning, drawing on the relevant work by Kevin Connolly (2019), to see to what extent we really need the distinction between seeming and sensation to counter the relevant problems. The upshot is that perceptual learning undermines the idea that there is a neutral “image”-like sensation that gives rise to an interpretive perceptual seeming.9 Let’s start by getting a better handle on what the phenomenon of perceptual learning consists in. According to Connolly (2019), perceptual learning is best understood as involving (1) long-term changes (2) in perception due to (3) practice or experience. I’ll briefy go over these three aspects of perceptual learning in turn. First, the changes resulting from perceptual learning are long-term. This is meant to distinguish genuine perceptual learning from mere perceptual adaptation efects, where prolonged exposure to a certain type of stimulus briefy afects a subject’s perception of other stimuli (such as having after-images). It’s also meant to underline that genuine perceptual learning should be distinguished from merely perceiving a certain stimulus a bit diferently on occasion: if genuine perceptual learning has occurred, then stimuli of a certain type will automatically appear diferently for a long period of time. Second, the changes resulting from perceptual learning are genuinely perceptual rather than cognitive. Connolly (2019, pp. 48–58) supports this thesis by presenting a collection of neuroscientifc and behavioral evidence that points in this direction.10 To briefy summarize the neuroscientifc evidence: frst, there are diferences in early visual processing between novices and experts; second, patterns of activation within the primary visual cortex change after perceptual learning has taken place; and third, disrupting these patterns of activation (by means of transcranial magnetic stimulation) leads to decreased performance on the classifcation of trained stimuli. In combination with behavioral evidence that shows that perceptual learning can fail to generalize across stimulus orientation, retinal position, retinal size, and across eyes (Sinha & Poggio, 2002, p. 275 as cited in Connolly, 2019, p. 58), there is good reason to think that the changes resulting from perceptual learning are genuinely perceptual rather than cognitive. Third, the changes from perceptual learning result from prolonged practice with, or experience of, certain types of perceptual stimuli. This practice can take place either supervised or unsupervised: mere exposure to diferent kinds of stimuli can already be sufcient to learn to distinguish between those kinds on the basis of their appearances, but supervised learning can certainly speed up the process (by, for instance, helping you to attend to those features that are indicative of belonging to a certain kind). The relevant contrast here is with cases of long-term perceptual change that are
Perceptual Seemings and Perceptual Learning
207
due to other factors than prolonged experience of certain stimuli, such as eye injuries or brain lesions. With this understanding of perceptual learning in mind, which types of examples of perceptual learning can we distinguish? Connolly (2019, pp. 20–27) distinguishes between three diferent types: diferentiation, unitization, and attentional weighting. Diferentiation and unitization both change the perceptual objects of one’s experience, albeit in diferent directions. In cases of diferentiation, what frst phenomenally appears to subjects as single “objects,”11 later appears as multiple objects. In contrast, in cases of unitization, what frst phenomenally appears to subjects as multiple objects, later appears as distinct parts of a certain complex whole. An example of the former (diferentiation) would be learning the sounds of a foreign language: where one frst has difculty distinguishing between diferent sentences, words, and even phonemes, one later learns to diferentiate between these diferent items in the language. An example of the latter (unitization) would be learning to perceive chess confgurations: where beginners only perceive individual pieces on the board, experts are able to perceive the complex whole of a chess confguration. Especially relevant for our purposes are cases of attentional weighting. In those cases of perceptual learning, how an object phenomenally appears changes due to an increase (and/or decrease) in attention to specifc perceptual features of those objects. Connolly (2019, pp. 24–25) mentions cases in sports, where expert soccer players, goalkeepers, and fencers attend to diferent parts of their opponent’s bodies in comparison to non-experts. But, as Connolly argues, it also appears to be exactly what is going on in the case where one learns to distinguish pine trees from other trees (Siegel, 2006, 2010), or, perhaps more to our point, where one learns to distinguish great spotted woodpeckers from other birds. By learning to attend to those features of great spotted woodpeckers that are indicative of their specifc kind, great spotted woodpeckers gain a specifc phenomenal appearance by means of which they can be automatically distinguished from other birds. This means that great spotted woodpeckers will actually look diferent to experts than they do to novices.12 13.3.2.
Consequences for Perceptual Dogmatism
This last point speaks immediately to the earlier problem of non-inferential perceptual identifcation. Recall that the problem for perceptual dogmatism was to explain the diference between novice and expert perceiver, such that only the expert comes out as having immediate justifcation for the belief that the bird is a great spotted woodpecker. The earlier response invoking the distinction between perceptual sensation and perceptual seeming claimed that although novice and expert might have the same perceptual sensation, only the
208
Harmen Ghijsen
expert has a perceptual seeming that the bird is a great spotted woodpecker. But with the idea that perceptual learning, and more specifcally, attentional weighting, can lead to genuine perceptual changes, perceptual dogmatism is no longer forced to make this distinction. Instead, it can simply maintain that the relevant perceptual seemings (in the sense of perceptual experiences) of novice and expert are diferent due to perceptual learning, without having to introduce the metaphysically suspect notion of a neutral visual image (the perceptual sensation) that is the same for both novice and expert. The moral of this solution to the problem of non-inferential perceptual identifcation also helps to deal with the problem of the speckled hen. Recall that here the challenge for perceptual dogmatism lies in explaining why perception provides us with immediate justifcation for believing of a three-speckled hen that it has three speckles, and no such justifcation for believing of a 47-speckled hen that it has 47 speckles. The earlier response again invoked the distinction between perceptual sensation and perceptual seeming, claiming that although we might have a perceptual sensation representing 47 speckles, we still don’t have a perceptual seeming that represents that the hen has 47 speckles. However, instead of accepting that we have a perceptual sensation that represents all of the 47 speckles of the hen, we might be more skeptical of the whole idea that such a detailed visual image is present for any normal perceiver. Being a 47-speckled hen might simply not be the kind of thing that perceptual learning could be applied to. Instead of focusing on a specifc number of speckles, our perceptual system seems geared towards the more general feature of numerosity—at least in cases where the relevant numbers are higher than six (Munton, 2021). In line with that idea, our perceptual system could be able to learn to distinguish sets with diferent numerosities without being able to diferentiate each of the specifc elements within those sets. If that view of perception is correct, then perceptual dogmatism can simply accept that our perceptual seemings are far less specifc when we are dealing with perceptual stimuli that have large numerosities (a 47-speckled hen) in comparison to perceptual stimuli with very small numerosities (a three-speckled hen). And, again, it can accept this without accepting that there is any level of perception at which all of the speckles are somehow individually represented. This means that we can, again, eliminate the suspect notion of a visual image that is similarly constructed for three-speckled hens and 47-speckled hens. This leaves the fnal problem for perceptual dogmatism, that of cognitive penetration. Here, the distinction between perceptual sensation and perceptual seeming was thought to be helpful as it allowed the perceptual dogmatist to claim that perceptual seemings need to be properly grounded in perceptual sensations. Thus, the novice birdwatcher’s perceptual seeming that the bird is a great spotted woodpecker was unable to provide
Perceptual Seemings and Perceptual Learning
209
immediate justifcation if it was caused by wishful thinking, as it was then too insensitive to the contents of the relevant perceptual sensation—had the sensation represented a pileated woodpecker, the seeming would still have represented that the bird was a great spotted woodpecker. However, with the phenomenon of perceptual learning on the table, one can also provide a diferent interpretation of what goes on in purported cases of cognitive penetration. First, in the imagined case of the novice birdwatcher, it is unlikely that the phenomenology of their perception changes in exactly the same way that it would have changed had they undergone the expert’s process of perceptual learning. Thus, even though the novice might judge that the bird is a great spotted woodpecker due to wishful thinking, there’s little chance of the novice having a perceptual seeming (in the sense of perceptual experience) with the same phenomenology as that of the expert. That already provides perceptual dogmatism with an important counter to the problem of cognitive penetration. However, there also is a second point that can be made in relation to cognitive penetration: many of the purported empirical cases of cognitive penetration could actually turn out to be cases of perceptual learning (Arstila, 2016; Connolly, 2019). For instance, instead of maintaining that one’s belief about the normal color of a specifc type of object infuences one’s perceptual experience of that object (as in the memory color efect discussed by Fiona Macpherson (2012)), one could just as easily claim that one’s perceptual experience of stereotypically colored objects has changed due to perceptual learning. If the strongest empirical examples of cognitive penetration can be explained as cases of perceptual learning, then the perceptual dogmatist seems within rights to be skeptical of the hypothetical cases of cognitive penetration that are supposed to constitute a problem for the theory. Either these will be cases in which the perceptual seeming is simply absent, or they should somehow be redescribed as cases of perceptual learning—in which case we can accept that the resulting beliefs are justifed on the basis of the relevant perceptual seemings. Thus, again the upshot is that sufcient attention to the phenomenon of perceptual learning provides us with adequate responses to purported problems for perceptual dogmatism without appealing to a distinction between perceptual sensation and perceptual seeming. 13.4. The Problems of Perceptual Learning 13.4.1.
Grounding Justifcatory Force
Although the phenomenon of perceptual learning seems to bolster a specifc version of perceptual dogmatism that interprets perceptual seemings simply as perceptual experiences, it also presents some older worries for
210
Harmen Ghijsen
perceptual dogmatism in a new light. The frst of those worries is about the special justifcatory force of perceptual seemings: what is it about perceptual seemings that allow them to provide immediate justifcation for belief without being in need of any justifcation themselves?13 As mentioned before, many perceptual dogmatists uphold the idea that the special phenomenology of perceptual seemings is what explains their special justifcatory powers. Without going into detail about the specifc phenomenal feature that is supposed to ground this justifcatory power, there is still some reason to be skeptical of the answer in general. This reason stems from what Connolly (2019, p. 79) takes to be a plausible mechanism for perceptual learning: the blind failing model. According to the blind failing model of perceptual learning, perceptual learning takes place by means of random changes to attentional weighting. Once those random changes lead to important perceptual discriminations, they are reinforced and selected. Importantly, Connolly takes the relevant discriminations here to be present at the phenomenal level: if certain random variations in attention lead to better discriminations at the phenomenal level—if random variations lead to perceptual objects’ acquiring distinctive appearances—then those variations will be preserved. But this means that distinctive phenomenal appearances come before having successful recognitional dispositions: it is because an object acquires a distinctive phenomenal appearance that a subject is able to develop a successful recognitional disposition. If this blind failing model of perceptual learning is on the right track, then this seems to spell trouble for perceptual dogmatism. In cases of non-inferential perceptual identifcation, the distinctive appearances in the expert’s perceptual experiences will already be present before that expert actually has the disposition to recognize whatever they are (going to be) an expert on. This means that a lot of the relevant phenomenology is present even before the expert is able to have immediately justifed beliefs about, say, great spotted woodpeckers. But then having a perceptual experience with a specifc type of phenomenology seems insufcient for providing immediate justifcation, in contrast with the thesis of perceptual dogmatism. An obvious response here is to claim that the full justifcation-conferring phenomenology of perceptual seemings is still absent if the expert has not yet learned to make use of the distinctive appearances of great spotted woodpeckers. If the expert has not yet formed a recognitional disposition for great spotted woodpeckers, then they will not have the perceptual seeming that, say, phenomenally presents that the bird is a great spotted woodpecker. The distinctive appearance of great spotted woodpeckers might be present, the “assertiveness” or “presentationality” with regard to the bird’s being a great spotted woodpecker could still be lacking.14 In the absence of a perceptual seeming of this specifc kind, the developing expert
Perceptual Seemings and Perceptual Learning
211
will lack immediate justifcation for the belief that the bird is a great spotted woodpecker. Natural as this response might be, we still need to tread carefully. If what explains the expert’s ability of having a perceptual seeming that phenomenally presents a bird as a great spotted woodpecker is their knowledge that great spotted woodpeckers have this specifc phenomenal appearance, then we seem to lose the immediacy of perceptual justifcation. The justifed perceptual belief that the bird is a great spotted woodpecker now seems to depend on the justifed belief that great spotted woodpeckers look like this. And this was not the picture of immediate perceptual justifcation that perceptual dogmatism wanted to uphold.15 I will get back to this point when we discuss the problem of immediacy in the next subsection. On the other hand, if the expert’s ability to have a perceptual seeming that phenomenally presents a bird as a great spotted woodpecker is due to their development of a recognitional disposition—a kind of know-how perhaps—then phenomenology no longer seems to play a crucial role in explaining the special justifcatory force of perceptual seemings. If anything, this suggestion seems to move perceptual dogmatism in a reliabilist direction: perceptual seemings are able to provide immediate justifcation because they are the result of a process in which perceivers come to be reliable identifers of certain types of objects. There thus appears to be a challenge for perceptual dogmatism to explain why, if Connolly’s blind failing model of perceptual learning is correct, phenomenology is still crucial in explaining perception’s ability to provide immediate justifcation. 13.4.2.
The Immediacy of Perceptual Justifcation
Let’s now return to the problem of the immediacy of perceptual justifcation that we touched upon briefy in the previous subsection. If perceptual learning somehow involves connecting beliefs about appearances to beliefs about the identity of objects, then the justifcation of the latter might depend on the justifcation of the former.16 In that case, there is no immediate justifcation to be had from perceptual seemings—or at least not for the run-of-the-mill type of beliefs that motivated perceptual dogmatism in the frst place, such as the belief that there is a table in front of me. However, the perceptual dogmatist could make use here of the distinction between psychological dependence and epistemic dependence. Before I can form any beliefs at all, I need to acquire all kinds of beliefs about concepts and their meanings. For instance, I cannot form the belief that there is a table in front of me if I don’t have any beliefs about what tables are and what they look like. But this kind of psychological dependence between beliefs does not yet show that they also epistemically depend on each other.
212
Harmen Ghijsen
Although there is something to be said for this point, we might still contend that the epistemic dependence between beliefs about appearances and beliefs about object-identities is an inference to the best explanation (McGrath, 2018). Even in cases of primary qualities, such as colors, the justifcation of beliefs about what those colors look like will correlate with the justifcation of beliefs about the presence of that color. For instance, take the following set of cases by Michael Pace (2017, pp. 10–11): Periwinkle mix-up: A web designer acquires concepts of the web colours “periwinkle” and “powder blue” by learning their defnitions (given in terms of the RGB triples used to program them). She learns this while using a computer monitor that consistently displays the colour inaccurately: it displays a colour that looks powder blue rather than periwinkle. She has ample reason to trust the monitor’s colour accuracy and no way to detect the mistake. Very soon, she becomes adept at recognizing the look of the colour by sight. Accidental periwinkle recognition: An epistemically defective learning process leads the web designer, for no good reason, to associate her concept of periwinkle with what happens to be the correct look. In the case of periwinkle mix-up, the web designer appears to have a justifed belief that, for example, this color is periwinkle, even though that belief is false. In contrast, in the case of accidental periwinkle recognition, the web designer appears to have an unjustifed belief that this color is periwinkle, even though that belief is true. These facts can both be explained by positing that there is an epistemic dependence between the belief about the look of periwinkle and the belief about color-categorization. If the looks-belief is justifed, then the belief about color-categorization is also justifed; if the looks-belief is unjustifed, then the belief about color-categorization is also unjustifed. Although perceptual dogmatists might still want to maintain that there is no genuine epistemic dependence here, they will then have to meet the challenge of explaining the justifcatory correlations that are present in cases like this.17 13.5.
Conclusion
I have argued that the phenomenon of perceptual learning can undercut some of the epistemic motivation for introducing a distinction between perceptual sensations and perceptual seemings. The problem of non-inferential perceptual identifcation, the problem of the speckled hen, and the problem of cognitive penetration can all be dealt with by paying more attention to perceptual learning. With respect to the frst two problems, we need not introduce perceptual sensations to maintain that the relevant
Perceptual Seemings and Perceptual Learning
213
perceptual seemings are diferent. With respect to the third, we can simply deny that purported cases of cognitive penetration revolve around perceptual seemings that lead to intuitively unjustifed beliefs, in which case there is no need to maintain that perceptual seemings need to be properly grounded in perceptual sensations. However, even if all of this is true, perceptual learning still (re)introduces some familiar worries for perceptual dogmatism. First, if the blind failing model of perceptual learning is correct, then objects’ distinctive appearances are already present before immediate justifcation can occur, undermining the idea that perception’s special justifcatory force could stem from that aspect of perceptual phenomenology. Second, if perceptual learning also involves acquiring beliefs about the typical appearances of objects, then this puts pressure on the idea that perceptual seemings can provide immediate justifcation. The phenomenon of perceptual learning thus provides fertile ground for further discussion of perceptual dogmatism. Acknowledgments Thanks to Kevin McCain, my colleagues at the Center for Cognition, Culture and Language, and the participants of the Rhein-Ruhr Epistemology Meeting 2022 for their helpful comments and suggestions. Notes 1 Versions of this view have been defended by (Berghofer, 2020; Brogaard, 2013; Chudnof, 2012; Huemer, 2001, 2007; Kriegel, 2021; McGrath, 2013; Moretti, 2015, 2020; Reiland, 2015; Tucker, 2010), though many of these focus on seemings in general rather than just perceptual seemings, and many also add further restrictions on the type of phenomenology that should accompany those seemings. 2 All these problems (and more) are discussed in (Tucker, 2010), though Tucker doesn’t use the third problem to motivate the distinction. 3 In (Ghijsen, 2015), I discussed the frst two problems and argued that a richer notion of perceptual content could solve them. Here, I take a closer look at what could explain how such richer content comes about: perceptual learning. 4 Some proposals about the relevant type of phenomenology are that perceptual seemings have (1) a phenomenology of assertiveness (Tucker, 2010), assuring us of the truth of their contents; (2) a phenomenology of presentationality (Bengson, 2015), presenting to us that their contents are true; (3) a presentational phenomenology (Chudnof, 2012, 2018a), representing that P and making us seemingly aware of a truthmaker for P; (4) a phenomenology of felt evidence insensitivity (Brogaard & Gatzia, 2018), making it feel to us as if the experience would remain even in the face of defeating evidence; or (5) a phenomenology of givenness (Berghofer, 2020), seemingly making us aware of the object of experience. 5 But see Ghijsen (2014), Ghijsen (2016a, pp. 128–131), and Teng (2022) for the idea that the relevant phenomenal feature might not actually be perceptual.
214
Harmen Ghijsen
6 A similar case can be found in (Markie, 2005). 7 Matt McGrath (2013) also provides a view according to which the problem with cases of cognitive penetration has to do with high-level seemings whose contents have not been properly (quasi-)inferred from the low-level seemings on which they were based. 8 But see (Tucker, 2010) and (Huemer, 2013) for the thought that blind seemings can also provide immediate justifcation. 9 Cf. Tucker (2010, p. 530). 10 Connolly also cites the introspective evidence of a multitude of philosophers to support this point, but such evidence appears pretty thin: why expect that anyone (expert philosopher or otherwise) can really distinguish between perceptual diferences and cognitive diferences? 11 I’m using “objects” here in a broad sense to also encompass, say, features and events. 12 Also see (Landers, 2021) for the idea that perceptual learning leads to experts having diferent visual experiences. But note that Landers cashes this out in terms of experts’ experiences representing gestalt properties. 13 See, for example, (Ghijsen, 2014, 2016b; Lyons, 2009) for versions of this problem. 14 See footnote 4 for some suggestions of what this further aspect of experience comes down to. 15 But see (McGrath, 2017, 2018) for a “looks view” of perceptual justifcation that is a lot like this. 16 But see (Ghijsen, 2020) for some reasons to be skeptical of this idea. 17 See (McCain & Moretti, this volume) for a dogmatist account that takes on this challenge.
References Arstila, V. (2016). Perceptual learning explains two candidates for cognitive penetration. Erkenntnis, 81(6), 1151–1172. https://doi.org/10.1007/s10670-0159785-3 Bengson, J. (2015). The intellectual given. Mind, 124, 707–760. Berghofer, P. (2020). Towards a phenomenological conception of experiential justifcation. Synthese, 197, 155–183. https://doi.org/10.1007/s11229-0181744-5 BonJour, L., & Sosa, E. (2003). Epistemic justifcation: Internalism vs. externalism, foundations vs. virtues. Malden, MA: Blackwell Pub. Brogaard, B. (2013). Phenomenal seemings and sensible dogmatism. In C. Tucker (Ed.), Seemings and justifcation: New essays on dogmatism and phenomenal conservatism (pp. 270–289). New York: Oxford University Press. Brogaard, B., & Gatzia, D. E. (2018). The real epistemic signifcance of perceptual learning. Inquiry, 61(5–6), 543–558. https://doi.org/10.1080/00201 74X.2017.1368172 Cecchi, A. S. (2014). Cognitive penetration, perceptual learning and neural plasticity. Dialectica, 68(1), 63–95. https://doi.org/10.1111/1746-8361.12051 Chisholm, R. (1942). The problem of the speckled hen. Mind, 51(204), 368–373. https://doi.org/10.1093/mind/LI.204.368 Chudnof, E. (2012). Presentational phenomenology. In S. Miguens & G. Preyer (Eds.), Consciousness and subjectivity (pp. 51–72). Frankfurt: Ontos Verlag.
Perceptual Seemings and Perceptual Learning
215
Chudnof, E. (2018a). Epistemic elitism and other minds. Philosophy and Phenomenological Research, 96(2), 276–298. https://doi.org/10.1111/phpr.12308 Chudnof, E. (2018b). The epistemic signifcance of perceptual learning. Inquiry: An Interdisciplinary Journal of Philosophy, 61(5–6), 520–542. https://doi.org/1 0.1080/0020174x.2017.1357496 Connolly, K. (2014). Perceptual learning and the contents of perception. Erkenntnis, 79(6), 1407–1418. https://doi.org/10.1007/s10670-014-9608-y Connolly, K. (2019). Perceptual learning: The fexibility of the senses. Oxford: Oxford University Press. Ghijsen, H. (2014). Phenomenalist dogmatist experientialism and the distinctiveness problem. Synthese, 191(7), 1549–1566. https://doi.org/10.1007/s11229-0130348-3 Ghijsen, H. (2015). Grounding perceptual dogmatism: What are perceptual seemings? Southern Journal of Philosophy, 53(2), 196–215. https://doi.org/10.1111/sjp.12104 Ghijsen, H. (2016a). The puzzle of perceptual justifcation. Cham, Switzerland: Springer. Ghijsen, H. (2016b). The real epistemic problem of cognitive penetration. Philosophical Studies, 173(6), 1457–1475. https://doi.org/10.1007/s11098-015-0558-2 Ghijsen, H. (2020). Do looks constitute our perceptual evidence? Philosophical Issues, 30(1), 132–147. https://doi.org/10.1111/phis.12176 Gibson, E. J. (1963). Perceptual learning. Annual Review of Psychology, 14(1), 29–56. Huemer, M. (2001). Skepticism and the veil of perception. Lanham, MD: Rowman & Littlefeld Publishers. Huemer, M. (2007). Compassionate phenomenal conservatism. Philosophy and Phenomenological Research, 74(1), 30–55. Huemer, M. (2013). Phenomenal conservatism über alles. In C. Tucker (Ed.), Seemings and justifcation: New essays on dogmatism and phenomenal conservatism (pp. 328–350). New York: Oxford University Press. Jenkin, Z. (2022). Perceptual learning and reasons-responsiveness (online frst). Noûs, 1–28. https://doi.org/10.1111/nous.12425 Kriegel, U. (2021). The structure of phenomenal justifcation. Australasian Journal of Philosophy, 1–16. https://doi.org/10.1080/00048402.2021.1978513 Landers, C. (2021). Specialized visual experiences. The Philosophical Quarterly, 71(1), 74–98. https://doi.org/10.1093/pq/pqaa018 Lyons, J. (2009). Perception and basic beliefs: Zombies, modules and the problem of the external world. Oxford: Oxford University Press. Macpherson, F. (2012). Cognitive penetration of colour experience: Rethinking the issue in light of an indirect mechanism. Philosophy and Phenomenological Research, 84(1), 24–62. https://doi.org/10.1111/j.1933-1592.2010.00481.x Markie, P. (2005). The mystery of direct perceptual justifcation. Philosophical Studies, 126(3), 347–373. https://doi.org/10.1007/s11098-004-7795-0 McCain, K., & Moretti, L. (this volume). Phenomenal explanationism and the look of things. In K. McCain, S. Stapleford, & M. Steup (Eds.), Seemings: New arguments, new angles. New York: Routledge. McGrath, M. (2013). Phenomenal conservatism and cognitive penetration: The “bad basis” counterexamples. In C. Tucker (Ed.), Seemings and justifcation (pp. 225–247). New York: Oxford University Press.
216
Harmen Ghijsen
McGrath, M. (2017). Knowing what things look like. Philosophical Review, 126(1), 1–41. https://doi.org/10.1215/00318108-3683602 McGrath, M. (2018). Looks and perceptual justifcation. Philosophy and Phenomenological Research, 96(3), 110–133. https://doi.org/10.1111/phpr.12289 Moretti, L. (2015). Phenomenal conservatism. Analysis, 75(2), 296–309. https:// doi.org/10.1093/analys/anu153 Moretti, L. (2020). Seemings and epistemic justifcation: How appearances justify beliefs. Cham, Switzerland: Springer. Munton, J. (2021). Visual indeterminacy and the puzzle of the speckled hen. Mind & Language, 36(5), 643–663. https://doi.org/10.1111/mila.12296 Pace, M. (2017). Experiences, seemings, and perceptual justifcation. Australasian Journal of Philosophy, 95(2), 226–241. https://doi.org/10.1080/00048402.201 6.1202991 Reiland, I. (2015). Experience, seemings, and evidence. Pacifc Philosophical Quarterly, 96(4), 510–534. https://doi.org/10.1111/papq.12113 Siegel, S. (2006). Which properties are represented in perception? In T. S. Gendler & J. Hawthorne (Eds.), Perceptual experience (pp. 481–503). New York: Oxford University Press. Siegel, S. (2010). The contents of visual experience. Oxford: Oxford University Press. Sinha, P., & Poggio, T. (2002). High-level learning of early visual tasks. In M. Fahle & T. Poggio (Eds.), Perceptual learning (pp. 273–299). Cambridge, MA: MIT Press. Teng, L. (2022). A metacognitive account of phenomenal force. Mind & Language, Online First, 1–21. https://doi.org/10.1111/mila.12442 Tucker, C. (2010). Why open-minded people should endorse dogmatism. Philosophical Perspectives, 24(1), 529–545. Vaassen, B. M. K. (2016). Basic beliefs and the perceptual learning problem: A substantial challenge for moderate foundationalism. Episteme, 13(1), 133–149. https://doi.org/10.1017/epi.2015.58
14 Phenomenal Explanationism and the Look of Things Kevin McCain and Luca Moretti
Matthew McGrath (2018) has challenged all theories that allow for immediate perceptual justifcation, that is, theories that allow that S can have perceptual justifcation for p that is not dependent upon S’s justifcation for believing some other proposition q.1,2 His challenge comes by way of arguing for what he calls the “looks view” of visual justifcation.3 Roughly, the looks view says that our simple visual beliefs are mediately justifed based on the looks of things. What are simple visual beliefs? They are (visual) perceptual beliefs that satisfy two conditions: (1) They are manifestations of a stable disposition to categorize a perceived object as F upon having certain sorts of visual experience, where F may be a sensible quality or a kind of property, and (2) they enjoy the phenomenology of “just seeing that the thing is F.”4 (McGrath 2018: 111) McGrath refers to any view that allows for immediate perceptual justifcation as “dogmatism.”5 Simple visual beliefs are prima facie good candidates for immediate justifcation—or so the dogmatist contends. McGrath’s looks view maintains, in contrast, that when a simple visual belief is justifed, it is always mediately justifed based on the looks of things. What does it mean that S has justifcation for a visual belief on the basis of looks? It means S has the reason—identifed by McGrath with a fact or proposition—that this looks like an F. For example, according to McGrath, when S has justifcation for believing that the thing that she apparently sees is a tree, it is because S has “looks-related reasons” for believing it is a tree—that is, she has it looks like a tree as a reason for her belief. Reasons of this type are—according to McGrath—public with respect to both their content and their possession.6 Furthermore, importantly, S must have justifcation for these reasons in order for them to work as reasons, and their justifcation constitutes, at least partly, S’s justifcation DOI: 10.4324/9781003299349-18
218
Kevin McCain and Luca Moretti
for the relevant simple perceptual beliefs. Hence, S has “looks-related reasons” for believing that the thing she apparently sees is a tree only if S is justifed in believing that that thing looks like a tree, where this justifcation is a component part of S’s justifcation for believing that it is a tree. Specifcally, McGrath (2018: 119) argues for what he calls “The looks view concerning propositional justifcation”: In cases in which one has a justifed simple visual belief that an object is F, one has a looks-related reason to believe that it is F that justifes one in believing that the thing is F. Further, McGrath claims that when our simple visual beliefs are justifed, it is because they are held on the basis of looks-related reasons. In other words, doxastic justifcation for such beliefs is a result of believing on the basis of the sort of propositional justifcation described by the looks view. Thus, the looks view of perceptual justifcation is a full account of the justifcation of our simple visual beliefs—it provides an analysis of both propositional and doxastic justifcation. Here is a sketch of McGrath’s overall argument against immediate justifcation for simple visual beliefs: (1) Simple visual beliefs (when justifed) are mediately justifed. (The looks view is true.) (2) Simple visual beliefs (when justifed) are not justifed both mediately and immediately. (3) Therefore, simple visual beliefs (when justifed) are not justifed immediately. Our plan for this chapter is simple: show that McGrath’s argument is unsound or, at the very least, that (3) is not sufciently well supported to cause genuine concern for dogmatists—at least, there is no problem for the species of dogmatism that we have elsewhere called Phenomenal Explanationism (hereafter “PE”).7 We limit our focus to premise (1). Premise (2) is necessary for McGrath’s argument because the looks view is compatible with the claim that justifed simple visual beliefs are also immediately justifed. (In forming them, we might rely on two distinct bases, one providing mediate and the other immediate justifcation.) While we are not fully convinced by McGrath’s arguments in support of premise (2), they do not seem implausible if one grants that simple visual beliefs when justifed are always justifed mediately by looks-related reasons. The arguments for premise (1), however, are fawed. We proceed by looking at both the motivations McGrath gives for the looks view and how he attempts to defend it from a straightforward
Phenomenal Explanationism and the Look of Things
219
objection. Ultimately, neither the positive motivations for the looks view nor its defense succeeds. In Section 14.1, we argue that considering their respective accounts of defeat gives no reason to prefer the looks view to PE. In Section 14.2 we take on McGrath’s central argument for the looks view and show that it is unsound. We then raise a serious problem for the looks view in Section 14.3 and argue that McGrath’s attempt to avoid this problem fails. In light of the considerations raised in Sections 14.1–14.3, premise (1) of McGrath’s argument appears to lack adequate support. So, McGrath’s case for (3) is unpersuasive. Section 14.4 concludes the chapter. 14.1. Explaining Defeat McGrath begins his defense of the looks view by arguing that it provides good explanations of intuitive examples of defeat. He contends that while dogmatist views can account for the defeat in these cases, their doing so comes at a signifcant dialectical cost—it commits the dogmatist to acknowledging that ordinary agents have “substantial knowledge of the relation between how things are, how they look and what our experiences are” (McGrath 2018: 122). The problem is that once committed to acknowledging this, the dogmatist will be unable to charge the looks view with overintellectualizing perceptual justifcation, which is—as we will see in Section 14.3—a major objection to the looks view. McGrath ofers an example in which he claims that a defeater does its work by way of attacking the relevant “looks” proposition. Here is the example: Oboes8
Suppose I love the sound of an oboe and can recognize it well from recordings and in concerts (though from my cheap seats I can’t usually see the oboes at all well). But suppose I’ve gotten them mixed up, by sight, with clarinets. We can imagine this is due to reading an otherwise reliable book that contains a mislabeled photograph. Show me a clarinet and I’ll think it is an oboe. My mistake isn’t verbal. I do not use “oboe” to mean clarinet. I use it to mean oboe. Now, suppose, you’ve setup an identifcation task for me. You’ve showed me an instrument, about which I had no previous information, and asked me what it is. I declare it is an oboe, expressing my visual belief that it is an oboe, presumably a justifed belief. You, who know the looks of oboes, tell me: “you’ve mixed up clarinets and oboes; this is not what oboes look like.” (McGrath 2018: 120)
220
Kevin McCain and Luca Moretti
McGrath claims that in this case, his justifcation for thinking that the instrument is an oboe is defeated. This seems correct, assuming that McGrath has reason to think that the person who has set up the identifcation task is not purposely trying to trick him, knows what the various instruments look like, and so on. However, as McGrath notes, one might think that this case does not really do what he needs it to do. To involve the looks view in explaining the defeat, McGrath needs a case where information about how things look undercuts the justifcation that one would normally get from a visual experience—that is, he needs an example with an undercutting defeater. But, in this case it is plausible that what McGrath receives is a rebutting defeater—he is given reason to think that what he sees is not an oboe rather than reason to think that how things look to him fails to provide good evidence for thinking the instrument is an oboe. With this in mind, McGrath (2018: 120) ofers a slight modifcation: “add that you preface your remark with ‘This just so happens to be an oboe disguised to look like a clarinet, but.’” Now we have a case where the justifcation provided by McGrath’s visual experience is undercut. (Of course, he receives evidence that he does not really know what oboes look like together with evidence that what he sees is in fact an oboe. So, he may still have non-perceptual justifcation for believing that it is an oboe.) McGrath claims that the defeat occurs in this case because the relevant looks proposition—[this looks like an oboe]—becomes unjustifed. Once he is no longer justifed in believing that this looks like an oboe, he loses his perceptual justifcation for believing that the instrument is an oboe. McGrath suggests that similar considerations apply in instances where one only gains a partial defeater. For example, if one were to learn that “many of the clarinets in this shop have been made to look like oboes,”9 then one’s justifcation for believing that the instrument one is looking at is an oboe will be reduced from what it was prior to learning about the disguised clarinets. How does this and consideration of the previous example support the looks view? McGrath points out that if part of one’s justifcation for the simple visual belief that this is an oboe came from one’s reason that this looks like an oboe, we would expect exactly the sort of defeat that we fnd in these cases. So, the looks view gets the right results in cases where one’s simple visual beliefs are defeated, and it provides an explanation for how the defeaters in these cases do their work. It is important to note that this does not in itself pose a problem for dogmatist views. As McGrath acknowledges, the dogmatist can simply allow that evidence about how things look, or the connection between how things look and how they are, can defeat immediate justifcation too. After all, dogmatists generally allow that immediate justifcation is defeasible. Consequently, dogmatists can claim that McGrath, in the above examples, has immediate perceptual justifcation for believing that what he sees is an
Phenomenal Explanationism and the Look of Things
221
oboe, but once he learns that, say, this is not what oboes look like (even if it is an oboe), his immediate perceptual justifcation is defeated. Yet, as said at the beginning of this section, McGrath contends that if dogmatists respond to his cases in this way, there is a signifcant dialectical cost for them. We disagree. To clarify the reasons for our disagreement, we frst need to address a related but more direct challenge McGrath presents for dogmatism. Drawing from his discussion of Oboes, McGrath emphasizes that it takes cognitive sophistication to be able to use evidence about looks to provide defeat for immediate justifcation. Specifcally, One needs to be able to see that one could easily have had one’s experience while not looking at an oboe when one is informed of the likes of: (a) that’s not what oboes look like (even though it is an oboe); (b) this is a situation in which many non-oboes look like oboes; or (c) here is what an oboe looks like (showing picture). (2018: 121) McGrath rightly notes that most ordinary agents can easily appreciate the power that (a), (b), and (c) have to generate defeat. However, he claims that whereas the looks view “easily” explains this defeat, dogmatist views require “more complicated inferences or background knowledge” (2018: 121). McGrath’s claim seems false, at least if the dogmatist view in question is something like our preferred theory of epistemic justifcation, PE. Among all the forms of dogmatism available in the literature, only PE allows for a clear explanation of defeat.10 Take the sort of case McGrath is interested in—S has a visual experience of an instrument, and she has justifcation for thinking that the instrument is an oboe. PE, roughly, says that S has justifcation for believing that this is an oboe just in case [this is an oboe] is part of the best explanation of her evidence and that explanation is itself suffciently good.11 Assuming that S’s evidence is a presentational appearance (an appearance that apparently presents the truth maker for its content) that this is an oboe, S will have immediate justifcation for believing that this is an oboe.12 Now, what if S receives (a)? Supposing that the evidence for accepting (a) is itself sufciently strong, S will still have justifcation for believing that this is an oboe. This is because (a) includes the claim that the instrument is in fact an oboe. However, in this case it will not be S’s original appearance that is doing the justifying. Rather, whatever gives S justifcation for (a) will give her justifcation for believing that the instrument is an oboe. Given PE, it is easy to see how this sort of defeat works. (a) makes it so that [this is an oboe] is no longer part of the best sufciently good explanation of S’s
222
Kevin McCain and Luca Moretti
appearance. After all, S’s evidence now includes the trustworthy information that appearances of this sort are not typically caused by oboes. So, [this is an oboe] is not part of a good explanation of the visual appearance that S has in this case. It is, however, part of the best sufciently good explanation of S’s overall evidence in this case, which includes both S’s appearance and testimony that this is in fact an oboe. There does not seem to be any complicated inference or signifcant background knowledge at play here. What if S receives (b)? Here the situation is very similar to when S gets (a). Given PE, [this is an oboe] is no longer part of the best sufciently good explanation of S’s appearance. After all, S’s evidence now includes the trustworthy information that appearances of this sort in the situation that S is in are likely caused by something other than oboes. So, [this is an oboe] is not part of the best sufciently good explanation of the visual appearance that S has in this case. The rival explanation provided by the claim that this is an instrument disguised to look like an oboe is as good (or nearly so) as the explanation provided by [this is an oboe]. As a result, S’s justifcation for believing that this is an oboe is signifcantly decreased once she has sufciently strong evidence for (b). Unlike the previous situation where S receives (a), S’s overall evidence does not support believing that this is an oboe because her evidence no longer includes the testimony that the instrument she is looking at is in fact an oboe. In other words, when S receives (b) she only gains a defeater; she does not gain any additional evidence that the instrument is an oboe like she does when she receives (a). Again, there does not seem to be any complicated inference or signifcant background knowledge at play here. What if S receives (c)? As expected, it is again the case that [this is an oboe] is no longer part of the best sufciently good explanation of S’s evidence. Assuming that S has sufciently strong evidence to trust that the picture actually represents what oboes look like, and this is diferent from her original appearance, [this is an oboe] is not part of the best sufciently good explanation of her evidence. S now has evidence that appearances like the one she has when looking at this instrument are unlikely to be caused by oboes. So, a better explanation of her total evidence is that her appearance is caused by some other instrument. Yet again, there does not seem to be any complicated inference or signifcant background knowledge at play here. In conclusion, PE can account for the patterns of defeat that McGrath appeals to without positing any exceedingly complex cognitive machinery. In fact, PE’s account does not seem any more cumbersome than that of the looks view. Of course, this does not show that the looks view is mistaken. Yet since PE can explain the defeat in McGrath’s cases at least as easily as the looks view, these cases provide no reason to prefer the looks view over
Phenomenal Explanationism and the Look of Things
223
PE. Since PE is a kind of dogmatism, McGrath’s defeater cases fail to provide reason to think that we need to posit mediate, rather than immediate, justifcation of simple visual beliefs in order to properly account for what is going on in these sorts of situations. Let us turn to the issue of the dialectic cost of the dogmatist’s explanation of defeaters. McGrath claims that this explanation commits the dogmatist to acknowledging that ordinary agents have substantial knowledge of the relation between “how things are, how they look and what our experiences are.” The above accounts of cases (a)—(c), which rely on PE, cast doubts on McGrath’s claim. Phenomenal explanationists can explain defeat in these situations without attributing to S substantial knowledge of the relation between how things are, how they look, and what her experiences are. What matters is that the new information S gains by coming to possess (a), (b), or (c) changes the quality of the explanations provided by the previously justifed proposition. And this can happen without S possessing the sort of substantial knowledge to which McGrath refers. 14.2.
The Straightforward Argument
Following up on his discussion of defeat, McGrath (2018) ofers what he calls the “straightforward argument” for the looks view. According to McGrath, the dogmatist is forced to respond to what he said about defeaters in a way that “leaves open the possibility that there is an important sort of epistemic dependence relation between the relevant looks propositions and [the] proposition that the thing is an F” (2018: 122). McGrath intends to say that the dogmatist must grant that it is at least possible that simple visual beliefs are mediately justifed by relevant looks propositions. We can concede that this very generic claim is true. (Although we saw that the phenomenal explanationist can account for the defeat that McGrath discusses without adducing this sort of epistemic dependence.) McGrath contends that, once the dogmatist grants this possibility, she will be committed to concluding that simple visual beliefs are actually mediately justifed by relevant looks propositions. In this section we show that this contention is false. Let us begin by getting clear on McGrath’s straightforward argument. Here is his description of it: All this gives us a straightforward argument for the looks view concerning propositional justifcation. Take a case of a justifed simple visual belief that a thing is an F. There will be an appropriate looks proposition that one must be independently justifed in believing in order to be justifed in believing it is an F. Moreover, this looks proposition will support the belief that the object is an F (this support relation will not be
224
Kevin McCain and Luca Moretti
defeated). But if one is independently justifed in believing such a looks proposition, which supports the target proposition that the thing is an F, and this support is undefeated, then one will have a mediate justifcation from a looks-related reason to believe the thing is an F. Extending the same argument to beliefs attributing sensible qualities, we arrive at the looks view concerning propositional justifcation. (2018: 123) The linchpin of McGrath’s argument is the claim that when one has a justifed simple visual belief, there must be “an appropriate looks proposition that one must be independently justifed in believing.” Why does McGrath think there must be an independently justifed looks proposition in order to have justifcation for simple visual beliefs? He appeals to a scenario like Oboes. This time McGrath asks us to suppose that he is justifed from the very beginning in suspending judgment about whether the object that he sees looks like an oboe. McGrath correctly points out that, in both versions of the case, it seems he does not have justifcation for believing that what he sees is an oboe. McGrath elaborates on the second alternative. To make his point explicit, he appropriately emphasizes how strange it would be to say that he forms a justifed simple visual belief that the instrument is an oboe at the same time that he suspends judgment on whether or not the instrument looks like an oboe. McGrath claims that when it comes to his simple visual belief and his suspension of judgment about the corresponding looks proposition, “without the assumption that one must be justifed in a relevant looks proposition in order to be justifed in believing it is an oboe, it is difcult to see why this combination of states would make no sense, i.e., would be irrational” (2018: 122). Thus, McGrath concludes that one must be justifed in believing the relevant looks proposition in order to have justifcation for a simple visual belief. Of course, McGrath has to supplement his argument to get from this to the claim that the looks belief must be independently justifed. However, if he can get to the conclusion that one must be justifed in believing the relevant looks proposition, it is not hard for him to get to the claim that the justifcation must be independent of one’s current visual experience. So, we grant that he can get to the independence claim from the conclusion that believing the relevant looks proposition must be justifed. We do not think that he can get to the conclusion that believing the relevant looks proposition must be justifed, though. Hence, the move to the independence claim is irrelevant. The problem with McGrath’s argument is that it misconstrues a negative requirement as a positive requirement. Let us explain. We agree with McGrath that if Oboes is modifed so that he has justifcation for disbelieving that, or suspending judgment about whether, the instrument
Phenomenal Explanationism and the Look of Things
225
looks like an oboe, he does not have justifcation for his simple visual belief that this is an oboe. This is so because his disbelief or suspending judgment would give McGrath a defeater. That said, it is a mistake to think that we can move from the fact that one’s having justifcation for disbelieving or suspending judgment about the relevant looks proposition provides one with a defeater to the claim that one must have justifcation for believing the looks proposition in order to have justifcation for the simple visual belief. One can readily accept a no-defeater condition without also accepting a corresponding higher-order requirement on justifcation.13 For example, a non-reductionist about testimony might accept that S’s having justifcation for disbelieving that, or suspending judgment about whether, S* is a reliable informant provides a defeater for S*’s testimony without also accepting that S must, therefore, have justifcation for believing that S* is reliable in order for accepting S*’s testimony to be justifed.14 Instead, S can justifedly accept S*’s testimony even if S has no justifcation for adopting any doxastic attitude about whether or not S* is a reliable informant. So, we can grant that having justifcation for disbelieving or suspending judgment on the looks proposition provides a defeater without accepting that in order to have justifcation for a simple visual belief one must have justifcation for believing the relevant looks proposition. That is to say, we can plausibly accept a negative higherorder requirement on justifcation (one cannot have a defeater for the looks proposition) without committing to a corresponding positive highorder requirement (one has to have independent justifcation for believing the looks proposition). In order to make this point clearer, let us consider a couple ways in which McGrath might fail to have justifcation for believing a looks proposition about oboes while having justifcation for believing the related simple visual belief.15 Suppose o is “this looks like an oboe” and p is “this is an oboe.” In each case our goal is to determine (1) if McGrath has or does not have justifcation for adopting any doxastic attitude toward o and (2) whether McGrath’s relation to o negatively afects the justifcation for his simple visual belief that p. First, let us imagine that McGrath has a simple visual belief p caused by a presentational appearance that p and he is just now, for the frst time, considering o. As McGrath is looking for evidence for o, he is interrupted by an emergency that requires his full attention. McGrath’s simple visual belief that p is retained, but he never fnishes his deliberations concerning o, and let us assume this results in his having no available evidence for or against it. McGrath does not have justifcation for believing or disbelieving o, but it is not clear that he has justifcation for suspending judgment about o either. After all, in this case McGrath has no relevant evidence when it comes to o, so it is not unreasonable to think that McGrath is still in a state where no doxastic attitude toward o is justifed.16
226
Kevin McCain and Luca Moretti
In spite of this, it does not seem that McGrath’s justifcation for his simple visual belief that p disappears at any point during this process. Now, let us consider an additional case. Suppose McGrath is afected by a form of amnesia that prevents him from—among other things—retrieving evidence for looks propositions.17 McGrath still has a presentational appearance that p (“this is an oboe”). So, he does have justifcation for believing p. Yet as soon as he tries to substantiate o (“this looks like an oboe”), he cannot fnd any independent evidence for o. McGrath cannot recall any past episode in which something similar to the oboe he is looking at has been identifed as an oboe. He does not even have an appearance that o. In this case too it seems that McGrath does not have justifcation for believing, disbelieving, or suspending judgment about o. Nevertheless, it seems clear that prior to his failure to retrieve memorial evidence for o he has justifcation for believing p.18 In each of these situations, McGrath does not have justifcation for believing o, and yet his simple visual belief that p seems to be justifed. Now, one might complain that in these cases McGrath is not justifed in suspending judgment concerning o, so these are not the sort of situation that McGrath initially described. True, but in both cases, we do have someone who plausibly has justifcation for a simple visual belief that p without being justifed in believing the relevant looks proposition. This, of course, straightforwardly rebuts McGrath’s claim that one must be justifed in accepting the relevant looks proposition in order to have justifcation for the simple visual belief. Furthermore, consideration of these cases helps to illustrate our general point that the fact that having justifcation to suspend judgment on a higher-order proposition might provide a defeater for a lower-order belief does not entail that one must have justifcation for believing the higher-order proposition in order to have justifcation for the lower-order belief. Applied to this particular issue, the key point is that while McGrath is correct that being justifed in disbelieving that, or suspending judgment about, o does defeat his justifcation for the simple visual belief that p, it is a mistake to move from this to the conclusion that one must, therefore, be justifed in believing o in order to have justifcation for the simple visual belief that p. Consequently, the straightforward argument is unsound because its central premise (stating that when one has justifcation for a simple visual belief, there must be “an appropriate looks proposition that one must be independently justifed in believing”) is false. Before moving on, let us briefy address McGrath’s claim that when it comes to his simple visual belief that p and his suspension of judgment about the corresponding looks proposition, without the looks view “it is difcult to see why this combination of states would . . . be irrational.” Appealing to PE is helpful in showing that McGrath is mistaken in this case too. PE provides an explanation of why this combination would be
Phenomenal Explanationism and the Look of Things
227
irrational that is at least as simple as that aforded by the looks view. PE entails that in order for McGrath to have justifcation for believing that p (“this is an oboe”), p has to be part of the best sufciently good explanation of McGrath’s evidence. Plausibly, when he does not have evidence concerning o (“this looks like an oboe”), p can still be the best sufciently good explanation of his evidence. On our view, what is going on in this case is that it appears to McGrath that p. McGrath’s appearance provides him with justifcation for thinking that p because the best sufciently good explanation of its appearing that p includes p. Now, if we add to the case that McGrath is justifed in suspending judgment on o because his evidence for ~o is just as strong as his evidence for o, p is no longer part of the best sufciently good explanation of McGrath’s evidence. After all, once it is the case that McGrath should suspend judgment on o, his evidence includes that he cannot tell that an instrument is an oboe by its appearance—it includes evidence that makes it so that he should not think that this sort of appearance is one of an oboe. The truth of p does not provide a very good explanation of McGrath’s appearance coupled with his evidence that the appearance fails to give him a good reason to believe p. Given this, it is not difcult to see why the combination of suspending judgment on o and believing that p is irrational. 14.3.
Beliefs About Looks
Having taken himself to have shown that the looks view is correct when it comes to propositional justifcation, McGrath (2018) turns toward making the case that the looks view also applies to doxastic justifcation. He acknowledges that in order for this to be correct, assuming that the looks view is correct about propositional justifcation, it has to be that we ordinarily base our simple visual beliefs on beliefs we actually have about how things look. Of course, this is where the looks view faces a problem—it does not seem true that we ordinarily form all the looks beliefs that we would need if the looks view is to accurately account for our justifed simple visual beliefs. Recall from above that McGrath thinks that his discussion of defeaters shows that the dogmatist is committed to claiming that ordinary agents “know about the relations between how things are, how they look and how we experience them.” We have seen, nevertheless, that McGrath fails to establish this point. For the phenomenal explanationist is not committed to the idea that ordinary agents have such knowledge.19 Because of this, the phenomenal explanationist could charge advocates of the looks view with overintellectualizing perceptual justifcation without the fear of receiving a tu quoque reply. The phenomenal explanationist can insist that since it is implausible that ordinary people generally know about the relations
228
Kevin McCain and Luca Moretti
between how things are and how they look, we should not expect that they generally form the looks beliefs they would need in order to acquire doxastic perceptual justifcation according to the looks view. Nevertheless, McGrath makes a case for thinking that ordinary agents do have the required looks beliefs. McGrath asks us to consider a scenario where one sees an apple and forms the simple visual belief that that is an apple. According to him, In ordinary cases of seeing an apple, one doesn’t form the belief [that] the thing looks like an apple in the sense of making a conscious judgment to this efect. Still, one obviously knows and so believes it looks like an apple. When I ask you, “does it look like an apple?” you[r] answer is “of course it does.” It doesn’t seem to you that you are forming a belief; you already had the belief. Whether we want to say the belief/knowledge is implicit, tacit, or whatever, it is there. And it can do epistemic work. (2018: 123) In order to help illustrate how beliefs not explicitly manifested in one’s thinking can do epistemic work, McGrath describes an example (not concerning a looks belief) that he draws from Senor (2008): “I look at a sunset and judge that it’s a beautiful sunset. I don’t consciously judge that it’s evening. But I know it’s evening, and this is surely part of my justifcation for believing it is a beautiful sunset” (2018: 124). According to McGrath, this is just one example among many. The question here is whether these sorts of examples genuinely show that one has the relevant belief rather than merely a disposition to form the belief. Let us take the sunset example frst. If one knows that it is evening, that knowledge helps justify believing that it is a sunset that one is seeing rather than a sunrise. Since knowledge presumably entails justifed belief, in this case, one’s justifed belief that it is evening is part of the justifcation one has for thinking that it is a beautiful sunset that one is seeing. That said, would one have to have knowledge (or justifed belief) that it is evening in order for this belief about the sunset to be justifed? It seems not. Consider a variation on the case where one does not actually believe, and so does not know, that it is evening, but one is sufciently justifed in believing that it is evening. In this case it still seems like one is justifed in believing that one is seeing a beautiful sunset. It seems that all that is required is that one have justifcation for believing that it is evening—especially since one is not actually inferring from a belief that it is evening. Similar considerations apply in McGrath’s apple case. The fact that one would answer that it looks like an apple, if asked whether it does, gives us very strong grounds for thinking that one has the disposition to
Phenomenal Explanationism and the Look of Things
229
afrm and believe that it looks like an apple. And, it is very plausible that one has propositional justifcation for this belief as well. What is unclear, however, is why we should think that one already has the belief in question. The nature of the question being asked may explain why one does not seem to oneself to be forming a belief. One is being asked to afrm or deny that the object looks like an apple. Being asked in this way puts the proposition [this looks like an apple] directly before one’s mind for conscious consideration, which makes it easy to form a belief about it almost immediately. Consider, if one is looking at a bookshelf full of books one might not believe (or disbelieve) that there is a red book on the shelf. However, when asked “do you see a red book?” one might almost immediately answer “yes.” This does not show that one believed all along that there is a red book on the shelf. Rather, it shows that either one already believed there is a red book on the shelf or one was disposed to form this belief almost immediately when asked.20 These are equally good explanations of what is going on in this case. Similarly, in McGrath’s example it seems that there are two equally good explanations for what happens. One is, as McGrath suggests, that the person already believes that the object looks like an apple. The other explanation is that the person does not yet believe the object looks like an apple, but instead is only disposed to form this belief almost immediately when asked. In light of this, at the very least it is clear that McGrath’s case does not obviously support his claim. Things become even less congenial to McGrath’s claim when we think about very young children or other unrefective agents. Plausibly, such agents can have a justifed belief that this is an apple, but they do not seem to have beliefs about how things look. Of course, McGrath may want to insist that his looks view is only applicable to agents capable of having refective justifcation. Fair enough. But, even if we limit our focus to refective agents, it does not seem that we have enough looks beliefs to do the work that McGrath needs. For the looks view to be workable at the level of doxastic justifcation, it would have to be that for every justifed simple visual belief that S forms, she also has a belief about how things look. The looks view must posit at least two beliefs in each case of a justifed simple visual belief,21 where dogmatist views, such as PE, only have to posit one. And, as we have mentioned, it is unclear that one needs to posit these additional beliefs to account for the data that McGrath appeals to when trying to motivate the looks view. It is plausible, instead, that all that is needed is a disposition to form looks beliefs when prompted in the appropriate way. Therefore, when it comes to the problem of whether or not we ordinarily form looks beliefs, there is not a decisive objection to the looks view, but it is far from clear that we do form such beliefs. The looks view appears questionable on this point.
230 14.4.
Kevin McCain and Luca Moretti Conclusion
The considerations adduced in the previous three sections reveal that McGrath’s supporting strategy for the frst premise of his argument against immediate perceptual justifcation—which asserts that the looks view is true—fails. The cases he constructs in support of the looks view are in fact unpersuasive. First, McGrath contends that the looks view provides a good explanation of defeat, more straightforward than that provided by dogmatism. This is false: PE supplies a simpler and so preferable account of defeat that—unlike the looks view—cannot be charged with overintellectualizing perceptual justifcation. Second, McGrath contends that since justifcation to disbelieve or withhold concerning looks propositions would defeat justifcation for the related simple visual beliefs, the justifcation of the latter beliefs must rest in part on justifcation for the former beliefs. This conclusion does not follow from the premise, so this argument is invalid. Third, McGrath contends that only the looks view can easily explain why it would be irrational to claim that one has justifcation for a simple visual belief when one is justifed in suspending judgment about the corresponding looks proposition. This is untrue: PE ofers an explanation that seems equally simple. Finally, the looks view appears false, or at least not sufciently plausible, when it is applied to doxastic justifcation. This leaves us without reason to think that the justifcation for simple visual beliefs is mediate. Hence, dogmatism—at least in the form of PE—does not look to be in any trouble.22 Notes 1 McGrath (2018) is explicit that by “justifed” he is noncommittal on whether this is a necessary component for knowledge, though he thinks that plausibly it is. He says that by “justifed” he means “reasonable in believing”. Presumably, he accepts that when it is reasonable to believe p, it is not also reasonable to believe ~p or suspend judgment concerning p. 2 Also see his (2017), which makes similar moves, challenging the possibility of immediate perceptual knowledge. 3 Although McGrath restricts his discussion to justifcation for visual beliefs, he notes that he thinks the look view is extendable to perceptual justifcation more generally. We will assume that what he says about visual beliefs, if correct, would be generalizable in this way. So, we will speak of McGrath’s argument as an attack on immediate perceptual justifcation in general (aside from the justifcation of beliefs about looks, which McGrath has explained in correspondence may be immediately justifed), rather than just visual perceptual justifcation. 4 When you have this phenomenology, you do not have the feeling of reasoning from evidence but of simply seeing that the thing is an F. 5 We follow McGrath in referring to such views as “dogmatism” or “dogmatist views.” 6 They are public with respect to content because they are not about us or our experiences but instead about worldly objects. They are public with respect to
Phenomenal Explanationism and the Look of Things
7 8 9 10 11 12
13 14 15 16
17 18
19
20
21
231
their possession because many people can share the very same looks-related reasons. For a full discussion and defense of Phenomenal Explanationism, see McCain and Moretti (2021). We have named McGrath’s example. McGrath (2018: 121). See in particular McCain and Moretti (2021: sections 2.1 and 4.5.1). Note that this does not require S to consider this explanation or conduct any sort of evaluation of competing explanations (see McCain and Moretti 2021: 97). At frst glance, there may seem to be a problem here because one might worry that if justifcation depends on explanation, then justifcation must be inferential and not immediate. However, closer inspection reveals that this worry is ultimately no cause for concern (see McCain and Moretti 2021: 87–88). See Bergmann (2005) for discussion. For discussion of this point, see Adler (2017) and Lackey (2008). For discussion of various ways that one might hold doxastic attitudes other than believing or disbelieving, see Feldman and Conee (2018) and Friedman (2013, 2017). One might worry that this situation is one in which McGrath faces an epistemic dilemma. This worry is misplaced though. Since McGrath is still deliberating, it is not clear that McGrath is actually in a position to adopt a doxastic attitude toward o at all, and plausibly one can be in an epistemic dilemma only if the situation is such that one must adopt a doxastic attitude and no such attitude is justifed. For discussion of this point about epistemic dilemmas see Stapleford and McCain (2021). Perhaps humans are not afected by this sort of amnesia. The relevant fact is, however, that the situation that we describe is at least metaphysically possible. We allow that it might be the case that if McGrath tries and fails to recall evidence in support of o, his recognition of failing in this way gives him new evidence that makes it so that he should suspend judgment about o. If he does end up in a situation where he is justifed in suspending judgment about o, it is plausible that this gives him a defeater for believing p. However, prior to gaining this new evidence for suspending judgment about o McGrath has justifcation for believing that p. One might worry that Phenomenal Explanationism abandons internalism by not requiring for justifcation awareness of the connection between a’s looking like an F and a’s being an F. This worry is misplaced, however. Phenomenal Explanationism is a form of mentalism that respects key internalist intuitions, such as the intuition at the heart of the New Evil Demon problem. Of course, there are other more demanding forms of internalism that one might prefer. But, to our minds, Phenomenal Explanationism is internalist enough. Our point here relies upon the very intuitive idea that one can have evidence for p without forming the belief that p. Hence, one can have evidence that there is a red book on the shelf prior to forming the belief. In such a case, one’s being asked whether there is a red book can prompt one to form a belief supported by the evidence one already has. It might be many more than two. Consider again the example that McGrath draws from Senor. Suppose s is “it is a beautiful sunset” and e is “it is evening.” McGrath agrees with Senor that his belief that e provides part of the doxastic justifcation for his belief that s. So the belief that e appears to be an additional doxastic basis for his simple visual belief that s. The other basis is the belief that
232
Kevin McCain and Luca Moretti
this looks like a beautiful sunset. Clearly, McGrath’s simple visual belief that s may have—simultaneously—many other doxastic bases of this sort. If this is what McGrath has in mind, the looks view must posit far more than two beliefs in many cases of justifed simple visual beliefs. 22 We are grateful to Matt McGrath, Declan Smithies, and Matthias Steup for helpful comments on earlier drafts.
References Adler, Jonathan. 2017. Epistemological problems of testimony. In Edward N. Zalta (Ed), Stanford encyclopedia of philosophy. https://plato.stanford.edu/archives/ win2017/entries/testimony-episprob/ Bergmann, Michael. 2005. Defeaters and higher-level requirements. Philosophical Quarterly, 55: 419–436. Feldman, Richard & Conee, Earl. 2018. Between belief and disbelief. In Kevin McCain (Ed), Believing in accordance with the evidence (pp. 71–89). Cham: Springer. Friedman, Jane. 2013. Suspended judgment. Philosophical Studies, 162: 165–181. Friedman, Jane. 2017. Why suspend judging? Noûs, 51: 302–326. Lackey, Jennifer. 2008. Learning from words: Testimony as a source of knowledge. Oxford: Oxford University Press. McCain, Kevin & Moretti, Luca. 2021. Appearance and explanation: Phenomenal explanationism in epistemology. Oxford: Oxford University Press. McGrath, Matthew. 2017. Knowing what things look like. Philosophical Review, 126: 1–41. McGrath, Matthew. 2018. Looks and perceptual justifcation. Philosophy and Phenomenological Research, 96: 110–133. Senor, Thomas. 2008. Epistemological problems of memory. In Edward N. Zalta (Ed), Stanford encyclopedia of philosophy. https://plato.stanford.edu/archives/ fall2008/entries/memory-episprob/ Stapleford, Scott & McCain, Kevin. 2021. Epistemic dilemmas, epistemic quasidilemmas, and quasi-epistemic dilemmas. In Kevin McCain, Scott Stapleford, and Matthias Steup (Eds), Epistemic dilemmas: New arguments, new angles (pp. 77–85). Routledge.
Part 4
Intellectual Seemings and Intuitions
15 A Priori vs. A Posteriori Justifcation The Central Role of Rational Intuitions Bruce Russell 15.1.
Is There a Signifcant Diference Between A Priori and A Posteriori Justifcation?
Timothy Williamson argues that there is no signifcant diference between a priori and a posteriori knowledge and justifcation. He does not deny that there is a diference, but he argues that it is not signifcant. Through examples, he gives us an idea of what he means by an insignifcant diference: a classifcation of plants and animals according to color; a classifcation of plants as bushes or non-bushes; a distinction between red and non-red bicycles (Roeber/Steup, forthcoming: 1). Sometimes he says that those distinctions “don’t cut nature at its joints,” or more recently, “does not cut at a cognitive joint” (Roebers, et al: 2). When Williamson writes of not cutting at a joint, I take him to mean that the distinction does not mark an essential diference. The ones that don’t mark an essential diference do not cut at the joints. For instance, relative to some better biological theories, there is no essential diference between bushes and non-bushes, or between plants or animals of diferent colors. Bicycles have an essential nature; they must have two wheels. It doesn’t matter what color they are. But Williamson’s examples of insignifcant diferences are examples of things that have essences. Justifcation isn’t a thing. Are there examples of diferences in the way that beliefs can be justifed that are insignifcant? Some features of the world can be seen with the naked eye. Others require using a microscope or telescope. These diferences seem insignifcant because they all rely on vision as the source of justifcation, whether enhanced by the use of instruments or not. Microscopes and telescopes are like a very powerful pair of glasses. How about the diference between modes of perception, say, between vision, hearing, and touch? That does not seem signifcant either. We could be justifed in believing there is a bell nearby by seeing one, hearing its gong, or touching it. If echolocation were a source of evidence for us (as it is for bats) or electrolocation were (as it is for duck-billed platypuses), there would not be a signifcant diference DOI: 10.4324/9781003299349-20
236
Bruce Russell
between them and the perceptual sources of evidence we have. They all allow us to learn about things in the world outside of us. A case could be made that there is a signifcant diference between how these perceptual modes can provide justifcation and how, say, introspection and proprioception provide it because perception provides justifcation about the external world and the other two about ourselves, either about our inner lives or the movement and position of our bodies, by means signifcantly diferent from the way that vision or touch can tell us about those aspects of our bodies. But a priori justifcation is supposed to be diferent from all these other modes of justifcation. To tell how many teeth are in a horse’s mouth, you should look and count them.1 But to tell how many primes there are between zero and one hundred, you don’t have to look; you can just count. Counting teeth is similar to counting primes but also diferent. Is that diference signifcant? Williamson proposes that we consider paradigm cases of a priori knowledge to see if that sort of knowledge is signifcantly diferent from paradigm cases of a posteriori or empirical knowledge.2 I will put the dispute in terms of justifcation rather than knowledge, and Williamson has said that his dispute with Paul Boghossian on this issue does not depend on taking his “knowledge frst” approach to justifcation. In the past, he compared the following pair of propositions to see whether there was a signifcant diference between a priori and a posteriori justifcation. A. If two marks had been nine inches apart, they would have been at least nineteen centimeters apart. B. If two marks had been nine inches apart, they would have been further apart than the front and back legs of an ant. (Williamson, 2007: 165–167) And later he compared the following two propositions with the same aim in mind. C. All crimson things are red. D. All recent volumes of Who’s Who are red. (Boghossian and Williamson, 2020: ch. 8: 120–125) In both pairs of cases, Williamson assumes that we learned one concept independently of the other, that is, inches apart from centimeters and crimson apart from red. In the frst case, that would enable us to form an image of, say, two marks nine inches apart and another of two marks 19 cm apart, and then compare them in our imagination and see that being nine inches apart is greater than being 19 cm apart. We could justify (B) in a similar way. Having acquired an idea of how far apart the front and
A Priori vs. A Posteriori Justifcation
237
back legs of an average ant are, we could form an image of that distance, compare it to the image of a nine-inch line, and then see that nine inches is greater than the average distance between the front and back legs of an ant. Williamson concludes that these ways of justifying (A) and (B) are not signifcantly diferent and assumes that if (A) is justifed, it is justifed a priori, and if (B) is justifed, it is justifed a posteriori, that is, empirically. He seems to assume that the justifcation of (A) in this way is a paradigm case of a priori justifcation, and the justifcation of (B) in this way is a paradigm case of a posteriori justifcation. From this, he concludes that a priori justifcation is not signifcantly diferent from a posteriori justifcation. But it’s not clear that the justifcation of (A) that Williamson ofers is a priori. Suppose someone claimed to have justifed 2 + 2 = 4 by flling two two-quart bottles with water and dumping them into a fve-quart beaker with marks at every half-quart. Assume they found that the beaker was flled to the four-quart mark. That would not be an a priori justifcation of 2 + 2 = 4. If it were, then if we poured two quarts of water and two quarts of carbon tetrachloride into the beaker and found it resulted in slightly less than four quarts of liquid, that would disconfrm that 2 + 2 = 4. Clearly, it does not. So the example of adding two quarts of water to two quarts of water does not confrm that 2 + 2 = 4. There is an a priori justifcation that 2 + 2 = 4, but adding quarts of water together is not such a justifcation. We could cut a stick nine inches long and another 19-cm long and then place them next to each other with one end of both sticks starting at the same mark. We could then see with our eyes that the nine-inch stick is longer than the 19-cm stick, which could then be used, at most, to justify (A) a posteriori. Forming images of nine-inch and 19-cm distances is like putting nine-inch and 19-cm sticks next to each other. If we assume that our images accurately represent the relevant distances, as accurately as the two sticks would, then we would have a justifcation for (A) that is similar to the justifcation in the two-sticks scenario. That’s grounds for saying that the justifcation in imagination of (A) that Williamson ofers is an a posteriori justifcation. A paradigm example of an a priori justifcation of (A) would start with being a priori justifed in believing 1 inch = 2.54 centimeters and then multiplying 2.54 by 9 to get 22.86 centimeters, which is clearly greater than 19 centimeters. Williamson’s example of an a priori justifcation of (A) seems more like an a posteriori than an a priori justifcation of (A). It certainly is not a paradigmatic example of a priori justifcation. Further, Williamson’s justifcation of (B) is not a paradigmatic example of a posteriori justifcation. Normally, we would measure the distance between the front and back legs of ants, compute the average, and compare that number given in inches to nine inches to justify that (B) is true. No scientifc journal would accept that forming images of things and then
238
Bruce Russell
comparing them in the imagination constitutes empirical justifcation of the claimed result, but it is more like empirical than a priori justifcation. The method of comparing the lengths in imagination provides justifcation only if the corresponding perceptual comparison would and the images accurately map the perceptual evidence. To show that there is no signifcant diference between a priori and a posteriori justifcation, Williamson needs to ofer a paradigm example of each where there does not seem to be any signifcant diference between the methods of justifcation, but his examples involving the justifcation of (A) and (B) do not ft the bill. Perhaps the examples (C) and (D) serve Williamson’s purposes better. For (C), he imagines that Norman has learned “crimson” by examples in the country and “red” by examples in the city and has never heard “red” used in the country or “crimson” in the city (Williamson, 2020: ch. 10, 164). So Norman is well justifed in believing that some things are crimson and well justifed in believing other things are red, and he has a clear grasp of the concepts crimson and red. One day Norman is just letting his imagination wander and forms the image of a crimson patch and then of a red patch right beside it in mental space. Perhaps on this basis, he has a kind of “aha” moment and comes to believe that crimson is a shade of red. Is Norman a priori justifed in believing all crimson things are red? Not unless he is a priori justifed in believing some relevant and necessary truth, say, that: C1) Necessarily (If something is a shade of some color, it is that color). This general claim is not about crimson and red; it’s about the relationship between shades of a color and its color. Let’s grant that the episode in his imagination that Williamson attributes to Norman justifes Norman in thinking that crimson is a shade of red.3 Then, if he is justifed a priori in accepting (C1), he is in a position to be justifed in believing (C): Necessarily, all crimson things are red. Let’s contrast how Norman is justifed in believing (C) with how someone I’ll call “Lerner” is justifed in believing (D) according to Williamson’s view. Suppose Lerner’s father has a pile of books and asks Lerner to pull out a subset of those books that have some feature in common, other than color, and put them in a pile. Lerner goes through the large stack of books and notices that many of them have the title Who’s Who written on their covers along with a recent publication date. He puts all those books into a pile and is aware that what they have in common is the title and a recent publication date. He was not paying attention to what color the books were and initially had no idea of their color. Like Norman, Lerner later lets his imagination wander. He forms an image of the books that he sorted according to title, and now notices that the books in that image are also
A Priori vs. A Posteriori Justifcation
239
red. Suppose, on that basis, he has an “aha” moment and comes to believe that, “All recent volumes of Who’s Who are red.” Hasn’t Lerner put twoand-two together, just like Norman, and thereby reached a conclusion he is justifed in believing? By using his imagination, Norman comes to realize something he had never realized before. By using his imagination, Lerner also comes to realize something he never realized before, namely, that his image of recent volumes of Who’s Who is also an image of red books. Let’s grant that both of them are justifed in believing the conclusion they reached, but Norman is a priori justifed and Lerner a posteriori justifed. Williamson argues that here there is not a signifcant diference between a priori and a posteriori justifcation since the methods they used to reach their conclusions are so similar. The same sorts of questions arise here as in the frst pair of examples (A) and (B). Norman is a priori justifed in believing that all crimson things are red only if he is a priori justifed in believing some proposition like (C1). Otherwise, his justifcation for believing that all crimson things are red would be like believing that proposition on the basis of observing lots of crimson things in the world that are red and coming to the same conclusion via induction alone. That would be like coming to believe that all crows are black by observing lots of black crows. That is clearly a case of a posteriori justifcation. Williamson’s case of Norman is a paradigm case of a priori justifcation only if Norman is a priori justifed in believing some relevant necessary truth like (C1).4 But then, is Norman’s a priori justifcation for what he believes similar to Lerner’s justifcation for what he believes? I don’t think it is. Lerner frst pays attention to one feature of an image that stems from his observation of the pile of books he pulled out from the larger stack: the image that they’re all recent publications of volumes of Who’s Who. He then pays attention to the color of those books in his image of them and sees that they are all red. He then has an “aha” moment and comes to believe that all recent volumes of Who’s Who are red. This process in imagination is like what would have happened if Lerner’s father had asked him to look again (with his eyes) at the books in the stack he had pulled out, paying attention this time to their color. Lerner would then have had empirical evidence that all recent volumes of Who’s Who are red (at least all recent volumes in the sample he examined). Lerner’s exercise in imagination yields justifcation if and only if he has good reason to believe that his images are accurate representations of what he observed. Lerner’s justifcation for believing that would then be empirical, but not a paradigm case of empirical justifcation. The paradigm case would be where Lerner sees with his eyes that all the books have the same title and that they are red. Williamson still has not given paradigmatic cases of a priori and a posteriori justifcation, and when his examples are clarifed so that they are
240
Bruce Russell
such cases, there seems to be a signifcant diference between those types of justifcation. Defenders of a priori justifcation can readily grant that the justifcation in (A) and (B) is not signifcantly diferent but also plausibly deny that the justifcation in (A) is a paradigm case of a priori justifcation. They can also allow that there is an interpretation of the justifcation in (C) that makes it a case of a priori justifcation, but then plausibly hold that the justifcation in (D) is signifcantly diferent from that in (C). Williamson has not ofered a paradigm case of both a priori and a posteriori justifcation where there is no signifcant diference in the types of justifcation. Either the pair of cases does not involve a clear case of a priori vs. a posteriori justifcation or it does, but there is also a signifcant diference in the assumptions the justifcation rests upon. Williamson’s latest attempt to establish the superfciality of the distinction between a priori and a posteriori justifcation focuses on logical and mathematical proof (Roeber et al., forthcoming). The underlying idea is that if anything involves a priori justifcation, it is logical and mathematical proof. But then Williamson notes how an a posteriori element is involved in at least long and complex proofs of that sort. To be justifed in believing that such a proof is valid, someone has to check the steps in the proof, its author, others, or both. But to do that there must be some written version of the proof and the person must see that each step in the proof is a valid step. Suppose some step is an instance of modus ponens. To check that it is, a person must recognize that it is by literally looking at the written sentences to see that they manifest the relevant logical form. But this sort of recognition is like recognizing that a square, not a circle, has been written on a white board. If you are justifed in believing that the fgure on the board is a square, that is clearly a case of a posteriori justifcation. So the a priori justifcation involved in a long and complicated mathematical or logical proof must involve the same sort of a posteriori justifcation that stems from the perceptual recognition that the sentences that constitute the segments of the proof are instances of a valid logical form. Williamson is right in thinking that a recognitional capacity (which is a perceptual capacity) is needed for a person to be a priori justifed in believing that a complex argument is valid. And he rightly denies that recognitions of valid argument forms are enabling experiences in Boghossian’s new account of a priori justifcation. But here is an example to show that recognitional experiences are enabling experiences. We have to learn from our parents and relatives what it is to be a mother, father, child, brother, sister, aunt, uncle, niece, nephew, etc. Once we learn these concepts, we can form a family tree, and once we learn the concepts, we can answer questions like, “Starting from one mother and one father, how many children would be in the fourth generation if that mother and father had two children, each of which had two children, each of which
A Priori vs. A Posteriori Justifcation
241
had two children, and so on?” There are three ways we might answer this question. We might construct on paper a family tree that meets the description of the case. Two boxes for generation one represent the two children. Four boxes for generation two; eight boxes for generation three, and so on. We would then count the number of boxes in generation four: 16! After doing this, we may come to realize that there is a formula that gives the answer for the number of children in any generation in this sort of “double-the-children” scenario. It is N = 2n, where N is the number of children in a generation and n is the number of the generation. In the future, we could use this formula to compute the number of children in, say, the 15th generation. But if we didn’t have a calculator, we’d have to write down: 2; 2 × 2 = 4; 4 × 2 = 8; 8 × 2 = 16, and so on 11 more times because most of us can’t calculate 215 in our heads. The third way to answer the question is to actually fnd families that ft the description of the “double-the-children” scenario, fnd the children (or their birth certifcates) in the fourth generation, and count how many there are. We would do this for many families and then by induction conclude that the number is 16: all families with the relevant structure have 16 children in the fourth generation. The second method is clearly a priori even if we have to write down on paper all the multiplications by 2 in the case where we are looking for the number of children in the 15th generation. Writing those multiplications down is just a mental aid to prevent us from making a mistake given our limited mental capacities. This case is like Williamson’s involving a long and complex mathematical proof where the mathematicians have to write down various segments of the proof to check its validity. Writing it down is an aid used to prevent them from making a mistake. Both the second method and Williamson’s case are instances of a priori justifcation. The third method is clearly empirical or a posteriori. People fnd the relevant families, count the number of children in the relevant generation, and then use induction on their results to conclude what that number is: 16 children in the fourth generation. The frst method of constructing a family tree has something in common with the third, but no actual families or children are observed: boxes that represent children are not children, nor are they birth certifcates of actual children. More importantly, induction is not used to reach the conclusion. Once the family tree for four generations is written down, it just seems obvious that the answer has to be 16 for any family with the “double-the-children” structure! Counting the boxes is like counting the primes between one and one hundred. Constructing the family tree is a method for eliciting the rational intuition that the answer has to be 16 children in the fourth generation. So far, Williamson has not established that there is no signifcant diference between a priori and a posteriori justifcation. That there are some
242
Bruce Russell
similarities between some instances of those diferent types of justifcation is not enough to establish his claim. 15.2. What Makes a Priori Justifcation Signifcantly Diferent from A Posteriori Justifcation? A plausible answer to this question is that a priori justifcation rests on rational intuitions and a posteriori justifcation rests on empirical observations. Paul Boghossian defends this sort of view, but he must overcome a couple of objections. First, because he takes rational intuitions to be a type of seeming, he must distinguish them from what George Bealer calls physical intuitions, which are also a type of seeming. Bealer says that a person might have the intuition that a house undermined will fall, but that is different from the intuition that in Gettier cases a person has a justifed true belief but lacks knowledge.5 And Boghossian must address Williamson’s example, where he has an intuition that he should not attend a birthday party for Donald Trump to which he has been invited. Is that a rational intuition?6 Boghossian takes rational intuitions to be intellectual seemings. These seeming are of two types. First, there are propositions that seem necessarily true. Second, there are propositions that just seem true to a person but would seem necessarily true to that person if she had the relevant concept of necessity and considered whether they are necessarily true (Boghossian, 2022a: 4, where Boghossian quotes Joel Pust). To simplify, assume that rational intuitions are apparent insights into necessary truths = what seems to be necessarily true at least on refection. But what specifcally are those insights about? Boghossian tentatively ofers that they are seeming insights into the nature of the properties they concern and how they are related (say, the relationship between the property of being a justifed true belief and the property of being an instance of knowledge or the relationship between the property of having size and the property of having shape). He thinks that in the case of analytic propositions, “our insights are grounded in our understanding of the relevant concepts,” but in the case of synthetic a priori propositions, they are “grounded in our grasp of the natures of the properties themselves” (Boghossian, 2022a: 5; Boghossian and Williamson, 2020: ch. 13, esp.: 186–188; 192–194). Williamson does not think there are any intuitions, only inclinations to believe the propositions that are thought by some to be the objects of those intuitions. So Boghossian frst gives an existence argument to justify his claim that there are rational intuitions. He calls the following proposition (No Torture): It is morally wrong to infict pain upon a human being merely for one’s own amusement. Suppose you believe (No Torture). Why do you believe this? Either it’s because there is some non-conscious mental
A Priori vs. A Posteriori Justifcation
243
state that is analogous to what causes people with blind sight to believe what they do or it is because of some conscious mental state.7 If it were like blind sight, then you would feel that your belief is a stab in the dark or a guess, but you don’t. If it were a conscious mental state, then it must be that it simply seems true to you or because of some substantive reason you have, say, that this sort of torture violates a human right or is “an impermissible trampling of human dignity,” etc. (Boghossian, 2022a: 11; 9–12). But sometimes people believe (No Torture) is true and have no substantive reason for believing it. So by elimination, the only source of their belief that (No Torture) is true is the intellectual seeming that it’s necessarily true (or that it just seems true but would seem necessarily true, if they considered it). But that sort of seeming is just what a rational intuition is. This is Boghossian’s argument that there are rational intuitions and sometimes they cause us to believe what we do, for example, that (No Torture) is true.8 Boghossian also ofers an example given to him by Crispin Wright of a proposition that can seem true even though you have no inclination to believe that it is true because you see that accepting it leads to a sorites paradox. The proposition is: If two things look exactly alike, then if one looks red, so does the other (Boghossian, 2022a: 11). This can lead you to believe that things that are clearly orange and not red are red. The example that I often give is from the Monty Hall case. It can seem true to someone that once Monty has opened one of the doors, the probability that the big prize is behind your door or the other unopened door is 50–50 even though you don’t believe this and even have no inclination to believe it. Contra Williamson, something can seem to be the case even if you have no inclination to believe that it is the case. 15.3. Do Rational Intuitions Justify? Boghossian begins his defense of the justifcatory force of rational intuitions by endorsing what has been called “dogmatism” about seemings, roughly, the view that if something seems true to you, you are prima facie justifed in believing it is true. Boghossian ofers three versions of dogmatism: (Perceptual Dogmatism, PD): If it visually seems to you that P, you are thereby prima facie justifed in believing that P. (Intellectual Dogmatism, ID): If it intellectually seems to you that P, you are thereby prima facie justifed in believing that P. (Rational Intuition Dogmatism, RID): For any necessary proposition P, if you rationally intuit that P, you are thereby prima facie justifed in believing that P. (Boghossian, 2022a: 13 for the statement of all three forms of Dogmatism)
244
Bruce Russell
Boghossian acknowledges that “there are many deep issues” that surround all forms of dogmatism but says that they are beyond the scope of his chapter (Boghossian, 2022a: 13). Some have to do with the interpretation of “prima facie.” Does that mean that the relevant beliefs would be all things considered justifed by the relevant seemings unless they were overridden (rebutted) by other considerations, or that they would be justifed in that way unless they were overridden OR undercut by other considerations? The frst interpretation would yield what I call a “smidgen” view of the justifcation the seemings provide because a “smidgen” of justifcation would always remain no matter how weighty the overriding considerations may be. On the second interpretation, the justifcatory weight of the seemings could be completely destroyed by undercutting considerations.9 But putting aside these deeper issues, why does Boghossian think that Perceptual and Rational Intuition Dogmatism are true? Boghossian says that perception “is a source of a posteriori justifcation” because the perceptual seemings “present themselves as being [a] causally sensitive presentations of the shifting contingent facts in its environment” (Boghossian, 2022a: 14). Simplifed, I think this means that in perception, it seems like the perceptual seemings are a response to a changing external world that causes them. He contrasts rational intuition with perceptual intuition as follows: “rational intuition presents itself as a source of non-causal insight into the timeless natures of the properties it treats, into the timeless modal relations of compatibility, incompatibility, and entailment that hold between them” (Boghossian, 2022a: 14–15). Simplifed, I think this means that rational intuitions are responses to how necessary relations between properties seem to be. But then, couldn’t someone defne spiritual dogmatism as follows: (Spiritual Dogmatism, SD): For any proposition about spiritual beings, if it seems to you that you have encountered a spiritual being, you are thereby prima facie justifed in believing that you have encountered one. And if someone asks for further clarifcation about what these spiritual seemings present themselves as being responses to, their defenders might say, “they present themselves as being a causally sensitive presentation of the contingent spiritual facts in our environment.” So after a loved one has died, it might seem to you that their spirit is nearby watching over you, or on certain occasions that God is watching over you. Then, on SD, you would thereby be prima facie justifed in believing that these things are true. This sort of dogmatism has gone too far, and the explanation of why seemings can provide evidence doesn’t help. You do not have any
A Priori vs. A Posteriori Justifcation
245
justifcation for believing that there are spiritual beings, or even intelligent, material aliens, just because it seems to you that there are.10 15.4.
Why Does Boghossian Reject the Understanding-Based Account of Rational Intuitions According to Which They Provide A Priori Justifcation?
Boghossian frst distinguishes between what he calls epistemically analytic claims and substantive ones.11 In the moral realm, he distinguishes the epistemically analytic claim: (12) If an act A is morally wrong, then it merits moral disapproval. from the substantive claims: (13) Inficting severe pain on babies just for fun is morally wrong. (14) A person’s sufering is a prima facie moral reason for someone to help relieve it. (2020: ch. 13: 197 for all three statements) G.E. Moore would have said it is an open question whether (13) is true, so “inficting pain just for fun” cannot be part of the defnition of “morally wrong,” and not part of the concept morally wrong. He would have said the same about (14). It’s not clear if someone fully understood “morally wrong” the truth of (13) and (14) would be an open question. So it’s not clear that Moore’s test would result in rejection of all naturalistic proposals about the meaning of moral terms. Boghossian rejects Moore’s test because he thinks that on certain naturalistic conceptions of, say, “morally wrong,” Moore’s relevant question, “Is N wrong?” would be closed just as the question, “Is a bachelor male?” is closed. On Moore’s test, since these questions are closed, it follows that N is part of the defnition of “wrong” and “male” is part of the defnition of “bachelor,” or are at least entailed by those defnitions. But Boghossian thinks that it is an open question whether any naturalistic conception of “morally wrong” is the correct concept of wrong.12 This motivates him to propose a diferent Open Question test, which he calls the Concept Correctness Open Question: CCOQ for short. CCOQ assumes that there is a correct concept of morally wrong. Boghossian assumes that it is (12). CCOQ asks whether a specifc conception of wrong correctly specifes “those acts that are genuinely deserving of moral disapproval” (Boghossian, 2020: ch. 13: 198, 2021: 376–379). For example, we can ask does “acts that fail to maximize utility” correctly specify those acts that are genuinely deserving of moral disapproval? If the answer is yes, then “failing to maximize utility” would be the correct substantive
246
Bruce Russell
concept of what is morally wrong. If the answer is no, then it would not be. Boghossian seems to think that no such proposal can meet the CCOQ test. Still, we are justifed in believing that (13) and (14) are true based on our rational intuitions. Here’s his argument to show that those rational intuitions must not be based on our understanding of “morally wrong.” 1. BOG: If those rational intuitions were based on our understanding of “wrong,” they would have to be based on our understanding of some correct substantive concept of wrong. 2. BOG: But failure of all such substantive proposals to pass the CCOQ test shows that there is no correct substantive concept of wrong. 3. BOG: Therefore, those rational intuitions are not based on our understanding of “wrong.” Someone might reject what is assumed as background to this argument, namely, that we are justifed in believing (13) and (14) on the basis of the rational intuitions we have. But even granting this background assumption, 2BOG is not obviously true. I believe Boghossian is mistaken about the role of rational intuitions. They play both a role in justifying some minimalist concept of wrong and a more substantive one. They are epistemically prior to any minimalist moral concept, and so they do not have to meet some test that presupposes the truth of that minimalist concept. A Kantian might ofer a minimalist concept of wrong that is a competitor to (12). She might hold that it is morally wrong for someone to perform an action if she cannot will that everyone in her circumstances perform that sort of action.13 Examples might be produced that elicit rational intuitions that constitute counterexamples to this proposal. Perhaps a rugged individualist or a political libertarian cannot will that everyone relieve another’s sufering when that can easily be done. This person does not will that he help others in those circumstances nor that others help him. So on this Kantian account of the concept of morality (14) is false: a person’s sufering is not a prima facie moral reason to relieve it. But we have a rational intuition that (14) is true. So that rational intuition gives us reason to reject this Kantian proposal for the concept of morally wrong. (12) might also be criticized by appeal to rational intuitions. Aren’t there actions that are morally wrong that don’t merit moral disapproval because the person has a legitimate excuse for performing them? Perhaps he was non-culpably ignorant of their consequences, perhaps he was grief stricken and out of his grief said something cruel or insulting, perhaps he is schizophrenic and thinks people are out to get him, perhaps he is obsessive compulsive and was driven to use up all the sanitizing gel, or perhaps determinism is true and though it was wrong for him to kill an innocent person he could not have done otherwise. In all these cases the person did
A Priori vs. A Posteriori Justifcation
247
what is wrong, but it seems that moral disapproval is not merited. That gives us reason to reject (12). Perhaps there is another proposal close to (12), say, (12*) that says that if a person is morally blameworthy, then they merit moral disapproval. But to be justifed in accepting that proposal, it must also be put to the test of standing up to possible counterexamples that elicit relevant rational intuitions. (Could a person be morally blameworthy for doing something yet not merit moral disapproval because his mild negligence resulted in such good consequences? Perhaps he failed to throw a switch and the train then did not go down a track where it would have killed hundreds of people on the track.) Rational intuitions play a similar role in epistemology. Someone might think that the correct minimalist concept of knowledge is given by the following statement: S knows that P if and only if S has a non-accidentally justifed true belief that P. They might arrive at this concept of knowledge by rejecting more substantive conceptions of knowledge on the basis of rational intuitions, say, by rejecting the proposal that a person has knowledge if and only if he has a justifed true belief and the proposal that he has knowledge if and only if he has a reliably produced true belief. Rejection of more substantive epistemic or moral conceptions on the basis of rational intuitions can be the epistemic ground for accepting a more minimalist statement of the concept. In place of Moore’s Open/Closed question test, and Boghossian’s CCOQ, the correct test seems to be what I will call the Counterexample Test. It says that for any substantive or non-substantive proposal for an account of moral concepts like morally wrong, you are justifed in believing that the proposal is correct if and only if there are no relevant rational intuitions available to you that constitute evidence against the proposal. Rational intuitions will be relevant if they are based solely on your understanding, for example, “morally wrong,” “morally obligatory,” “morally good/ bad,” etc. That understanding need not be complete or full, but it must come up to some hard-to-specify minimum before the rational intuitions that come from that understanding provide evidence. If Boghossian’s CCOQ test is not a good test of whether rational intuitions are evidence, why can’t rational intuitions based on understanding synthetic a priori propositions provide evidence that they are true? Here are some examples of synthetic a priori propositions. (SAP 1): If spatial area X is wholly contained in spatial area Y, and spatial area Y is wholly contained in spatial area Z, then spatial area X is wholly contained in spatial area Z. (Bealer, 1998a: 212) (SAP 2): No object can appear red all over and appear green all over at the same time. [the incompatibility of phenomenal colors.] (Bealer, 1998a: 211)
248
Bruce Russell
(SAP 3): Something has size if and only if it has shape. (Bealer, 2002: 75) (SAP 4): X is a closed plane fgure every point on which is equidistant from a common point if and only if every arc of X has equal curvature. (This combines two proposed defnitions of “circle.” See Bealer, “Analyticity,” 1998b, sec. 4, paragraph 2) (SAP 5): A is congruent to B if and only if B is congruent to A. (Bealer, 1998b, sec. 5, paragraph 2) (SAP 6): If x is warmer than y, then y is not warmer than x. (Boghossian, 2020: “Analyticity Reconsidered,” ch. 1: 6) (SAP 7): If z is downstream from y, and y is downstream from x, then z is downstream from x. (Boghossian, 2022b: 3) (SAP 8): Happiness is an intrinsic good. The frst fve examples are from George Bealer. As we’ve seen, Boghossian argues that if we have a rational intuition that any of these eight propositions is true, that intuition is not based on understanding them. But that argument is fawed because it tacitly assumes that minimalist proposals of statements of concepts themselves do not rely on justifcation, or at least non-defeat, by rational intuitions. But George Bealer thinks that our intuition that (SAP 1) is true is based on our concept of spatial part (Bealer, 1998a: 231) and so it is based on our understanding of the proposition, even though it is not analytic.14 I would say something similar about (SAP 8). If we have a rational intuition that it is true, it is based on our understanding of the concept happiness and the concept intrinsically good and their relationship. Those understandings go beyond understanding the defnitions of “happiness” and “intrinsically good” or the analyses of the concepts they express. Those understandings need not be complete, infallible, or indefeasible. We might think of an example where someone is made happy by seeing other people sufer and then reject, or revise (SAP 8) because we have the intuition that happiness based on the sufering of others would not be intrinsically good. We can understand propositions, but the basis of our understanding need not be what it is when we understand analytic propositions. Understanding can include understanding necessary connections between diferent concepts where those connections are not meaning connections, as it is, say, with bachelor and unmarried male. Williamson thinks that there is no signifcant diference between a priori and a posteriori justifcation and knowledge. I argued that there is. Boghossian thinks that there is a signifcant diference between the a priori justifcation of analytic and synthetic a priori propositions. I’ve argued that there is not. Both kinds of justifcation are based on rational intuitions that are based on understanding the propositions that are their objects.15
A Priori vs. A Posteriori Justifcation
249
Notes 1 In Rationality (Viking, 2021: 94–95), Steven Pinker cites a story attributed to Francis Bacon in which a group of old friars beats and then banishes a young friar who had the audacity to suggest that how many teeth there are in a horse’s mouth should be determined by counting them, not by consulting ancient texts. The method the young friar proposed was considered “unholy and un-heard of,” surely inspired by Satan! 2 “Is the A Priori/A Posteriori Distinction Superfcial?” in Roeber, et al, Contemporary Debates in Epistemology 3rd ed., (Malden, MA: Wiley, forthcoming), pp. 1–2. 3 It seems that all Norman is justifed in believing is that this shade of crimson is red, or maybe only that this shade of crimson is a shade of red. But we can assume that Norman repeats his experiment with other shades of crimson and shades of red, sees that each shade of crimson is a shade of red, and then, by induction and C1, concludes that all crimson things are red. 4 It’s not clear that the Norman case is a case of a priori justifcation, even if something like C1) is implicitly assumed. That may be necessary, but it is not sufcient for it to be a case of a priori justifcation. 5 George Bealer (1998a: 207, 211, 213); George Bealer and P. F. Strawson (1992: 102, 104) 6 Williamson (2020: ch. 14, 211–212). 7 You might think that Boghossian has left out a possibility, namely, that some neural state causes you to believe (No Torture). But Boghossian is looking for a cause that might also justify you in believing what you do, and he thinks that no external considerations could provide justifcation. 8 I have interpreted Boghossian’s argument to be for the existence of rational intuitions, though he just couches it in terms of intuitions. 9 See Matthias Steup’s two essays (2018) and (2019) for a thorough discussion of the options and for a rejection of both these forms of dogmatism. 10 Perhaps the example of intelligent aliens can best make my point because someone might maintain that there are defeaters of which the person is aware that cancel the epistemic weight of spiritual seemings. That might explain why, intuitively, it may appear that a seeming that spirits exist does not provide any reason to believe they do. The reply is that such a seeming does not provide reason to believe they exist, but that is because it is overridden or undercut by such a defeater. But we can imagine a situation where you have no reason to believe that there are intelligent aliens and no reason to believe there are not. If, in those circumstances, it just seems to you that there are aliens, that does not tip the balance at all in favor of believing there are. You should suspend judgment. 11 By an “epistemically analytic” claim, Boghossian means “a proposition’s being justifably believable merely in virtue of its being understood” (Boghossian, 2021: 368); also, 2020: ch. 1, esp: 3, 6, 22–23). On this account of analyticity, many propositions that would traditionally be called synthetic a priori turn out to be epistemically analytic. See (SAP1)–(SAP8) below. 12 Boghossian thinks that what’s behind Moore’s Open Question argument is the idea that “To say of an act that it is good analytically entails endorsing it, approving it, or taking some other kind of positive attitude towards it” and no naturalistic defnition of “good” can capture that (Boghossian, 2021: 373; his italics). This seems false. A sociopath might sincerely say that an act is good that he in no way endorses, approves, or is motivated to perform. An act’s
250
Bruce Russell
being good, or morally obligatory, might entail that there is some reason to perform it, but that is diferent from endorsing it or being motivated to perform it. Normativity is about reasons, not motivation or attitudes. Further, “saying” something is good, etc., is diferent from its “being” good, etc. The former may have pragmatic implications that the latter lacks. It is important to Boghossian’s account of normative concepts that they capture their normative role (Ibid: 378, 381). If he is mistaken about their normative role, his argument will be unsound. 13 Boghossian assumes that the concept of wrong is given by its normative role and that its normative role is tied to endorsing, approving, motivating, and the like. That is not its normative role; its normative role is to provide reasons to refrain from acting (see note 12, above). Further, a Kantian might say that moral concepts are tied to universalizability, even if that does not explain their normativity. Still, that would be a minimalist view of moral concepts. 14 Of course, Boghossian might say that it is thereby epistemically analytic, which means it is justifably believable merely in virtue of its being understood. This understanding of “analytic” makes it impossible to be justifed in believing synthetic a priori propositions simply on the basis of understanding them, because if that were true of them, they would be epistemically analytic. To allow for the possibility of synthetic a priori propositions that are justifed merely on the basis of being understood, analytic propositions should not be understood in terms of epistemic analyticity. Instead, “analytic” should be taken to refer to propositions expressed by sentences that express logical truths once relevant terms in those sentences are replaced by relevant synonyms. This is what Boghossian calls Frege-analytic. None of (SAP1)–(SAP8) is Frege-analytic. Synthetic a priori propositions are necessarily true but not Frege-analytic. See note 11, above. 15 Thanks to Dan Korman, Kevin McCain, Scott Stapleford, and Matthias Steup for comments on earlier drafts of this chapter.
References Bealer, George. (1998a) Intuition and the Autonomy of Philosophy. In Michael R. DePaul and William Ramsey (eds.), Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry. Lanham, MD: Rowman & Littlefeld. ——— (1998b) Analyticity. In E. Craig (ed.), Routledge Encyclopedia of Philosophy. London: Routledge. ———. (2002) Modal Epistemology and the Rationalist Renaissance. In Tamar Szabo Zendler and John Hawthorne (eds.), Conceivability and Possibility. Oxford: Oxford University Press. Bealer, George and Strawson, P. F. (1992) The Incoherence of Empiricism. Aristotelian Society Supplementary Volume 66 (1): 99–138. Boghossian, Paul. (2021) Normative Principles Are Synthetic A Priori. Episteme 18: 367–383. ——— (2022a) Intuition and a Priori Justifcation. In Dylan Dodd and Elia Zardini (eds.), Beyond Sense? New Essays on the Signifcance, Grounds, and Extent of the a priori. Oxford: Oxford University Press.
A Priori vs. A Posteriori Justifcation
251
——— (2022b) Minimalism and the Synthetic A Priori. Rough Draft for Participants in Normative Realism Workshop, NYU, May 24–26, 2022. Reworking of an argument in his “Normative Principles are Synthetic A Priori.” ———. (forthcoming a) Is There a Priori Justifcation? In B. Roeber et al. (eds.), Contemporary Debates in Epistemology, 3rd ed. Malden, MA: Wiley. ——— (forthcoming b) Do We Have Reason to Doubt the Importance of the Distinction Between A Priori and A Posteriori Knowledge? A Reply to Williamson. In B. Roeber et al. (eds.), Contemporary Debates in Epistemology, 3rd ed. Malden, MA: Wiley. Boghossian, Paul and Williamson, Timothy (2020) Debating the A Priori. Oxford: Oxford University Press. ——— Boghossian. Chpt 1. Analyticity Reconsidered: 1–27. Reprinted from Nous 30(3) (1996): 36–91. ——— Boghossian. Chpt. 13. Intuition, Understanding, and the A Prior: 186–207. ——— Williamson. Chpt. 8. How Deep Is the Distinction Between A Priori and A Posteriori Knowledge?: 117–136. Reprinted from Albert Casullo and Joshua Thurow (eds.), The A Priori in Philosophy. Oxford: Oxford University Press, 2013: 291–312. Steup, Matthias. (2018) Destructive Defeat and Justifcational Force: The Dialectic of Dogmatism, Conservativism, and Meta-Evidentialism. Synthese 195: 2907–2933. ——— (2019) Easy Knowledge, Circularity, and the Puzzle of Reliability Knowledge. Episteme 16 (4): 453–473. Williamson, Timothy. (2007) The Philosophy of Philosophy. Oxford: Blackwell.
16 Thought Experiments as Tools of Theory Clarifcation Grace Helton
Among contemporary philosophers, it is a pervasive view that intuitions about thought experiments can sometimes help justify the rejection of relevant philosophical theories. It is controversial whether intuitions are judgments, dispositions to judge, sui generis mental states, or some other kind of attitude.1 For present purposes, we can characterize intuitions as, minimally, mental states that are generated in a non-inferential fashion and that have a certain phenomenal quality, or “feel.”2 In the relevant sense, thought experiments (sometimes called cases) are scenarios or mental models of scenarios, typically introduced via a vignette.3 An example will help illustrate the mainstream view about how intuitions about cases can help overturn philosophical theories. Consider act utilitarianism, the ethical view according to which some action is morally permitted just in case it maximizes well-being. Utilitarianism is an initially appealing view, and as anyone who has taught an introductory ethics course can attest, it is a view whose appeal to the novice philosopher can seem undeniable. One way of criticizing utilitarianism is to appeal to thought experiments, such as the following: Transplant Case
A doctor oversees six patients, fve of whom will die if they do not quickly receive organ donations. The sixth patient has organs which could be distributed to the other fve to save all fve of them. Without the consent of this sixth patient, the doctor kills him and distributes his organs to the others, thus saving fve people.4 Many people respond to the doctor’s action with disapproval, even horror. They think that by killing the sixth patient, she does something morally wrong, even though her action saves several lives. Drawing on this intuition, critics of act utilitarianism develop the following argument: DOI: 10.4324/9781003299349-21
Thought Experiments as Tools of Theory Clarifcation
253
Argument Against Act Utilitarianism (1) The doctor’s action in the transplant case is morally wrong.5 (2) The doctor’s action (the same action as in (1)) maximizes well-being. (3) According to act utilitarianism, the doctor’s action in the transplant case is not morally wrong. (4) Act utilitarianism is false. Most ethicists—utilitarians and non-utilitarians alike—take this style of argument to provide a powerful challenge to utilitarianism. But how exactly does the thought experiment, understood as the scenario about the doctor, relate to the argument? The mainstream view goes something like this: The thought experiment elicits a powerful intuition that the doctor’s action is morally wrong. This intuition is itself evidence for or else helps generate evidence for (1), the claim that the doctor’s action is morally wrong. (1) in turn combines with other claims about the doctor’s action, namely that it is one that maximizes well-being and is therefore sanctioned by utilitarianism, to help buttress the overall conclusion that utilitarianism is false.6 In this picture, the intuition about the wrongness of the doctor’s action plays an epistemic role in the argument against utilitarianism, and relevantly similar intuitions play an epistemic role in arguments aimed at other philosophical views. Call this view the epistemic view of intuition. This is the view that intuitions can help overturn philosophical theories, not that intuitions are the only route by which such theories might be overturned. Many theorists defend the epistemic view of intuition overtly, ofering myriad and competing views about how intuitions might play a relevant epistemic role in arguments such as the argument against utilitarianism.7 Both critics and advocates of the epistemic view of intuition tend to implicitly endorse an additional claim, namely that generating intuitions is the sole epistemic function thought experiments play in overturning philosophical theories. Thus, we fnd the most prominent critics of thought experiments focusing their attacks on intuitions, for instance, by claiming that intuitions: do not exist, play little or no role in extant philosophical arguments, or are systematically unreliable.8 Throughout most of this chapter, I accept that intuitions play an epistemic role in explaining how thought experiments can help overturn philosophical theories. But I will argue that intuitions are only part of the story of how thought experiments can help overturn philosophical theories. Thought experiments also help clarify relevant theories, and it is only in concert with theory clarifcation that intuitions about cases can help overturn philosophical theories. In the relevant sense, a thought experiment clarifes a theory for some reasoner when that thought experiment deepens
254 Grace Helton a reasoner’s understanding of a theory, for instance, by improving a reasoner’s grip on what the theory’s concrete predictions are and how those predictions are generated by elements of the theory. Here is the plan for the chapter: I will argue that, in some cases, theory clarifcation can help justify the rejection of the theory clarifed (Section 16.1). I will further claim that thought experiments help clarify theories without the help of intuition, even when those same thought experiments also elicit powerful intuitions (Section 16.2). Drawing on these points, I argue that thought experiments do not overturn philosophical theories by intuition alone. They do so in conjunction with theory clarifcation (Section 16.3). I close by sketching how a more radical view might be true, on which thought experiments help justify the rejection of philosophical theories exclusively by clarifying theories, not by any intuitions those thought experiments might generate (Section 16.4).9 If I am correct that intuitions about thought experiments overturn philosophical theories only in conjunction with theory clarifcation, then the mainstream story about how thought experiments work is incomplete. It leaves out a critical epistemic role, one not played by intuition. As a result, the recent literature about thought experiments is misguided on both sides. On the one side, critics who decry the method of cases on the grounds that intuitions are epistemically inert or inapt have not yet given us a reason to think that thought experiments are epistemically inert. These arguments overlook the epistemically signifcant role of theory clarifcation. On the other side, advocates of thought experiments who have focused wholly on the epistemic value of intuition have neglected an additional route by which cases might play a role in overturning philosophical theories, namely that of theory clarifcation. 16.1.
Two Routes to Rationally Discarding a Theory: Disconfrmation and Clarifcation
Disconfrmation and theory clarifcation are two (non-exhaustive) routes by which theorists might come to rationally discard their theories. First, a theorist might come to discard some theory by way of evidence that disconfrms a theory’s frst-order predictions. This is disconfrmation. For instance, suppose a theorist—let’s call them Xander—provisionally accepts the theory that ants do not use tools. Xander then discovers empirical evidence that ants sometimes stack stones to prevent competition with other colonies.10 So, Xander rationally discards the theory in light of evidence that is frst-order relative to the domain of that theory. That is, Xander discards the theory because of something they learn about ants. And in general, disconfrmation requires evidence about objects in the domain of the theory.
Thought Experiments as Tools of Theory Clarifcation
255
A second route by which a reasoner might come to rationally discard a theory, one that does not require new evidence about the objects in the theory’s domain, is theory clarifcation. It occurs when a reasoner gains an improved understanding of a theory by coming to appreciate certain of the theory’s predictions and how those predictions are generated by elements of the theory. This improved understanding can occur even for the reasoner who already knows what the theory says.11 Consider a diferent theorist, call her Maya, who provisionally accepts the theory that birds do not use tools. Maya explicitly adopts a view of tool use that is relatively minimal, having to do with a creature’s skillfully employing some object outside of its body towards some end. Maya simultaneously implicitly conceives of tool use in terms of certain anthropocentric prototypes, or exemplars, such as swinging a hammer and pouring liquid from a pot, and other activities involving the grasping of objects manipulated near the tool user’s body.12 As a result of Maya’s reliance on anthropocentric paradigms of tool use, she fails to consider that prey dropping might be a form of tool use, where prey dropping is an activity employed by some birds wherein they drop prey from extreme heights to kill them. Prey dropping involves learning, skill, and repeated attempts.13 On the minimal conception of tool use Maya endorses, prey dropping is a kind of tool use, one in which the ground itself is employed as a tool. This is so even though prey dropping involves a non-graspable object exploited far from the tool user’s body. Maya might fail to appreciate that prey dropping is, by her own lights, a form of tool use, even if she knows everything there is to know, empirically speaking, about prey dropping. At the same time, once Maya deepens her understanding of her theory of tool use in the relevant way, she can rationally discard her view that birds don’t use tools. She can do this by relying on this improved understanding of the theory combined with the evidence she already possesses about prey dropping.14 At this point, one might object that theory clarifcation plays merely a psychological role in theory rejection, not an epistemic role. One way of developing this line of thought is to appeal to the familiar distinction between propositional justifcation and doxastic justifcation. A reasoner is propositionally justifed in believing some claim when she has good evidence for the claim. She is doxastically justifed in believing some claim when she believes that claim on good evidence. One can be propositionally justifed without being doxastically justifed. For instance, if Yoli has good evidence that Simon is generally a reliable person and that Simon said he’d be home by now, then Yoli is propositionally justifed in believing that Simon is home by now. But if Yoli hasn’t considered the matter and thus has formed no belief concerning whether Simon is home, Yoli isn’t doxastically justifed with respect to this claim.15
256
Grace Helton
Along these lines, one might suggest that Maya is propositionally justifed in thinking prey dropping is a form of tool use even before she has considered the matter, and thus, before she is doxastically justifed in this view. If this is right, then perhaps Maya’s deepened understanding of the theory plays merely a psychological role, and not an epistemic one, in her rejection of the theory that birds don’t use tools. For, perhaps Maya was already justifed in rejecting the theory, even before she considered the matter. It’s just that considering the matter made it psychologically possible for her to reject the theory. This objection refects fraught and large questions about the nature of justifcation, and it’s not one I can do full justice to here. But to sketch a kind of reply: There is reason to think that Maya isn’t even propositionally justifed in rejecting her theory until she deepens her understanding of the theory in the relevant way. To see this, consider the contrast with Yoli. Yoli needs only to consider whether Simon is home to form the judgment that he is. The simplicity of Yoli’s reasoning contributes to the sense that Yoli was already propositionally justifed in thinking Simon is home by now, even before she had considered the matter. In contrast, Maya will likely need to engage in a sustained process of reasoning to reach the judgment that prey dropping is a form of tool use. Specifcally, she will have to check each element of her theory of tool use to see whether each is present in prey dropping. That is, she will have to check whether prey dropping involves (1) a skill (2) employed towards an end (3) partly carried out through some extrabodily object. The fact that Maya must engage in substantial reasoning is at least weak evidence that she wasn’t justifed, before doing this reasoning, in thinking that prey dropping is tool use. For, the way in which Maya becomes justifed in thinking prey dropping is a form of tool use plausibly proceeds via Maya’s insights about how the elements of her theory map onto the example of prey dropping. If this is right, Maya’s deepened understanding of her theory of tool use plays an epistemic role, and not merely a psychological one, in her rejection of the theory that birds don’t use tools. I conclude that in at least some cases, theory clarifcation can epistemically contribute to the rational rejection of the theory clarifed. In the next section, I will argue that thought experiments can clarify philosophical theories and that they can do this without the aid of intuition. Sometimes these thought experiments also elicit intuitions, but the way in which the thought experiments clarify theories is independent of those intuitions. More ultimately, I will argue that when thought experiments help overturn philosophical theories, they do so via intuition and theory clarifcation working together. So, for instance, the transplant case can help justify a philosopher’s rejection of utilitarianism partly because it clarifes for that
Thought Experiments as Tools of Theory Clarifcation
257
philosopher what utilitarianism says about the case and why it says it, and not (merely) by eliciting the intuition that non-consensual organ harvesting is wrong. 16.2.
Theory Clarifcation from Thought Experiment, Without Intuition
The claim of this section is that reasoners can deepen their understanding of a theory by refecting on relevant cases. Moreover, in at least some cases, this theory clarifcation doesn’t proceed by way of intuition.16 To see this, consider the following thought experiment: Apple Case
Siobhan has three apples and can either give all three to Jeremiah or else can give one to Jeremiah, one to Chandra, and one to Monique. Distributing the apples in the latter way would result in more well-being than giving them all to Jeremiah, due to the diminishing marginal returns of any one person’s receiving more than one apple. According to act utilitarianism, what should Siobhan do? This case is a kind of thought experiment, but it is not one whose function is to challenge utilitarianism. Utilitarianism recommends that Siobhan distribute the apples evenly, and this is not a counter-intuitive result; indeed, reasoners might have no intuition whatsoever about whether this is the right thing to do. At the same time, this thought experiment clarifies utilitarianism in the sense that it can deepen a reasoner’s appreciation of what the view predicts about the case and how the elements of the theory are tied to that prediction. So, even if thought experiments do have an epistemic function of generating intuitions, this is not their only function. They can also help clarify theories, and this latter function is in principle separable from their intuitiongenerating function. In at least some cases, such as the apple case, these functions come apart. Here is another reason to think that thought experiments can clarify theories without the aid of intuition: Consider the small fraction of reasoners who, when considering the transplant case, lack a strong intuition about whether the doctor’s action is morally permissible. For such reasoners, refecting on this case might nevertheless clarify utilitarianism, insofar as it can help such reasoners appreciate what the view predicts in the case and how the elements of the view generate this prediction. So, here too, we have some evidence that thought experiments can clarify theories without the help of intuition.
258
Grace Helton
So far, I’ve argued that thought experiments can clarify theories without the aid of accompanying intuitions. But one might object that theory clarifcation itself proceeds via intuition. For instance, perhaps one just “sees” that utilitarianism requires that Siobhan distribute the apples equally, in the sense that when considering the question, one generates a non-inferential assessment accompanied by a certain “feel.” Or, to extend this thought to the transplant case: Perhaps reasoners just “see” via intuition that utilitarianism predicts that the doctor should kill the patient. If this is right, theory clarifcation from cases is either itself an intuition or is generated by one. Since intuitions are, at a minimum, non-inferential mental states accompanied by a “feel,” it will sufce to show that the relevant forms of clarifcation do not proceed via intuition that they proceed via substantial inference (regardless of whether they also have a distinctive “feel”). Along these lines, I will argue that the relevant assessments are inferential, even if those inferences are rapid and implicit. My argument will focus on the transplant case, but these points can be extended mutatis mutandis to the apple case. Let’s refect: Does one consider the transplant case and just “see” that the doctor’s action maximizes overall well-being and hence, is what utilitarianism requires? For some reasoners, this assessment might be made extremely rapidly, such that to them it might seem to them that they simply “see” the point without inference. Nevertheless, there are reasons to think this assessment is in fact inferential, even if, for some reasoners, this inference occurs quickly. To make this point, I will draw on anecdotal evidence about some of the strategies commonly employed by ethics instructors who teach this thought experiment. Many of us who have taught this thought experiment will be familiar with the phenomenon that not all students immediately appreciate that the doctor’s action maximizes well-being. To make this point, many of us will have attempted to motivate the point by, for instance, drawing on a chalkboard “units” of well-being, represented, for instance, by squares—we might have resorted to jokingly calling these units of wellbeing “utiles.” We might have lined up squares representing the units of well-being of the fve patients whose lives were saved against squares representing the units of well-being of the patient whose life was taken, in a kind of macabre exercise of addition. This activity looks like something designed to scafold inference. It helps students undertake a certain reasoning process so that they can reach the conclusion that the doctor’s action maximizes overall well-being.17 Moreover, many of us will have encountered from some of our students the objection that the doctor’s action does not maximize well-being because her behavior dramatically threatens the psychic well-being of the public, who will no doubt be alarmed to learn that their local doctors are
Thought Experiments as Tools of Theory Clarifcation
259
sometimes in the business of non-consensually carving up some of their patients. This objection is a good one in that the case is not a counterexample to utilitarianism unless one can refute it. Many of us in our role as instructors will have at this point refned the case to stipulate that the doctor’s action is secret and one-of. It is not, we will stress, a medical policy, one the public is eventually bound to discover. As with the adding up of “utiles,” this activity looks like something designed to scafold inference. For the sake of making the strongest case possible against utilitarianism, we are attempting to lead the students to the conclusion that the doctor’s action maximizes well-being and thus, is what utilitarianism requires. One might object to these points that intuitions can be guided in certain verbally cued, inference-like ways without themselves being inferred. Elijah Chudnof (2017) develops this general point that intuitions can sometimes be guided by verbal cues, some of which might seem to scafold inference. For instance, he suggests that one can help a viewer “see” both fgures in this bistable stimulus in Figure 16.1 via instruction. For instance, one
Figure 16.1 A stimulus viewable as an old woman or a young woman.
260
Grace Helton
might say, “the old woman’s nose is the young woman’s jaw line and the old woman’s mouth is a necklace on the young woman,” and in this way help a viewer see that the fgure can be seen either as an old woman or as a young woman (Chudnof 2017: 381). However, as Chudnof argues, this guidance doesn’t mean that seeing itself is inferential. It is merely that these verbal guides help to alter the viewer’s attention in a way that triggers the relevant visual experience. Adopting this suggestion for the case of the transplant case (and extending it beyond Chudnof’s intended usage), one might suggest that the kinds of considerations ethics instructors adduce in the classroom—for instance, considerations from numbers of “utiles”—function in a similar way. These considerations help students “see” that the doctor’s action maximizes overall well-being, but they do so by triggering an intuition, not by scafolding inference. I cannot rule out the bare possibility that these kinds of activities—viz., those employed by ethics instructors to help students appreciate that the doctor’s action maximizes well-being—function merely to trigger an intuition that the doctor’s action maximizes well-being and not to scafold inference. However, the best explanation of these practices is that they scafold inference, not that they merely trigger intuition via some non-inferential route. Consider that the claims ethics instructors adduce together entail the conclusion that the doctor’s action maximizes well-being. So, it would be very surprising if the fact that these claims can together help justify the relevant conclusion were somehow incidental to the fact that they help students form the relevant assessment. Moreover, the ethics instructor’s promptings are plausibly unlike the kinds of verbal guides one might use to try to help someone see both renderings of a bistable stimulus, such as that in Figure 16.1. For, the prompts employed to help one “see” a particular rendering of a bistable stimulus likely function by directing one’s visual attention in such a way so as to trigger the other visual rendering of the stimulus.18 These promptings needn’t constitute premises in an inference whose conclusion is (say) “this is a young woman.” I conclude that thought experiments can clarify theories and that their ability to do so does not proceed via intuition. First, the co-presence of certain common intuitions is not required for theory clarifcation. For instance, a reasoner might appreciate that in the apple case, utilitarianism requires that Siobhan distribute the apples evenly without having any intuition whatsoever about whether this action is morally wrong or right. Second, the assessments by which thought experiments clarify theories are not themselves intuitions; they are inferences. For instance, the assessment that, in the transplant case, utilitarianism requires that the doctor kill the patient is not itself an intuition. Rather, this assessment is an inference
Thought Experiments as Tools of Theory Clarifcation
261
about an element of the case, one derived by mapping the core element of utilitarianism, the one to do with maximizing well-being, onto the case. In the next section, I will argue that thought experiments can sometimes help overturn philosophical theories partly because they clarify the relevant theories. At the same time, theory clarifcation does not play this role on its own; theory clarifcation plays this role in concert with relevant intuitions. 16.3.
Theory Clarifcation Can Help Overturn Philosophical Theories
As mentioned at the chapter’s outset, most theorists implicitly presume that if thought experiments help overturn philosophical theories, they do through generating intuitions. As I will now argue, this picture is, at best, incomplete. When thought experiments help overturn philosophical theories, they do so at least partly through theory clarifcation. Consider once more the novice philosopher who: learns what utilitarianism is for the frst time, provisionally accepts the view, and then, after some line of reasoning that involves the transplant thought experiment, comes to reject the view. Here are three (potentially non-exhaustive) routes by which this philosopher’s rejection of utilitarianism might be justifed by the thought experiment: Disconfrmation Alone
Via the thought experiment, the philosopher comes to appreciate something about morality, and this fact alone explains the fact that the thought experiment helps to justify her rejection of utilitarianism. Theory Clarifcation Alone
Via the thought experiment, the philosopher comes to appreciate something about utilitarianism, and this fact alone explains the fact that the thought experiment helps to justify her rejection of utilitarianism. Both Disconfrmation and Theory Clarifcation
Via the thought experiment, the philosopher comes to appreciate both something about morality and something about utilitarianism and these facts together explain the fact that the thought experiment helps to justify her rejection of utilitarianism. I will argue against the “disconfrmation alone” explanation. The fact that the transplant case helps justify the philosopher’s rejection of utilitarianism
262
Grace Helton
is not explained wholly by the fact that the case helps her appreciate something about morality. Rather, this explanation must appeal, at least in part, to the fact that the case helps the philosopher appreciate something about utilitarianism. Specifcally, I will develop the view that the case helps the philosopher appreciate both something about morality and something about utilitarianism and that these epistemic improvements together help justify her rejection of utilitarianism. Here is how the transplant case helps to justify the philosopher’s rejection of utilitarianism. First, the case helps to generate the philosopher’s intuition that the doctor’s killing of the patient is morally impermissible. Second, the case helps to elicit the philosopher’s inference that utilitarianism requires that the doctor kill her patient. Notice that while the frst assessment is intuitive and the second is inferred, both are about the very same action, the doctor’s killing of her patient. Moreover, it is essential for the thought experiment to do its work that the very same of the doctor’s actions be both morally impermissible and mandated by utilitarianism. For it is only because the case justifes the reasoner in thinking that the very same action has both of these traits simultaneously that the case constitutes a counterexample to utilitarianism. To see this, consider once more the argument against utilitarianism: Argument Against Act Utilitarianism (1) The doctor’s action in the transplant case is morally wrong. (2) The doctor’s action (the same action as in (1)) maximizes well-being. (3) According to act utilitarianism, the doctor’s action in the transplant case is not morally wrong. (4) Act utilitarianism is false. In this formulation of the argument, (1) and (2) must be about the same action. Otherwise, the argument would not be valid. While this point is trivial enough, it is not trivial that reasoners can become justifed in holding (1) and (2) simultaneously through refection on the transplant case.19 When this happens, it is because of the intuitive justifcation of (1) and the inferred justifcation of (2). In other words, it is through disconfrmation and theory clarifcation, working together, that the case somehow helps to justify the argument. So, it is false that the case buttresses the argument solely by enhancing the philosopher’s appreciation of moral facts. Of at least equal importance, the case permits a kind of inference about utilitarianism, and this latter inference is also critical to the philosopher’s being justifed in rejecting utilitarianism. What is it about the thought experiment that explains how it can justify a reasoner in thinking that the same action is simultaneously morally
Thought Experiments as Tools of Theory Clarifcation
263
impermissible and mandated by utilitarianism? This question is entangled with multiple empirical and epistemological questions, and it is not one I could hope to answer here. But, as a frst attempt at an answer: Plausibly it is something about the specifcity of the thought experiment that justifes this assessment, much as some models, in virtue of their specifcity, permit reasoners to see how multiple traits can inhere simultaneously in the same entity. Compare: One might point to a topographical map of the Earth to see how a location could both be ~4,000 feet in elevation and located in the Northern Hemisphere. The Mojave Desert is such a location, and reading this kind of map can justify a reasoner in thinking that the Mojave has both such traits. This isn’t to suggest that a topographical map is the only way to be so justifed; testimony could also play this role. Rather, the present claim is that when it is the map that justifes one in thinking that the Mojave has both these traits, there is something about the specifcity of the model that explains how it justifes this. Likewise, something about considering a particular action of the doctor helps reasoners appreciate that the very same action that is morally impermissible is also mandated by utilitarianism.20 In its specifcity, the thought experiment facilitates seamlessly coordinated reasoning about features simultaneously had by the same action.21 I conclude that the fact that thought experiments can help to overturn philosophical theories is not wholly explained by the intuitions that those thought experiments generate. Rather, thought experiments help to overturn philosophical theories at least partly through helping to clarify the very theories that are overturned. Think once more of Maya, the theorist who provisionally accepted that birds do not use tools, even though she had evidence that birds engage in prey dropping. It was only after coming to appreciate that her own theory counted prey dropping as tool use that Maya was justifed in revising her view. In other words, Maya’s empirical evidence about the objects in the theory’s domain was inadequate to justify the rejection of her view until she deepened her understanding of the theory. Likewise, for the philosopher who contemplates the organ harvesting case, it is partly because the case reveals that sometimes non-consensual organ harvesting is a way of maximizing well-being and thus is in those cases required by utilitarianism that the case can help justify the philosopher’s rejection of utilitarianism. 16.4. Is Intuition a Red Herring? I have argued that thought experiments can facilitate both disconfrmation of a theory and clarifcation of the same theory and that it is only when these functions work together that thought experiments can help overturn philosophical theories. For instance, the transplant thought experiment
264
Grace Helton
helps to facilitate an intuition that the doctor’s killing of the patient is morally impermissible, and it also helps to ground an inference that utilitarianism morally requires that the doctor kill the patient. And both this intuition and this inference, made about one and the same action, are essential in explaining how the case can help to justify the rejection of utilitarianism. These claims have important implications for metaphilosophy. First, as previously mentioned, some theorists criticize the role of thought experiments in philosophy by way of criticizing the epistemic role of intuitions. These theorists implicitly presume that if thought experiments play any epistemic role in theory revision, it is exclusively by way of intuitions, such that criticizing the epistemic status of intuitions amounts to a criticism of the method of cases. But if the points of this chapter are correct, even if critics of the method of cases are right that intuitions never epistemically matter in overturning philosophical theories, it wouldn’t follow that thought experiments are epistemically inert in overturning such theories. To make this further claim, they must show that thought experiments play no epistemic role in theory clarifcation or else that theory clarifcation is epistemically inert in the overturning of philosophical theories. By the same token, advocates of the method of cases, in focusing almost exclusively on the epistemic role of intuition, have neglected a distinct way in which cases can help justify the rejection of philosophical theories. I close with a further speculative and more radical thought. I have been supposing that the reason thought experiments can help overturn philosophical theories is partly because of the intuitions that such cases can elicit. For instance, I have been supposing that the intuition it is wrong for the doctor to kill her patient partly explains how the transplant case justifes a rejection of utilitarianism. But now, I would like to briefy entertain a rival view according to which intuitions about cases play no epistemic role in the overturning of philosophical theories. Here is one way intuitions might be epistemically inert in the overturning of philosophical theories. Suppose that the typical reasoner who considers the transplant case is already justifed, before refecting on the case, in thinking that non-consensual transplant is morally wrong. In a certain light, this supposition isn’t just coherent; it is nearly undeniable. For, how could a typical adult fail to possess good evidence that it’s wrong to carve someone up without their consent? If this is right, then this reasoner’s intuition about the wrongness of the doctor’s action does not obviously play an epistemic role in her rejection of utilitarianism, even though it does play a psychological role in explaining her rejection. For, the intuition she has about the case does not constitute new evidence for her. It merely encodes or refects evidence she already possesses. At the same time, this intuition plausibly is psychologically necessary in her rejection of the theory since
Thought Experiments as Tools of Theory Clarifcation
265
she presumably needs to token the relevant judgment in order to be able to use it in the service of an argument against utilitarianism. Suppose further that the core points of this chapter concerning theory clarifcation are correct: Thought experiments can play the epistemic role of helping philosophers clarify theories; theory clarifcation in turn plays an epistemic role in helping to justify the rejection of certain philosophical arguments; and theory clarifcation is not driven by intuition. Taken together, these points would suggest that the mainstream picture of thought experiments as functioning primarily in terms of intuitions is not just incomplete. It is entirely misguided. For, on this view, thought experiments play an epistemic role in overturning philosophical theories, but not through whatever intuitions they might generate. On this more radical view I have tentatively sketched, we should think of cases such as the transplant case as primarily functioning to tell us something about utilitarianism, the ethical theory, not about morality itself. Likewise, we should take cases such as the Gettier case as primarily functioning to tell us something about the tripartite theory of knowledge, not about knowledge itself. I leave the viability of such a view as a further question.22 Notes 1 For the belief view, see Gopnik and Schwitzgebel (1998) and Ludwig (2007); For the disposition to believe view, see Pust (2000) and Erlenbaugh and Molyneux (2009). For the sui generis state view, see Bengson (2015), Brogaard (2014), Chudnof (2011), and Koksvik (2011). 2 Even the intuition skeptic Cappelen (2012) and the intuition advocate Brogaard (2014) can agree on this much. For a helpful overview, see Pust (2019). 3 See Brown and Yiftach (2022) for a helpful overview of the nature of thought experiments. 4 Thomson (1976). 5 It is contentious whether (1) should be interpreted as a necessity claim, a counterfactual claim, a possibility claim, or something else. See Williamson (2007), Ichikawa and Jarvis (2009), Malmgren (2011), and Pust (2019) for discussion and competing perspectives. 6 Often, the method of cases is construed as part of a broader strategy of moving between less general and more general considerations. In this chapter, I’m neutral on how less general considerations, such as those raised by the argument against utilitarianism ought to be balanced against more general considerations. See McGrath and Kelly (2015) and McGrath (2019b) for the claim that levels of generality are irrelevant to philosophical theorizing. 7 See Bengson (2014, 2015), Brogaard (2014), Chudnof (2013, 2017), Devitt (2015), and Neta (2012). 8 See Alexander and Weinberg (2014), Cappelen (2012), Deutsch (2015), and Williamson (2007). 9 In contrast to the view of philosophical thought experiments I sketch in the chapter’s conclusion, I fnd it implausible that scientifc thought experiments
266
10 11
12
13 14
15 16
17 18 19
20
Grace Helton
function only to clarify theories. Instead, I fnd Clatterbuck’s (2013) suggestion that scientifc thought experiments can justify simulation-scafolded inductive judgments to be a compelling one. Here, as elsewhere, I suspect that an overly close comparison with the natural sciences has misled philosophers about the project of their own discipline (Helton forthcoming). Möglich and Alpert (1979). For present purposes, I am neutral on whether understanding is a kind of cognitive map (Gopnik, Glymour, & Sobel 2002; Gopnik et al. 2004; Grimm 2016), a skill (Hills 2016; Khalifa 2017; Cf. Sullivan 2018), a kind of relevance matching (Roush 2016, 2017), or a propositional attitude. I am employing “understanding” in a factive way (Cf. McSweeney 2023) and am presuming that understanding is usable for the reasoner (see Elgin 2009 and Roush 2017: 404). For helpful recent overviews, see Baumberger, Breisbart, and Brun (2017), Grimm (2021), and Hannon (2021). My notion of clarifying a theory bears important comparisons with Henk de Regt’s notion of making a theory intelligible (de Regt 2017, de Regt and Dieks 2005) and with Michaela McSweeney’s (2020) notion of a theory’s being higher-order virtuous. Prototypes for concepts aren’t typically thought to determine the extension of those concepts, so it’s plausible that the exemplars Maya uses might dissociate from the extension of her concept. See Hampton (2006), Rosch (2011), and Del Pinal (2016) for recent discussion. Beck (1982) and Boire, Nicolakakis, and Lefebvre (2002). See Hunt, Gray, and Taylor (2013) for tool use in general. Conceptual engineering is plausibly an additional route by which theories can be rationally revised. See, for example, Chalmers (2020), Eklund (2021), Haslanger (2020), Isaac, Koch, and Nefdt (2022), and Plunkett (2015). Silva and Oliveira (forthcoming). By arguing that thought experiments permit theory clarification without intuition, I do not mean to deny that imagination—both of the case itself and of how the theory might be true—might play a substantial role in theory clarification. Specifically, I take my claim to be consistent with McSweeney’s (2023) important and rich development of the view that philosophy (especially metaphysics) aims at helping us grasp a theory by imagining how the theory might be true. Intuition is a kind of rapid and non-inferential assessment, not a form of imagining. My claim is that cases permit theory clarification through a non-intuited inference, but this is consistent with this inference’s itself requiring a kind of imaginative rendering of those cases and also with the resultant form of theory clarification itself being a kind of subjectively seeming to be true of the kind McSweeney champions as an aim of philosophy. See also McSweeney (2016) for the view that theories should be individuated at least partly by occupiable perspectives. Alternatively, these activities might partly constitute reasoning, as defended by Dutilh Noaves (2013). Either way, the activity involves inference. Thanks to Scott Stapleford for discussion on this point. Tsal and Kolbet (1985) and Long and Toppino (2004). I do not claim that it is only through thought experiment that one might be simultaneously justifed in holding (1) and (2). For instance, perhaps testimony could confer this kind of justifcation. See McGrath (2019a: 59–105, 2021) for a defense. Philosophical use of thought experiments is perhaps comparable to the use of concrete examples in math and science learning, which can help facilitate more
Thought Experiments as Tools of Theory Clarifcation
267
abstract forms of reasoning. See Fife et al. (2014), Goldstone and Son (2005), Koedinger, Alibali, and Nathan (2008), Novak et al. (2014), and Schwartz et al. (2011). 21 To suggest that thought experiments have a kind of specifcity is not to suggest that they are tokens as opposed to types. Even if thought experiments are types, they still might permit us to focus on specifc aspects of that type. Special thanks to Elijah Chudnof for discussion on this point. 22 For extremely helpful feedback on this paper, I am indebted to Josh Armstrong, Elijah Chudnof, Kevin McCain, Michaela McSweeney, Chris Register, Scott Stapleford, and audience members at the Workshop on Seemings and Intuitions. Special thanks to Mark van Roojen for inspiring this paper by way of a remark, made ofhand in an undergraduate ethics course of his in which I was a student, that in doing ethics, we use moral knowledge we already have.
References Alexander, J., & Weinberg, J. M. (2014). The “unreliability” of epistemic intuitions. In Current controversies in experimental philosophy (pp. 128–148). Routledge. Baumberger, C., Beisbart, C., & Brun, G. (2017). What is understanding? An overview of recent debates in epistemology and philosophy of science. In S.R. Grimm, C. Baumberger, & S. Ammon (Eds.), Explaining understanding: New perspectives from epistemology and philosophy of science, 1–34. Beck, B. B. (1982). Chimpocentrism: Bias in cognitive ethology. Journal of Human Evolution, 11(1), 3–17. Bengson, J. (2014). How philosophers use intuition and “intuition.” Philosophical Studies, 171(3), 555–576. Bengson, J. (2015). The intellectual given. Mind, 124(495), 707–760. Boire, D., Nicolakakis, N., & Lefebvre, L. (2002). Tools and brains in birds. Behaviour, 139(7), 939–973. Brogaard, B. (2014). Intuitions as intellectual seemings. Analytic Philosophy, 55(4). Brown, J. R., & Yiftach, F. (2022). Thought experiments. In Edward N. Zalta & Uri Nodelman (Eds.), The Stanford encyclopedia of philosophy (Winter 2022 ed.). https://plato.stanford.edu/archives/win2022/entries/thought-experiment/. Cappelen, H. (2012). Philosophy without intuitions. Oxford University Press. Chalmers, D. J. (2020). What is conceptual engineering and what should it be? Inquiry, 1–18. Chudnof, E. (2011). What intuitions are like. Philosophy and Phenomenological Research, 82(3), 625–654. Chudnof, E. (2013). Intuition. Oxford University Press. Chudnof, E. (2017). The reality of the intuitive. Inquiry, 60(4), 371–385. Clatterbuck, H. (2013). The epistemology of thought experiments: A non-eliminativist, non-platonic account. European Journal for Philosophy of Science, 3(3), 309–329. Del Pinal, G. (2016). Prototypes as compositional components of concepts. Synthese, 193, 2899–2927. de Regt, H. W. (2017). Understanding scientifc understanding. Oxford University Press. de Regt, H. W., & Dieks, D. (2005). A contextual approach to scientifc understanding. Synthese, 144, 137–170. Deutsch, M. E. (2015). The myth of the intuitive: Experimental philosophy and philosophical method. MIT Press.
268
Grace Helton
Devitt, M. (2015). Relying on intuitions: Where Cappelen and Deutsch go wrong. Inquiry, 58(7–8), 669–699. Dutilh Noaves, C. D. (2013). Mathematical reasoning and external symbolic systems. Logique et Analyse, 56(221), 45–65. Earlenbaugh, J. & Molyneux, B. (2009). Intuitions are Inclinations to Believe, Philosophical Studies, 145, 89–109. Eklund, M. (2021). Conceptual engineering in philosophy. In J. Khoo & R. Sterken (Eds.), Routledge Handbook of Social and Political Philosophy of Language (pp. 15–30). New York: Routledge. Elgin, C. (2009). Is understanding factive? Epistemic Value, 322–330. Fyfe, E. R., McNeil, N. M., Son, J. Y., & Goldstone, R. L. (2014). Concreteness fading in mathematics and science instruction: A systematic review. Educational Psychology Review, 26, 9–25. Goldstone, R. L., & Son, J. Y. (2005). The transfer of scientifc principles using concrete and idealized simulations. The Journal of the Learning Sciences, 14(1), 69–110. Gopnik, A., Glymour, C., & Sobel, D. M. (2002). Causal maps and bayes nets: A cognitive and computational account of theory-formation. In The cognitive basis of science (pp. 117–132). Cambridge University Press. Gopnik, A., Glymour, C., Sobel, D. M., Schulz, L. E., Kushnir, T., & Danks, D. (2004). A theory of causal learning in children: Causal maps and Bayes nets. Psychological Review, 111(1), 3. Gopnik, A. & Schwitzgebel, E. (1998). Whose concepts are they, anyway? The role of philosophical intuition in empirical psychology. In M. DePaul and W. Ramsey (Eds.), Rethinking intuition (pp. 75–91). Rowman & Littlefeld. Grimm, S. (2016). Understanding and transparency. In Explaining understanding (pp. 228–245). Routledge. Grimm, S. (2021). Understanding. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy. https://plato.stanford.edu/archives/sum2021/entries/understanding/ Hampton, J. (2006). Concepts as prototypes. In B. H. Ross (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol. 46, pp. 79–113). Elsevier. Hannon, M. (2021). Recent work in the epistemology of understanding. American Philosophical Quarterly, 58(3), 269–290. Haslanger, S. (2020). How not to change the subject? In T. Marques & A. Wikforss (Eds.), Shifting Concepts: The Philosophy and Psychology of Conceptual Variability (pp. 235–259). Oxford: Oxford University Press. Hills, A. (2016). Understanding why. Noûs, 50, 661–688. Wikforss (Eds.), Shifting concepts: The philosophy and psychology of conceptual variation (pp. 235–259). Oxford University Press. Helton, G. (forthcoming). Viewpoint convergence as a philosophical defect. In Sanford Goldberg & Mark Walker (Eds.), Attitude in philosophy. Oxford University Press. Hunt, G. R., Gray, R. D., & Taylor, A. H. (2013). Why is tool use rare in animals. In Sanz C., Call J. & Boesch C. (Eds.), Tool use in animals: Cognition and ecology (pp. 89–118). Cambridge University Press. Ichikawa, J., & Jarvis, B. (2009). Thought-experiment intuitions and truth in fction. Philosophical Studies, 142(2), 221–246.
Thought Experiments as Tools of Theory Clarifcation
269
Isaac, M. G., Koch, S., & Nefdt, R. (2022). Conceptual engineering: A road map to practice. Philosophy Compass, 17(10), e12879. Khalifa, K. (2017). Understanding, explanation, and scientifc knowledge. Cambridge: Cambridge University Press. Koedinger, K. R., Alibali, M. W., & Nathan, M. J. (2008). Trade-ofs between grounded and abstract representations: Evidence from algebra problem solving. Cognitive Science, 32(2), 366–397. Koksvik, O. (2011). Intuition, Ph.D. Thesis, Australian National University. Long, G. M., & Toppino, T. C. (2004). Enduring interest in perceptual ambiguity: Alternating views of reversible fgures. Psychological Bulletin, 130(5), 748. Ludwig, K. (2007). The epistemology of thought experiments: First person versus third person approaches. Midwest Studies in Philosophy, Philosophy and the Empirical, 31(1), 128–159. Malmgren, A. S. (2011). Rationalism and the content of intuitive judgements. Mind, 120(478), 263–327. McGrath, S. (2019a). Moral knowledge. Oxford University Press. McGrath, S. (2019b). Philosophical methodology and levels of generality. Philosophical Perspectives, 33(1), 105–125. McGrath, S. (2021). Epistemic autonomy for social epistemologists: The case of moral inheritance. In Epistemic autonomy (pp. 271–287). Routledge. McGrath, S., & Kelly, T. (2015). Soames and Moore on method in ethics and epistemology. Philosophical Studies, 172, 1661–1670. McSweeney, M. M. (2016). An epistemic account of metaphysical equivalence. Philosophical Perspectives, 30, 270–293. McSweeney, M. M. (2020). Theories as recipes: Third-order virtue and vice. Philosophical Studies, 177(2), 391–411. McSweeney, M. M. (2023). Metaphysics as essentially imaginative and aiming at understanding. American Philosophical Quarterly, 60(1), 83–97. Möglich, M. H., & Alpert, G. D. (1979). Stone dropping by Conomyrma bicolor (Hymenoptera: Formicidae): A new technique of interference competition. Behavioral Ecology and Sociobiology, 105–113. Neta, R. (2012). Knowing from the armchair that our intuitions are reliable. The Monist, 95(2), 329–351. Novack, M. A., Congdon, E. L., Hemani-Lopez, N., & Goldin-Meadow, S. (2014). From action to abstraction: Using the hands to learn math. Psychological Science, 25(4), 903–910. Plunkett, D. (2015). Which concepts should we use?: Metalinguistic negotiations and the methodology of philosophy. Inquiry, 58(7–8), 828–874. Pust, J. (2000). Intuitions as evidence. Garland/Routledge. Pust, J. (2019). Intuition. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Summer 2019 ed.). https://plato.stanford.edu/archives/sum2019/entries/ intuition/. Rosch, E. H. (2011). Slow lettuce: Categories, concepts, fuzzy sets, and logical deduction. In R. Belohlavek & G. J. Klir (Eds.), Concepts and fuzzy logic (Chap. 4). The MIT Press. Roush, S. (2016). Simulation and understanding other minds. Philosophical Issues, 26(1), 351–373.
270
Grace Helton
Roush, S. (2017). The diference between knowledge and understanding. In Rodrigo Borges, Claudio de Almeida, & Peter D. Klein (Eds.), Explaining knowledge: New essays on the gettier problem (p. 384). Oxford University Press. Schwartz, D. L., Chase, C. C., Oppezzo, M. A., & Chin, D. B. (2011). Practicing versus inventing with contrasting cases: The efects of telling frst on learning and transfer. Journal of Educational Psychology, 103(4), 759. Silva, P., & Oliveira, L. R. G. (forthcoming). Propositional justifcation and doxastic justifcation. In M. Lasonen-Aarnio & C. M. Littlejohn (Eds.), Routledge handbook of the philosophy evidence. Routledge. Sullivan, E. (2018). Understanding: not know-how. Philosophical Studies, 175, 221–240. Thomson, J. J. (1976). Killing, letting die, and the trolley problem. The Monist, 204–217. Tsal, Y., & Kolbet, L. (1985). Disambiguating ambiguous fgures by selective attention. The Quarterly Journal of Experimental Psychology Section A, 37(1), 25–37. Williamson, T. (2007). The philosophy of philosophy. Routledge.
17 Lessons from Commonsensism for Religious Epistemology Michael Bergmann
In this chapter, I argue that commonsense responses to radical skepticism can provide helpful lessons for religious epistemology—in particular, for thinking about how best to defend, and respond to, religious skepticism. Section 17.1 provides a brief summary of some of the main elements of the Reid-inspired epistemic-intuition-based commonsense response to radical skepticism developed in my 2021 book, Radical Skepticism and Epistemic Intuition. Section 17.2 highlights fve important lessons that can be drawn from the position presented in that book and applies them to religious epistemology. Section 17.3 explains how these lessons can help us account for our religious disagreements in more plausible ways (i.e., in ways that don’t implausibly disparage the rationality of those with whom we disagree). Section 17.4 addresses some potential concerns about Sections 17.2 and 17.3. 17.1.
Radical Skepticism, Commonsensism, and Epistemic Intuition
Chisholm taught us the diference between methodist and particularist approaches to skeptical worries.1 Methodists start with epistemic principles articulating criteria for rationality and use them to determine which of our beliefs are rational. Particularists start with clear cases of beliefs that are rational (and clear cases of beliefs that are irrational) and use these as a basis for formulating epistemic principles stating correct criteria for rationality—that is, criteria that correctly classify those clear cases. Those who fnd arguments for radical skepticism compelling typically proceed as methodists. Their skeptical arguments rely on a premise that endorses an epistemic principle. And these arguments conclude that large swaths of our ordinary beliefs fail to satisfy the criteria for rationality mentioned in the principle and, hence, aren’t rational. Commonsense responses to skepticism (responses of the sort given by Thomas Reid and
DOI: 10.4324/9781003299349-22
272
Michael Bergmann
G.E. Moore) proceed in a particularist way. Commonsensists endorse the following thesis: Commonsensism: (a) We rationally believe the most obvious things that most humans take themselves to rationally believe (this includes the truth of simple perceptual, memory, introspective, mathematical, logical, and moral beliefs); and (b) we also rationally believe (if we consider the question) that we are not in some skeptical scenario in which we are radically deceived in these widely held obvious beliefs.2 And because commonsensists are particularists, they insist that if the epistemic principles used in skeptical arguments imply that certain clear cases of rational beliefs are irrational, the principles should be rejected or modifed.3 My 2021 book flls out this account in terms of epistemic intuitions.4 Epistemic intuitions are seemings about epistemic value just as moral intuitions are seemings about moral value. Seemings are the distinctive mental states we’re in when things seem to us a certain way. Seemings have a presentational feel; they feel as if they’re revealing to us what reality is like. We very often treat them as evidence for their propositional content; this is how we treat perceptual seemings, memory seemings, seemings about simple mathematical and logical truths, and moral seemings. Some epistemic intuitions (or seemings) are treated as evidence for the epistemic principles that are their contents. Other epistemic intuitions are treated as evidence that a particular belief is rational or that another belief is irrational.5 Skeptically inclined methodists rely on epistemic intuitions as evidence in support of the epistemic principles they employ as premises in their skeptical arguments. And commonsense particularists rely on epistemic intuitions as evidence in support of their judgments that particular beliefs (targeted by skeptical arguments) are rational. Further details about my 2021 book’s account of the dispute between commonsensism and radical skepticism will be given in Section 17.2 as I present the lessons for religious epistemology. What I’ve said so far is intended merely to provide a rough initial account of the sort of commonsensism I’ll be relying on in drawing those lessons. Two caveats before considering the application to religious belief: First, in saying that there are lessons from commonsensism for religious epistemology, I am not saying that religious beliefs are commonsense beliefs. Nor am I saying that there are no diferences between the dispute between radical skeptics and commonsensists, on the one hand, and the dispute between religious skeptics and religious believers, on the other hand. Instead, I’m contributing to the decades-long tradition in contemporary
Lessons from Commonsensism for Religious Epistemology
273
philosophy of religion of emphasizing signifcant similarities between these two disputes.6 Second, I’m not claiming that the religious believer’s response to religious skepticism—the one that is highlighted and recommended in this chapter —is the only or best response to religious skepticism for all religious people. The claim is just that it is a rational response that many will fnd realistic and attractive. This can be so even if, given the intuitions and evidence that some religious people in fact have, they cannot honestly adopt the response to religious skepticism recommended in this chapter.7 17.2. 17.2.1.
Five Lessons Anti-skepticism and Noninferentially Rational Belief
Lesson 1: there is an anti-skeptical advantage for epistemologies that allow for beliefs targeted by skeptical objections to be properly basic (i.e., noninferentially rational) in the way that Reid, in opposition to Hume, allowed for perceptual and testimonial belief to be properly basic. Rather than accept an epistemic principle saying that one’s perceptual beliefs are rational only if one is able to infer their probable truth via a good argument from one’s sensory experience, commonsensists endorse a less demanding and more plausible standard—one that accommodates the view that our perceptual beliefs can be noninferentially rational. Since the average person’s perceptual beliefs are not based on arguments, and since sufciently good (noncircular) arguments in support of our perceptual beliefs haven’t been discovered even by the best philosophers, there is an enormous anti-skeptical advantage for those holding the Reidian view that perceptual beliefs are properly basic. Of course, this is an advantage only if this view is plausibly true. But Reid, Moore, and many others think it is plausibly true. It’s common knowledge to anyone familiar with developments in philosophy of religion over the past four decades that a parallel point applies in religious epistemology: given the difculty of identifying sufciently good arguments in support of religious beliefs, there is an anti-skeptical advantage for epistemologies endorsing the view that religious belief can be noninferentially rational in the ways proposed by Plantinga (2000), Alston (1991), and others. Here too, this is an advantage only if the view in question is plausibly true. But many philosophers (including me) think it is plausibly true. Thus, for the remainder of this chapter, all discussions of religious epistemologies will focus on religious epistemologies of this sort, calling them “noninferentialist” religious epistemologies.8
274 17.2.2.
Michael Bergmann Autodidactic vs. Proselytizing Responses to Skepticism
Lesson 2: anti-skeptics are wise to take an autodidactic rather than a proselytizing approach in responding to skepticism. The gold standard in dealing with disagreement is to defend one’s own position via arguments from compelling premises in such a way that rational and well-informed people who previously disagreed with you will be forced, on pain of irrationality, to concede that your view is the correct one and their opposing view is mistaken. Adopting this standard in dealing with skepticism is the proselytizing approach—that is, showing skepticismsympathizers, to their own satisfaction, that your nonskeptical view is the correct one. It has not met with success.9 Commonsensists don’t take this approach. Instead, their goal is to fully appreciate and appropriately respect the force of skeptical objections and to determine honestly—by their own lights, relying on their own epistemic intuitions—what rationality requires, belief-wise, in response to those objections. Commonsensists are aware that those inclined to skepticism (including many who are well-informed and philosophically competent) might not arrive at the same conclusions; that diference will have to be factored in when determining what rationality requires. But the question for the commonsensist is not “how can I convince the skeptically inclined to give up their skepticism?”; that’s the question for those who take the proselytizing approach. Instead, it’s “what is the principled rational response for me, in light of all the evidence I have, to these skeptical objections?”; this is the autodidactic approach that is typical of commonsensism.10 The autodidactic approach has often been taken by those who endorse noninferentialist religious epistemologies. Alston (1991) and Plantinga (2000) explain in principled ways, to their own satisfaction (in light of the evidence they have), why it is that their religious beliefs are rational or warranted. They realize that what they’re saying won’t persuade non-religious people to convert to religious belief. In fact, it may not even convince non-religious people that the religious person’s beliefs are rational (even if false). But since Alston and Plantinga are taking the autodidactic approach rather than the proselytizing approach, this doesn’t count as a failure to achieve their goal. Of course, nonskeptics have other-regarding hopes when they explain their autodidactic refections in their published writings (this is why they publish them). One such hope is that they may help fellow nonskeptics see how they too can rationally resist the force of skeptical objections. A second hope is that skeptics will come to see how it can rationally seem (from the perspective of nonskeptics) that their nonskeptical views are rational. A third, albeit dimmer, hope is that skeptics will become (1) more open to
Lessons from Commonsensism for Religious Epistemology
275
the kinds of evidence on which the nonskeptic relies and (2) more skeptical of the relative strength of the evidence supporting skepticism, thereby becoming more open to a nonskeptical view. Obviously, persuading skeptics would be nice. But since that isn’t the goal, not achieving it isn’t a failure to do what one was trying to do. Moreover, the skeptic is also failing to persuade the nonskeptic of the irrationality of her nonskeptical position (using evidence and premises that the nonskeptic views as rationally acceptable). So if this is a problem, it goes both ways.11 17.2.3.
Epistemic Intuitions on All Sides
Lesson 3: both skeptics and anti-skeptics rely heavily on epistemic intuitions, which contributes to leveling the playing feld. Epistemic intuitions—that is, seemings about rationality, either about epistemic principles identifying criteria of rationality or about whether particular beliefs are rational—are often treated as evidence. Importantly, this is true even of the skeptically inclined. In arguing for the irrationality of beliefs, they standardly rely on epistemic principles stating criteria for rationality that they think the targeted beliefs fail to satisfy. But their evidence in support of these principles is (plausibly) epistemic intuition. In fact, even those who think it isn’t rational to treat epistemic intuition as having evidential value are, it seems, relying on epistemic intuition as evidence in their endorsement of that very thought. There are multiple ways that epistemic intuitions factor into the exchange between commonsensists and skeptics. Both sides have epistemic intuitions in support of epistemic principles, though there are diferences regarding exactly which principles their epistemic intuitions support and how strongly. Both sides have epistemic intuitions in support of the rationality of particular beliefs and the irrationality of others; here too, there are differences in strength of intuition and degree of epistemic value attributed (whether positive or negative). In addition, both sides have higher-level epistemic intuitions (also of difering strength and attributing diferent degrees of epistemic value) about the epistemic quality and evidential value of the previous two kinds of epistemic intuitions just mentioned. That the skeptically inclined, and not just commonsensists, rely on epistemic intuition is worth noting because acknowledging this contributes to leveling the playing feld. It can be tempting for those impressed by skeptical arguments to view the commonsensists’ reliance on particularist epistemic intuitions (exemplifed famously by G.E. Moore holding up his hands and announcing his knowledge of their existence as
276
Michael Bergmann
external world objects) as stubbornly, dogmatically, fat-footedly, and simplistically holding their ground in the face of principled, refective, more sophisticated, and more philosophically serious objectors wielding impressively developed skeptical arguments. But when the dust settles, what we really have are the epistemic intuitions of the commonsensists in confict with the epistemic intuitions of the skeptics. The appearance of an elite or privileged higher ground occupied by the more sober-minded skeptical objector evaporates when we see that she and the commonsensist are relying on the same kind of evidence. Turning to religious epistemology,12 we can make the very same sorts of points. First, it’s important to highlight the fact that both religious skeptics and religious believers rely on epistemic intuitions. It’s true that noninferentialist religious epistemologies often highlight religious experience (including various religious seemings)13 as the evidence on which religious beliefs are based. And it’s also true that nonreligious people seem to either lack this sort of evidence or lack any inclination to treat it as evidentially valuable. But that isn’t the full story. In addition, both religious and nonreligious people often have (1) epistemic intuitions about the evidential value of the religious experience on which religious beliefs are based and (2) higher-level epistemic intuitions about the evidential value of the epistemic intuitions mentioned in (1). To religious people, it often seems—because of (1)—that it is epistemically appropriate to rely on their religious seemings in forming their religious beliefs. And if they’re refective, it will also often seem—because of (2)—that it is epistemically appropriate for them to have that positive view about the evidential value of their religious seemings. But for nonreligious people, it often seems—because of (1) in their own case— that the religious seemings of religious believers lack sufcient evidential value. And if these nonreligious people are refective, they’ll often think—because of (2) in their own case—that this negative view of the evidential value of the religious seemings of religious believers is itself entirely reasonable. Recognizing the relevance of epistemic intuitions of kinds (1) and (2) and the pervasive evidential role they play in the beliefs formed by those on both sides of a dispute between skeptics and anti-skeptics is crucial to understanding disagreements about skepticism, whether radical skepticism or religious skepticism. Both sides in these disputes are relying, often without realizing it, on evidence of the very same kind, namely, epistemic intuition. And this evidence often plays an unrecognized role in buttressing one’s own position and resisting opposing views. In the case of religious skepticism, this is important because it undermines the impression that the religious skeptic avoids relying on “questionable”
Lessons from Commonsensism for Religious Epistemology
277
evidence of the sort employed by the religious believer to whom she is objecting. 17.2.4.
The Importance of Intuitions About Particular Cases
Lesson 4: disagreements between skeptics and anti-skeptics in their epistemic intuitions about the rationality of particular beliefs play a signifcant and under-appreciated role in accounting for their overall disagreements. In disputes about radical skepticism with respect to perception and memory, there are at least four kinds of evidence: (a) sensory experience and memory seemings, (b) epistemic intuitions about the evidential value of the evidence mentioned in (a), (c) higher-level epistemic intuitions about the evidential value of the epistemic intuitions mentioned in (b), (d) epistemic intuitions about epistemic principles stating what the criteria for rationality are. The point I want to emphasize here is that the evidence mentioned in (b) plays a very signifcant role in determining whether one sides with antiskepticism or skepticism in this disagreement. And the evidence mentioned in (c) is nearly as important when people are more refective. It’s often the case that the epistemic principles to which skeptics appeal in their arguments—supported by the evidence mentioned in (d)—are viewed by all sides as being at least in the ballpark of some relevant truth. But the particularists in the dispute will tend to tweak and modify statements of these principles so that they accommodate the evidence mentioned in (b). And when challenges are raised against the legitimacy of using the evidence mentioned in (b), the evidence mentioned in (c) comes to the rescue.14 From the perspective of the commonsensist, the (b)- and (c)-evidence she highlights signifcantly outweighs the (b)- and (c)-evidence highlighted by the skeptically inclined and provides strong support for the rationality of our ordinary perceptual and memory beliefs. Perhaps there are also differences in the strength of the (d)-evidence had by these two groups, but there needn’t be, or at least such diferences needn’t be signifcant. Ultimately, the reason commonsensists aren’t moved by skeptical arguments is that—in light of their (b)- and (c)-evidence—they can’t take seriously the skeptical view that most of their ordinary perceptual and memory beliefs are irrational.
278
Michael Bergmann
A similar point can be made about disputes between religious believers and religious skeptics. Here too, there is (d)-evidence, but in place of (a) through (c) we have: (a*) religious seemings and religious experience, (b*) epistemic intuitions about the evidential value of the evidence mentioned in (a*), (c*) higher-level epistemic intuitions about the evidential value of the epistemic intuitions mentioned in (b*). And here too (b*)-evidence and (c*)-evidence play a very signifcant role. Many religious believers are intelligent, sincere truth-seekers who are very well-informed about skeptical objections to religious belief and who know of many other similarly qualifed people who hold religious beliefs similar to their own. A deep and personal familiarity with these people, in which their intelligence, erudition, and intellectual honesty are evident in abundance and completely intertwined with their religious beliefs and practices, makes it very difcult to take seriously the suggestion that their religious beliefs are manifestations of irrational belief-forming tendencies. In light of this knowledge of other people, these religious believers—who recognize that the religious beliefs of those in their community are based on (a*)evidence—have strong (b*)-evidence supporting the epistemic appropriateness of these religious beliefs.15 In addition, they have strong (c*)-evidence supporting this evaluative judgment about the epistemic appropriateness of these religious beliefs. As a result, the religious believer’s (d)-evidence in support of the epistemic principles used in skeptical objections to religious belief—including principles focused on the non-universality of the religious believer’s (a*)-evidence—is outweighed by the religious believer’s (b*)-evidence and (c*)-evidence, which can be accommodated by tweaking these epistemic principles.16 Of course, things are diferent for the religious skeptic. But this diference is not due solely to the fact that the religious skeptic doesn’t have the (a*)-evidence had by the religious believer (and, in fact, sometimes the religious skeptic has some evidence of this very same kind). Nor is it due to a signifcant diference in (d)-evidence between the religious believer and the religious skeptic (especially if they both endorse commonsensism in response to radical skepticism).17 Instead, the religious skeptic difers in having quite diferent (b*)-evidence—that is, she does not have a positive view of the relevant (a*)-evidence. Likewise, the religious skeptic’s (c*)evidence does not refect well on the religious believer’s (b*)-evidence, as described in the previous paragraph. As a result, the skeptical objections ofered to religious belief seem much more compelling to the religious skeptic than they do to the savvy and well-informed religious believer. It’s in
Lessons from Commonsensism for Religious Epistemology
279
large part the diference between the religious skeptic and the religious believer with respect to their (b*)- and (c*)-evidence that explains this.18 Four decades ago, Plantinga made a point in the same neighborhood when he was discussing the relevance of particularism to the dispute between the religious believer and the religious skeptic: criteria for proper basicality must be reached from below rather than above; they should not be presented ex cathedra but argued to and tested by a relevant set of examples. But there is no reason to assume, in advance, that everyone will agree on the examples. The Christian will of course suppose that belief in God is entirely proper and rational; if he does not accept this belief on the basis of other propositions, he will conclude that it is basic for him and quite properly so. Followers of Bertrand Russell and Madelyn Murray O’Hare may disagree; but how is that relevant? Must my criteria, or those of the Christian community, conform to their examples? Surely not. The Christian community is responsible to its set of examples, not to theirs.19 In this passage, Plantinga is taking for granted Lessons 1 and 2 (about religious belief being noninferential and about the wisdom of taking the autodidactic approach), and he is highlighting Lesson 4, under discussion here. Plantinga doesn’t put it in precisely this way, but his point is essentially that religious believers have very diferent (b*)-evidence and (c*)-evidence than religious skeptics have, and this plays a signifcant role in their view on the rationality of their religious belief. Why is this old news worth highlighting? Because those involved in disagreements about the rationality of religious belief often seem not to recognize the very signifcant role played by the diferences in the (b*)and (c*)-evidence possessed by those involved in the dispute. There is a tendency to emphasize the diference in (a*)-evidence instead, as if this is the only relevant evidence at issue in the dispute. But this can distract us from the fact that disagreements about religious skepticism often pit one set of epistemic intuitions about cases against another set of epistemic intuitions about cases. And that distraction leads to misunderstandings about what’s actually going on in these religious disagreements and can distort assessments of the rationality of those who hold an opposing viewpoint. 17.2.5.
The Strength of the Anti-skeptical Web
Lesson 5: anti-skeptics disagree with skeptics in a complex and coherent set of ways composing a larger anti-skeptical web of positions that is impressively resistant to skeptical attack.
280
Michael Bergmann
It’s clear from the previous lesson that in disagreements between commonsensists and those inclined to radical skepticism, there are multiple kinds of evidence—that is, (a)- through (d)-evidence—and that this evidence can be diferent for those on opposite sides of this dispute. And Section 17.1 made it clear that there is often a methodological disagreement between methodist-leaning skeptics and particularist-leaning anti-skeptics regarding how best to treat and weight that evidence. As a result, the full evidential picture from the commonsensist perspective will be quite diferent than it is from the skeptical perspective. Importantly, there may well be multiple points of diference—for example, diferences in the content or strength of the epistemic intuitions making up one’s (b)-evidence, (c)-evidence, or (d)-evidence, in addition to diferences in the strength of one’s leanings toward methodism or particularism (which will, perhaps, be intertwined with the diferences in epistemic intuitions just mentioned). Objections targeting just one point of disagreement (e.g., disagreements about epistemic principles) will run into an interconnected web of positions on the other side, which hold together well and jointly stand in opposition to multiple diferent positions held by the objector. Because so many seemingly plausible views supported by diferent kinds of evidence will need to be overturned (for the objection to be successful), objections need to be stronger than they would need to be if they were targeting only an isolated position. In the religious case, parallel remarks apply. There is diferent evidence on both sides of disputes about the rationality of religious belief, including a diference in (a*)-evidence consisting of religious seemings and experience. But just as important is the diference in (b*)- and (c*)-evidence on both sides, which will interact diferently with the (d)-evidence on both sides, even if the latter is fairly similar. In addition, there may be diferences in how much each side leans toward particularism or methodism. As a result, both sides can have a coherent set of positions supported by multiple kinds of evidence. Trying to object to the rationality of religious belief by hammering home an epistemic principle is unlikely to be efective against an anti-skeptical web of interconnected views, including views about the rationality of particular religious beliefs thought of as paradigm cases that must be accommodated by a revision of the epistemic principle being used against those beliefs. In short, in disputes between skeptics and anti-skeptics (over radical or religious skepticism), the best proponents of each side have a web of views with evidential support from multiple sources that can make one’s position remarkably and rationally resilient in the face of objections.20 This fact is commonly acknowledged by anti-skeptics who are comfortable sticking with an autodidactic defense of their own perspectives. But it is less often appreciated by skeptics (about perception or religion), who sometimes seem
Lessons from Commonsensism for Religious Epistemology
281
to think that their skeptical arguments highlighting specifc statements of epistemic principles as premises should make it obvious to all concerned that their skeptical conclusion wins the day. An important caveat: although these anti-skeptical views (whether they’re opposed to radical skepticism or religious skepticism) are impressively strong, they are not invincible. Many proponents of these anti-skeptical views acknowledge that there are possible circumstances in which their anti-skeptical responses would fail.21 17.3.
Plausibly Explaining Religious Disagreement
These fve lessons and the approach they recommend for religious epistemology provide the basis for explaining religious disagreement without implausibly attributing irrationality to those with whom we disagree. It’s a striking fact—evident to those who spend time conversing with enough people who disagree with them on religious topics—that some of those who disagree with you on religious matters are extremely impressive people, both morally and intellectually. For example, consider the most morally virtuous, intellectually competent, and well-informed adherents and defenders of the following views: atheism, agnosticism, Judaism, Christianity, and Islam. It is highly implausible that what you view as the errors of some or all of these people on religious matters are, in all cases, due to their inferiority—relative to the most impressive people who agree with you—in terms of rationality, moral decency, emotional maturity, practical wisdom, intelligence, or erudition. Christians can, of course, acknowledge that there are many adherents and defenders of Christianity (and other views) who are inferior in these ways. Likewise, atheists can admit the same about many adherents and defenders of atheism (and other views). But those acknowledgements aren’t relevant when our question is whether a well-informed, intellectually competent, and morally decent person can rationally endorse atheism (or Christianity, etc.). It’s common to raise objections to atheism, agnosticism, Judaism, Christianity, or Islam—identifying reasons to think that they are false or that it isn’t rational to believe them. Objections of this sort are helpful in our philosophical refection on these views and on the fact that we disagree about them. But often those ofering these objections think that, in light of such objections, one cannot be rational while endorsing the views they target. This thought fies in the face of our experience of those who hold these views. I’m not here arguing that there are well-informed, intellectually competent, and morally decent adherents of these various views on religious matters. Instead, I’m taking it as an exceedingly plausible starting point
282
Michael Bergmann
supported by the personal experience of many of us. The lessons discussed in Section 17.2 provide a way of accommodating this plausible starting point. They do so by supporting the following thought: the reason that the best of those who disagree with you on religious topics are mistaken is that—even if they are your equals or superiors in terms of moral decency, intelligence, etc.—they have misleading evidence. And it’s a commonplace that highly intelligent, well-informed people with diferent evidence can rationally disagree. Thus, objectors to religious belief—who think that noninferential religious belief lacks sufciently strong evidence to withstand the defeating efects of widespread disagreement on religious matters—often have different evidence than refective adherents of a particular religion who think of their religious belief as, in some ways, noninferentially supported. The evidence of these religious believers and their detractors can difer in multiple ways: (a) they might not have the same frst-order direct evidence (with the same strength) for or against religious claims (e.g., evidence consisting of religious seemings supporting religious claims or seemings connected with whether a perfect being could have a good reason for permitting the evils we know of); (b) they might not have the same epistemic intuitions (with the same strength) about the evidential value of the evidence mentioned in (a); (c) they might not have the same epistemic intuitions (with the same strength) about certain epistemic principles (e.g., principles about the rational way to respond to awareness of widespread disagreement); (d) they might not have the same higher-order epistemic intuitions (with the same strength) about the evidential value of the epistemic intuitions mentioned in (b) and (c).22 It’s an interesting question why there are these diferences in the evidence of religious believers and their detractors. Is it because, in the religious case or in the case of disagreements about radical skepticism, either the skeptical objectors or those who resist them are in a misleading environment or subject to some sort of departure from proper cognitive function? I don’t have the space here to address these questions. But because there often seem to be these diferences in evidence in the case of religious disagreement—which in some ways mirror the diferences in evidence in the dispute between commonsensists and their skeptical detractors—we can see how religious belief can be rational in the face of arguments for religious skepticism in something like the way that commonsensism can be rational in the face of arguments for radical skepticism.
Lessons from Commonsensism for Religious Epistemology 17.4.
283
Potential Concerns
I have space to deal only briefy with three potential concerns.23 The frst is that many non-religious philosophers will be inclined to think it’s ridiculous to hold and defend religious belief in the way I’ve been recommending in this chapter. But notice: that very reaction has everything to do with epistemic intuitions about (1) the evidential value of religious seemings and of epistemic intuitions attributing positive evidential value to those religious seemings, and about (2) the truth of the particular epistemic principles used as premises in arguments for religious skepticism. This confrms the very points I’ve been making. Another potential concern is that there is reason to doubt that those involved in religious disputes are being honest and refective about what their evidence in fact is.24 Do religious people and their opponents really have epistemic intuitions that would help their case? Can they, in all honesty, say that these epistemic intuitions of theirs are compelling and strong, even after considering carefully the perspectives and objections of those who disagree? In response to arguments for radical skepticism about our perceptual and memory beliefs, perhaps our commonsense epistemic intuitions remain strong; perhaps something like this is also true in connection with our commonsense moral beliefs in the face of arguments for moral nihilism. But is it true for our religious or atheistic beliefs in the face of arguments against them? Maybe in some cases, the answer is “no.” But in many other cases (including my own case), I think that the honest answer is “yes.” This can be so for a particular person, even if skeptical objections undermine that person’s confdence in some of their religious beliefs but not others. A third potential concern is that taking seriously what I’ve said in Sections 17.2 and 17.3 forces us to approve of people who use what I’ve said there to defend truly bizarre religious or political beliefs, such as that God has told them that Donald Trump won the 2020 US election.25 This concern is not well-founded. In thinking about bizarre beliefs, on the one hand, and Christianity and atheism, on the other, we need to consider at least these three things in each case: (1) what the beliefs in question are based on, (2) whether that basis seems to have positive evidential value, and (3) what impressions we have concerning how virtuous, intelligent, wise, and informed those who hold these beliefs are. While I agree that it’s important to be open to having my horizons expanded by becoming more familiar with people who disagree with me, my current experience leads me to take very seriously that many atheists and Christians are virtuous, intelligent, wise, and well-informed, whereas this isn’t the case for those holding bizarre religious or political views such as the one just mentioned earlier. Hence, it makes good sense to treat atheism and Christianity diferently
284
Michael Bergmann
than bizarre religious or political views, despite the fact that defenders of each may appeal to seemings about religious matters and epistemic intuitions in favor of those religious seemings.26 Notes 1 See Chisholm (1982). 2 Commonsensism also endorses points (a) and (b) when “rationally believe” is replaced with “know.” 3 See Bergmann (2021: 119 & 128–129) for a more nuanced discussion of particularism and methodism in connection with skepticism. 4 See especially Bergmann (2021: chs. 6 & 7). 5 For more on epistemic intuitions and seemings, see Bergmann (2021: 84–86, 122–126, & 131–145). 6 See work by William Alston (e.g., 1991), John Greco (e.g., 2000), and myself (e.g., Bergmann 2017b) in this vein. 7 A similar point could be made about the commonsensist response to radical skepticism—that is, not all inclined to resist radical skepticism will share the intuitions and evidence of the commonsensist. 8 Reformed Epistemology—see Plantinga (1983)—is a noninferentialist religious epistemology in this sense. 9 See Alston (1993) and Bergmann (2021: ch. 3). For an impressive recent attempt to provide an anti-skeptical argument that can satisfy skeptics, see Rinard (2019). For objections to Rinard’s argument, see Bergmann (2021: 64–72). 10 And it’s the approach taken in Bergmann (2021). 11 For further discussion of how this problem goes both ways in the dispute between radical skeptics and commonsensists, see Bergmann (2021: 145–150). 12 Recall that our focus is on noninferentialist religious epistemologies. 13 See Bergmann (2017b: 35–37) for a summary description of this sort of evidence, which consists of theistic or religious seemings triggered by a wide variety of ordinary experiences, including experiences of nature or reading sacred texts (which is to say, it doesn’t consist solely or even mainly of dramatic religious experience reported by religious mystics). See also Rea (2018: chs. 6 & 7) for some discussion of the way in which ordinary instances of perception can be rationally experienced as divine encounters. These too can be cases where ordinary experience triggers religious seemings. 14 People can have (a)-, (b)-, and (c)-evidence without thinking of it as such. (For the following example, suppose that (a) speaks of “evidence we have for simple mathematical beliefs such as 2 + 3 = 5” rather than sensory experience and memory seemings.) Suppose a skeptic suggests to an intelligent person, S (without familiarity with epistemology), that a demon could get a person to feel confdent about a necessary falsehood (e.g., 2 + 3 = 6). And suppose the skeptic says that, for all S knows, this might be happening with S’s belief that 2 + 3 = 5 (which is in fact false). S might dismiss the suggestion that she could be making that sort of mistake as ridiculous and remain confdent that her evidence in support of her belief that 2 + 3 = 5 is excellent. In doing this, she will be relying on (b)-evidence, even if she doesn’t think of it as such. Next, suppose that the skeptic says that S’s persistent confdence in the high quality of her evidence for thinking that 2 + 3 = 5 might itself be a demon-produced deception. In response, S might dismiss this further suggestion as ridiculous too. In doing so, she is relying on (c)-evidence, without thinking of it as such.
Lessons from Commonsensism for Religious Epistemology
285
15 One might have a high level of confdence in some beliefs based on (a*)-evidence—or (a)-evidence—and not others. 16 It may be that some people have a kind of (a*)-, (b*)-, and (c*)-evidence without thinking of it as such. In support of this claim, points similar to those made in footnote 14 about mathematical belief could be made in the religious case as well. 17 At least this is the case if the religious skeptic endorses the same epistemic principles in dealing with religious skepticism as she endorses when dealing with radical skepticism (instead of holding religious beliefs to higher standards). Alston (1991: 199) and van Inwagen (1996) suggest that some religious skeptics do arbitrarily hold religious beliefs to higher standards. 18 Keith DeRose (2018)—an epistemologically sophisticated theist—claims that neither he nor anyone else with whom he’s acquainted knows that God exists. This suggests that, despite being a theist, DeRose doesn’t have (b*)-evidence indicating that his theistic beliefs count as knowledge; or if he does, he doesn’t trust it, which suggests that his (c*)-evidence does not support the rationality of any (b*)-evidence he has in support of the claim that theistic beliefs count as knowledge. But the fact that a theist like DeRose doesn’t ft the pattern I’ve sketched in this section doesn’t count against the claim that many theists do ft that pattern. Moreover, it’s not clear that DeRose doesn’t ft the pattern, since he would ft the pattern if he had (b*)-evidence supporting the claim that his (a*)-evidence provided some positive epistemic status or other for his theistic beliefs or acceptances (e.g., that they are rational) and if, in addition, he had (c*)-evidence supporting that evaluative claim just mentioned about his (b*)-evidence. DeRose doesn’t say whether he has such (b*)- and (c*)-evidence, though he accepts theism, which suggests that he does have such (b*)- and (c*)evidence. (DeRose has said—in personal correspondence—that he thinks his acceptance of theism is justifed.) 19 Plantinga (1983: 77). 20 My emphasis on this web of views is not intended to support coherentism over foundationalism (to say that coherence is relevant to rationality doesn’t commit one to coherentism). See Bergmann (2017a) for a defense of foundationalism against coherentist objections. 21 See Bergmann (2021: 129–130) for further discussion. 22 Although religious believers and their detractors can get evidence (via testimony from each other) that their evidence difers in these ways, this is not the same thing as having the evidence that the other has. For example, consider the evidence you have consisting of (i) your strong epistemic intuition that p, (ii) your sense of the evidential value of your intuition that p, (iii) the testimony you get from me that I have a strong epistemic intuition that q (which is incompatible with p), and (iv) your sense of the evidential value of my intuition that q. This is quite diferent (in terms of how strongly it supports p and how strongly it supports q) from my evidence consisting of (i*) the testimony I get from you that you have a strong epistemic intuition that p, (ii*) my sense of the evidential value of your intuition that p, (iii*) my strong epistemic intuition that q, and (iv*) my sense of the evidential value of my intuition that q. In particular, your evidence consisting of (i) and (ii) supports p more strongly than does my evidence consisting of (i*) and (ii*); and your evidence consisting of (iii) and (iv) supports q less strongly than does my evidence consisting of (iii*) and (iv*). In the end, your evidence may quite reasonably leave you more open than I am to the possibility that my evidence is misleading; and my evidence may quite reasonably leave me more open than you are to the possibility that your evidence is misleading.
286
Michael Bergmann
23 One concern not addressed here is whether externalists (and not just internalists) can accommodate an epistemic-intuition-based response to radical and religious skepticism. For some discussion of this point, see Bergmann (2021: 159–164). 24 Keith DeRose (2018) raises concerns in this neighborhood. 25 See Alberta (2022) for some discussion of these sorts of bizarre beliefs. 26 For helpful comments on previous versions of this chapter, I’m grateful to Jef Brower, Perry Hendricks, Kevin McCain, Scott Stapleford, and audience members when this chapter was presented at the Center for Philosophy of Religion at the University of Notre Dame and at the 14th Midwest Epistemology Workshop at the University of Iowa.
References Alberta, Tim. 2022. “How Politics Poisoned the Evangelical Church.” The Atlantic. www.theatlantic.com/magazine/archive/2022/06/evangelical-church-pastorspolitical-radicalization/629631/ Alston, William. 1991. Perceiving God. Ithaca, NY: Cornell University Press. ———. 1993. The Reliability of Sense Perception. Ithaca, NY: Cornell University Press. Bergmann, Michael. 2017a. “Foundationalism.” In William Abraham and Frederick Aquino (eds.) The Oxford Handbook of the Epistemology of Theology. New York: Oxford University Press. Pp. 253–273. ———. 2017b. “Religious Disagreement and Epistemic Intuitions.” Royal Institute of Philosophy Supplement 81: 19–43. ———. 2021. Radical Skepticism and Epistemic Intuition. New York: Oxford University Press. Chisholm, Roderick. 1982. “The Problem of the Criterion.” In The Foundations of Knowing. Minneapolis: University of Minnesota Press. Pp. 61–75. DeRose, Keith. 2018. “Delusions of Knowledge Concerning God’s Existence.” In Matthew Benton, John Hawthorne, and Dani Rabinowitz (eds.) Knowledge, Belief, and God. Oxford: Oxford University Press. Pp. 288–301. Greco, John. 2000. Putting Skeptics in Their Place. New York: Cambridge University Press. Plantinga, Alvin. 1983. “Reason and Belief in God.” In Alvin Plantinga and Nicholas Wolterstorf (eds.) Faith and Rationality. Notre Dame, IN: University of Notre Dame Press. Pp. 16–93. ———. 2000. Warranted Christian Belief. Oxford: Oxford University Press. Rea, Michael. 2018. The Hiddenness of God. New York: Oxford University Press. Rinard, Susanna. 2019. “Reasoning One’s Way Out of Skepticism.” In Kevin McCain and Ted Poston (eds.) The Mystery of Skepticism. Leiden: Brill. Pp. 240–264. van Inwagen, Peter. 1996. “It Is Wrong, Everywhere, Always, and for Anyone, to Believe Anything Upon Insufcient Evidence.” In Jef Jordan and Dan HowardSnyder (eds.) Faith, Freedom, and Rationality. Lanham, MD: Rowman and Littlefeld. Pp. 137–153.
Index
agency 139–141 a posteriori justifcation of necessary truths 3, 235–242, 244, 248 appearance 8, 23, 57–58, 63–73, 76, 78–79, 84–87, 89, 94, 98, 101–102, 106, 113, 131, 136–138, 141, 142, 206–207, 210–213, 221–222, 225–227 attraction to assent 172 bare seeming statements 40, 43, 50–52 Bergmann’s dilemma 34 BonJour’s dilemma 34 cognitive penetration 25, 148, 151, 158, 202–203, 205–206, 208–209, 212–214 Concept Correctness Open Question (CCOQ) 245 conception vs. concept of morally wrong 245–250 conditions: cognitive 170; good 170–171 contentful 174–177, 179–181 data 119, 151, 162–167, 169, 171, 177–178, 229 data collection 163–164, 177 defeat/defeater 2, 10–11, 15, 23–25, 29–30, 32–33, 35, 54, 57, 59, 67, 72–73, 94, 112–113, 115–116, 147–148, 157–159, 171, 189–191, 199, 219–223, 225, 226–227, 230–231, 248–249 deliberation 141, 148, 153, 225 Descartes, René 24, 26–27, 179 deviant causal chains 144–145, 175 dogmatism 1–2, 9–12, 15–16, 18–21, 78, 91–92, 111–113, 115–116, 120, 122–123, 125–126, 147,
152, 187, 202–205, 207–211, 213, 217–218, 221, 223, 230, 243–244, 249 epistemic practices 165–167, 169, 177 experiential view of seemings 8, 14, 65–66, 73, 111–113, 115–117, 119–126, 136, 148–149, 151, 155–156, 208–210 expert cognition 148, 151–152, 158–159, 202, 204, 207–210 externalism 15, 17, 25–26, 35, 286 forcefulness (feeling of truth) 9, 15, 31, 66, 173 foundationalism 8–9, 30, 34–35, 285 hedging use (or hedge) 51, 171 imagination 66, 141, 176, 179–180, 196, 236–239, 266 inclination 50, 79, 94, 98, 102, 106, 107, 119, 149, 155, 171, 172, 178, 180, 200, 242, 243 inferentialism 167, 178 inquiry 162, 163, 164, 165, 167, 169, 177 internalism 30, 34, 35, 63, 231 justifcation: doxastic 8, 10, 36, 73, 112, 218, 227, 229–231, 255; immediate 2, 24, 111–113, 115–116, 119–120, 124–126, 202–205, 207–211, 213–214, 217–218, 220–221, 223, 230–231; inferential 2, 111–112, 118–124, 126, 130–132, 135–136, 139, 231, 258, 260; mediate 112, 115,
288
Index 218, 223–224; non–inferential 139, 147, 203, 279, 282; propositional 8, 12, 36, 68, 73, 112–113, 116, 120, 218, 223–224, 227, 229, 255
many property problem 176 metajustifcation 30, 34 modal 19, 164, 165, 178, 195, 196, 199, 244 Monty Hall 150, 160, 243 morally wrong vs. morally blameworthy 247 naive realism 115, 168, 169, 171, 176 non-subjectivism about how things seem 40, 44–52 Norman the clairvoyant 29–33, 36 open/closed question test 247 perspectives 40, 44–52 phenomenal conservatism (PC) 1, 10, 21, 23–26, 28, 32, 34, 35, 59, 147, 152, 179, 187 Phenomenal Explanationism (PE) 2, 217–218, 231
Plato 169, 179 presentational phenomenology 88–89, 92, 162, 173–177, 180, 181 proper goals 163 responsibility 137 Sellar’s dilemma 34; see also BonJour’s dilemma Simpson’s paradox 46–47, 49 sophisticated subjectivism about how things seem 39–40, 43–44 speckled hen 202–204, 206, 208, 212 stored beliefs 2, 63–65, 67–73 subjectivism about how things seem 38–40, 44, 50 synthetic a priori propositions 242, 247, 248, 249, 250 taking condition 2, 130–131, 133, 135–145 thought experiments 164, 165–166, 252, 253–254, 256, 257, 260–261, 263–267 zetetic 2, 148–149, 152–159