Common Sense Metaphysics: Essays in Honor of Lynne Rudder Baker 2020037421, 2020037422, 9780367333218, 9780429319976

Citation preview

Lynne Rudder Baker was one of the foremost metaphysicians of her time. Opposing the contemporary trend to explain aspects of our world in terms of posits of fundamental physics, Baker defended important nonreductive views about a range of phenomena including persons, minds, and ordinary material objects. In this new collection, eminent scholars in philosophy discuss various aspects of Baker’s thought—from her Constitution View of human persons to her positions on arguments about the existence of God. Whether or not one is persuaded by Baker’s common-sense approach, it is clear that her ideas warrant deep attention. This volume will be of interest to theorists and students of metaphysics for years to come. —Jacob Berger, Lycoming College, USA

Common Sense Metaphysics

This book celebrates the research career of Lynne Rudder Baker by presenting sixteen new and critical essays from admiring students, colleagues, interlocutors, and friends. Baker was a trenchant critic of physicalist conceptions of the universe. She was a staunch defender of a kind of practical realism, what she sometimes called a metaphysics of everyday life. It was this general “common sense” philosophical outlook that underwrote her famous constitution view of reality. Whereas most of her contemporaries were in general given to metaphysical reductionism and eliminativism, Baker was unapologetic and philosophically deft in her defense of ontological pluralism. The essays in this book engage with all aspects of her unique and influential work: practical realism about the mind; the constitution view of human persons; the first-person perspective; and God, Christianity, and naturalism. Common Sense Metaphysics will be of interest to scholars of Baker’s work, as well as scholars and advanced students engaged in research on various topics in metaphysics, philosophy of mind, philosophy of language, and philosophy of religion. Luis R.G. Oliveira is assistant professor of philosophy at the University of Houston. He has published numerous articles in refereed journals, on topics in epistemology, ethics, and religion. He is also the director of the LATAM Bridges in the Epistemology of Religion, an international project focused on connecting Latin American philosophers to the Anglophone philosophical world. Kevin J. Corcoran is professor of philosophy at Calvin University. He is the author of Rethinking Human Nature (2006), the co-author of Church in the Present Tense (2011), and the editor of Soul, Body and Survival (2001). He has also published numerous articles in refereed journals, on topics in metaphysics, mind, and religion.

Routledge Festschrifts in Philosophy

Mind, Language and Morality Essays in Honor of Mark Platts Edited by Gustavo Ortiz Millán and Juan Antonio Cruz Parcero Sensations, Thoughts, Language Essays in Honor of Brian Loar Edited by Arthur Sullivan Common Sense Metaphysics Essays in Honor of Lynne Rudder Baker Edited by Luis R.G. Oliveira and Kevin J. Corcoran

For more information about this series, please visit: https://www.routledge. com/Routledge-Festschrifts-in-Philosophy/book-series/RFSP

Common Sense Metaphysics Essays in Honor of Lynne Rudder Baker

Edited by Luis R.G. Oliveira and Kevin J. Corcoran

First published 2021 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 Taylor & Francis The right of Luis R.G. Oliveira and Kevin J. Corcoran to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or 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 identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Names: Baker, Lynne Rudder, 1944–2017, honoree. | Oliveira, Luis R. G., editor. | Corcoran, Kevin, 1964– editor. Title: Common sense metaphysics : essays in honor of Lynne Rudder Baker / edited by Luis R.G. Oliveira and Kevin J. Corcoran. Description: New York, NY : Routledge, 2021. | Includes bibliographical references and index. Subjects: LCSH: Baker, Lynne Rudder, 1944–2017. | Realism. | Metaphysics. | Constitution (Philosophy) | Perspective (Philosophy) Classification: LCC B945.B184 C66 2021 (print) | LCC B945.B184 (ebook) | DDC 110—dc23 LC record available at https://lccn.loc.gov/2020037421 LC ebook record available at https://lccn.loc.gov/2020037422 ISBN: 978-0-367-33321-8 (hbk) ISBN: 978-0-429-31997-6 (ebk) Typeset in Sabon by codeMantra

This book is dedicated to Lynne. Teacher, Mentor, Friend.

Contents

Introduction

1

K E V I N J . C O RC O R A N A N D L U I S R .G . O L I V E I R A

PART I

On Practical Realism about the Mind

11

1 What Is a Concept?

13

C H R ISTOPH ER H I LL

2 Practical Realism about the Self

39

C A RO LY N D I C E Y J E N N I N G S

3 Propositional Attitudes as Self-Ascriptions

54

A N G E L A M E N D E L OV I C I

4 Saving Physicalism

75

JA N ET LEV I N

PART II

On the Constitution View

93

5 Constitution, Non-reductivism, and Emergence

95

DE R K PE R E BOOM

6 The Threat of Thinking Things into Existence

114

K AT H R I N KO S L I C K I

7 Unkind Persons: A Critique of Baker’s Constitution View

137

K E V I N J . C O RC O R A N A N D PAU L M A N ATA

8 Constitution and Personal Identity M A RYA S C H E C H T M A N

158

x Contents PART III

On the First-Person Perspective

175

9 On Baker on the First Person

177

JOSEPH LEV I N E



11 Naturalism and Non-qualitative Properties

209

S A M C OW L I N G



PART IV

On God, Christianity, and Naturalism

253









Notes on Contributors Index

331 333

Introduction Kevin J. Corcoran and Luis R.G. Oliveira

Lynne Rudder Baker (1944–2017) was, by all meaningful metrics, an outstanding philosopher. She authored five books: Saving Belief: A Critique of Physicalism (Princeton, 1989); Explaining Attitudes: A Practical Approach to Mind (Cambridge University Press, 1995); Persons and Bodies: A Constitution View (Cambridge University Press, 2000); The Metaphysics of Everyday Life: An Essay in Practical Realism (Cambridge University Press, 2007); and Naturalism and the First-Person Perspective (Oxford University Press, 2013). She authored more than 100 academic articles and book chapters over a career spanning more than three decades. In addition to having her work translated into many different languages, Lynne was also the recipient of numerous prestigious academic honors and accolades, including delivering the esteemed Gifford Lectures in 2001. The evidence is unambiguous. But Lynne was unique among preeminent Anglo-analytic philosophers of the latter part of the 20th and early part of the 21st centuries. She brought a kind of humanity to her work, a humanity that undergirded it, inspired it, and somehow managed to seep through even the hard-nosed, sometimes painstakingly technical, logical machinery that is the stock-in-trade of analytic philosophy. Indeed, what Roberta De Monticelli says of Lynne’s The Metaphysics of Everyday Life could also be said of the whole body of Lynne’s published work: A paradigm of rigor and clarity in its ontology of ‘the world that we live and die in, the world where our plans succeed or fail, the world we do or do not find love and happiness in – in short, the world that matters to us’. The latter words are, of course, from the opening pages of The Metaphysics of Everyday Life, and they reveal precisely the sort of humanity that infused all of Lynne’s work and set her apart from most of her ontologically reductionist and deflationary colleagues in analytic philosophy of mind and metaphysics. She was a staunch defender of what she called “commonsense metaphysics”—a metaphysics that includes myriads of kinds of things as well as myriads of things of those kinds, from cats,

2  Kevin J. Corcoran and Luis R.G. Oliveira cabbages, and combustion engines to planets, dinner parties, art objects of various sorts, nations, economies, weddings, funerals, tables, chairs, points scored in football games, beliefs, hopes, plans, etc. According to Lynne’s commonsense metaphysics, all of these things exist in their own right, in addition to bosons and gluons. This volume represents a collection of original essays written to honor the memory and legacy of Lynne Rudder Baker. It is a testament both to the very high esteem in which Lynne was held by some of the world’s most celebrated analytic philosophers (many of whom contribute essays to this volume) and to Lynne’s enduring philosophical legacy. Some contributors to this volume are (we dare say) regarded as giants in their field; others are former students of Lynne’s or those she mentored from afar; still others are doing cutting-edge research in areas explored by Lynne, finding that their own work must contend with hers. What all of the essays have in common is that they honor Lynne by taking her arguments for various positions seriously and subjecting them to rigorous analysis and critique, which is the ultimate praise one might offer any philosopher’s work. Some of the essays defend aspects of Lynne’s views, while others challenge those views. But all of the essays pay due homage and respect to the impressive arsenal of arguments that Lynne advanced in defense of her robust sort of ontological realism. Doing justice to the breadth of Lynne’s work, the contributors to this volume address a variety of topics, from her trenchant critique of a reductionistic physicalism in the philosophy of mind to her staunch defense of the ontological significance of the first-person perspective; from her powerful arguments for a constitution view of human persons to her musings on the ontological argument for the existence of God and the peculiarly Christian doctrine of the resurrection. Each of these topics, as well as others, is taken up, examined, critiqued, and evaluated.

Part 1 The essays that make up Part I of the volume concern what might be called practical realism about the mind. In ordinary life and speech, we human beings take ourselves to have beliefs, desires, hopes, fears, etc., and take ourselves to plan and act on the basis of them. Lynne was an ardent realist about intentionality and mental states such as belief, desire, and the like. She strenuously objected to eliminativist and deflationary views that rid the world of such “folk-psychological” entities. In the first chapter of this section, “What Is a Concept?,” Christopher Hill defends what he calls Baker’s “hardcore realism” about intentionality. He does so by responding to three different sorts of objection to that kind of realism, objections that Baker herself did not consider. First, he considers a Quinean objection that goes after propositions as genuine entities. If propositions are eliminated from our ontology, then realism

Introduction  3 about the attitudes disappears too, as propositions are precisely what the attitudes are said to be aimed at. The problem centers on individuating propositions. If we cannot identify a principle of individuation for propositions, then statements attributing propositional attitudes will turn out to be too vague to play any meaningful role in the sciences. Hill carefully considers Quine’s arguments for this propositional pessimism but ultimately argues that actually there is a quite plausible and promising way of individuating propositions. The second sort of objection to hardcore realism concerns the claim that intentional state attributions are radically underdetermined, marking a stark and qualitative difference between our folk practice of attributing propositional attitudes and scientific theorizing. As Quine would put it, “the true and ultimate structure of reality, the canonical scheme for us is the austere scheme that knows no quotation but direct quotation and no propositional attitudes but only the physical constitution and behavior of organisms.” However, Hill (p. 15) contends that there is no qualitative difference here. Our folk heuristics for attributing propositional attitudes and for testing such attributions are quite powerful – perhaps less so than the methods of hypothesis formation and testing in mature sciences, but nonetheless sufficient to be compatible with intentional realism. Finally, Hill faces down a third objection to Baker’s hardcore realism about intentionality. This criticism concerns intentional holism, the claim that all of the intentional states possessed by an individual are interdependent, in a catastrophic sort of way for hardcore realism. The central idea behind intentional holism is that if a given attitude (a belief, a desire, etc.) is removed from the whole interconnected and interdependent system of propositional attitudes of which it is a part, the subject ceases to possess all of the other members of the system as well. The original system, in other words, ceases to be and is replaced with a different system of attitudes with different essential natures. If true, holism spells doom for intentional realism. One way that has been offered for rejecting holism relies in part on the claim that in most cases, the beliefs that are constitutive of a concept are comparatively few in number. But the claim presupposes that for any concept C, there is a principled basis for distinguishing between C-involving beliefs that are constitutive of C and C-involving beliefs that are merely adventitious or nonessential. And this has been rejected on the basis of a Quinean critique of analyticity. Here Hill (p. 18) suggests that [t]here is in fact a principled distinction between the essential or constitutive parts of the role of a concept and the parts of the role that are non-essential or adventitious. And if this is so, then the claim

4  Kevin J. Corcoran and Luis R.G. Oliveira that in most cases, the beliefs that are constitutive of a concept are comparatively few in number is vindicated and holism rejected. In the second chapter, “Practical Realism about the Self,” Carolyn Dicey Jennings argues for what—following Chris Hill in the last chapter—we might call “hardcore realism” about the self, the subject of the attitudes. Against illusionists about the self, like Daniel Dennett and Keith Frankish, who argue that there is no self or subject, no “I” who has mental content, Jennings argues for what she calls a “substantive” view of the self. The existence of the self, Jennings claims, is assumed in our everyday interactions with others and its productive use in everyday contexts counts as evidence of its existence. Moreover, Jennings argues that data in brain sciences provide additional evidence for the causal power of a self over phenomena of mental attention. On her view, the self corresponds to a “…pattern of wave activity… [in] the brain…” (p. 46). Or, as she says later, “the self is a set of interests or tendencies….” (p. 47). These claims, of course, may lead one to wonder just how “substantive” such a view of the self actually is when characterized as a “pattern” of neural activity or a “set” of interests or tendencies. It is left to the reader to judge whether he or she finds Jennings’ arguments ultimately persuasive. In Chapter 3, Angela Mendelovici argues for a “self-ascriptivist” view of propositional attitudes in her “Propositional Attitudes as SelfAscriptions.” She claims that the attitudes exist insofar as and precisely because we ascribe them to ourselves. It is the very features of selfascribing the attitudes, i.e., their role in our self-conception, that “makes them real.” She is concerned only with those propositional attitudes that form “an integral part of our commonsense conception of ourselves and others …” (p. 55). Mendelovici’s argument for the reality of these commonsense attitudes, like Baker’s own, does not depend either on the claim that introspective acquaintance with them vindicates their reality or on the claim that science does. Rather, as she puts it, ascribing contents to our immediate contents, our internal states, or ourselves as persons is necessary and sufficient for… derivatively represent[ing] those contents—where derived representation is representation that is at least partially constituted, grounded in, or identical to other instances of representation—and the propositional attitudes of commonsense psychology are nothing over and above such self-ascriptions (p. 59, emphasis ours). In short, “propositional attitudes are products of derived representation, and, when it comes to derived representation, thinking it is so makes it so… and so, self-ascribing an attitude is all it takes for that attitude to really, actually exist” (p. 72).

Introduction  5 Janet Levin’s “Saving Physicalism” rounds out the first part of the volume. In this chapter, Levin argues that Baker’s “practical realism” about the attitudes (what Chris Hill called her “hardcore realism”) faces a stiff problem in the form of so-called “belief-discordant” behaviors. The disconnect between a subject’s behavior and that same subject’s ­ self-avowals makes it difficult, if not impossible, to answer questions of the form “does S believe that P?” But role-functionalism, on the other hand—a view Baker labeled, among others, as the Standard View—may actually be better situated than practical realism to handle the problem of belief-discordant behaviors. Moreover, Levin argues, many of the problems that Baker levels against what she calls the Standard View of the attitudes, while real and important and serious, may actually be worse for Baker’s own practical realism about the attitudes than for role-functionalism (or some other non-reductive version of physicalism).

Part 2 The chapters that make up Part II of the volume focus specifically on the putative relation of constitution. Constitution is said to be a kind of unity or sameness relation that numerically distinct objects (states, properties, etc.) stand in to one another. Since it is distinct objects (states, properties, etc.) that make up the relata, the constitution relation is not identity. Examples of objects that are said to stand in the putative relation of constitution are statues and lumps of bronze, flags and pieces of cloth, dollar bills and pieces of blended cotton and linen cloth, and, according to Baker, human persons and their bodies. In this section’s first chapter, “Constitution, Non-reductivism, and Emergence,” Derk Pereboom presents an alternative account of constitution to that of Baker. In it, he takes a fine-toothed comb to Baker’s account of constitution and demonstrates where his own account, which is both indebted to Baker’s and similar to it in fundamental ways, nevertheless differs from hers. It is an illuminating chapter. And as it turns out, perhaps the most important difference between the accounts concerns the role physicalism plays in the metaphysics of nonfundamental levels of reality, those levels at which, for example, statues, flags, and currency exist. Whereas for Baker, physicalism plays no significant role in a metaphysics of such everyday objects and indeed is largely independent of it, for Pereboom, physicalism here plays a much more central role. In the second chapter in this section, “The Threat of Thinking Things into Existence,” Kathrin Koslicki examines a common objection to a Baker-style metaphysics of constitution, namely, that to the extent that some of what exists is “intention-dependent,” her view commits her to the worst sort of anti-realism, where it is we “minded” human beings who are the creators of the world and its inhabitants, and it is a relatively

6  Kevin J. Corcoran and Luis R.G. Oliveira easy thing to bring into existence both new things of various kinds and new kinds of things. Koslicki examines in this regard the objections of Ted Sider and Dean Zimmerman. Koslicki demonstrates the degree to which “practices and conventions” actually mitigate against the objections of Sider and Zimmerman. It will be up to the reader to determine whether he or she judges that the mitigation is sufficient to the task of rendering Baker’s view ultimately invulnerable to charges of anti-realism. The third and fourth chapters in this section focus on Baker’s constitution view of human persons. On a constitution account of them, human persons are constituted by bodies without being identical to the bodies that constitute them. According to Baker, when the human organism that is a particular person’s body acquired a rudimentary capacity for a first-person perspective, a new kind of thing came into existence—a human person. Also, according to Baker, the human person that comes into existence is most fundamentally a person. Their “primary kind” is not human person, or Martian person or silicon person or any kind of person at all. In Chapter 7, “Unkind Persons,” Kevin Corcoran and Paul Manata argue that Baker’s version of the constitution view commits her to an unkind view of persons and, they argue, this creates problems for her when it comes to her account of the persistence conditions for human persons over time. They argue that person simpliciter is not a primary kind, and offer an alternative constitution view of human persons. According to their alternative, our primary kind is human person and, they claim, human persons are essentially constituted by human organisms, something Baker’s version of the constitution view denies. They contend that their version of a constitution view offers a more satisfying account of the persistence conditions for human persons than Baker’s own. In Chapter 8, “Constitution and Personal Identity,” Marya Schechtman examines Baker’s constitution view in light of the two most prominent views of personal identity over time—the biological view, as defended by Eric Olson, and the psychological view, as defended by nearly everyone else (including Baker). Taking her cues from Baker, whose metaphysics of everyday objects emphasizes the ontological significance of social environments, relations, and intentions, Schechtman tries to show how, despite their differences, Olson’s biological view of personal identity and Baker’s view share a common essentialist framework that allows for a sharp dichotomy between persons and the animal bodies that, according to Baker, constitute them. But the resources for a replacement framework, Schechtman contends, are present in Baker’s insight into the ontological significance of relational-social-practical features. She argues that it is precisely these relational, social features that ought to have ontological priority when it comes to understanding how persons come into existence and persist and not possession of a first-person perspective, which as Baker sees it, never would have emerged absent a linguistic community. The differences Schechtman points out between Baker’s

Introduction  7 view and her own are instructive and present a challenge, we think, to those attracted to a constitution view of human persons.

Part 3 Part III focuses on the Baker’s notion of the first-person perspective. It is the capacity for a robust first-person perspective that Baker believes makes something a person and, therefore, carries ontological significance. Moreover, it is the presence or reality of a strong first-person perspective that, Baker claims, entails the falsehood of “scientific naturalism,” the claim that a complete inventory of reality is exhausted by the entities and properties invoked by scientific theories. Since such theories leave out the first-person perspective, it follows that the entities and properties invoked in scientific theories are not exhaustive of reality. Baker’s claim that naturalism leaves something out of its inventory of reality is referred to variously as the “problem of the missing self” (Perry) or the “location problem” (Cowling). The suggestion, more exactly, is that naturalism has no room for thisnesses or haecceities, those non-qualitative properties such as being Lynne Baker or being Abraham Lincoln Whereas qualitiative properties such as mass, charge, and spin have a place in the naturalist ontology, non-qualtitative properties do not. And, therefore, scientific naturalism is false. In the first chapter of this part of the volume, Joseph Levine argues in “On Baker on the First Person” that her arguments against naturalism fail, that, in fact, the sorts of phenomena she uses to argue for the ineliminability of the first-person perspective can actually be accommodated by scientific naturalism. Moreover, Levine argues that the claim that scientific naturalism does leave something of ontological significance out of its inventory of reality requires a more radical sort of break with naturalism than Baker is willing to accept. At the end of the day, if Levine is correct, Baker’s position turns out to be not so significantly different than the scientific naturalism she claims to be opposing. Really, it will take some sort of dualism to make true the claim that scientific naturalism leaves out of the world something of ontological significance. But Baker is staunchly opposed to dualism. John Perry’s chapter, “The Missing Self,” argues for the very same conclusion as Levine’s, but via a different route. Again, what is at issue is the naturalist’s claim that their inventory of all the individuals and properties exhausts all there is, and that every fact about every individual can be explained just by what shows up on that inventory. But Baker argues that this cannot be right, since the fact that one of those individuals on that inventory is me is not an individual nor a natural property of any individual. And since being Lynne Baker will not show up on the naturalist’s inventory, there is a fact—a haecceitistic fact—that cannot be accommodated by the naturalist.

8  Kevin J. Corcoran and Luis R.G. Oliveira Perry wants to claim that a first-person perspective and robust self-knowledge give us all we need to account for having a first-person property and, therefore, no non-qualitative properties such as haecceities are needed to account for a “missing self.” And if the question is what makes this individual Lynne Baker or that individual John Perry, then the answer, again, is not a non-qualitiative property of Lynne Baker or John Perry, but rather a very complicated series of historical-contingent properties involving large numbers of historical-contingent facts that involved other people’s histories, world, and regional events that culminated in the birth of two children, one of whom is the author of “The Missing Self” and the other of whom is the person this anthology seeks to honor. And all of those facts can be explained just by what shows up on the naturalist’s inventory of reality’s contents. In the third chapter in this section, however, Sam Cowling addresses the same issue but from an ontological perspective that is open to haecceities. In his “Naturalism and Non-qualitative Properties,” Cowling disagrees with Levine and Perry by first arguing that non-qualitative properties—most notably, haecceities such as being Lynne Baker—are ineliminably tied to first-person perspectives. Next, Cowling considers whether naturalism is incompatible with haecceities by addressing Shamik Dasgupta’s argument against individuals and, in turn, nonqualitative properties. He considers several strategies against Dasgupta, drawing on de re laws and haecceitistic possibilities. Finally, Cowling draws an analogy between naturalism and platonism regarding mathematical entities, on the one hand, and naturalism’s parallel commitment to individuals, on the other. He concludes that naturalists are obliged to posit non-qualitative properties after all. The result is that Baker is right to insist in the ineliminability of non-qualitative properties— contra Levine and Perry—but wrong to see this as a challenge to naturalism. In the final chapter of this section, “Persons First Metaphysics,” Einar Duenger Bohn examines Baker’s account of persons as objects with a capacity for a first-person perspective. He identifies four distinct problems for Baker’s account. In solving the four problems that plague Baker’s account, an alternative account comes into focus: what the concept of a person picks out in the world isn’t an object with a capacity for a first-person perspective, as Baker claims, but rather “a distinctive kind of fundamental plural collective property.”

Part 4 The fourth and final part of the volume brings together chapters on God, Christianity, and Naturalism. In the first chapter, Peter van Inwagen takes us to a corner of Baker’s published work that is much less explored than the philosophical neighborhoods explored in the previous chapters. van Inwagen’s chapter, “Speaking About Things Independently of

Introduction  9 Whether They Exist,” is a trenchant analysis of and critique of Baker’s take on Anselm’s ontological argument for the existence of God. In several papers, Baker has contended that criticisms of Anselm’s ontological argument (in Proslogion II) are mistaken, owing to the fact that Anselm is speaking about that than which nothing greater can be conceived, independently of whether it exists. She contends that the idea of speaking of that than which nothing greater can be conceived independently of whether it exists does not require a Meinongian interpretation of the premises that make up her argument. Van Inwagen, however, offers a reason to believe that the idea cannot be used to give an acceptable non-Meinongian interpretation of the premises that make up her argument. And, moreover, van Inwagen argues that even if a non-Meinongian account of the premises were available, it still would not follow that the argument itself could be given an acceptable nonMeinongian interpretation. In the second chapter in this section, “Constitution, Persons, and the Resurrection of the Dead,” Thomas Senor examines Baker’s constitution account of the Christian doctrine of the resurrection. According to Baker, any account of that doctrine must satisfy three conditions: a person that exists post-resurrection must be identical to the person who existed pre-resurrection; the post-resurrected person must be embodied (and the body must be of a different kind than the pre-resurrected body); and finally, the resurrection itself must be miraculous. On these grounds, Baker contends that her constitution view of human persons is preferable to its closest competitors—dualism and animalism. Indeed, Baker levels various sorts of objection to dualism and materialism, objections her view avoids. Senor contends, however, that Baker’s objections lack substance. And he considers what he takes to be a very serious problem confronting Baker’s constitution view of the resurrection: the problem of specifying the conditions of identity of post- and pre-resurrection persons. The problem for Baker comes down to specifying what the difference is between God’s duplicating an embodied first-person perspective in the hereafter and God’s resurrecting that embodied first-person perspective in the hereafter. And for Baker’s view to succeed, she needs to specify that difference in a way that does not run afoul of Noonan’s only x and y principle, according to which whether or not something y, that exists at some later time, is identical to something x, that exists at some prior time, cannot depend on facts extrinsic to x and y, and the relations that hold between them. As far as Senor can tell, Baker has not done so. In “Putnam and Baker on Naturalism,” Mario De Caro examines two alternatives to the reductionist or eliminativist doctrine of scientific naturalism, which sets the boundaries for both ontology and epistemology, namely the “Liberal Naturalism” of Putnam and the “Near Naturalism” of Baker. The problem for the scientific naturalist is that of fitting—or not being able to fit—into the scientific image of the world all sorts of

10  Kevin J. Corcoran and Luis R.G. Oliveira facts and entities that simply do not appear in a strictly physical description of the world, a description strictly and solely in terms of bosons and gluons. For example, facts about and features of consciousness, moral facts and features, mathematical facts, sociocultural facts and properties, etc., do not appear on the scientific naturalist’s ontological inventory. De Caro’s chapter is an extremely illuminating cartography of the geography of versions of scientific naturalism, liberal naturalism, and Baker’s near naturalism. We see both what unites the liberal naturalists and Baker and what divides them. In the final chapter in this section and in this volume, Louise Antony addresses Baker’s claim that there are important aspects of human life and experience that cannot be dealt with scientifically. Antony traces this conclusion back to a linguistic challenge to Impersonalism: the claim that reality is completely describable in language that contains no tenses or indexical expressions. Impersonalism is false, according to Baker, because the indexicality of our robust first-personal concept “I*” picks out a property that is irreducible to non-indexical properties and ineliminable. According to Antony, however, there is no unique and special essentially indexical property that is expressed by our robust first-personal concept “I*.” As Antony sees it, not only are there independent reasons to think there is no such unique property—general reasons related to the contextual semantics of indexicals—there are also reasons to think Baker’s argument is deficient in any case. Baker’s appeal to the irreducibility and ineliminability of “I* sentences” either begs the question or undermines the very point she is trying to make (by providing explanations that reduce and eliminate “I* sentences” anyway). Antony goes on to outline a naturalistic account of the central features Baker thinks naturalism cannot accommodate, namely the computational representational theory of mind, and then defend it against Baker’s own criticisms.

Conclusion Lynne Baker’s contributions to metaphysics, philosophy of mind, and philosophy of religion are many. What unites all of her writing is a relentless defense of the ontology that makes up the world we all inhabit, a physical world made out of the entities and properties that physics studies, as well as the psychological, cultural, and social world that includes persons, beliefs, presidential elections, university committee meetings, disease pandemics, art exhibitions, board meetings, dinner parties, protest marches, marriages, and divorces. Lynne’s is truly “the world that we live and die in, the world where our plans succeed or fail, the world we do or do not find love and happiness in – in short, the world that matters to us.”

Part I

On Practical Realism about the Mind

1

What Is a Concept?1 Christopher Hill

Lynne Baker wrote two books (1987, 1995) and a number of articles defending a very strong version of intentional realism – that is, realism concerning beliefs and other intentional states. One of her principal claims was that statements ascribing intentional states are objectively true or objectively false, no less than statements about the density of fluids, the orbits of planets, and the charges of particles. In other words, according to Baker, there are objective facts that serve as truth-makers for statements ascribing intentional states. Another principal claim was that the reality of intentional states is not relative to any interests, stances, or explanatory practices of human beings. For example, the statement that Columbus believed that Earth is round is true simpliciter. There is no need to prefix an operator such as “According to folk psychology” or “According to the intentional stance.” These two doctrines define a position I will call hardcore intentional realism. The present paper is a defense of this view. Hardcore intentional realism is just one component of a family of views that Baker developed and collected under the label practical realism. Although I fully share her commitment to hardcore realism, I am less enthusiastic about the other components, so I will not be concerned with them here. In my view, and perhaps also in Baker’s, hardcore realism is the most important component, and it can easily be pried loose from the rest of her position. Baker’s defense of hardcore realism covers a lot of territory, but there are three objections to the view that to my knowledge she never addressed. My defense will be complementary to hers in that it will focus on these three challenges. The first is Quine’s complaint that there is no principled basis for individuating propositions (Quine 1960). If legitimate, this complaint would also pose a threat to propositional attitudes, since propositions are the objects on which intentional states are directed. Is the proposition that a fortnight is a period of 14 days identical with the proposition that a fortnight is a fortnight? If not, then in virtue of what are the propositions different? Unless there is a principled basis for answering questions of this sort, Quine maintained, statements

14  Christopher Hill attributing propositional attitudes are too vague to play a substantial role in science. Now it can seem that Quine was presupposing a general principle to the effect that realism about a domain is permissible only insofar as the domain is governed by a precise principle of individuation for its members. If that were true, his objection to propositions could be challenged, for it is by no means obvious that the principle in question holds in all cases. In fact, however, Quine’s objection can be seen as depending on the considerably less tendentious idea that the domains that figure in the designs of scientific experiments require reasonably precise principles of individuation. So far from being trivially false, reflection shows that this idea has considerable merit. Without some way of individuating the entities that define an experiment, it would be unclear what would count as different trials of the experiment, and unclear also what would count as a replication of it. To appreciate the relevance of this point to intentional realism, consider an experiment concerning problem-solving that is being conducted by a cognitive scientist. It is clearly desirable for the scientist to have a solid basis for determining whether different subjects are bringing the same or different beliefs to bear on the experimental task, and a solid basis for determining whether a single subject is bringing the same beliefs to bear on different trials of the task. If the basis for making such determinations were as slender and soupy as Quine maintained – if the determinations were as empirically unconstrained as he maintained – there would be correspondingly slender grounds for thinking that the subjects had been involved in a single experiment. Equally, claims that the experiment had been replicated would become problematic. One of my main goals in the paper is to show that there is a promising proposal for a principle of individuation for propositions. The proposal I have in mind is due to Chisholm (1981, 1989) and is the topic of the next section. The second objection to hardcore realism is also associated with Quine (1960), though it has many echoes in contemporary literature. This is the claim that attributions of intentional states are radically underdetermined by all possible evidence. Quine argues for this view at length in Word and Object, and then goes on to infer the following antirealist conclusion: If we are limning the true and ultimate structure of reality, the canonical scheme for us is the austere scheme that knows no quotation but direct quotation and no propositional attitudes but only the physical constitution and behavior of organisms. (1960, p. 221) Quine also expressed this conclusion by saying that there is no fact of the matter as to whether attributions of attitudes are true or false. The

What Is a Concept?  15 truth values of such attributions are indeterminate. Chomsky responded to Quine by pointing out that underdetermination by evidence does not entail indeterminacy of truth values (Chomsky 1969). All theoretical claims in the sciences are underdetermined by the evidence that is respectively relevant to them, but we hardly regard this as a ground for rejecting scientific realism. This objection is prima facie legitimate. In fact, however, a number of passages make it clear that Quine was not inferring indeterminacy from underdetermination. Rather, he was inferring it from a presumed radical underdetermination – an underdetermination so extreme as to amount to a qualitative difference between our folk practice of attributing attitudes and scientific theorizing. My second goal is to make a case that there is no qualitative difference here. Our folk heuristics for attributing propositional attitudes and for testing such attributions are quite powerful – perhaps less so than the methods of hypothesis formation and testing in mature sciences, but nonetheless sufficient to be compatible with intentional realism. The third objection to hardcore realism comes from arguments for intentional holism, which asserts that all of the intentional states possessed by an individual are interdependent. For example, according to intentional holism, if a given belief is removed from a system of propositional attitudes in which it is embedded, the agent ceases to possess all of the other members of the system as well. They are automatically replaced by attitudes with different essential natures. Versions of holism have been defended by Davidson (1984), Stich (1985), Dennett (1989), and Churchland (1991), among others. Holism poses a serious threat to intentional realism because it implies (i) that few if any individuals share intentional states, and also (ii) that few if any individuals have intentional states that persist across time. The first consequence is alarming because it would be difficult to maintain that different individuals attach the same meanings to their words if the individuals never use their words to express the same beliefs. That is, (i) threatens to undercut our understanding of communication. Further, (ii) is alarming because it challenges our understanding of diachronic rationality. Diachronic rationality is what occurs when individuals work out plans and make choices in advance of acting, and then at appropriate later times perform the actions that have been preselected. In such cases, the actions are made rational by the reasoning that produced the earlier plans. According to holism, however, at the time of acting, the individual agent may no longer possess the intentional states that figured in the original reasoning and may no longer possess the resulting plan. Various philosophers have responded to these problems by invoking similarity, maintaining that communication and diachronic rationality are not threatened as long as it makes sense to speak of whole systems of attitudes as similar to other whole systems. But this is not nearly enough. Any two systems of attitudes will be similar in a number of

16  Christopher Hill respects and also dissimilar in a number of respects. To make the suggestion work, it would be necessary to specify the respects of similarity that are relevant to communication and diachronic rationality, and to describe a scheme for pairing up individual attitudes on the basis of these respects. There have been proposals for doing this (e.g., Churchland 1998, Schroeder 2007), but they have run into difficulties. This is not surprising, since, as Fodor and Lepore pointed out (1992), there is a serious underlying problem. If one were to explain the relevant form of similarity, one would presumably want to say that two beliefs B1 and B2 in different systems are relevantly similar if they stand in similar inferential relations to other beliefs. But these other beliefs cannot be beliefs that are shared by the different systems, for by hypothesis there are no such beliefs. So, what must be said is that B1 and B2 are relevantly similar if they bear similar inferential relations to other members of their respective systems that are relevantly similar. Clearly, we are on our way to a regress or a circle. In view of this, it is not hard to see that on the present approach, similarity of individual beliefs would collapse into overall similarity of total systems of attitudes – a fatal flaw, since overall similarity of total systems is comparatively rare, except in individual agents across short intervals of time. (Each of us has countless beliefs that others lack – about our upbringing, about the literature we have read, about our acquaintances, about what we did yesterday, about what we are seeing now, and so on.) To simplify the discussion, let us focus for the moment on the part of holism that is concerned with beliefs. This is the thesis that an agent would perforce cease to hold every member of a system of beliefs if the agent ceased to hold any one member of that system. The most natural way of opposing this thesis is to embrace the following pair of claims: (C1) In order to hold a given belief, B, an agent need hold no beliefs other than the ones that are constitutive of the concepts that figure in B. (C2) In most cases, the beliefs that are constitutive of a concept are comparatively few in number. In combination, these claims provide a sufficient rationale for rejecting holism. But are the claims acceptable? (C1) is initially plausible and appears to survive reflective vettings, but as Fodor and Lepore have emphasized, (C2) faces a problem that is on the face of it quite serious. The claim presupposes that for any concept C, there is a principled basis for distinguishing between the C-involving beliefs that are constitutive of C and C-involving beliefs that are merely adventitious or nonessential. As Fodor and Lepore pointed out, this presupposition is strongly analogous to, and therefore stands or falls with, the claim that there is a principled basis for distinguishing between the statements containing a word that are constitutive of the meaning of the word and the statements containing the word that play no role in

What Is a Concept?  17 fixing its meaning. In other words, the presupposition stands or falls with the idea that there is a well-motivated analytic/synthetic distinction. It follows that (C2) is called into question by Quine’s critique of analyticity (or, more precisely, by a counterpart of Quine’s critique that is concerned with concepts rather than meanings of words). Since they were persuaded by Quine’s critique, Fodor and Lepore rejected (C2) and also the defense against holism that is based on it. Believing that the defense fails, but appropriately horrified by the prospect of having to embrace holism, Fodor developed an alternative approach to the problem of holism that involves (i) reducing concepts to words in a representational system that he calls the language of thought, and (ii) reducing beliefs to computational relations to sentences composed of such words. 2 Given these reductions, he maintained, we can see how it is possible to possess one belief without possessing many others. To possess a certain belief, one need only have performed some computations on a specific sentence in the language of thought. Computation being what it is, it is possible to have performed the relevant computations on one sentence without having performed those same computations on others. Unfortunately, Fodor ran into serious problems when he turned to the task of developing his language-of-thought proposal. The proposal depends crucially on the claim that there is a universal language of thought – one that is shared at least by all human beings at all times of their lives. Otherwise the proposal could not explain communication or diachronic rationality. Now to vindicate the claim of universality, it is necessary to individuate words in the language of thought in a way that is independent of intersubjective differences in human cognitive powers and faculties, and also in a way that is independent of intersubjective changes in powers and faculties across time. In fact, however, Fodor was unable to do this. He tried to accomplish it by claiming that identity of a word in the language of thought is determined by its effects on all of the high-level processing mechanisms in the human cognitive architecture. (Fodor 1995) But it is clear that not all humans have the same mechanisms, and also that, when two individuals do happen to have the same mechanisms, there can be significant differences in the mechanisms’ manners of operation. Similar points apply to the mechanisms possessed by a single individual at different times. In effect, then, Fodor’s proposal for individuating words in the language of thought confines words to individual minds and significantly restricts the temporal extent of words within individual agents. (For an extended treatment of these points, see Schneider 2011.)

2 See, e.g., Fodor (2008, p. 25, p. 69). I am simplifying. Fodor actually maintained that concepts are reducible to pairs consisting of words in the language of thought and objects or properties that serve as the referents of such words.

18  Christopher Hill Since Fodor’s own proposal for solving the problem of holism seems not to work, there is reason to hope that there is something wrong with the first part of his line of thought – the part consisting of his objection to (C2). I will try to show that this is the case. More specifically, I will try to show that there is after all a principled distinction between the essential or constitutive parts of the role of a concept and the parts of the role that are nonessential or adventitious. I have almost finished setting the stage for the main business of the paper, but before moving on, I need to acknowledge a few assumptions. First, I will be assuming that concepts are the building blocks of propositions and therefore the most fundamental constituents of intentional states. Second, I will assume that concepts are the meanings of words and that propositions are the meanings of sentences: words express concepts and sentences express propositions. My third assumption is that propositions have logical structures that are reflected in the syntactic structures of the sentences that express them. And my fourth and final assumption is that concepts and propositions normally have semantic characteristics. More specifically, concepts normally have referents, and propositions normally have truth conditions. In sum, while I will not be assuming that concepts and propositions are associated with any particular natural language, nor with a language of thought, I will be assuming that they are quasi-linguistic entities. This view is, I believe, widely shared by philosophers, and it also seems to be well entrenched in cognitive science. Anyone who speaks of logical relations among beliefs and attributes truth and falsity to beliefs is committed to some version of the picture. As both my title and these last few remarks no doubt suggest, concepts will be in the foreground in much of the paper. Since they are the most fundamental units of intentionality, they deserve pride of place in a defense of hardcore intentional realism. But we will also be much concerned with beliefs and other intentional phenomena that have concepts as their constituents.

1 Individuating Propositions and Concepts Chisholm taught a seminar at the University of Pittsburgh in the spring of 1974.3 In the course of the seminar, he put forward the following principle of individuation for propositions: For any propositions x and y, x is identical with y just in case, necessarily, for every agent z, z believes x just in case z believes y.



What Is a Concept?  19 As far as I know, Chisholm never published this proposal, though there are related ideas in writings that appeared in the 1980s (1981, 1989). Perhaps he thought the principle too obvious to be worth publishing. After all, it seems inevitable when we reflect that belief is by definition a propositional attitude – that is, a relation between minds and propositions. Chisholm’s doctrine counts as a principle of individuation because it provides a compact condition that propositions satisfy exactly when they are identical, but it provides no basis for determining or finding out whether the condition is satisfied by specific propositions. Thus, for example, it does not provide a basis for determining whether the proposition that a fortnight is a period of 14 days is identical with the proposition that a fortnight is a fortnight. For answers to such questions, we must consult the principles that guide us in attributing beliefs and other intentional states. That is to say, we must consult the laws of folk psychology, together with the superstructure of additional laws that cognitive science has erected on the basis of folk psychology. It may be that the laws in question are not sufficiently powerful to provide answers to all the questions of propositional identity that interest us, even when they are supplemented by an ideally complete body of evidence about individual agents. This would be true if the package consisting of all of the laws and all of the evidence failed to settle an important range of questions of the form, “Does agent x believe that p?” But if a range of questions of this form were to be left open, with the result that a correlative range of questions about propositional identity were to be left open, would that challenge Chisholm’s principle of individuation for propositions? No. To serve its purpose, that principle need only specify the type of facts that would have to obtain in order for pairs of propositions to be identical. It need not provide a rule for determining whether the relevant facts actually obtain in a given case. The adequacy or inadequacy of a rule for determining whether facts obtain is an epistemological issue, not a metaphysical issue bearing on individuation. What if Quine were right in claiming that attributions of belief are radically underdetermined by evidence? In that case, Chisholm’s principle would have little metaphysical importance. It would purport to adjudicate questions about the identity of objects of belief in a situation in which there was little or no motivation for thinking that we should recognize beliefs when we are “limning the true and ultimate structure of reality.” But this still would not be a mark against the principle itself. It would have the right form to do an adequate job if there were a job to do. To rephrase this point, if Quine were to raise questions about the individuation of propositions on the basis of his claim that attributions of belief are radically underdetermined by evidence, he would no longer be stating an independent objection to intentional realism. His individuation objection would have collapsed into his underdetermination objection.

20  Christopher Hill As noted earlier, I am assuming that concepts are the building blocks of propositions. Given this assumption, it is possible to leverage the Chisholmian account of the individuation of propositions into an account of the individuation of concepts. That is, we can say that two concepts are identical just in case it is impossible to convert a proposition containing one of the concepts into a distinct proposition by replacing the given concept with the other.

2 Heuristics for Attributing Intentional States I turn now to Quine’s claim that attributions of intentional states are radically underdetermined by all possible evidence. In my view, this claim is quite wrong. Quine erred because he overlooked the power of our commonsense attributive practice, and also because he failed to appreciate the extent to which that practice is governed by empirical constraints. When one corrects these errors, his radical underdetermination thesis is no longer believable. Our attributive practice has four dimensions. First, there is a battery of low-level algorithms that are sensitive principally to various forms of biological motion, to direction of gaze, and to facial expressions. These algorithms are quick, automatic, inflexible, informationally encapsulated, and moderately reliable, at least when used in a restricted range of contexts. Second, there is a large class of methods that are grounded in laws of folk psychology that are not concerned with language. Processes involving these methods are comparatively slow, but they are also comparatively flexible, and they are reliable across a wider range of contexts. Their reliability is due in part to the fact that they are less informationally encapsulated, and in part to the fact that they are sensitive to a wider range of inputs. Third, there are methods that are grounded in laws of folk psychology that link possession of intentional states to various kinds of linguistic behavior. And fourth, there are methods that are grounded in laws that are concerned with the members of specific categories of concepts, such as the category of natural kind concepts and the category of normative concepts. To experience the low-level algorithms at work, take a look at the classic short video created by the social psychologists Heider and Simmel (1944). It shows very simple shapes (a dot and two triangles) moving around on a screen; but because the movements of the shapes conform to several standard patterns of motion that are common to animals, it is irresistible to see the shapes as having a range of psychological characteristics, including anger, fear, dislike, affection, and joy. These irresistible first impressions are overridden by the higher-level perception that two-dimensional shapes on a screen just are not capable of highlevel psychological states, but when similar impressions occur in a context that is ecologically relevant, they are generally consonant with our

What Is a Concept?  21 general beliefs about intentional states. Another example is afforded by our tendency to take prolonged looking as evidence for attention, and attention as evidence for interest. When you see me gazing at an object or event, you automatically infer that I find the object or event interesting. (See, e.g., Carey 2009) For another illustration, consider the fact that there is a standard facial expression for registering surprise. When you witness this expression on a face, you automatically infer that the owner of the face is surprised, and this leads, in turn, to the hypothesis that the current object of the owner’s attention has in some way violated a preexisting expectation. As noted, the second dimension of our attributive practice is grounded in laws of folk psychology that make no reference to linguistic capacities. I will just note a few familiar facts about these laws. To begin with laws of desire, folk psychology contains some assumptions about basic biological and psychological needs and also principles to the effect that these needs give rise to desires for things that satisfy them. Thus, it seems that folk psychology recognizes that human beings generally need fluids, food, air, safety, warmth, shelter, health, sex, sleep, affection, esteem, and a sense of accomplishment. It also recognizes relationships between these needs and desires that can be illustrated as follows (Cf. Churchland 2013): L1. Generally speaking, agents denied fluids for some time feel thirst. L2. Generally speaking, agents who feel thirst desire to drink fluids. In combination, these principles support inferences from environmental conditions to hypotheses about intentional states. In addition, there are principles linking needs and past experience to desires, for example: L3. Generally speaking, if in the past an agent x has found that a substance F has a pleasant taste, x perceives an instance of F in the current environment, and x is experiencing hunger, x will form a desire to eat F. L4. Generally speaking, if in the past an agent x has found situations of type S to be dangerous, x has a desire to avoid situations of type S in the future. Here is a more general principle that belongs to the same family as L3 and L4: L5. Generally speaking, if N is a need such that in x’s experience N has been consistently fulfilled by operating on instances of the property P in way W, then on those occasions when N is particularly pressing and x’s perceptual state carries information to the effect

22  Christopher Hill that there is an instance O of P in the current environment, x forms a desire to operate on O in way W. In addition to laws that support attributions of desires, there are also laws that support the testing of such attributions. The most important of these is the following principle, which is often cited in the literature: L6. Generally speaking, if an agent desires that p and believes that doing A is an effective means of bringing p about, then, barring conflicting desires or preferred strategies, the agent does A. To apply this law in testing a hypothesis about desire, it is necessary to have on hand a number of hypotheses about an agent’s other desires, and also hypotheses about relevant beliefs, but that should not raise any alarms about radical underdetermination. As Quine himself emphasized, scientific testing rarely seeks confirmation for a single hypothesis. In most cases, it takes a cluster of hypotheses to generate a prediction. Here is another such principle: L7. Generally speaking, if an agent desires that p and discovers that p, then the person is pleased. This principle supports testing because there are postural, facial, and physiological manifestations of pleasure. The laws of desire are matched in power by the laws that support attribution of beliefs and the testing of such attributions. Here are a couple: L8. Generally speaking, if an agent attentively acquires perceptual information indicating that a perceived object O has a certain perceptible property P, the agent forms a belief to the effect that O has P. L9. Generally speaking, if an agent has over time acquired evidence indicating that objects with the perceptible property P1 also possess the perceptible property P2, the agent will form the belief that everything that possesses P1 and possesses P2. L9 rests on an assumption about the language-independent reasoning capacities of agents. Folk psychology contains a number of assumptions of this sort. Here is another one: L10. Generally speaking, if an agent believes that everything that possesses P1 also possesses P2, and the agent attentively perceives an object with P1, then the agent expects that the object also has P2.

What Is a Concept?  23 There are various ways of testing hypotheses about beliefs. As we recently noticed, such hypotheses can be tested in tandem with hypotheses about desires by employing L6 to generate behavioral predictions. Another test consists in looking for facial expressions indicating surprise in contexts in which the hypothesized beliefs would be manifestly false, and by looking for prolonged visual attention to the falsifying object or event. (Cf. Carey 2009.) I hope that in view of (L1)–(L10), the reader will feel that as far as the nonlinguistic portion of our attributive practice is concerned, the burden of proof is very much on philosophers, such as Quine, who maintain that attributions of intentional states are radically underdetermined by evidence. The point to remember in evaluating this claim is that it is not necessary to justify our attributions by showing that they are in some way mandated by non-intentional facts. It suffices that there are powerful heuristics for inferring intentional states from non-intentional phenomena, provided that there are also principles that provide for the testing of such inferences. This is how justification is obtained in science, and there is no reason to demand higher standards here. “But the inferences depend on laws of folk psychology – how do we know that the laws are true?” The answer is that they are tested simultaneously with, and by the same procedures as, hypotheses about particular intentional states and processes. And they are confirmed by the same data that confirm the particular hypotheses. This brings us to the third dimension of our attributive practice, which involves using heuristics to infer intentional states from linguistic behavior, and then testing those attributions by combining them with laws to generate predictions of further linguistic behavior. The main folk principle supporting language-based attributions to speakers of English seems to be the law that Kripke called the disquotation principle (Kripke 2011): L11. If a normal English speaker x reflectively and sincerely assents to “p,” then x believes that p. The folk principle supporting language-based testing of attributions of belief is presumably the converse of L11: L12. If x believes that p, then, assuming (i) that x is a normal English speaker, (ii) that x believes that p, (iii) that x is not reticent, (iv) that x is in a reflective frame of mind, and (v) that x is prepared to speak sincerely, x will assent to “p.” It is also possible to make and test language-based attributions to speakers of languages other than English, but those attributions require

24  Christopher Hill support from different laws. I will not be able to discuss this additional aspect of our practice here. Now, reflection shows that attributions to English speakers require much more than L11. Specifically, they require heuristics for inferring four conditions from non-intentional data: x is a normal English speaker, x is being reflective, x is being sincere, and x assents to “p.” L12 also requires heuristics of these sorts. Are we able to meet these requirements? Yes, as long as we think of heuristics as conferring probabilities on hypotheses instead of guaranteeing their truth. Let us start with assent. Here we can follow Quine, who was prepared to be more generous in discussing assent than he was in discussing other intentional phenomena. He offered this heuristic for identifying expressions as signs of assent: “A speaker will assent to a sentence in any circumstance in which he would volunteer it.” (Quine 2008, p. 342) In other words, to identify an expression as a sign of assent, one watches an informant until an utterance is volunteered. Perhaps the informant utters “p.” One then turns “p” into a question and addresses it to the informant. The informant’s response will probably be a term that expresses assent. Quine does not regard this process as providing a guarantee: the resulting hypothesis must be tentative. But it will be confirmed if it is found to yield correct predictions about similar cases, and also if it generates correct predictions when it is combined with principles such as L12. It is worth noting that Quine’s procedure for identifying signs of assent presupposes a principle of folk psychology – that agents will assent to sentences that they are also disposed to volunteer. It would be inconsistent to allow this folk principle while disallowing others. Are there heuristics for the condition x is a speaker of English? Of course! In fact, folk psychology provides us with several. The most basic one simply authorizes us to believe that an agent satisfies the condition if an enquiry shows that the agent is capable of producing grammatical sentences containing words from the English lexicon. Simple though it is, this procedure is no doubt highly reliable. But there are more demanding heuristics that can be used to increase the probability that, in addition to mastering the vocabulary and syntax of English, the agent has also learned the meanings of the words in that vocabulary. Let W be an English word, and let it be the case that an agent participates in W-involving patterns PP such that (i) PP are shared across the population of users of W; (ii) PP are highly predictive of successful communication involving W, where measures of success include fluency of conversations, acceptance of assertions by listeners, achievement of goals in linguistically guided cooperative ventures, and transmission of perceptually acquired information; (iii) PP are normative, in the sense that speakers are disposed to correct W-involving behaviors that conflict with the patterns, simply in virtue of awareness of that conflict (as opposed to

What Is a Concept?  25 awareness of extralinguistic fact); (iv) PP are intrasubjectively sturdy, in the sense that they are comparatively stable across changes in interests and extralinguistic information; and (v) PP are rewarded by other speakers independently of context, in ways that are both subtle and pronounced. If an agent participates in patterns with these properties, we can infer with reasonably high probability that the agent has learned the meaning of W. Of course, W may be ambiguous. If so, we can credit the agent with knowledge of additional meanings of W if the agent participates in additional sets of W-involving patterns with the given features. (Linguists have devised tests of ambiguity that we can use to confirm hypotheses that attribute multiple meanings to W. [See, e. g., Sennet 2016]) Given that a speaker knows the meanings of the terms in an English sentence, other English speakers will be able to assess the speaker’s utterance for truth and falsity in various contexts. Investigations of this sort will suggest hypotheses about the degree to which the speaker is reflective and sincere. I turn at last to a fourth way in which folk psychology supports attributions of intentional states. This dimension of our attributive practice presupposes that our lexicon of concepts can be divided into broad categories in a natural way, yielding such categories as logical concepts, perceptual concepts, natural kind concepts, and normative concepts. Folk psychology contains a number of principles governing possession of members of these categories – principles that help us to arrive at and test attributions of intentional states. These principles connect possession of concepts with intentional states of various other kinds. It is presupposed that principles such as some of the ones given earlier provide us with ways of arriving at and testing hypotheses about the latter states. I will give several examples of such principles, beginning with what I will call nominal concepts – concepts that are counterparts of linguistic proper names. My main claim about these concepts can be expressed as follows: L13. If a nominal concept C refers to an object O, an agent cannot possess C unless the agent uses C to encode and store information about O. We can appreciate the merits of the first part of L13 by recalling Kripke’s work on what is involved in knowing the meaning of a proper name in a natural language. (Kripke 1980) Let O be an object, P be a proper name that refers to O, and S be any speaker whose lexicon contains P. According to Kripke, if S is to be credited with knowledge of the meaning of P, S’s use of P must be connected to O by a causal chain of some kind, usually involving a series of speakers each of whom transmits P to later members of the series. This can also be expressed, I suggest, by saying that if S is to be credited with knowledge of the meaning of P, S’s

26  Christopher Hill use of P must show the influence of information flowing from O to S. Kripke argued convincingly for this claim, using thought experiments to show that the claim holds not only for actual speakers but for possible speakers as well. But why is Kripke’s work relevant to L13? The answer is given by the fact that nominal concepts are the meanings of proper names. Suppose that P expresses the concept C. Given this assumption, if there is an informational constraint on knowledge of the meaning of P, there must be a parallel constraint on possessing the concept C. Concepts of natural kinds are similar to nominal concepts, so it is not surprising that they are governed by a principle that is similar to L13. The principle I have in mind is L14: L14. If a concept C refers to a natural kind K, an agent cannot possess C unless the agent uses C to encode and store information about members of K. Just as L13 is supported by Kripke’s work on the meanings of proper names, so also L14 is supported by Putnam’s work on terms that stand for natural kinds. (Putnam 1975) As the reader may recall, Putnam used thought experiments to establish that an agent can count as knowing the meaning of a term that refers to a kind only if the agent’s use of the term reflects the influence of information flowing from members of the kind. Since kind terms express kind concepts, it follows that L14 is correct. In addition to containing nominal concepts and kind concepts, our descriptive conceptual vocabulary contains concepts whose use is grounded in empirical theories. L15 is a principle that governs members of this category: L15. If C is a concept that belongs to an empirical theory, and the theory is concerned with a domain D, then an agent cannot possess C autonomously unless the agent is disposed to use C in predicting and/or explaining facts about members of D. As the reader will have observed, L15 is restricted to agents who possess theoretical concepts autonomously – that is, without being disposed to defer to others with regard to questions of proper deployment. (Autonomous possession is explained further below.) It is of course true that many agents may possess a theoretical concept without having any command of the theory that supports predictions or explanations of the relevant domain. The concept of a black hole is an example. In all such cases, however, possession of a concept is grounded in deference to experts. In contrast, it is plausible that an expert’s command of a theoretical concept is grounded in an ability to appreciate the predictive and explanatory power that the concept bestows.

What Is a Concept?  27 I turn now briefly to logical concepts. Here is a restricted principle that applies to an important subset of this category: L16. If C is a logical concept that expresses a Boolean operation, O, an agent cannot possess C unless the agent’s dispositions to use C in inference are constrained by in/elim rules that are valid for O. This gives expression to the familiar idea that the semantic interpretation of a logical concept depends on the role that the concept plays in reasoning. To appreciate the merits of the idea, consider a speaker of English whose use of “or” is guided by disjunction introduction, but who has no disposition to conform to disjunctive syllogism, preferring instead to perform inferences of the following form: p or q, p Ⱶ q. Suppose that the latter disposition is incorrigible. I believe I speak for everyone in saying that we would deny that the person’s use of “or” expresses the concept of classical disjunction. This is not to say that reasoners must always conform to valid rules of inference in order to possess logical concepts. Everyone is guilty of performing fallacious inferences. L16 should be seen as concerned with dominant tendencies, not with universal patterns. For a final example, let us consider what is involved in possessing normative concepts. In general, it seems that an essential feature of normative concepts is to guide choice and decision. To illustrate this point, let us focus on a comparatively simple relationship of this sort, the relationship between the concept of all-things-considered obligation and decision. The following principle seems correct: L17. If C is the concept of all-things-considered obligation, an agent cannot possess C unless the agent is disposed to transition from judgments of the form All things considered, I ought to perform A to decisions of the form I will perform A.4 L17 has an obvious appeal, but one might worry that it is challenged by cases of akrasia. There are agents who have continuing akratic tendencies: time and again they judge that they ought to perform an action and yet fail to perform it. Now in some cases of akrasia, the agents do, in fact, move from judgments of obligation to decisions to act. They just fail to act in accordance with their decisions. Cases of this sort pose no threat to L17. It is no doubt true, however, that there are also cases of a different sort – cases in which agents fail to make decisions that accord with their judgments of obligation. Is there any way of protecting L17 from the challenge posed by cases of the latter sort? I think the answer is

4 I am indebted here to Jamie Dreier.

28  Christopher Hill “yes,” for reflection suggests that we would not credit agents with making judgments of all-things-considered obligation unless they showed remorse or regret for failing to act in accordance with their judgments and recognized their failings as somewhat incapacitating. If this is true, L16 is on the right track, but it should be modified by adding that exceptions to the indicated judgment/decision pattern must recognize that their ability to make decisions is impaired. When we make this qualification, we seem to have a principle that holds for actual agents and also for their merely possible colleagues. To conclude this portion of our inquiry, there are quick and dirty algorithms and also a number of laws of folk psychology that enable us to arrive at attributions of intentional states. There are also laws that enable us to test such attributions. Accordingly, it is plausible that Quine substantially overestimated the extent to which hypotheses about intentionality are underdetermined by evidence.

3 Fodor on Essential Properties of Concepts As we observed, Fodor maintains that there is no principled basis for distinguishing between the cognitive states, processes, and dispositions that are constitutive of concepts and the states, processes and dispositions that are nonessential or adventitious. In defending this thesis, he simply cites Quine’s critique of analyticity. Quine argued that there is no principled basis for distinguishing between the statements containing a term that are constitutive of its meaning and the statements containing the term that have no bearing on its meaning. Evidently, Fodor thinks that Quine’s discussion settles the question. This is, however, a serious mistake. Quine’s “Two Dogmas of Empiricism” contains two main arguments against analyticity (Quine 1980). The first argument maintains that all of the traditional attempts to define analyticity have made essential use of concepts that are either equivalent to analyticity or no less in need of clarification. It is widely agreed that this argument was successfully challenged by Grice and Strawson (1956), who pointed out that a concept can be legitimate, and indeed, in good working order, even if there is no satisfactory definition of it in terms of more basic concepts. In the second argument, Quine presupposes that the only hope of explaining analyticity in an illuminating way is to equate it with the property of being immune to empirical revision. The argument attempts to dash this hope by maintaining that every statement is empirically revisable. This is the argument that philosophers have found deeply challenging. Now in my view, Quine’s argument for universal revisability fails (Hill 2019), but that is not the point that is most pertinent to our present concerns. For our purposes, the important point is that when analyticity is explained in terms of unrevisability, a critique of it has no real bearing

What Is a Concept?  29 on the idea that beliefs and other cognitive phenomena can be constitutive of concepts. A cognitive state or disposition could be constitutive of a concept even if Quine was right and there was no such thing as immunity to empirical revision. What matters for constitutivity is not whether an item is immune to revision but whether an agent would automatically cease to possess the concept if the item were revised. Accordingly, it is a mistake to suppose that the following principle is correct: (P1) Where X is a state or disposition involving the concept C, X is constitutive of C (that is, is essential to C) only if X could not be revised on empirical grounds. The correct principle is actually this, quite different, one: (P2) Where X is a state or disposition involving the concept C, X is constitutive of C (that is, is essential to C) only if, necessarily, an agent would cease to possess C if the agent ceased to possesses X. The problem with Fodor’s argument against essential properties of concepts is that it depends on thinking of essentiality in terms of (P1) rather than (P2). On its face, (P2) is a thesis about the conditions under which agents lose possession of concepts. Reflection shows, however, that it is equivalent to (P3): (P3) Where X is a state or disposition involving the concept C, X is constitutive of C (that is, is essential to C) only if, necessarily, every agent who possesses C also possesses X. Even though the two principles are equivalent, (P3) seems more fundamental, perhaps because its truth is so transparent. After all, how could an item be essential to C if there were possible agents who possessed C but lacked the item?5 Reflection shows, however, that it is necessary to qualify (P3). We have been taught by Putnam (1979) and Burge (1979) to distinguish between deferential and non-deferential possession of concepts. Roughly speaking, an agent possesses a concept deferentially if the agent is disposed to seek and be guided by the advice of others in his or her use of the concept. No less roughly, an agent possesses a concept non- deferentially if the agent uses the concept autonomously – that

5 The idea that there is a deep connection between the essential natures of concepts and their possession conditions is due to Peacocke (1992).

30  Christopher Hill is, without being disposed to correct his or her use of the concept in response to the advice of others. This distinction reminds us that in many cases, at least, there are no intentional states or dispositions that all users of a concept share. Instead, there will be a set of states and/ or inferences that are shared by all non-deferential users of a concept – that is, all users who are “experts” in its use. Other users will not share this cognitive equipment, but will instead be disposed to defer to those who do. These reflections indicate that (P3) should be replaced with (P4): (P4) Where X is a state or disposition involving the concept C, X is constitutive of C (that is, is essential to C) only if, necessarily, every agent who possesses C either possesses X or is disposed to defer to an agent who possesses X. It is this principle, and not (P3), that anchors the notion of a constitutive state or inference.

4 More on Essential Properties We are seeking a principled basis for distinguishing between properties of concepts that are essential to them and properties that are adventitious. Fodor argued that no such basis is available. Thus far we have seen that his argument fails, and have found that (P4) seems to provide a necessary condition for a property of a concept to count as essential. Is the condition also sufficient? If we were to answer yes, we would have an account of the essential properties of concepts that can be captured by (P5): (P5) Where X is a state or disposition involving the concept C, X is constitutive of C (that is, is essential to C) just in case, necessarily, every agent who possesses C either possesses X or is disposed to defer to an agent who possesses X. (P5) is simple and elegant, but the conception of the essential properties of concepts that it offers is quite thin. I think our intuitions about what is essential to a concept point toward a more robust conception of essence.6

6 To illustrate, notice that (P4) implies that the following trivial property is an essential property of the concept of a dog: being so disposed that one would believe that everything either is a dog or isn’t a dog if one possessed the concepts of negation, disjunction, and universal quantification. Presumably every agent who possesses the concept of a dog has this dispositional property, but it would be counterintuitive to claim that it is constitutive of the concept. It has no bearing on what makes the concept distinctively useful to those who possess it.

What Is a Concept?  31 I will here assume a conception of essence that implies that an essential feature of an entity must satisfy two conditions: (i) it must be a necessary feature, just as (P4) requires; and (ii) it must explain other features of the entity. This conception was anticipated by Locke (1975, III.iii.15), and perhaps even by Aristotle (1994, 75a42-b2; Charles 2000); and it appears as a full-blown idea in the writings of a number of contemporary authors, including (paradigmatically) Putnam (1975) and Kripke (1980). There is also considerable experimental evidence that something like this Putnamian conception is built into our commonsense conceptual scheme. (Keil 1989, Gelman 2003) In line with the Putnamian principle, I will be assuming that the essential features of concepts must explain other features of concepts that count as appropriate explananda, according to some reasonable criterion of appropriateness. But how exactly should this rather abstract principle be applied? It is natural to suppose the answer should come from the explanatory agenda of cognitive science. That is, in seeking the essential properties of concepts, we should begin by identifying the explanatory goals of cognitive science that are especially deep and fundamental. This will tell us what features of concepts count as appropriate explananda. We can then try to identify what the essential features of concepts are by checking to see what actually explains the given features. When we reflect, we find, I suggest, that from the perspective of cognitive science, the most important explanatory questions concerning a concept are questions about the causes of the concept’s contribution to the survival and flourishing of the agents who non-deferentially possess it – that is to say, questions about the causes of the benefits that accrue to agents in virtue of non-deferentially possessing the concept, where the benefits include contributing to the fulfilment of biological needs, enabling the satisfaction of desires, causing positive emotions, and increasing pleasure in a range of contexts. These questions are intrinsically important, but they also have an instrumental importance in that the causes that explain the given benefits will also explain why the concept is acquired, retained, deployed, and passed on to others. In short, they will explain why agents possess the concept. This seems to be an excellent reason for taking them to be essential. To summarize, when the foregoing Putnamian principle about the relationship between essence and explanation is combined with a plausible hypothesis about the explanatory agenda of cognitive science, we arrive at a principle concerning the essential features of concepts that can be expressed as follows: (P6) Where X is a state, process, or disposition involving the concept C, X is constitutive of C (that is, is essential to C) only if X is

32  Christopher Hill deeply explanatory of the most important benefits that come from possessing C. Now of course, in referring to the most important benefits bestowed by possessing a concept, I am using a vague idea. It follows that the idea of properties that explain important benefits is also vague. Does it follow in turn that any account of the essences of concepts that includes (P6) will be quite vague, perhaps unacceptably so? No. When (P6) is combined with (P4), the pair provides a test for essentiality that is perfectly precise, for (P4) in effect restricts the relevant explanatory properties to properties that a concept instantiates in all of the possible agents who non-deferentially possess the concept. In other words, while (P6) is by itself too vague and open-ended to stand alone as a criterion of essentiality, (P4) compensates for these shortcomings.7 I will take it as a tentative working hypothesis that in addition to stating necessary conditions for a feature of a concept to count as essential – or in other words, as constitutive of the concept – (P4) and (P6) combine to provide a sufficient condition. Assuming this hypothesis is right, we can say that the essence of a concept C consists of all and only the C-involving states and dispositions that satisfy (P4) and (P6).8 So far so good; but to apply the foregoing hypothesis to obtain conclusions about the essential properties of a specific concept, we must make use of a theory of mind that specifies the relations of casual dependence among the properties of the concept and also identifies the aspects of the use of the concept that are significantly beneficial. The best candidate for such a theory is clearly folk psychology, as corrected and extended by cognitive science. This theory can be said to represent the mind as an information processing system that takes in information about the environment that is provided by perception and testimony; uses concepts to encode, store, and classify that information; and then deploys those concepts in various mental activities that are to some extent ends in





What Is a Concept?  33 themselves, but that mainly serve to guide behaviors that promote the survival and flourishing of the agent. Some of the assumptions of the theory are articulated in Section 2.

5 Examples of Essential Properties In Section 2, we noticed that concepts can be sorted into categories that seem to reflect deep similarities. We also noticed that there are categoryrelative principles linking autonomous possession of concepts with intentional states and dispositions of various kinds. To illustrate these points, I cited principles governing possession of concepts that belong to five categories – nominal concepts, kind concepts, theoretical concepts, logical concepts, and normative concepts. Perhaps it will be useful to cite the principles again: L13. If a nominal concept C refers to an object O, an agent cannot possess C unless the agent uses C to encode and store information about O. L14. If a concept C refers to a natural kind K, an agent cannot possess C unless the agent uses C to encode and store information about members of K. L15. If C is a concept that belongs to an empirical theory, and the theory is concerned with a domain D, then an agent cannot possess C autonomously unless the agent is disposed to use C in predicting and/or explaining facts about members of D. L16. If C is a logical concept that expresses a Boolean operation, O, an agent cannot possess C unless the agent’s dispositions to use C in inference are constrained by in/elim rules that are valid for O. L17. If C is the concept of all-things-considered obligation, an agent cannot possess C unless the agent is disposed to transition from judgments of the form All things considered, I ought to perform A to decisions of the form I will perform A. Now as we observed in Section 2, it is plausible that these principles are necessary truths about membership in the categories: they seem to apply to all possible agents. Assuming that this is right, we can conclude that they state possession conditions for concepts that satisfy (P4): (P4) Where X is a state or disposition involving the concept C, X is constitutive of C (that is, is essential to C) only if, necessarily, every agent who possesses C either possesses X or is disposed to defer to an agent who possesses X.

34  Christopher Hill It follows that we are already halfway toward identifying essential properties of concepts that belong to the five categories. In this section, I will urge that these principles about possession of concepts from the five categories dovetail with plausible conjectures about the benefits that accrue to possessors of the concepts in virtue of possessing them. That is to say, I will try to establish the following proposition: if a concept C belongs to one of the five categories, say category Z, then the C-involving states and dispositions that are most fundamentally responsible for the benefits conferred by non-deferential possession of C are ones that are, according to the foregoing principles, essential to members of Z. To the extent that this effort is successful, we can claim to have identified essential properties of members of the five categories – that is, properties of members that satisfy both (P4) and (P6): (P6) Where X is a state, process, or disposition involving the concept C, X is constitutive of C (that is, is essential to C) only if X is deeply explanatory of the most important benefits that come from possessing C. This will give us a constructive argument for the thesis that concepts have distinctive essential natures. Members of the five categories may of course have essential properties beyond those we identify, but further exploration of the scope of essentiality lies beyond the scope of the present essay. To begin with nominal concepts, let C be a member of this category, and suppose that C is used to encode and store information about an object O. We must ask whether this fact about the use of C is deeply explanatory of important benefits that accrue to possessors of C in virtue of possessing it. Now in many cases the answer is obviously affirmative. Suppose, for example, that C is being used by an agent A, and O is a person with whom A is personally acquainted. In this case, the fact that C is used to encode and store information about O will clearly be of considerable practical value to A. In virtue of having a store of information about O, A will be able to predict O’s actions, plan for joint ventures with O, know how to help O, and so on. Similar observations apply if O is a city or a building or a natural landmark with which A is personally acquainted. But what about cases in which O stands at a considerable remove from A, in space or in time or in both, and there is no question of direct interaction between A and O? Suppose, for example, that O was an ancient king of a minor realm. It is still quite possible that A uses C to encode and store information about O, but what benefits does this information confer on A? The benefits will of course be quite modest, unless A is a professional historian, but we are less concerned here with the absolute quantity of benefits that possession of a concept confers, than with the proportion of that absolute quantity that is due to A’s informational connections with O. And it seems that the proportion will

What Is a Concept?  35 be high. Principal among these is that A’s use of C will be reinforced to the extent that it is perceived by A’s associates and by A personally to be conditional on information flowing from O. After all, both A and A’s associates prefer truth to error, and they are all aware that beliefs based on information are much more likely to be true than beliefs that are not. Accordingly, it appears that across the spectrum of possible cases, informational connections between nominal concepts and their referents play a fundamental role in explaining the benefits conferred by possession of nominal concepts. Reflection shows that the same observations apply, mutatis mutandis, to informational relationships between kind concepts and the kinds to which they refer. Turning now to theoretical concepts and L15, and remembering that L15 is restricted to agents who possess such concepts autonomously, let us suppose that C is a concept whose home base is a theory that is concerned with the domain D. What are the benefits that accrue to agents who possess C autonomously, and what facts involving the use of C are responsible for those benefits? It is pretty clear that the benefits come from the predictive power and the explanatory power that C contributes to the theory in question. The ability to predict is a crucial component of the ability to control nature and the ability to adapt to it and is, therefore, of great value. Further, explanation promotes understanding, and as we all appreciate, the human mind yearns to understand how things work. Explanations purport to provide understanding. To be sure, explanations lose most of their appeal if they turn out to be false; but as long as an explanation seems to be succeeding, we feel that it affords a cognitive purchase on items we care deeply about – specifically, causal relations among external phenomena, and causal relations linking external phenomena to our perceptual evidence. We value this sense of cognitive purchase partly for its own sake but mainly because it affords a basis for such necessary activities as predicting, planning, and deciding. The basis may turn out to be misleading, but until it does, we feel that we are right to trust it and we are grateful to have it. These observations about the value of theoretical concepts are of course in line with L15, which claims that use of such concepts in prediction and explanation is necessary for autonomous possession of the concepts. This brings us to logical concepts and L16. In recent years, psychologists have maintained that much of the use of logical concepts is governed by probabilistic heuristics rather than valid logical deduction. (Oaksford and Chater 2010) This raises the question of whether the heuristics are in some sense fundamental in human reasoning or are instead just shortcuts that are dominated by and somehow dependent upon inferential commitments that can be codified as deductive rules. I cannot discuss this interesting question here. I will just assume that there exist logical concepts that are governed, at the most fundamental level, by principles such as L16. If there are concepts that are grounded more

36  Christopher Hill fundamentally in probabilistic heuristics, they will require principles of a different kind. That said, I think it is clear that the benefits of possessing logical concepts are principally due to the inferential commitments that govern the use of the concepts in deductions. Yes, it is important to have concepts that can be used to construct complex propositions, such as the ones that express laws of nature, but those complex propositions would be inert and useless were it not for their deductive relationships to other propositions, such as the predictions that follow from laws of nature. The value of logical concepts stems from their role in reasoning, and ultimately, from the psychologically basic in/elim rules that ground those roles. What about normative concepts? L17 claims that the concept of allthings-considered obligation is necessarily linked to decisions, at least in properly functioning cognitive systems. Can we also say that this link contributes to the value of the concept in a fundamental way? Yes. Take away the connection between judgments of obligation and decisions and what do you have left? Not much, beyond the self-criticism that would be a psychological staple of a completely and incorrigibly akratic agent. And that self-criticism would have no benefit, for it would have no tendency to effect a reform. Accordingly, it is plausible that the link between possession and role described in L17 is largely responsible for the value of the concept.

Conclusion I have tried to extend Baker’s defense of hardcore intentional realism by responding to three objections that she did not directly address. Like Baker, I have devoted quite a bit of attention to beliefs and other propositional attitudes, but the focus has been a bit different than in her work because I have tried to keep concepts in the foreground. If the foregoing arguments are sound, we can conclude that concepts are governed by a natural principle of individuation, that the procedures for attributing them are empirically well grounded, and that there is a principled basis for distinguishing between the states and dispositions that are constitutive of a concept and the states and dispositions that are merely adventitious.

References Aristotle (1994). Posterior Analytics. In J. Barnes (Ed.), The Complete Works of Aristotle (pp. 39–113). Princeton: Princeton University Press. Baker, L. R. (1987). Saving Belief. Princeton: Princeton University Press. Baker, L. R. (1995). Explaining Attitudes. Cambridge: Cambridge University Press. Burge, T. (1979). Individualism and the Mental In P. A. French, T. F. Uehling, and H. K. Wettstein (Eds.), Midwest Studies in Philosophy, IV (pp. 73–121). Minneapolis: University of Minnesota Press.

What Is a Concept?  37 Carey, S. (2009). The Origin of Concepts. Oxford: Oxford University Press. Charles, D. (2000). Aristotle on Meaning and Essence. Oxford: The Clarendon Press. Chisholm, R. M. (1981). The First Person. Minneapolis: University of Minnesota Press. Chisholm, R. M. (1989). On Metaphysics. Minneapolis: University of Minnesota Press. Chomsky, N. (1969). Quine’s Empirical Assumptions. In D. Davidson and J.  Hintikka (Eds.), Words and Objections (pp. 53–68). Dordrecht: D. Reidel. Churchland, P. M. (1998). Conceptual Similarity across Sensory and Neural Diversity: the Fodor/Lepore Challenge Answered. Journal of Philosophy 95: 5–32. Churchland, P. M. (1991). Some Reductive Strategies in Cognitive Neurobiology. In P.M. Churchland (Ed.), A Neurocomputational Perspective (pp. 77–110). Cambridge, MA: MIT Press. Churchland, P. M. (2013). Matter and Consciousness, 3rd ed. Cambridge, MA: MIT Press. Davidson, D. (1984). Radical Interpretation. In D. Davidson (Ed.), Inquiries into Truth and Interpretation (pp. 125–154). Oxford: Clarendon Press. Dennett, D. C. (1989). Three Kinds of Intentional Psychology. In D. Dennett (Ed.), The Intentional Stance (pp. 43–68). Cambridge, MA: MIT Press. Fodor, J. A. (1995). The Elm and the Expert. Cambridge, MA: MIT Press. Fodor, J. A. (2008). LOT 2. Oxford: Oxford University Press. Fodor, J. and Lepore, E. (1992). Holism. Oxford: Blackwell. Gelman, S. (2003). The Essential Child. Oxford: Oxford University Press. Grice, H. P. and Strawson, P. F. (1956). In Defense of a Dogma. The Philosophical Review 65: 141–158. Heider, F. and Simmel, M. (1944). Untitled video. Retrieved from https://www. youtube.com/watch?v=VTNmLt7QX8E Hill, C. S. (2011). Review of Paul Horwich, Truth – Meaning – Reality. Mind 120: 1262–1270. Hill, C. S. (2019). Unrevisability. Synthese. doi: 10.1007/s11229-019-02262-1 Horwich, P. (1999). Meaning. Oxford: Clarendon Press. Keil, F. P. (1989). Concepts, Kinds, and Conceptual Development. Cambridge, MA: MIT Press. Kripke, S. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press. Kripke, S. (2011). A Puzzle about Belief. In S. Kripke (Ed.), Philosophical Troubles, Volume I (pp. 125–161). Oxford: Oxford University Press. Locke, J. (1975/1689). An Essay Concerning Human Understanding. P. H. Nidditch (Ed.). Oxford: Clarendon Press. Oaksford, M. and Chater, N. (2010). Cognition and Conditionals: Probability and Logic in Human Thinking. Oxford: Oxford University Press. Peacocke, C. (1992). A Study of Concepts. Cambridge, MA: MIT Press. Putnam, H. (1975). The Meaning of Meaning. In H. Putnam (Ed.), Philosophical Papers, Volume II (pp. 215–271). Cambridge: Cambridge University Press. Quine, W. V. (1960). Word and Object. Cambridge, MA: MIT Press.

38  Christopher Hill Quine, W. V. (1980). Two Dogmas of Empiricism. In W.V. Quine (Ed.), From a Logical Point of View 3rd ed. (pp. 20–46). Cambridge, MA: Harvard University Press. Quine, W. V. (2008). Confessions of a Confirmed Extensionalist and Other Essays. Cambridge, MA: Harvard University Press. Schneider, S. (2011). The Language of Thought. Cambridge, MA: MIT press. Schroeder, T. (2007). A Recipe for Concept Similarity. Mind and Language 22: 68–91. Sennet, A. (2016). Ambiguity. Stanford Encyclopedia of Philosophy. Retrieved from https://plato.stanford.edu/entries/ambiguity/ Stich, S. (1985). From Folk Psychology to Cognitive Science. Cambridge, MA: MIT Press.

2

Practical Realism about the Self Carolyn Dicey Jennings

Baker’s later writings on the self and her property constitution view inspired my own work on a brain-based account of the self, or the subject of attention (see Baker, 2009, 2016; Jennings, 2012, 2020). While we differ on some of the details, we share a non-reductive materialist approach that takes as central the role of the body and its environment. In this chapter, I wish to go further in support of my account of the self, using some of Baker’s earlier work (e.g., Baker, 1995). Namely, I will use her work on practical realism as further evidence in support of my view. This work takes seriously the commitments of our everyday interactions, which she sees as including a role for beliefs. More fundamental than our commitment to beliefs, however, is our commitment to the holders of those beliefs—persons or selves. I will thus argue that if we should be practical realists about anything, it is the self, or the subject of attention. Along the way I address the opposing view, which I call the “illusion view”—the idea that the existence of a self with its own causal powers is an illusion. My account of the self, supported by some of Baker’s reasoning, provides a remedy to such skepticism.

1 Baker on Practical Realism In her early work, Baker espouses a theory that she dubs “Practical Realism.” It is “practical” because it concerns our everyday practices (Baker, 1995, p. 19). Specifically, it holds that we should be realists about the existence of that which is implied by our everyday practices, even in the absence of validation by the sciences (Baker, 1995, p. 21). Baker focuses her discussion on the existence of beliefs, contrasting her position with what she calls “the standard view.” According to the standard view, Baker says, beliefs are reducible to brain states; if a belief cannot be so reduced, then it does not exist (Baker, 1995, p. 6). She claims, instead, that our everyday practices serve as independent evidence for beliefs. Moreover, this independent evidence supports an understanding of beliefs as not reducible to brain states. Baker’s description of the standard view is intended to capture the most popular theories in philosophy of mind at the time of writing. These are

40  Carolyn Dicey Jennings listed as including type and token identity theories, constitution theories, functionalism, eliminative materialism, and certain relational views (Baker, 1995, pp. 8–12). Baker sees such theorists as committed to the standard view because they embrace naturalism: they see the standard view as the only alternative to dualism, they see science as supporting the standard view, and they think the standard view is the only way to explain how beliefs can have causal power (Baker, 1995, pp. 12–17). Baker rejects these claims, while nonetheless embracing naturalism. She calls her own view “radical relationism,” which she takes to be a type of “non-reductive materialism” (Baker, 1995, p. 156). Take, for example, what Baker calls “the most detailed attempt to provide an explanatory role for belief”—Dretske’s relational account (Baker, 1995, p. 62). Dretske’s view is a version of the standard view because beliefs are taken to supervene on states or properties of the brain (Baker, 1995, p. 57); it is a relational view because beliefs have meaning only in virtue of their role in bringing about a behavior by indicating something (Baker, 1995, p. 58). Despite its virtues, Baker finds Dretske’s account to fail due to a vicious circularity. Specifically, in Baker’s description, Dretske finds beliefs to have causal power because they have meaning, and he finds them to have meaning because of their causal role in behavior (Baker, 1995, p. 59). For Baker, this is one example of how the standard view has failed to find a noncircular explanation of beliefs. Yet, she finds that instead of abandoning beliefs we should abandon the standard view, in favor of practical realism. If we are practical realists—that is, if we are realists about the existence of that which is implied by our everyday practices—then we should accept the existence of beliefs, according to Baker. This is because our everyday practices depend on belief attributions, which are committed to the existence of beliefs. When we, for example, use a crosswalk to cross the street, we are committed to the existence of nearby drivers’ beliefs: the belief that we are in the crosswalk, that we have right of way, and that they should stop. If we were not committed to the existence of these beliefs, then it would be difficult to explain our behavior. For Baker, the commitments of these everyday practices trump our inability to explain the physical basis of belief; even if we cannot make sense of belief within a naturalistic worldview, we should accept the existence of beliefs. Further, Baker finds that our everyday practices support a radically relational understanding of belief. This form of relationism is “radical” because it rejects the microphysical reduction inherent in the standard view. Baker thinks we have focused on microphysical reduction due to a conflation of constitution with supervenience. That is, we think of beliefs as microphysically reducible because we believe they are constituted by brain states and so must supervene on brain states. Baker argues forcefully against this idea using the concept of multiple realizability: certain processes or mechanisms can be realized in multiple ways, such

Practical Realism about the Self  41 that even if an instance of such a process or mechanism is constituted in a particular way it need not supervene on that particular constitution (Baker, 1995, p. 132). Take the example of physical currency: the United States dollar may be constituted by paper, but it does not supervene on paper, since it might be constituted in other ways (e.g., plastic). As with currency, Baker sees beliefs as multiply realizable. This is because “one’s state of believing that p depends on global properties… of whole organisms” (Baker, 1995, p. 22). In other words, beliefs are states held by persons, not their parts: “belief is a global state of a whole person, not of any proper part of the person, such as the brain” (Baker, 1995, p. 153). We do not say that necks or knees believe, we say that people believe. We might hold that specific beliefs are constituted by specific parts of a person, such as specific neural circuits in the brain, while also holding that they could be constituted by other neural circuits. Put another way, the reason that currency and beliefs are multiply realizable with respect to their constituent parts is that they concern phenomena at a different scale than those parts, that engage with other phenomena at their own scale. That is, Baker sees beliefs as relational, holding between persons and their environments. It is because beliefs are relational at this scale that they “do not supervene on local microstructure” (Baker, 1995, p. 63). That is, they may be constituted by particular neural circuits, but they do not supervene on those circuits, since they have to do with the relationship between those circuits as part of a person and the environment of the person in question. As Baker sees it, constitution has to do with objects and their parts, whereas supervenience has to do with the properties of those objects, which may depend on other objects: thus, “supervenience relations (among properties) diverge from constitution relations (among things)” (Baker, 1995, p. 132). So, for Baker, beliefs may be constituted by brain states, but they are not reducible to brain states. This is because the distinctive properties of beliefs depend on the relationship of those brain states with other objects and states. This is what it means for beliefs to be held by persons in particular environments. This is a naturalistic view of beliefs, but one that differs significantly from the standard view that beliefs depend only on brain states; it is an early instance of externalism, the view that mental phenomena may depend on or be constituted by events that occur outside of the brain and body. Inspired by Baker, I apply her ideas to attention and the self. That is, I find that the self is constituted by brain states but is not microphysically reducible. I argue for this position below, first confronting an opposing position on the self: that it is an illusion. Like Baker, I see my project as an important remedy for skepticism, albeit in this case for skepticism about the existence of a self, rather than beliefs. I see practical realism about the self to be more basic than practical realism about beliefs, since beliefs are held by persons or selves. Thus, if we should be practical

42  Carolyn Dicey Jennings realists about anything, it is the self. I believe this coheres with Baker’s own philosophical commitments, as I will explain below.

2 Illusion View The illusion view, as I am calling it, has to do with the concept of a self with causal power. That is, it has to do with the concept of mental causation by a self. Many have argued that this is an illusion, in part due to empirical evidence. They argue that the self is a mental construct, and that there is no reason to think of the self as something that does things. Instead, as Baker puts it, this view is that “selves are illusory products of the brain” (Baker, 2016, p. 8). Take, for instance, Wegner and Wheatley’s now classic paper on the experience of mental causation (Wegner & Wheatley, 1999). They show that when two people have control over a process, a participant can be fooled as to how extensive their own role was in that process merely by inducing a thought that aligns with the outcome of that process. In Wegner and Wheatley’s view, which is inspired by Hume, causation is inferred from the conjunction of two events that are close to one another in time, when other explanations of the “caused” event are not available. In the case of mental causation, we might infer that a thought that occurred before an action is a cause of that action if it is related to that action and other potential causes are not apparent. In their experiment on the topic, Wegner and Wheatley found that participants who were induced to have thoughts regarding a particular outcome were more likely to judge themselves as having a larger role in the process of bringing about that outcome, depending on the timing of that induction. Wegner and Wheatley think that in the normal case, we mistake our conscious thoughts as being effective, when really only unconscious mechanisms have an impact on our behavior: “The real causal mechanisms underlying behavior are never present in consciousness. Rather, the engines of causation are unconscious mechanisms of the mind” (Wegner & Wheatley, 1999, p. 490). Wegner and Wheatley are likely inspired, in part, by Libet’s research a decade earlier (Libet, 1985). Namely, Libet found a delay between unconscious mechanisms that predict our choices and our own awareness of making those choices: “The brain ‘decides’ to initiate or, at least, to prepare to initiate the act before there is any reportable subjective awareness that such a decision has taken place” (Libet, 1985, p. 536). Given the ability for others to predict our choices before we are subjectively aware of them, it seems reasonable to see unconscious mechanisms as having the real causal power here. Yet, some have argued that the apparent neural tendencies that these predictions are based on are an artefact of the data processing in this study: Libet lines up the neural oscillations prior to the act, and background neural oscillations tend to ramp up

Practical Realism about the Self  43 before an act of this kind (i.e., a “spontaneous” act that is allowed to occur only when background oscillations have ramped up in this way; cf. Schurger, et al., 2012). Further, Libet found that these tendencies can be vetoed by the participant (Libet, 1985; see also Schultze-Kraft, et al. 2016). So Libet’s finding shows that there are unconscious influences on our choices before we make them but not that our conscious experience is ineffectual. Thus, even if we align the self with consciousness, Libet’s findings do not show us that the self does not have causal power. For that we would need further evidence, such as that put forward by Wegner and Wheatley. From within philosophy, Metzinger likewise argues that the self is a construct, drawing in part on Wegner and Wheatley’s findings (Metzinger, 2004, p. 406). For Metzinger, this illusion extends to all phenomenal objects: “What you subjectively experience as an integrated object possessing a transtemporal identity (e.g., the book you are holding in your hand) is being constituted by an ongoing process” (Metzinger, 2004, p. 23). Similarly, he thinks we reify the self in our experience, seeing it as something substantive, rather than something we are in the process of creating. Yet, the illusion of self is different from that of other objects, Metzinger argues, since the self does not exist: “There is no one whose illusion the self could be, no one who is confusing herself with anything” (Metzinger, 2004, p. 634). The existence of the self is rejected by Metzinger, in part, because of the homunculus problem posed by Dennett (Metzinger, 2004, p. 307). As Schneider puts it, Dennett is clearly opposed to… theories purporting to explain cognitive capacities by generating a decomposition that ultimately boils down to an internal agent, or homunculus… the consciousness of the homunculus would itself need to be explained, so this sort of explanation is circular. (Schneider, 2007, p. 317) In other words, posing a new internal agent cannot by itself help us to explain phenomena such as conscious perception, since the problem is how any agent can consciously perceive, including an internal one. But note that this problem does not show that the self does not exist, or even that an “internal agent” does not exist, only that introducing new selves does not by itself explain other phenomena. In arguing against the existence of a self with causal power, Dennett goes further than the homunculus problem, citing evidence from split-brain studies: “A central clue comes from the sort of phenomena uncovered by Michael Gazzaniga’s research. According to Gazzaniga’s view, the mind is not beautifully unified, but rather a problematically yoked together bundle of partly autonomous systems” (Dennett, 1992,

44  Carolyn Dicey Jennings p. 111). Instead of saying that a new self has been created in the process of severing the corpus callosum in these patients, he argues that it makes more sense to see these studies as revealing that a person’s “center of narrative gravity,” or self-concept, can be split in two (see also Nagel, 1971). Dennett argues that split-brain phenomena reveal that the brain is normally disunified, and that a unified self is an illusion. Yet, these patients also retain some unity, as Dennett illustrates through pain: “pain stimuli go to both hemispheres,” so a painful stimulus is easily reported by these patients, regardless of the side of stimulation (Dennett, 1992, p. 113). In fact, more recent work has claimed that “severing the cortical connections between hemispheres splits visual perception, but does not create two independent conscious perceivers within one brain” (Pinto, et al. 2017, p. 1231).1 Thus, we might simply say that the unity of selfhood is relative and that split-brain patients have a relatively disunified self, rather than two selves, avoiding the metaphysical problem (see also Schechter, 2014). In my view, we have ample evidence for the existence of a self that is not well countered by the above. Like any substantive thing, we can be wrong about the self, mistaking something as due to the self when it is due to someone or something else, which I think is the best explanation of Wegner and Wheatley’s findings. Further, there is a way in which our understanding of the self is constructed, just as our understanding of other substantive things (e.g., this book) is constructed; this does not mean that the self is constructed by us, any more than other substantive things are constructed by us, which I think is the best way of responding to Metzinger. Finally, the self can be more or less disunified, but this does not by itself show that there is no self, against the claims of philosophers such as Dennett. We might, for instance, pin the question of whether there is a self on whether there are non-reducible causal powers attributable to the self, rather than on whether there is sufficient unity. In the next section, I will discuss the evidence I take to point to the existence of a substantive self, later comparing this to Baker’s own view of the self.

3 The Subject of Attention In line with Baker’s stance of practical realism regarding beliefs, I see a case for practical realism regarding the self. Namely, the existence of the self is assumed in the way we interact with others. Specifically, the existence of a self with causal powers is so assumed. Take, for instance, our



Practical Realism about the Self  45 hesitation at the crosswalk, even when we obviously have right of way, and even when we are sure the driver has seen us—we hesitate because we are unsure what the driver will do with this information. We treat this as up to the driver, to a certain extent. And if the driver chooses to speed on through, we might blame them for their behavior, perhaps even shout at them. This is because we think of them as responsible for this choice, and we want to have an impact on their future choices. We might think of the driver’s beliefs as having an impact on their decision, but we also think there is more to their decision, that it depends on the person, or self, in question. The everyday assumption of the existence of a self with causal powers is especially evident, I think, in the way we think and talk about attention. Attention is said to be something that can be both “paid” and “grabbed.” It is, for example, “paid” in the classroom but “grabbed” by advertising. The distinction between one’s attention being paid or grabbed depends on a self who is either doing the paying or whose attention is being grabbed. In both cases, attention is a way of bringing about behavioral change. In one case, we pay attention to change our behavior in a way that aligns with our goals; in the other case, our behavior is changed by others in order to better align with theirs. So, the assumption of a self that can pay attention is the assumption of a self with causal power. Beyond speech, our everyday practices distinguish between attention that is driven by the self and attention that is driven by external forces. When we want to grab attention, we resort to fast movements, bright colors, and loud sounds. Social media use, for example, the sudden onset of red notifications. When we want someone to pay attention, we try to be interesting to them, or to show them why they should be interested. Teachers do this, for example, by connecting classroom material with everyday life. We take something extra to be involved in the latter case due to the assumption that we are interacting with a self, who has the power to control attention. We tailor our behavior toward someone’s attention depending on whether we need to involve the self or undermine it. In this way, our everyday practices assume the existence of a self. Philosophers and scientists have long recognized that attention might be the place to look for mental causation by the self and have debated whether attention truly is evidence of a self. This comes up in a dispute between Bradley and James over one hundred years ago, and between Indian philosophers a couple millennia before that (Ganeri, 2017; Jennings, 2020; Jennings, in press). In contemporary scientific debates, mental causation by the self is replaced with “executive control,” and the question is whether there is a plausible account of attention based in executive control that is not reducible to local neural interactions. As I argue in my book, there is such an account: attention is directed by the self when the neural firing that occurs over a larger region and over a

46  Carolyn Dicey Jennings longer time period cumulates into oscillations with higher amplitudes than those of feedforward processing, such that the higher-frequency, lower-amplitude oscillations have power over the organism’s behavior only at certain intervals in the lower-frequency, higher-amplitude oscillations (i.e. in the troughs; Jennings, 2020). 2 To draw out this idea a bit more, my view is that when attention benefits from executive control this is best explained by a whole-brain phenomenon. This aspect of attention is typically picked out by feedback from the deeper layers of the prefrontal cortex, which requires a delayed response (Connor, et al. 2004; Lakatos, et al. 2005; Wilken, et al. 2009; Baluch & Itti, 2011). This delayed response allows for input from a broader spatiotemporal range of neural activity: more areas of the brain, covering a larger spatial area, and over a longer time span. Due to the scale-free relationship between neural oscillations that maintains smaller amplitudes for higher frequencies, a process that allows for input from a broader spatiotemporal range of neural activity will correspond with neural oscillations of larger amplitudes (see, e.g., Buzsáki & Draguhn, 2004). This then allows for such a process to control local, feedforward processing. That is, local, feedforward processing would not be able to have an impact on behavior unless it occurred at certain time intervals, due to the larger amplitude of the feedback oscillations (Lakatos, et al. 2008). In fact, when attention benefits from executive control it has been found to correspond with lower frequency oscillations (Buschman & Miller, 2007). The self corresponds, in this picture, to the pattern of wave activity that occurs for longer intervals and larger regions of the brain, sometimes called “global” brain activity. I see this pattern as comprised of the tendencies and interests of the organism. Insofar as this pattern is responsible for the direction of attention, I call it the “subject of attention.” I think that this understanding of the self can satisfy the assumptions of our everyday practices, described above. If the subject of attention corresponds with the whole-brain pattern of wave activity, one might wonder how it has control over local neural processing. It might seem absurd, or even paradoxical, to claim that something can have control over itself in this way. Yet, I think we can make sense of this using some conceptual tools developed by others, such as machretic determination and contextual emergence. These tools differentiate the view on offer from the circular causation of strong emergence and the causal collapse of weak emergence, as I explain below. Contextual emergence is a concept developed by Harald Atmanspacher (2007), who contrasts it with forms of emergence that rely on

2 This is closely related to Thomas Hills’ account of “neurocognitive free will,” except that his account involves a construct of self, rather than a true self (Hills, 2019).

Practical Realism about the Self  47 supervenience. Supervenience is a relationship between a whole and its parts, such that changes in the whole depend on changes in its parts. One might thus think of any causal power of the whole as being driven by the parts. Contextual emergence is a case of a whole with causal power that is not reducible to its parts, a denial of supervenience. Atmanspacher situates contextual emergence as in between a “weak” emergence view, in which a whole merely appears to have properties that go beyond those of its parts, and a “strong” emergence view, in which the parts are neither necessary nor sufficient for the whole that emerges from them. Both strong and weak emergence have been argued to have significant philosophical problems. In the case of strong emergence, the main problem is that it seems to rely on a form of causation that is circular, such that the parts bring about the whole that causes changes to its parts. In the case of weak emergence, the main problem is that any causal power is reducible to the parts, and so the whole seems metaphysically unnecessary (see the discussion on Kim, below). In contextual emergence, the parts are necessary for the whole that emerges from them, but not sufficient, since emergence only occurs in a particular context. Similarly, in my view, the self is the sort of thing that relies on context. The self is a set of interests or tendencies to seek out and respond to stimuli in a particular sort of way. Only embodied beings have interests. Thus, it makes sense to suppose the context necessary for the self to be the body. After all, the body is necessary to provide the stimuli that interests govern, and the body has limitations that make governance necessary. But I think it is this last thing, in particular, that is required for the emergence of the self: the constraints faced by a body in a world of its approximate spatiotemporal scale. Only in such a scenario would you have a self, or a set of interests with its own causal power. In this way of thinking, the parts of the self are the parts of the pattern of neural oscillations or individual neural oscillations that correspond with individual interests. So, a particular pattern of neural oscillations is necessary for the self, but that pattern is not sufficient for the self. To bring about something with its own causal power, like a self, the pattern of neural oscillations must be part of a body; the self exists only so long as there is a body with constraints that the self can serve. The self thus does not supervene on these neural oscillations but is nonetheless intimately connected to them. The self, in my view, is a solution to the problem of conflicting interests— when two or more interests, or tendencies to seek out and respond to stimuli in a particular way, come into conflict, the self is what resolves that conflict for the good of the organism (see Jennings, 2017). The self is, in this picture, the set of all interests, or tendencies to seek out and respond in a particular way to stimuli. So, the resolution of conflict occurs through a weighing of all interests or tendencies, such that one is

48  Carolyn Dicey Jennings supported at the expense of the other because it is a better fit with the rest. We can understand this process through the concept of machretic determination. Machretic determination was developed by Carl Gillett (2016) to explain how an emergent entity might control its parts in a noncausal way. Specifically, an emergent entity might determine the contribution of its parts insofar as the parts are only able to make that contribution if they are brought into a relationship that allows for a new property at the level of the emergent entity. So, the emergent entity does not cause the parts to have that relationship, but the emergent entity can control the parts by determining that they will only be part of an emergent entity if they have that relationship. Gillett encourages us to think of the directional forces of parts that must all be lined up in a particular way to get a new, larger force at the level of the whole. Thus, the parts make up that larger force only when lined up in that way, and so the whole “controls” the contributions of its parts. Applying this to the view of self on offer, I consider the self to have power insofar as it resolves the conflicts between its component interests. Those interests are thus only part of the self insofar as they “line up” with one another—the self “machretically determines” the contributions of its component interests. This need not be seen as a causal power, since the self need not have causal power over its parts to determine which parts are component parts. Yet, this control can be a way for the self to exert causal power: by controlling its parts, the self alters the landscape of neural activity, and by altering the landscape of neural activity, the self alters bodily responses to stimuli. The self thus has causal power over the body through machretic determination over its parts. (I discuss this view at greater length in Jennings, 2020.) That a self with causal power is responsible for resolving conflicting interests through attention is implied, I think, by our everyday understanding of the self and attention. In this way, my approach is one of practical realism with respect to the self. Further, it is a more basic practical realism than Baker’s, since beliefs are held by persons or selves. Yet Baker’s own view of the self was somewhat different, as I will explain in the next section.

4 Baker’s “Self” and Property Constitution In putting forward his illusion view of the self, Metzinger cites Baker as inspiration (Metzinger, 2004, p. 395). Baker had, several years prior, argued that the existence of a first-person perspective challenges reductive materialism (Baker, 1998). To answer this challenge, Metzinger provided a reductive materialist account of the first-person perspective, which Baker called “the most comprehensive theory of the first-person perspective that I know of” (Baker 2007, p. 204). Yet Baker goes on to

Practical Realism about the Self  49 argue that his illusion view is incoherent, in part because the illusion would have to be held by something, such as a subject (recall Metzinger’s claim that “there is no one whose illusion the self could be…”). Further, in her view, the “experience of being a conscious subject is evidence that I am a subject, and this evidence overwhelms any possible evidence that I may have for any scientific theory to the contrary” (Baker, 2007, p. 221). That is, much like Moore’s two hands, Moore’s certainty of which cannot be undermined by something of which he is less certain, Baker’s certainty of being a subject cannot be undermined by the commitment to reductive materialism, or other positions of which she has less certainty (Moore, 1939). Baker’s way of talking about the subject is, again, through the notion of a first-person perspective. As she sees it, a weak first-person perspective is one in which experience is perspectival, but a strong first-person perspective is one that is also conceptualized as such, and it is the latter that is essential to being a person. In Baker’s view, the first-person perspective is relational, much like beliefs, which is why it does not fit well with reductive materialism: it is “a perspective from which one thinks of oneself as an individual facing a world, as a subject distinct from everything else” (Baker, 2007, p. 203). Her view is thus also externalist, in that a first-person perspective would not exist without objects or persons apart from the subject. One distinction between Baker’s view and my own is that she does not think a first-person perspective requires a self: “the idea of a self is much richer than the idea of a first-person perspective. A self is the locus of personal integrity and coherence, but such a self is not required for a first-person perspective” (Baker, 1998, p. 342). She uses the example of someone with brain damage trying to discover who they are, which she describes as a case of first-person perspective without a “coherent and comprehensive” self. As I say above, a self might be more or less unified, and a relative lack of unity may not point to the absence of self. Further, the example in question appears to pertain to an absence of memory and one’s conception of self, and not necessarily an absence of self. In fact, Baker closes that passage by claiming that a first-person perspective is necessary, but not sufficient, for “an idea of a self.” As I said in response to the Wegner and Wheatley experiment, we should separate the self from any construct, model, or idea of the self. In my view, a construct, model, or idea of the self is based on an actual self, and it aids us in recognizing the impact of our own self on our behavior, as well as the impact of other selves on their behavior. Baker has a similar view in the vicinity regarding persons, which, for her, are necessarily embodied: “When I deny that there are selves, what I deny is that the bearers of psychological properties are proper parts of persons or animals” (Baker, 2016, p. 14). (Recall also her earlier claim: “Belief is a global state of a whole person, not of any proper part of

50  Carolyn Dicey Jennings the person.”) Given that my own, brain-based account of the self also requires constraints at the level of the body, it might be difficult to find light between our views. Yet, I essentially base the self in the brain, seeing it as constituted by patterns of brain activity, whereas Baker sees a person as constituted by the body. So, Baker and I diverge on certain points, while sharing some important commitments. The most central of those commitments is to non-reductive materialism. As Baker makes clear, she finds the existence of a first-person perspective to challenge reductive naturalism. She urges us to embrace non-reductive materialism, in its stead. She helpfully demonstrates in a book chapter on the topic how such a perspective might establish the possibility for mental causation, against Kim’s influential arguments to the contrary (Baker, 2009; Kim, 2009). Kim argues that we have to choose between a reductive physicalist account of the mind and a dualist account, since a superveniencebased account is untenable (see, e.g., Kim, 2009). He claims that a supervenience-based account is untenable because mental causation, as distinct from physical causation, is untenable. Without causal power over the physical, the supervening entity either collapses into a reductive physical one or is relegated to the role of Huxley’s whistle: an epiphenomenon, separate from the physical without influencing it (Huxley, 1874). Mental causation is untenable due to the causal closure of the physical, or the view that all physical events have a sufficient physical cause. Since all physical events have a sufficient physical cause, any additional mental cause would result in overdetermination of physical events, which Kim rules out as absurd given the required number of such events. While Kim’s argument may be used against supervenience-based accounts, I have argued so far for a different perspective on the mind. In this perspective, the emergent self does not supervene on its microphysical basis. Kim seems to have been unaware of this possibility, claiming that all accounts of emergence are forms of supervenience. For this reason, Kim’s argument fails. Baker was the first, I believe, to make this point. According to Baker, Kim’s argument depends on at least one false supposition (Baker, 2009). That is, Kim supposes that if the mental is not reducible to its microphysical basis and yet has causal power over, say, other mental events, then it must do so through downward causation. This is because the mental event in question must itself have a microphysical basis it supervenes over, so to change the mental event the mental would have to change the event’s microphysical basis, thus requiring downward causation from the mental to this microphysical basis. As I argue above, contextual emergence does not rely on supervenience, so this inference does not hold. Baker uses the language of “property constitution” to make the point: “A property’s constituter on a given occasion may be a proper part of a supervenience base for the property,

Practical Realism about the Self  51 but the constituting instance… does not suffice for the constituted instance” (Baker, 2009, p. 121). Certain circumstances are necessary to bring about the emergent property, just as a certain context is necessary in contextual emergence. Yet, as Baker also points out, this may only be a denial of local supervenience; global supervenience may yet hold. This is, in fact, what allows Baker to hold onto the causal closure of the physical. That is, it may be that all cause and effect within the physical world is contained, while the causes and effects regarding a particular object within that world are not: “The supervenience base will be very broad—too broad to be specified or to be useful in explanation—but it will be metaphysically sufficient for the constituted property instance” (Baker, 2009, p. 123). In the same vein, the subject of attention might be thought to deny local but not global supervenience, countering Kim’s argument that an emergent entity could not have causal power so long as we accept the causal closure of the physical. The subject of attention could be an emergent entity with causal power without violating that principle, in line with Baker’s reasoning. In my view, this is further support for a stance of practical realism regarding the self, or the subject of attention.

Conclusion I have argued for the existence of a self, or a brain-based subject of attention, to counter a skeptical position with regard to the self—the illusion view. Baker found herself in the vicinity of this view, denying the existence of a self while embracing the existence of persons and their first-person perspectives. Yet our views are closely aligned. We each embrace non-reductive materialism, finding persons or selves to depend on bodies in a world. For Baker, persons are constituted by bodies, whereas I see selves as constituted by brain activity. Baker nonetheless joins me in resisting those, like Metzinger and Dennett, who deny the existence of a self, calling it an illusion. Against the illusion view, I have proposed that we use Baker’s earlier work, on practical realism, to support the existence of a substantive self. While Baker used practical realism to support her account of beliefs, I see the self as more fundamental than beliefs, since beliefs are held by persons or selves. Further, the same argumentative tactics used by Baker in favor of beliefs can support the self: the existence of a self is assumed in our everyday interactions with others. Thus, we should not reject the self simply because we do not yet have a scientific understanding of the phenomenon. Instead, we should see its use in everyday life as evidence for its existence. This, together with brain evidence as to the causal power of the self over attention, should be enough to stymie such skepticism, at least for now.

52  Carolyn Dicey Jennings

References Atmanspacher, H. (2007). Contextual emergence from physics to cognitive neuroscience. Journal of Consciousness Studies, 14(1–2), 18–36. Baker, L. R. (1995). Explaining Attitudes: A Practical Approach to the Mind. Cambridge: Cambridge University Press. Baker, L. R. (1998). The first-person perspective: a test for naturalism. American Philosophical Quarterly, 35(4), 327–348. Baker, L. R. (2007). Naturalism and the first-person perspective. In Gasser, G. (Ed.), How Successful is Naturalism? (p. 203–226). Heusenstamm: Ontos-Verlag. Baker, L. R. (2009). Non-reductive materialism. In McLaughlin, B., Beckermann, A., & Walter, S. (Eds.), The Oxford Handbook of Philosophy of Mind. Oxford: Clarendon Press. Baker, L. R. (2016). Making sense of ourselves: self-narratives and personal identity. Phenomenology and the Cognitive Sciences, 15(1), 7–15. Baluch, F. & Itti, L. (2011). Mechanisms of top-down attention. Trends in Neurosciences, 34(4), 210–24. Buschman, T. J. & Miller, E. K. (2007). Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices. Science, 315(5820), 1860–2. Buzsáki, G., & Draguhn, A. (2004). Neuronal oscillations in cortical networks. Science, 304(5679), 1926–1929. Connor, C. E., Egeth, H. E., & Yantis, S. (2004). Visual attention: bottom-up versus top-down. Current Biology, 14(19), R850–2. Dennett, D. C. (1992). The self as a center of narrative gravity. In Kessel, F., Cole, P. & Johnson, D. (Eds.), Self and Consciousness: Multiple Perspectives (p. 103–115). Hillsdale, NJ: Erlbaum. Ganeri, J. (2017). Attention, Not Self. Oxford: Oxford University Press. Gillett, C. (2016). Reduction and Emergence in Science and Philosophy. Cambridge: Cambridge University Press. Hills, T. T. (2019). Neurocognitive free will. Proceedings of the Royal Society B, 286(1908), 20190510. Huxley, T. H. (1874). On the hypothesis that animals are automata, and its history. Nature 10, 362–366. Jennings, C. D. (2012). The subject of attention. Synthese, 189(3), 535–554. Jennings, C. D. (2017). I attend, therefore I am. Aeon. Retrieved from: https:// aeon.co/essays/what-is-the-self-if-not-that-which-pays-attention Jennings, C. D. (2020). The Attending Mind. Cambridge: Cambridge University Press. Jennings, C. D. (in press). Too much attention, too little self? Philosophy and Phenomenological Research. Kim, J. (2009). Mental causation. In McLaughlin, B., Beckermann, A., & Walter, S. (Eds.), The Oxford Handbook of Philosophy of Mind (p. 29–52). Clarendon Press. Lakatos, P., Shah, A. S., Knuth, K. H., Ulbert, I., Karmos, G., & Schroeder, C. E. (2005). An oscillatory hierarchy controlling neuronal excitability and stimulus processing in the auditory cortex. Journal of Neurophysiology, 94(3), 1904–1911.

Practical Realism about the Self  53 Lakatos, P., Karmos, G., Mehta, A. D., Ulbert, I., & Schroeder, C. E. (2008). Entrainment of neuronal oscillations as a mechanism of attentional selection. Science, 320(5872), 110–113. Libet, B. (1985). Unconscious cerebral initiative and the role of conscious will in voluntary action. Behavioral and Brain Sciences, 8(4), 529–539. Metzinger, T. (2004). Being No One: The Self-Model Theory of Subjectivity. Cambridge, MA: MIT Press. Moore, G. E. (1939/2008). Proof of an external world. In Sosa, E., Kim, J., Fantl, J., & McGrath, M. (Eds.) Epistemology: An Anthology (p. 26–28). Wiley-Blackwell. Nagel, T. (1971). Brain bisection and the unity of consciousness. Synthese, 22(3), 396–413. Pinto, Y., Neville, D. A., Otten, M., Corballis, P. M., Lamme, V. A., De Haan, E. H.,... & Fabri, M. (2017). Split brain: divided perception but undivided consciousness. Brain, 140(5), 1231–1237. Schechter, E. (2014). Partial Unity of Consciousness: A Preliminary Defense. In Bennett, D. & Hill, C. (Eds.), Sensory integration and the unity of consciousness (p. 347–374). Cambridge, MA: MIT Press. Schneider, S. (2007). Daniel Dennett on the nature of consciousness. In Schneider, S., & Velmans, M. (Eds.), The Blackwell Companion to Consciousness, (p. 313–324). Wiley-Blackwell. Schultze-Kraft, M., Birman, D., Rusconi, M., Allefeld, C., Görgen, K., Dähne, S.,... & Haynes, J. D. (2016). The point of no return in vetoing self-initiated movements. Proceedings of the National Academy of Sciences, 113(4), 1080–1085. Schurger, A., Sitt, J. D., & Dehaene, S. (2012). An accumulator model for spontaneous neural activity prior to self-initiated movement. Proceedings of the National Academy of Sciences, 109(42), E2904–E2913. Tovar-Moll, F., Monteiro, M., Andrade, J., Bramati, I. E., Vianna-Barbosa, R., Marins, T.,... & Moll, J. (2014). Structural and functional brain rewiring clarifies preserved interhemispheric transfer in humans born without the corpus callosum. Proceedings of the National Academy of Sciences, 111(21), 7843–7848. Wegner, D. M., & Wheatley, T. (1999). Apparent mental causation: Sources of the experience of will. American Psychologist, 54(7), 480. Wilken, P., Bayne, T. J., & Cleeremans, A. (2009). The Oxford Companion to Consciousness. Oxford: Oxford University Press.

3

Propositional Attitudes as Self-Ascriptions Angela Mendelovici

Much discussion in philosophy of mind and cognitive science in the 1980s and 1990s has centered on the question of realism about the propositional attitudes posited by commonsense psychology. On the one hand, realists such as Jerry Fodor (1987) argue that propositional attitudes— states such as a belief that grass is green, a desire to eat ice cream, and a hope that global warming does not kill us all—are physically realized in our brains. On the other hand, instrumentalists such as Daniel Dennett (1987) take them to be predictively useful posits with no more reality than centers of gravity. Lynne Rudder Baker offers a refreshingly downto-earth alternative to both positions: the propositional attitudes of commonsense psychology exist, but they need not be identified with brain states, states of immaterial souls, or anything else. They are practically indispensable, irreducible, and totally real. Although I do not entirely agree with all the details of Baker’s view, I do agree with one key idea: propositional attitudes form an integral part of our conception of ourselves and the world around us, and this is reason to accept them in some form. In this paper, I propose a “self-ascriptivist” view of the propositional attitudes posited by folk psychology, on which they exist insofar as and because we ascribe them to ourselves. The very features of the attitudes that make us want to accept them—their integral role in our self-conception—is what makes them real. I proceed as follows: Section 1 provides some background, Section 2 considers why the issue of realism about the propositional attitudes posited by folk psychology is particularly challenging, Section 3 argues for a self-ascriptivist view of occurrent propositional attitudes whose contents go beyond the contents we consciously entertain, and Section 4 extends this view to standing propositional attitudes.

1 Background Propositional attitudes are states such as believing that grass is green, desiring ice cream, and hoping that global warming does not kill us all. They are standardly thought to involve a “content” and an “attitude”

Propositional Attitudes as Self-Ascriptions  55 component. The content of an attitude is what it “says,” is “about,” is “directed at,” or, more generally, represents. For example, a belief that grass is green represents that grass is green. The attitude of a propositional attitude is the stance we take toward the content—believing, desiring, hoping, intending, fearing, etc. It is possible for two propositional attitudes to have the same content but difference attitudes (e.g., Vera might believe that Santa Claus exists while Eleni desires that Santa Claus exist), and two propositional attitudes might have the same attitude but different contents (e.g., Vera might believe that Santa Claus exists while Eleni believes that Pegasus exists). In this paper, I am concerned with the propositional attitudes that form an integral part of our commonsense conception of ourselves and others—the beliefs and desires we take ourselves and others to have and that we take to explain and predict their behavior—and set aside any propositional attitudes that do not play a role in commonsense psychology, such as inaccessible attitudes, subpersonal attitudes, and conscious attitudes with contents we cannot in any sense articulate. For convenience, I will restrict the use of the term “propositional attitudes” to apply only to the propositional attitudes of commonsense psychology. Propositional attitudes come in occurrent and standing varieties. Occurrent propositional attitudes are propositional attitudes that we count as having while they “occur” to us—a passing thought that grass is green, a sudden realization that you forgot your wallet at home, a prolonged rumination on the reality of climate change. In contrast, standing propositional attitudes are propositional attitudes that we count as having even while they do not “occur” to us, such as a belief that the Acropolis is in Athens, which you have even while you are not thinking it, a desire to invite your friend over for dinner that you count as having even when you are not thinking about your friend or dinner, and an intention to get some ice cream after work that you are not constantly running through your mind throughout the work day. A few notes: First, as I am using the term, propositional attitudes are “cognitive” as opposed to perceptual or other sensory states. This, of course, is not to deny that commonsense psychology posits perceptual or other sensory states nor that such states are representational.1 Second, as I am using the term, propositional attitudes include attitudes that might end up not having propositional contents, such as a desire for ice cream or a love of Santa Claus. 2 Third, I am using the notion of content loosely 1 Indeed, in Mendelovici (2018a, appendix D), I argue that the account of propositional attitudes I offer here can be extended to perception, and in Mendelovici (2013a and 2013b), I argue for a representationalist view of moods and emotions. 2 Whether there are such non-propositional attitudes is a matter of contention, but this does not affect my main points. See Grzankowski and Montague (2018) for a

56  Angela Mendelovici such that all representational states—states that consist in the representing of a content—need not form a natural kind.

2 The Trouble with Propositional Attitudes It has sometimes been thought that there is a live question as to whether propositional attitudes are real: propositional attitudes are part of our folk or commonsense conception of ourselves and the world. But this conception might be mistaken and so its posits might not exist. It is for this reason that much discussion on the attitudes assumed that realism about propositional attitudes requires that commonsense psychology be vindicated by more rigorous scientific investigation (see especially Fodor, 1987)—and it is precisely this assumption that Baker rejects. One might suggest that propositional attitudes are not mere posits of commonsense psychology. We are also introspectively acquainted with them: we can tell just by introspection that we have certain beliefs, desires, and other such states. If so, realism about propositional attitudes would not require any scientific vindication. Compare: we are introspectively acquainted with consciousness, so realism about consciousness needs no scientific vindication. Considering the analogous suggestion that realism about propositional attitudes can likewise be supported will not only help us see why realism about propositional attitudes is reasonably thought to depend on scientific vindication but also help us get clear on the kinds of content propositional attitudes are thought to have, which will give us a clearer idea of what exactly realism about propositional attitudes involves. The problem with the suggestion that we are introspectively acquainted with propositional attitudes is that, while I agree that we are acquainted with something, what we are acquainted with does not answer to the commonsense psychological notion of a propositional attitude. For one, we are not introspectively acquainted with standing propositional attitudes—the mental states we are acquainted with are all occurrent. But even the occurrent states we are acquainted with do not answer to the commonsense notion of a propositional attitude. The clearest way to see this is to consider the contents that we are introspectively acquainted with in having an occurrent state. For example, suppose you consciously and occurrently believe that the mental supervenes on the physical. In this case, you are arguably acquainted with something. This something might answer to the notion of content in that it is the “saying” or representing of something.3 It might include

collection of papers on the topic and Mendelovici (2018b) for my favored way of construing the debate. 3 For an argument for this claim, see Mendelovici (2018a, §7.2.1).

Propositional Attitudes as Self-Ascriptions  57 some perceptual or verbal imagery or even a gist or partial grasp of the notions involved in your belief, such as that of supervenience. Let us call this content that you are introspectively acquainted with, whatever it is, your immediate content. The problem is that your immediate content does not include the full content of the propositional attitude that is posited by commonsense psychology. For instance, it does not include your full understanding of supervenience, though it might include some gisty or partial understanding of the notion. Still, when I try to predict your behavior (say, which questions you will ask at the talk on dualism), I attribute to you a belief with the full understanding, and when I assess your beliefs as rational or not, I hold you accountable for a belief with the full understanding. If this is right, then commonsense psychology at least sometimes posits rich descriptive contents—e.g., your full understanding of supervenience, which might consist in a definition or other characterization of supervenience—that outrun the immediate contents we are introspectively acquainted with. There is a second way in which the contents posited by commonsense psychology outrun our immediate contents. Some of the contents of the attitudes posited by commonsense psychology are object-involving contents, contents that involve particular objects. But we arguably are not introspectively aware of such objects.4 For example, you might take me to believe that George is a bachelor, where George himself—the fleshand-blood person—is supposed to be a constituent of my belief’s content. Since, arguably, George is not part of whatever content I am introspectively acquainted with, this is another way in which the contents of the propositional attitudes posited by commonsense psychology go beyond our immediate contents. If we also assume—plausibly, I think—that immediate contents are narrow, in that which contents we are immediately aware of depends on our intrinsic properties, then there is a third way in which the contents of commonsense psychological propositional attitudes go beyond our immediate contents. Thought experiments such as those of Putnam (1975) and Burge (1986) are taken to be intuitively forceful, which suggests that our commonsense psychological content attributions are broad in that they are not narrow. If so, then they are not our immediate contents. If all this is right, then introspection does not provide independent access to the propositional attitudes of commonsense psychology. This makes it reasonable to assume that realism about the attitudes requires scientific vindication. From this, we can also get a better idea of what realism about the attitudes requires: we must, at a minimum, capture the

4 This is a contentious claim, which I have argued for elsewhere (Mendelovici, 2018a). Note that this claim is compatible with the claim that we are introspectively acquainted with contents with a singular form (Mendelovici, 2018b).

58  Angela Mendelovici rich descriptive, object-involving, and broad contents that commonsense psychology ascribes. Elsewhere, I have proposed an account of immediate contents in terms of phenomenal intentionality (Mendelovici, 2018a, §7.2.4). On this account, our immediate contents are phenomenal contents, contents the representation of which is nothing over above the having of certain phenomenal states. For example, the immediate content of a thought about George might include some perceptual imagery, which is nothing over and above certain perceptual phenomenal states. Similarly, the immediate content of a thought about supervenience might include a gisty or partial grasp of a definition of supervenience, which might be nothing over and above certain cognitive phenomenal states.5 For present purposes, however, we need not endorse the account of immediate contents in terms of phenomenal intentionality. Our concern is not with immediate contents but with the contents of the propositional attitudes of commonsense psychology, which, as we have seen, go beyond our immediate contents. In what follows, I will offer an account of propositional attitudes in terms of dispositions to entertain immediate contents. Immediate contents will play a role in this account, but how we ultimately account for them is left open.

3 Self-Ascriptivism about Occurrent Propositional Attitudes So far, we have seen that propositional attitudes are posited by commonsense psychology and that we are not introspectively acquainted with them (though we are introspectively acquainted with some contentful states—those of having certain immediate contents). The fact that we are not introspectively acquainted with the attitudes suggests that it is reasonable to think that realism about the attitudes must be vindicated by science. Baker staunchly denies this claim, arguing that propositional attitudes need no scientific vindication—their practical utility is vindication enough. Baker’s realism about propositional attitudes stems from a larger metaphysical perspective, which she calls “Practical Realism” (1995, 2001, 2007a). According to Practical Realism, our everyday practices provide knowledge of the world around us, allowing us to infer the real existence of items that are usefully assumed to be real by commonsense. Baker writes: “[T]here is no better mark of reality than the utility, reliability, and indispensability of the commonsense conception.” (1995, p. 228) Practical Realism finds application in the case of

Propositional Attitudes as Self-Ascriptions  59 the propositional attitudes of commonsense psychology. Not only do the attitudes play a role in our psychological explanations of cognition and behavior but, further, they permeate our entire understanding of much of the commonsense world. Baker (1995, p. 4–5) writes: The attitudes are woven into the fabric of all social, legal, political, and other institutions. Nothing would be a contract or an invitation to dinner or an election or a death sentence in the absence of beliefs, desires, and intentions. Without attribution of propositional attitudes, there would be no justifying, excusing, praising, or blaming one another.6 While I do not endorse Baker’s Practical Realism wholesale,7 I do agree that propositional attitudes form part of an integral conception of ourselves as thinking, reasoning, and epistemically assessable persons and that this gives us reason to accept their existence. The reason propositional attitude realism can be defended in this way, I want to suggest, stems from the very nature of the attitudes: they are self-ascriptions. On the view I will propose, ascribing contents to our immediate contents, our internal states, or ourselves as persons is necessary and sufficient for the aforementioned items to derivatively represent those contents—where derived representation is representation that is at least partially constituted, grounded in, or identical to other instances of representation—and the propositional attitudes of commonsense psychology are nothing over and above such self-ascriptions.8 We can divide the overall view as it applies to occurrent propositional attitudes into two claims: The first is that our immediate contents, internal states, or selves at least sometimes derivatively represent some contents. The second is that at least some of these derivatively represented contents are, ground, or constitute the occurrent propositional attitudes of commonsense psychology. I argue for the first claim in the next subsection and the second claim in the two subsections that follow. Section 5 turns to standing propositional attitudes.

6 Baker follows up on this theme in later work on the metaphysics of everyday objects. In her 2006 paper “Everyday Concepts as a Guide to Reality,” she identifies a class of everyday objects that are “intention-dependent,” dependent on commonsense psychological states.

60  Angela Mendelovici 3.1 Derived Mental Representation Here is a sketch of an argument for the first claim, the claim that our immediate contents or other internal items at least sometimes derivatively represent some contents: 1 2 3

Ascribing a content to something is metaphysically sufficient for that thing to derivatively represent that content. We at least sometimes ascribe a content to our immediate contents, internal states, or selves. Therefore, our immediate contents, internal states, or selves at least sometimes derivatively represent some contents.

Premise (1) is supported by paradigm cases of derived representation, such as those of stop signs and words. A stop sign means because we in some sense stipulate, endorse, or accept that it does. Likewise, the word “dog” means because we stipulate, endorse, or accept that it does. In short, we might say that in these cases, a vehicle of derived representation (a stop sign or a word) gets its derived content from our ascriptions, which include tacit or explicit stipulations, endorsements, acceptances, and so on. If this is right, then ascribing a content to a vehicle of representation is metaphysically sufficient for that vehicle to derivatively represent. More precisely, the following claim is true: (Ascriptivism) Necessarily, if S ascribes content C to O, O derivatively represents C (for S).9 In some cases, our ascriptions are what we might call direct: they ascribe a content that is contained within the ascription itself. For example, I might stipulate that the pen that I am holding stands for the proposition that grass is green. But in other cases, our ascriptions are indirect: they ascribe a content that is not contained within the ascription itself, e.g., by deferring to experts, the community, or even the world to specify which content is ascribed. For example, I might stipulate that the pen that I am holding stands for the tallest person within 20km of me, whoever it turns out to be, or that my word “arthritis” means whatever the experts mean by “arthritis.”10 In these cases, I have succeeded in ascribing a content to my pen, but I have not entertained that content myself. 9 I also think that ascribing a content is metaphysically necessary for derived representation—it is not possible for a vehicle to derivatively represent a content in the absence of such ascriptions. If so, then accounts of derived representation that do not appeal to ascriptions fail to deliver metaphysically sufficient conditions for derived representation. See Mendelovici (2018a, §8.2) for related arguments against views of non-phenomenal intentionality in terms of derived representation.

Propositional Attitudes as Self-Ascriptions  61 Premise (2) states that we at least sometimes ascribe a content to our immediate contents, internal states, or selves—in short, we at least sometimes self-ascribe contents. There are various ways in which we might do this. Here, I will describe one way, which I take to be fairly undemanding. A key notion in my account is that of a cashing-out thought, a thought (or other representational state) that specifies that a content is at least partly elucidated by, unpacked into, precisified as, expanded into, or more generally, cashed out into another content. For example, suppose you think to yourself that George is a bachelor. Suppose you then ask yourself what you mean by . You might have a cashing-out thought with the content . This thought specifies that one content, , cashes out into another content, . Note that I could have also broken this thought up into two separate cashing-out thoughts, each partially specifying what I take to cash out into, e.g.,