Epistemic Uses of Imagination 9780367480561, 9781032018935, 9781003041979

Table of contents :
Cover
Half Title
Series Page
Title Page
Copyright Page
Table of Contents
Introduction: The Epistemic Role of Imagination
SECTION I: Modality and Modal Knowledge
1 Why We Need Something Like Imagery
2 An Imaginative Person’s Guide to Objective Modality
3 Crossing Rivers: Imagination and Real Possibilities
4 Imagination, Metaphysical Modality, and Modal Psychology
SECTION II: Reasoning
5 Reasoning with Imagination
6 Equivalence in Imagination
7 How Imagination Can Justify
8 Imagination, Inference, and Apriority
SECTION III: Thought Experiments
9 Narratives and Thought Experiments: Restoring the Role of Imagination
10 Two Ways of Imagining Galileo’s Experiment
11 Attention to Details: Imagination, Attention, and Epistemic Significance
SECTION IV: Understanding Self and Others
12 Bridging the Divide: Imagining Across Experiential Perspectives
13 On Imagining Being Someone Else
14 “Imagine If They Did That to You!”: The Complexity of Empathy
15 Imagination, Selves, and Knowledge of Self: Pessoa’s Dreams in The Book of Disquiet
Notes on Contributors
Index
Untitled

Citation preview

Epistemic Uses of Imagination

“This is a stunning and original collection of essays on imagination. It will advance discussions in epistemology, aesthetics, metaphysics, philosophy of mind, and even philosophy of science.” – Neil Van Leeuwen, Georgia State University, USA This book explores a topic that has recently become the subject of increased philosophical interest: how can imagination be put to epistemic use? Though imagination has long been invoked in contexts of modal knowledge, in recent years philosophers have begun to explore its capacity to play an epistemic role in a variety of other contexts as well. In this collection, the contributors address an assortment of issues relating to epistemic uses of imagination, and in particular, they take up the ways in which our imaginings must be constrained so as to justify beliefs and give rise to knowledge. These constraints are explored across several different contexts in which imagination is appealed to for justifcation, namely reasoning, modality and modal knowledge, thought experiments, and knowledge of self and others. Taken as a whole, the contributions in this volume break new ground in explicating when and how imagination can be epistemically useful. Epistemic Uses of Imagination will be of interest to scholars and advanced students who are working on imagination, as well as those working more broadly in epistemology, aesthetics, and philosophy of mind. Christopher Badura is a PhD student in philosophy at the Ruhr University Bochum, Germany, working on logics of imagination. His research interest is philosophical logic and its application to philosophical issues concerning imagination. Amy Kind is Russell K. Pitzer Professor of Philosophy at Claremont McKenna College, where she also serves as Director of the Gould Center for Humanistic Studies. In addition to authoring the introductory textbooks Persons and Personal Identity and Philosophy of Mind: The Basics, she has edited Philosophy of Mind in the 20th and 21th Centuries, The Routledge Handbook of Philosophy of Imagination, and (with Peter Kung) Knowledge Through Imagination.

Routledge Studies in Contemporary Philosophy

Extimate Technology Self-Formation in a Technological World Ciano Aydin Modes of Truth The Unifed Approach to Truth, Modality, and Paradox Edited by Carlo Nicolai and Johannes Stern Practices of Reason Fusing the Inferentialist and Scientifc Image Ladislav Koreň Social Trust Edited by Kevin Vallier and Michael Weber Green Leviathan or the Poetics of Political Liberty Navigating Freedom in the Age of Climate Change and Artifcial Intelligence Mark Coeckelbergh The Social Institution of Discursive Norms Historical, Naturalistic, and Pragmatic Perspectives Edited by Leo Townsend, Preston Stovall, and Hans Bernard Schmid Epistemic Uses of Imagination Edited by Christopher Badura and Amy Kind Political Philosophy from an Intercultural Perspective Power Relations in a Global World Edited by Blanca Boteva-Richter, Sarhan Dhouib, and James Garrison For more information about this series, please visit: https://www.routledge. com/Routledge-Studies-in-Contemporary-Philosophy/book-series/SE0720

Epistemic Uses of Imagination

Edited by Christopher Badura and Amy Kind

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 Christopher Badura and Amy Kind 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 A catalog record for this title has been requested ISBN: 978-0-367-48056-1 (hbk) ISBN: 978-1-032-01893-5 (pbk) ISBN: 978-1-003-04197-9 (ebk) Typeset in Sabon by codeMantra

Contents

Introduction: The Epistemic Role of Imagination

1

C H R ISTOPH ER BA DU R A A N D A MY K I N D

SECTION I

Modality and Modal Knowledge

21

1 Why We Need Something Like Imagery

23

PET ER KU NG

2 An Imaginative Person’s Guide to Objective Modality

44

DER EK L A M

3 Crossing Rivers: Imagination and Real Possibilities

63

REBECCA HANRAHAN

4 Imagination, Metaphysical Modality, and Modal Psychology

79

M IC H A E L OMOGE

SECTION II

Reasoning

101

5 Reasoning with Imagination

103

J O S H UA M Y E R S

6 Equivalence in Imagination

122

F R A N C E S C O B E RT O

7 How Imagination Can Justify C H R ISTOPH ER BA DU R A

141

vi Contents 8 Imagination, Inference, and Apriority

160

A N TON ELL A M A LLOZZ I

SECTION III

Thought Experiments

181

9 Narratives and Thought Experiments: Restoring the Role of Imagination

183

M A RG H E R I TA A RC A N G E L I

10 Two Ways of Imagining Galileo’s Experiment

202

M A RG O T S T RO H M I N G E R

11 Attention to Details: Imagination, Attention, and Epistemic Significance

218

ER IC PET ERSON

SECTION IV

Understanding Self and Others

235

12 Bridging the Divide: Imagining Across Experiential Perspectives

237

AMY KIND

13 On Imagining Being Someone Else

260

J U L I A L A N G K AU

14 “Imagine If They Did That to You!”: The Complexity of Empathy

279

L U K E RO E L O F S

15 Imagination, Selves, and Knowledge of Self: Pessoa’s Dreams in The Book of Disquiet

298

N I C K W I LT S H E R A N D B E N C E N A N AY

Notes on Contributors Index

319 323

Introduction The Epistemic Role of Imagination Christopher Badura and Amy Kind

Oftentimes imagination is used for fanciful purposes, from games of pretense to daydreams to fantasies. Playful children imagine that the foor between the couches is covered with hot lava, angry teenagers imagine storming out of the house and escaping the tyranny of their parents, and overworked adults imagine themselves drinking mojitos on the beach as a respite from the tedium of their daily lives. But other times imagination is used for more down-to-earth purposes. It is employed in efforts to predict and explain other people’s behavior, in decision-making contexts, and planning for the future. As a lovestruck teen, one might imagine how a particular conversational salvo will go over with a prospective love interest; years later, that former teenager might draw on imagination in an effort to fgure out how best to manipulate a big couch they’ve just purchased so it will ft through the doorway to their frst apartment. This volume focuses on these latter uses of imagination, what we here refer to as epistemic uses of imagination.1 Philosophical interest in the epistemic usefulness of imagination has recently blossomed. 2 Although there are a number of philosophers who remain skeptical, the claim that imagination has an important role to play in the epistemic domain now enjoys considerable support. 3 How best to articulate that role and what explains the ability of imagination to play it remain largely open questions, however. It is these kinds of questions that are explored in this volume. We have chosen to organize the volume by epistemic context. The ffteen chapters have been divided into four sections, each of which focuses on a particular context in which imagination seems to have epistemic usefulness: modality and modal knowledge, reasoning, thought experiments, and understanding self and others. However, in this introduction, we provide a general discussion of three themes that unite the chapters across sections. The frst part focuses on the themes themselves; the second part focuses on the individual chapters in more detail and, in doing so, aims to highlight how they pick up these various themes.

2

Christopher Badura and Amy Kind

I.1 Key Themes One notion that has featured prominently in recent discussions of imagination’s epistemic usefulness is that of constraint. It is generally agreed that imagination must be in some way constrained in order to be epistemically useful. Here it may be helpful to compare perception. Perception is, by its nature, world-sensitive. It tracks changes in the world. Changes to an object that one is seeing will cause changes to one’s perceptions of it. One way to describe this world-sensitivity would be in terms of constraint: perception is, by its nature, constrained by the world. And it’s precisely in virtue of this constraint that perception is epistemically useful. In contrast, imagination is not world-sensitive. An act of imagining typically fails to track changes in the worldly objects with which it is concerned and can diverge dramatically from the actual facts about them. It’s this lack of world-sensitivity that, traditionally, led many philosophers to dismiss imagination as epistemically irrelevant.4 But many recent discussions have suggested that imaginings can be, and often are, governed by constraints – even if these constraints are not provided by the world in the same way as they are in the case of perception. It’s precisely in virtue of these constraints – some architectural and some set by the imaginer – that imaginings can be epistemically useful. 5 Imaginative projects typically start from some initial content that then unfolds to some further content. Although sometimes the unfolding process happens almost automatically, in many cases the imaginer takes active control of the process (see Langland-Hassan (2016) for discussion). In epistemic uses of imagination, constraints need to be operative at both stages of this process. The initial content must be appropriately constrained and the unfolding process also needs to be appropriately constrained. Throughout this volume, authors attend to these issues – some focusing on constraints of the frst sort, some on constraints of the second sort, and some on both. We see particular attention to the unfolding process in most of the chapters in Section II of the volume, a section dedicated to issues concerning reasoning. For example, although there is widespread agreement that unfolding must be constrained by logic in order for an imagining to be epistemically useful, there are questions about how the logic of imagination should best be specifed – questions that are addressed in the contribution by Franz Berto (Chapter 6). In contrast, there is a focus on the constraints governing the initial content in most of the chapters in Section IV of the volume – the section dedicated to issues concerning knowledge of self and others. For example, in focusing on how exactly we should understand perspective-taking, both Julia Langkau (Chapter 13) and Luke Roelofs (Chapter 14) address issues relating to how we should specify the kind of imaginative endeavor in which we are

Introduction

3

engaged. Nick Wiltsher and Bence Nanay (Chapter 15) discuss the way that one’s reality constrains what imaginative projects are available to one. In other sections of the volume, Peter Kung’s discussion of the importance of mental imagery for the epistemology of modality (Chapter 1) and Margot Strohminger’s discussion of how we can best understand Galileo’s thought experiment (Chapter 10) also take up issues related to the constraints that must be in place in setting the initial content of one’s imaginative project. The better one is at setting appropriate constraints on one’s imagination, the more likely one is to succeed in putting imagination to epistemic use. This fact brings us to a second theme that runs throughout the volume, namely that imagination is best thought of within a framework that treats it as a skill. Constraint-setting and obeying constraints are activities that one can be better or worse at, and imagining is correspondingly an activity that one can be better or worse at. This feature – that is, the fact that people differ with respect to how good they are at it – is one of the paradigmatic features of activities that are skills. Drawing upon an analysis recently offered by Amy Kind, we can say that skills are naturally thought of as activities that can be done more or less well, that are under one’s intentional control, and that can be improved via practice and training – all features that are shared by imagining (Kind, 2020a). The skill associated with epistemic uses of imagination may well be different from the skill associated with more fantastical uses of imagination. When we imagine in contexts of daydreaming or make-believe, for example, we are using imagination to move beyond the world in which we live. But when we imagine in contexts of thought experimentation, mindreading, and the epistemology of modality, we are typically trying to learn something about the world in which we live. Someone who is very skilled at imagination in the more fantastical contexts is good at unshackling their imagination and letting it roam free of reality, disconnecting from it. Someone who is very skilled at imagination in the epistemic contexts is good at tethering their imagination to reality in just the right way that they can learn from it. This again highlights the connection between the frst theme and this second one. With respect to fantastical uses of imagination, skill relies on the removal of constraints, and often these constraints are ones that other imaginers assume without question. With respect to epistemic uses of imagination, skill relies on the imposition of constraints, and often these constraints are ones that escape the attention of other imaginers. While the frst theme we discussed – that relating to constraints – is explicitly taken up in many of the chapters of the volume, this second theme is often more implicit. But there are several places where it comes to the forefront. Consider Amy Kind’s discussion (Chapter 12) about bridging epistemic divides. When it comes to the question of whether an imaginer can have epistemic access to experiential perspectives other

4

Christopher Badura and Amy Kind

than their own, the answer will depend at least in parts on the imaginer’s imaginative capabilities. Another place where this theme comes through clearly is in Derek Lam’s discussion (Chapter 2) of the consequences for the epistemology of modality that arise from people’s inability to imagine certain states of affairs and, more specifcally, the extent to which it matters how that inability is to be explained. Michael Omoge’s attempt to psychologize metaphysical modality (Chapter 4) also seems to presuppose a conception of imagination as a skill. On his view, the more knowledge one has in a specifc domain, the more disciplined and trained one’s imagination will be with respect to that domain, and this will affect the way one can draw on one’s imagination in drawing conclusions about metaphysical modal claims. But now suppose that someone is a skilled imaginer, and that they appropriately set and abide by the relevant constraints in their imaginative exercises. As almost every single contributor to this volume would agree, at least some such imaginative exercises seem to supply the imaginer with justifcation. Here Antonella Mallozzi, whose contribution (Chapter 8) focuses on Williamson’s criticism of the a priori/a posteriori distinction, is the one outlier. Though Mallozzi agrees that imagination can be epistemically helpful in our search for truth, she does not think that its epistemic role is best understood as one involving justifcation. Rather, she thinks it is best understood as a helpful but ultimately dispensable reasoning tool. But even here, in the most pessimistic contribution, the possibility is left open that in at least some contexts, there might be a more signifcant epistemic role for imagination – or at least, for some types of imagination. So even the most pessimistic among the contributors does not rule out the possibility of imagination sometimes playing a justifcatory role. And, as already noted, the other contributors are all optimists with respect to the ability of imagination to provide an imaginer with justifcation. But an important question remains. How does such justifcation work? This brings us to the third theme that runs throughout the volume, namely, an attempt to provide an explanation of what we might call imaginative justifcation. Although many of the contributions to this volume take up this theme, they do not all provide the same sort of answer. In some of the contributions – particularly those in Section II on reasoning – the kind of justifcation seems to be understood as analogous to that provided by logical reasoning. This comes up especially clearly in the chapter by Joshua Myers (Chapter 5), who argues that we can reason with imagination analogously to the way that we can reason with beliefs. But we also see it in Christopher Badura’s contribution (Chapter 7), which aims to provide a rigorous understanding of the unfolding process of imaginative episodes as well as an explanation of how imaginative justifcation is provided via this process.

Introduction

5

In other contributions, however, the kind of justifcation seems to be understood as analogous to perceptual justifcation. Consider the discussion by Rebecca Hanrahan (Chapter 3) of how we use sensory imagination to determine whether a river is crossable. As she argues, it’s in cases where my epistemic twin could take this imagining to be a veridical perception that the imagining would provide me with reason for belief. Eric Peterson also relies on an analogy to perception in explaining imagination’s epistemic signifcance (Chapter 11). As Peterson argues, just as the epistemic signifcance of perception depends crucially on what is being attended to, a proper accounting of the epistemic signifcance of imagination will also assign a crucial role to attention. On his view, we can see this particularly clearly in the case of thought experiments: the way that thought experiments direct our attention is what accounts, at least in part, for their ability to play an important epistemic role. Finally, we see yet another explanation of imaginative justifcation in the contribution by Margherita Arcangeli (Chapter 9), one that depends on seeing imagination as fundamentally a simulative activity. As we noted at the start of this introduction, the philosophical discussion of epistemic uses of imagination has at this point largely moved beyond the question of whether imagination can play an epistemic role to the question of how it can play that epistemic role. (That we have already reached this stage is itself somewhat remarkable, considering the relative youth of this debate.) In our view, as the discussion of this section suggests, any answer to this question will likely have to reckon with the three themes that we have outlined. By doing so, the chapters in this volume help address the role of imagination in epistemic contexts and thereby contribute to the development of an epistemology of imagination. Of course, it is clear that there is still considerable work to be done. The epistemology of imagination has not progressed to a stage at which it can be compared to the epistemology of perception or the epistemology of testimony. But it is our hope that this volume contributes to that progress and, in particular, that these chapters will advance our understanding of many of the critical issues surrounding the epistemic work that imagination can accomplish.

I.2 The Chapters As mentioned above, this volume is organized around four different contexts in which imagination seems to have epistemic usefulness: modality and modal knowledge, reasoning, thought experiments, and understanding self and others. We here set each chapter in its context and aim to summarize the central ideas. We also aim to show how the themes we have identifed are at work in each of these epistemic contexts.

6

Christopher Badura and Amy Kind

I.2.1 Modality and Modal Knowledge Modal claims have played a signifcant role in debates across a wide variety of philosophical subfelds, from philosophy of mind to ethics, from epistemology to metaphysics. Discussion of the nature of consciousness involves claims about the possibility of zombies; discussion of the nature of God involves claims about necessary existence; and discussion of the nature of objects involves claims about the necessity of their origins. Philosophers rarely agree on the truth-value of modal statements; even the ones about which there is wide agreement, e.g., that true identity claims hold by necessity, have been doubted; see, e.g., Priest (2016). An epistemology of modality is meant to explain when and how we can be justifed in believing modal claims and thus come to know their truth. One family of views about modal knowledge relies on an imagination-possibility link, according to which the imaginability of P provides evidence for the possibility of P.6 This is sometimes, but not always, conjoined with the claim that unimaginability of P provides evidence for the impossibility of P. Theories of this sort vary in how they specify the kind of evidence that imagination can provide, and this in turn leads to variations in their assessments of how justifed we are in believing modal claims. The contributions in this section all take up discussion of views in this family. Closely related to views that develop an epistemology of modality based on imaginability are views that develop an epistemology of modality based on conceivability, where conceivability is spelled out in a non-imagistic way. Proponents of such views typically claim that conceptual conceivability entails metaphysical possibility. But as Peter Kung argues in his contribution, in order to defend this entailment, proponents of conceivability views must rely on a highly idealized notion of conceptual conceivability. Focusing on two paradigmatic theories of this sort – those of David Chalmers and Frank Jackson (Chalmers & Jackson, 2001) and Jonathan Ichikawa and Benjamin Jarvis (Ichikawa & Jarvis, 2013) – Kung argues that the idealizations required by such theories are so demanding that the theories cannot plausibly be seen as describing our epistemology of modality. As Kung shows, the idealizations required by Chalmers and Jackson include: (1) that the conceiving agent has expert understanding of all their concepts; (2) that they have concepts that go beyond those of ordinary thinkers; (3) that they have ideal memory, infallible reasoning skills, and ideal attention; and (4) that they can perfectly evaluate scientifc theories against each other. Kung further argues that Ichikawa and Jarvis’s account likewise requires that the ideal conceiver has unlimited reasoning capacities and memory and that the ideal conceiver is able to detect any incoherencies in the set of conceived propositions. Arguing that these idealizations are more signifcant than typically realized, Kung concludes that neither of these accounts lives up

Introduction

7

to the  job  of providing an epistemology of modality for creatures like us. To achieve such an epistemology, Kung suggests that we will need to reassign an essential role to mental imagery. While Kung focuses specifcally on conceivability accounts of possibility, Derek Lam focuses more generally on accounts that are in either the conceivability or the imaginability camp (though he uses the term “imaginability,” this is meant to apply to views in both camps). In taking up these accounts, Lam aims to explain a puzzling asymmetry between the imaginability-possibility link and the unimaginability-impossibility link. Suppose, for example, that an evil scientist is tinkering with my brain in such a way that they not only cause me to imagine a winged tiger but also prevent me from being able to imagine a winged elephant. The fact that the scientist’s tinkering is what made a winged tiger imaginable doesn’t seem to block the conclusion that a winged tiger is possible. But at the same time, the fact that the scientist’s tinkering is what made a winged elephant unimaginable does seem to block the conclusion that a winged elephant is impossible. More generally, the imagination-possibility link is not easily defeated by information about our imagining’s causal origin, but the unimaginability-impossibility link is easily defeated by such information, and these facts jointly call out for explanation. Lam argues that we can best provide such an explanation by adopting a modal logic for alethic modality that assumes axiom 4 (“If possibly possibly P, then possibly P”) rather than axiom 5 (“If possibly P, then necessarily possibly P”). His argument relies heavily on Lewis’s principal principle, which states that a rational agent’s credence in a proposition should match what the agent believes to be the objective chance of that proposition. Moreover, a rational agent’s credences should satisfy the axioms of the probability calculus. Consequently, a rational agent updates their credences according to Bayes’ theorem. From axiom 4, the principal principle, and the Bayesianist assumption, it follows that imagination provides conclusive evidence for possibility if it provides any evidence for possibility in the frst place. This explains why the imagination-possibility link can’t be easily defeated. However, if one were to assume axiom 5 rather than axiom 4, it would follow that imagination provides conclusive evidence not only for possibility but for necessity. Since “necessarily P” is equivalent to “it is impossible that not-P,” this would entail that evidence for impossibility is conclusive evidence too. But this contradicts the observation that the unimaginability-impossibility link can be defeated. So, Lam suggests, to explain the asymmetry between the imaginabilitypossibility link and the unimaginability-impossibility link with respect to their defeasibility, we should give up axiom 5 and instead accept axiom 4 as one of the axioms governing modal reality. While the previous two contributions address views that aim to explain our knowledge about non-actual possibilities, Rebecca Hanrahan’s contribution is mostly concerned with modal knowledge from everyday

8

Christopher Badura and Amy Kind

life: how do we know that we can jump across a particular river on our hiking trip? Or, how do we know that this red house could be yellow? Her chapter begins with a discussion of Margot Strohminger’s account, according to which some modal knowledge is perceptual knowledge because we can train ourselves to immediately “see” that we could jump that river, or that the house could be yellow (Strohminger, 2015). On Hanrahan’s view, while Strohminger’s arguments establish that we can train ourselves to form our modal beliefs immediately, they cannot establish that the justifcation is immediate; rather, the justifcation is still mediated by a reasoning process. Hanrahan then considers a related view, one that sees modal knowledge as perceptual but locates the justifcation of our modal beliefs in evolution rather than in habits that we train ourselves to have. As she argues, while evolution might have enabled us to become good at fguring out mundane possibilities for our evolutionary kin, we are, in fact, often pretty bad at judging what is possible for us: we often overestimate our own real abilities. Having offered these arguments, Hanrahan concludes that we should adopt an account of modal knowledge that relies on imagination rather than perception. On her view, it’s only via imagination that we are given access to real possibilities, possibilities that relate us to the actual environments in which we fnd ourselves. Although there has been extensive philosophical discussion of the epistemology of modality, there has been considerably less philosophical discussion of the psychological processes at work in the formation of our modal judgments. In his contribution, Michael Omoge aims to provide an account of the psychology of how it is that we can use imagination to form metaphysical modal judgments – a process that he calls metaphysical modalizing. Building on the cognitive architecture developed by Shaun Nichols and Stephen Stich (Nichols & Stich, 2003), Omoge shows how imagination plays a crucial role in our modal psychology. He then argues that getting clearer on modal psychology has important payoff for trenchant disputes in modal metaphysics. Focusing specifcally on the debate about the possibility of zombies, Omoge attempts to locate the source of the disagreement in the scripts on which philosophers rely. This notion of script comes from seminal work by Roger Schank and Robert Abelson (Schank & Abelson, 1977); as they defne this notion, it is “a structure that describes appropriate sequences of events in a particular context” (1977, 41). We have scripts about all sorts of repeatable activities, from dining in restaurants to taking public transportation. Importantly, however, there are often states or actions that occur that affect the normal unfolding of a script, i.e., that function as interferences. Sometimes the interferences arise because of the occurrence of anomalous or unexpected events, sometimes they arise because of conficts among scripts, and sometimes they arise because of conficts with theoretical commitments. This last

Introduction

9

explanation is of particular importance in explaining modal disagreements such as the one about zombies, for on Omoge’s view there are theoretical interferences that come into play when philosophers’ zombie scripts are activated. Though philosophers on different sides of the zombie debate might be using scripts that are equally well developed, their differing theoretical commitments end up interfering in different ways with the normal script unfolding. In Omoge’s view, one does a better job of using one’s imagination in metaphysical modalizing when fewer interferences come into play. Thus, unraveling facts about these interferences can help us to better diagnose the philosophical disagreement about modal judgments. As these brief summaries suggest, all of the contributions in this section address the theme of the justifcatory force of imagination, but the contributors disagree in various ways about what that justifcatory force is. Lam and Kung both seem to have an expansive conception of this justifcatory force. On their views, imagination can give us conclusive evidence for substantial modal claims about possibility. On Hanrahan’s view, in contrast, the justifcatory force of imagination seems more limited. Though it doesn’t provide us with justifcation for interpersonal possibility statements, it can justify statements involving personal possibility, i.e., statements of the sort “P is possible for me.” Omoge’s picture is probably closer to the more expansive conception offered by Lam and Kung, but he offers reasons for caution about the justifcatory force of imagination in metaphysically modalizing. In particular, as noted above, imagination that faces less interference will have more justifcatory force than imagination that faces more interference. I.2.2 Reasoning Recall the case mentioned above of someone trying to fgure out whether a couch will ft through their apartment door. More generally, one is trying to fgure out what would be the case if something else were the case, and this kind of consideration of hypotheticals is commonplace in our everyday lives. In the couch case, as in other cases of this sort, one might try to fgure this out by way of an explicit process of reasoning – starting with certain premises (presumably involving details about the size of the couch, about angular measurements, and so on) and then reaching a conclusion. But one also might try to fgure this out by engaging in an act of imagination, i.e., by rotating the couch in one’s imagination and seeing whether it will ft. What is the relationship between these two different ways of reaching the conclusion? Do they work in tandem with one another? If so, should we see one as having priority over the other – in other words, should we best understand this as imagination playing a role in a process of reasoning, or as reasoning playing a role in a process

10

Christopher Badura and Amy Kind

of imagination? Or is this really just a single kind of process? These are some of the questions addressed by the contributors in this section. First, Joshua Myers argues that epistemically useful imagination is itself a form of reasoning. He begins by feshing out four markers of reasoning: (1) reasoning involves operations on contents, (2) justifcation by reasoning depends on the justifcatory force of the inputs, (3) justifcation by reasoning depends on how well one reasons, and (4) reasoning implicates epistemic responsibility. He then argues that when we consider paradigmatic cases of epistemic uses of imagination, the way that the imagining confers justifcation can be best understood in terms of these markers. Thus, though many philosophers have tried to model imagination’s epistemic usefulness by analogy to perception, Myers argues that this is a mistake. On his view, given that the epistemic usefulness of imagination can be explained in terms of the markers of reasoning, and given that perception does not exhibit these markers, the analogy between imagination and perception is misguided – or at least, it’s misguided with respect to an explanation of imagination’s epistemic usefulness. The next two contributions in this section – those by Franz Berto and Christopher Badura – can be best understood in line with the basic picture that we laid out earlier in our discussion of constraints, i.e., that an imaginative project begins with an initial input that is then unfolded in line with appropriate constraints. Both Berto and Badura discuss this unfolding process and, in particular, they provide logical frameworks for how it should be understood. In his contribution, Berto takes up how considerations of imaginative equivalence affect our understanding of how to logically account for the unfolding process in epistemically useful imaginings, what he calls reality-oriented mental simulations (ROMS). According to Berto, ROMS allow us to enlarge our stock of conditional beliefs. On the notion of imaginative equivalence that Berto develops, two sentences A and B are imaginatively equivalent if they play the same cognitive role for the imaginer. For example, when one understands A just in case one understands B, or when one concludes C from A just in case one concludes C from B, A and B play the same cognitive role and, correspondingly, are imaginatively equivalent. Berto then captures this idea by way of an important semantic condition: A and B are imaginatively equivalent just in case whenever one initially imagines A, this unfolds to imagining B and vice versa. Having developed this semantic condition, Berto then turns to the main task of the chapter, an attempt to show how this semantic condition plays out in the two different logics he goes on to develop. The frst logic employs a possible world semantics, while the second employs a semantics with impossible worlds. Though he determines that both semantics can handle logical equivalence, he shows that they end up doing so in subtly different ways. Though Berto does not argue for one

Introduction

11

semantics over the other, he concludes by assessing their comparative advantages and disadvantages. Like Berto, Badura focuses on the unfolding process and, in particular, on spelling it out in a way that explains when and how it can provide imaginative justifcation. Badura invokes the framework of conditional beliefs, a framework for which there is already a well-worked-out account of justifcation. By relying on conditional beliefs, Badura aims to transfer that account to imaginative justifcation. How might this transfer go? On his view, when we imagine some content A, we manifest some of our dispositions to believe C if we believed A, i.e., we manifest some of our conditional beliefs. Importantly, however, this manifestation occurs offine. In offine manifestation, the stimulus condition of believing that A does not occur; rather, there is some offine-analogue of this. Suppose I am disposed to believe that if the couch is rotated at a 30-degree angle, the upper end hits the top of the doorframe. But when I imagine that the couch is rotated at a 30-degree angle, I don’t actually come to believe that the couch is rotated at a 30-degree angle. Instead, the conditional belief is activated by an offine manifestation of that belief. Badura then considers an important complication: sometimes our initial imaginative content is one about which we do not have any conditional beliefs. When this content is unfolded, where, then, would the justifcation come from? To answer this question, Badura invokes Amy Kind’s notion of “imaginative scaffolding” (Kind, 2020b) – a process by which we can build out in imagination to new content. In the case in question, we might imaginatively scaffold to imagining a new content A by considering A′ – where A and A′ have a common topic. If we have dispositions to believe C′ if we believed A′, then in another step, one can scaffold out in imagination to a content C, where C has the same relationship to A that C′ stands to A′. As his discussion shows, the justifcatory force of an imaginative episode doesn’t depend only on the justifcatory force of one’s dispositions to believe but also on how skillful an imaginative scaffolder one is. Of all the contributors to the volume, Antonella Mallozzi is the most skeptical about the epistemic usefulness of imagination. Her challenge comes by way of criticizing Timothy Williamson’s invocation of imagination in undercutting the traditional analysis of the a priori/a posteriori distinction (Williamson, 2013). According to Williamson, imagination works by simulating certain cognitive processes offine and thereby enables us to know truths in a quasi-experiential way. Since there are typical examples of a priori and a posteriori knowledge that both rely on imagination for their justifcation, the ways we acquire knowledge in either case are similar enough to show that the traditional distinction should be dismissed. Mallozzi argues not only against Williamson’s treatment of imagination but also against the effectiveness of his examples. In particular, she claims that the examples are ad hoc and thus do not generalize. Ultimately, Mallozzi suggests that the epistemic

12 Christopher Badura and Amy Kind usefulness that Williamson attributes to imagination is better understood as deriving from inferential reasoning. Though she leaves open the possibility that imagination can be epistemically useful, she argues that the defense of that claim would require one to articulate clear criteria for when a process counts as distinctively imaginative as opposed to being broadly inferential. Our discussion thus far has already made explicit how the contributions in this section connect with the overarching themes that we’ve identifed relating to constraints and to imaginative justifcation. Our discussion of Badura’s chapter also illustrates how the notion of skill plays a key role in his argument. What’s perhaps less obvious is that the theme of skillfulness is important to several other contributions in this section. Insofar as these contributions understand imagination in the context of logical reasoning, and insofar as logical reasoning is generally treated as a skill, we can see that these contributions seem to be operating within a skills-based framework for imagination. I.2.3 Thought Experiments It is generally agreed that the way we engage with thought experiments is via imagination.7 It is also generally agreed that this role carries epistemic signifcance, i.e., that the involvement of imagination is essential to our ability to learn from thought experiments. But as is often the case, underneath general agreement we fnd both numerous points of disagreement and numerous points in need of further explication. The contributions in this section address some of these points of disagreement and aim to shed light on some of the unresolved issues. Margherita Arcangeli comes to the issue of imagination’s role in our engagement with thought experiments by way of consideration of imagination’s role in our engagement with narratives more generally. Her discussion aims specifcally to respond to a line of argument offered by Derek Matravers (Matravers, 2014). Against the common view that fctions provide us with an invitation to imagine while non-fctions provide us with an invitation to believe, Matravers argues that engagement with both fctional and non-fctional narratives involves the same cognitive processes, namely that of entertaining mental models. While conceding that Matravers may be right to claim that imagination cannot itself help us to distinguish fction from non-fction, Arcangeli disputes his claim that it does not play a key role in our engagement with narratives. Matravers’ argument proceeds largely by taking imagination to be a simulation of belief, and then showing that the simulation of belief is neither necessary nor suffcient for fctional narratives. In contrast to Matravers, Arcangeli takes imagination to be a much broader simulative process  – one that simulates mental states such as perception, desire, and emotions in addition to simulating belief. When we engage with narratives,

Introduction

13

we project ourselves into the world of the narrative and imaginatively simulate various mental states, where these imaginative simulations are phenomenologically and functionally similar to the mental states they simulate. This allows us to immerse ourselves into a narrative, be it fctional or not, and thus explains why we are sometimes surprised by a narrative, can come to believe certain things based on a narrative, or why we feel certain emotions when we engage with a narrative. As she then shows, because thought experiments have an important narrative dimension, the account she has given of the role of imagination in our engagement with narratives sheds important light on the role of imagination in our engagement with thought experiments. While Arcangeli is interested in big picture questions about the role of imagination in thought experiments, Margot Strohminger focuses on one thought experiment in particular, namely Galileo Galilei’s thought experiment about two falling stones. This scenario is put forth to challenge the Aristotelian principle that bodies of the same material fall in media at speeds directly proportional to their weight. In her chapter, Strohminger aims to show the implausibility of perhaps the most standard interpretation of this thought experiment, and she then offers an alternative interpretation in its place. The difference between these interpretations turns on two issues: frst, the conception of bodies the Aristotelian must be assumed to be operating with in order for the thought experiment to succeed, and second, the kind of imagination at work in engaging with the thought experiment. In developing her alternative interpretation, Strohminger relies on a distinction between an intuitive kind of imagination and a refective kind of imagination, where the former often involves imagistic or sensory/experiential imagination. While the standard interpretation of Galileo’s thought experiment sees the Aristotelian as engaging in two different acts of refective imagination when considering the falling stones, Strohminger’s interpretation sees the Aristotelian as relying on both intuitive and refective imagination. Moreover, as part of the employment of intuitive imagination, the Aristotelian sets aside and brackets off their belief in the principle that is being challenged. Importantly, though the discussion concentrates narrowly on a single thought experiment, it has broader implications for the role that intuitive imagination plays in our engagement with thought experiments. In his contribution, Eric Peterson aims to explain how sensory/experiential imagination comes to have the epistemic signifcance that it does. As we have noted throughout this introduction, the standard answer to this question involves the imposition of constraints. But Peterson suggests that there is more to the story that needs to be told. In particular, he believes that one crucial factor in accounting for imagination’s epistemic signifcance concerns the role that attention plays in our imaginative exercises. Peterson’s case relies on an analogy to perception. By considering

14

Christopher Badura and Amy Kind

the role that attention plays in perceptual exercises – a role that becomes especially clear when we look at cases where attention is missing, as in cases of inattentional blindness – we can then see the role that attention plays in imaginative exercises, or, more specifcally, in cases of imaginative exercises involving experiential imagination. Peterson develops this argument by using thought experiments as a case study. Focusing on the famous example of Mary in the black-and-white room, a thought experiment developed by Frank Jackson (Jackson, 1982), Peterson shows how what we imaginatively attend to, and what we do not imaginatively attend to, can make an epistemic difference. All of the contributors to this section are interested in the question of imaginative justifcation. Their contributions pick up on an issue mentioned earlier in our discussion of the contributions on modality and modal knowledge, namely that there may be reason to believe imagination’s justifcatory force owes in large part to the kind of imagination involved, and, in particular, that imagistic or experiential imagination may have special epistemic signifcance. For example, recall Arcangeli’s argument that engaging with a thought experiment typically requires us not only to simulate believing it but also to imagine, qua simulation of other mental states, certain experiential aspects. On her view, it’s because of the phenomenal and functional similarity between simulated belief/perception and actual belief/perception that our imaginative simulations have the justifcatory force that they do. These contributions also provide further evidence that imagination is a skill. To give just one example, it seems plausible that one can use one’s attention more or less skillfully, and that some people are better at focusing their attention than others. Thus, given Peterson’s treatment of the role that attention plays in imagination, it seems plausible to treat the skill of attention as an important determinant of the skill of imagination. Perhaps the best way to classify this would be to treat it as a subskill. Doing so also connects with several other possible subskills that come up in the contributions in Section III of the volume – subskills such as simulation (in Arcangeli’s chapter) and intuition (in Strohminger’s chapter). I.2.4 Understanding Self and Others Recall our discussion earlier of the role that imagination plays in modality and modal knowledge. Though much of our modal knowledge concerns the external world – whether it is possible for the couch to ft through the door, whether it is possible for philosophical zombies to exist – it is important to note that we also have modal knowledge concerning possibilities for our own selves. We could have taken up carpentry rather than becoming philosophers, and this is something that we know. We have related knowledge about possibilities for others. For example, someone might know that their best friend could have moved to the Netherlands

Introduction

15

rather than staying in California, or that they could have had biological children rather than adopting children from China. Like modal knowledge in general, our modal knowledge about self and others plausibly depends on imagination. But in addition to knowledge of these modal facts, we have a deeper kind of understanding of ourselves and of others that also depends on imagination. In considering which job we should take, we imagine different future selves, and come to know which career path makes the most sense. In trying to understand the perspective of someone who has undergone a particularly traumatic experience, we imagine ourselves in their situation in an effort to determine how they feel. In considering these issues about the role of imagination in coming to understand self and others, the contributions in this section develop different accounts of what such imagining amounts to and also what it means to engage empathetically with someone else via imagination. Can someone who hasn’t undergone a certain kind of experience understand what it’s like to have undergone such an experience? Can someone who doesn’t share a certain experiential perspective understand what it’s like to have that perspective? Though many philosophers are deeply pessimistic about our ability to cross these kinds of experiential divides in imagination, Amy Kind aims to show in her contribution that the case for pessimism is not as strong as it may appear. Her chapter centers on two arguments often put forward in support of the pessimistic view, what she calls the too big a gulf argument and the epistemic arrogance argument. In her view, neither argument succeeds in establishing the claim that experiential perspectives vastly different from our own are epistemically inaccessible to us. Her discussion underscores the difference between claiming that it is diffcult to succeed in certain imaginings and claiming that it is impossible to do so, and she also discusses ways that one’s imaginative resources may be more expansive than usually assumed – resources that range from fction and literary non-fction to really listening to what others tell us about their experiences and experiential perspectives. The next two contributions, one by Julia Langkau and one by Luke Roelofs, both explore in more detail what is involved in imaginative perspective-taking. Discussions of perspective-taking often rely on the distinction between imagining oneself in another person’s shoes (or what’s called in-their-shoes perspective-taking) and imagining being the other person (or what’s called empathetic perspective-taking). In the former kind of perspective-taking, the imaginer aims to imagine themself in the other person’s situation, i.e., to imagine the thinking to be done by themself. In the latter kind of perspective-taking, the imaginer aims to imagine being another person in a certain situation, i.e., to imagine the thinking to be done by the other person, whom they imagine themself to be. Peter Goldie, in discussing this distinction (Goldie, 2011), raises a series of worries about the viability of empathetic perspective-taking.

16

Christopher Badura and Amy Kind

In particular, because the other person may well have different psychological dispositions from those of the imaginer, and because there will often be non-rational infuences on the other person’s decision-making process or they may be confused, Goldie argues that it is conceptually impossible for an imaginer to successfully imagine being that other individual. Though the imaginer may take these things into account, in doing so they must adopt a certain empirical or theoretical stance, and that is not what the other person themselves does. In her contribution, Julia Langkau takes up this criticism and shows why it should be rejected. Though she accepts that there are challenges to successful perspective-taking, in her view any such challenges are merely epistemic or psychological in nature, not conceptual. To show this, she argues that the same kinds of psychological processes underlie both kinds of perspective-taking. In the fnal section of her chapter, she discusses the psychological challenges in more detail by way of a consideration of our engagement with fction. Interestingly, her refection on perspectivetaking in the context of fction shows that having explicit information about a subject’s psychological makeup may make empathetic perspectivetaking harder rather than easier. In their contribution, Luke Roelofs is concerned with empathetic role-reversal. Engaging in this kind of role-reversal is often encouraged in response to wrongdoing. For example, when one child A hits another child B on the head, we might suggest: “Imagine if B did that to you! How would you feel?” But there are other similar suggestions we might make: “Look at this situation from B’s perspective! Imagine how they must feel.” Or: “Imagine if you were B in that situation. How would you feel?” Roelofs argues that these suggestions don’t involve different activities but draw attention to different steps of a single complex cognitive process. In an effort to clarify these steps, Roelofs draws a distinction between what an imaginer targets by way of an act of imagination and what an imaginer mentally creates as part of an act of imagination. When an individual imagines their new couch, for example, the couch serves as their target and they form a certain sort of mental model in an effort to capture that target. This apparatus enables Roelofs to tease apart different projects that often get run together in discussions of empathetic role-reversal, and the resulting discussion thereby also sheds light on the interplay between imagining oneself and imagining others. Here Roelofs draws on the distinction between in-their-shoes imagining and empathetic imagining that we’ve just encountered in consideration of Langkau’s contribution. Like Langkau, Roelofs is critical of Goldie’s account of perspective-taking, but their criticism takes a different tack. Though Langkau’s criticism accepts the basic distinction that Goldie draws and then aims to show that he draws mistaken conclusions about the feasibility of the two kinds of perspective-taking, Roelofs wants to

Introduction

17

show that the very distinction is confused. More generally, Roelofs denies that there is a single, stable distinction to draw between the activity of imagining oneself and imagining others – rather, there are at least three orthogonal distinctions at work in these contexts. We come now to the fnal contribution in the volume: a chapter by Nick Wiltsher and Bence Nanay. While Kind, Langkau, and Roelofs are focused primarily on outward imaginings, i.e., those that target others, Wiltsher and Nanay are focused primarily on inward imaginings, i.e., those that target oneself. Their investigation proceeds by close examination of The Book of Disquiet, a work by the 20th-century Portuguese writer Fernando Pessoa that was published posthumously. Interestingly, Pessoa created numerous alternate selves throughout his career, and The  Book of Disquiet is written by one of these alternatives, Bernardo Soares. Given Pessoa’s interest in the self, it’s perhaps unsurprising that an examination of his work would yield important insight into the relationship between imagination, imagined selves, and knowledge of one’s own self. As their foil, Wiltsher and Nanay consider one popular interpretation of Pessoa’s work that sees him “as immersed in miasmatic boredom, but able to temporarily rise above it through the restorative powers of escapist imagination” (this volume, 300). Against this, Wiltsher and Nanay argue that Pessoa’s escapist imagination is unsuccessful. It is unable to offer him an enjoyable or restorative respite from his boredom. The problem, in their view, comes from Pessoa’s own self-conception. Given the properties that he considers essential to himself, he is prevented from being able to imagine himself being a different sort of self. More generally, excessive essentializing of the self leads to failures of imagination. To combat this tendency, Wiltsher and Nanay draw from Pessoa some compelling considerations about ironic detachment, and they also make some intriguing suggestions about the importance of self-love. Though all three themes that we have outlined play a role in the contributions in this section of the volume, it is the theme about the skillfulness of imagination that comes out most prominently. All of the  contributions point to ways that one can be more or less skilled at imagining something, and they also all point to ways that one can improve one’s imaginative skill. For example, both Kind and Langkau suggest that we can improve our ability to imagine others by engaging with works of fction. On Roelof’s account, how skilled one is at recreating mental states will determine how accurately one is able to imagine one’s target. Finally, following Nanay and Wiltsher’s interpretation of Pessoa, the better that we are at de-essentializing the self, and the better we become at refraining from importing our beliefs about the self into our imaginative projects, the more successful we can be at imagining being someone else.8

18

Christopher Badura and Amy Kind

I.3 Concluding Remarks In this introduction, we have focused on three themes running through the contributions to this volume – constraints, skill, and imaginative justifcation. We do not mean to suggest that these are the only themes that connect the various chapters. In fact, not only are there other themes running through the volume as whole, but there are also some additional themes that are more local to one or another of the four different sections. For example, the importance of “imaginative scaffolding” (Kind (2020b)) in epistemic uses of imagination is a general theme connecting the contributions by Badura, Langkau, and Kind and features implicitly in contributions by several others. And to give just one example of a local theme: several of the contributions in Section II (on reasoning) take up issues concerning the relationship between imaginative processes and inferential processes. Though the chapters in this volume make considerable progress in addressing the role of imagination in epistemic contexts, many questions remain in need of additional attention. To mention just a few: do all the different varieties of imagination – sensory imagination, experiential imagination, propositional imagination – succeed in providing imaginative justifcation, or is the epistemic usefulness of imagination limited to only one variety? Does the skills-based framework apply to all three of these types of imagination? If so, can they all be improved by similar means? Undoubtedly, these sorts of questions will continue to be much discussed in future, both by the philosophers featured in this volume and by the many other philosophers working on these topics. We hope and believe that the chapters here will prove important and useful in these future explorations.

Notes 1 Amy Kind and Peter Kung refer to these two uses, respectively, as the transcendent and the instructive uses of imagination (Kind & Kung, 2016b). 2 The publication of the edited collection Knowledge Through Imagination (Kind & Kung, 2016a) is both a sign of and a contributor to the increasing attention paid to the role of imagination in the epistemic domain. 3 For skepticism about imagination’s epistemic usefulness, see McGinn (2004), O’Shaughnessy (2000), and Spaulding (2016). 4 See, e.g., Sartre (1948) and Wittgenstein (1948/1980). For a detailed response to these worries, see Kind (2018). 5 For more on the different kinds of constraints governing imaginings, see Kind and Kung (2016b). 6 See Chalmers (2002), Yablo (1993), Kung (2010), and Ichikawa and Jarvis (2013). 7 See discussion in Brown and Fehige (2019). For a notable dissenting view, see Norton (2004). 8 We are grateful to Antonella Mallozzi, Frank Menetrez, and Nick Wiltsher for comments on a previous draft of this introduction.

Introduction

19

References Brown, J.R., & Fehige, Y. (2019). Thought Experiments (E.N. Zalta, Ed.). Retrieved from Stanford Encyclopedia of Philosophy: https://plato.stanford.edu/ archives/win2019/entries/thought-experiment/ Chalmers, D. (2002). “Does Conceivability Entail Possibility.” In T.S. G endler & J. Hawthorne (Eds.), Conceivability and Possibility (pp. 145–200). Oxford: Oxford University Press. Chalmers, D., & Jackson, F. (2001). “Conceptual Analysis and Reductive Explanation.” Philosophical Review, 110(3), 315–61. Goldie, P. (2011). “Anti-Empathy.” In Amy Coplan & Peter Goldie (Eds.), Empathy: Philosophical and Psychological Perspectives (pp. 302–17). Oxford: Oxford University Press. Ichikawa, J.J., & Jarvis, B.W. (2013). The Rules of Thought. Oxford: Oxford University Press. Jackson, F. (1982). “Epiphenomenal Qualia.” Philosophical Quarterly, 32(April), 127–36. Kind, A. (2018). “How Imagination Gives Rise to Knowledge.” In F. MacPherson & F. Dorsch (Eds.), Perceptual Imagination & Perceptual Memory (pp. 227–47). Oxford: Oxford University Press. Kind, A. (2020a). “The Skill of Imagination.” In E. Fridland & C. Pavese (Eds.), The Routledge Handbook of Philosophy of Skill and Expertise (pp. 335–46). London and New York: Routledge. Kind, A. (2020b). “What Imagination Teaches.” In J. Schwenkler & E. Lambert (Eds.), Becoming Someone New: Essays on Transformative Experience, Choice, and Change (pp. 133–46). Oxford: Oxford University Press. Kind, A., & Kung, P. (Eds.). (2016a). Knowledge Through Imagination. Oxford: Oxford University Press. Kind, A., & Kung, P. (2016b). “Introduction. The Puzzle of Imaginative Use.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 1–37). Oxford: Oxford University Press. Kung, P. (2010). “Imagining as a Guide to Possibility.” Philosophy and Phenomenological Research, 81(3), 620–63. Langland-Hassan, P. (2016). “On Choosing What to Imagine.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 61–84). Oxford: Oxford University Press. Matravers, D. (2014). Fiction and Narrative. Oxford: Oxford University Press. McGinn, C. (2004). Mindsight. Cambridge, MA: Harvard University Press. Nichols, S., & Stich, S. (2003). Mindreading. Oxford: Oxford University Press. Norton, J. (2004). “On Thought Experiments: Is There More to the Argument?” Philosophy of Science, 5(71), 1139–51. doi:10.1086/425238 O’Shaughnessy, B. (2000). Consciousness and the World. Oxford: Clarendon Press. Priest, G. (2016). Towards Non-Being: The Logic and Metaphysics of Intentionality (2nd ed.). Oxford: Oxford University Press. Sartre, J.-P. (1948). The Psychology of Imagination. New York: Philosophical Library. Schank, R., & Abelson, R. (1977). Scripts, Plans, Goals, and Understanding. Mahwah, NJ: L. Erlbaum Associates.

20

Christopher Badura and Amy Kind

Spaulding, S. (2016). “Imagination Through Knowledge.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 207–26). Oxford: Oxford University Press. Strohminger, M. (2015). “Perceptual Knowledge of Nonactual Possibilities.” Philosophical Perspectives, 29(1), 363–75. Williamson, T. (2013). “How Deep is the Distinction between A Priori and A Posteriori Knowledge?” In A. Casullo & J.C. Thurow (Eds.), The A Priori in Philosophy (pp. 291–312). Oxford: Oxford University Press. Wittgenstein, L. (1948/1980). Remarks on the Philosophy of Psychology (Vol. 2) (G. Von Wright & H. Nyman, Eds.) Chicago, IL: The University of Chicago Press. Yablo, S. (1993). “Is Conceivability a Guide to Possibility?” Philosophy and Phenomenological Research, 53(1), 1–42.

Section I

Modality and Modal Knowledge

1

Why We Need Something Like Imagery Peter Kung

In the epistemology of modality literature, there’s a family of views that I’ll call conceptual possibility views, which regard conceptual possibility—suitably qualifed—as a guide to metaphysical possibility. We access conceptual possibilities by conceiving them. While we can debate about the terms “conceive/conceiving/conceivable,” the rough idea is that conceiving is an exercise of rational capacities, particularly with an eye toward working out the implications of our concepts. Our concepts rule out certain possibilities; other possibilities are left open by our concepts. According to views in this family, once we put in place qualifcations to handle ways in which conceiving leads us astray, we have good reason to think that possibilities that our concepts leave open are metaphysically possible. Conceiving, suitably qualifed, is a guide to metaphysical possibility. In this chapter, I investigate how far conceivability gets us without imagery. Proponents of conceptual possibility views acknowledge that conceiving (as they understand it) sometimes can involve imagery, but they deny that it must involve imagery. Conceiving can simply be reasoning— without the aid of imagery—from proposition to proposition, or rationally working out—again, without the aid of imagery—whether a set of propositions leads to incoherence.1 I focus here on how far views in this family take us if we exclude imagery. I will not address border disputes about the relationship between imagination and conceiving. Henceforth when I use “conceiving” without qualifcation, I mean non-imagistic conceiving of the kind that interests proponents of conceptual possibility views. It’s no secret that some views in the conceptual possibility family rely on idealization. But I think the level of idealization is more extensive than has been acknowledged. In particular, some proponents in the conceptual possibility family argue that knowledge of conceptual possibilities requires exercising the same capacities that we use to form ordinary non-modal beliefs. They reason, “we trust our ordinary belief-forming capacities, even though we concede we might be making mistakes that an ideal agent would not; conceiving rests on those same fallible capacities.” I will argue this is a mistake. Conceiving requires far more than our ordinary belief-forming capacities. I tentatively suggest that the gap

24 Peter Kung between ordinary belief-forming capacities and conceiving is best flled by imagery. Here’s the plan. In Sections 1.1 and 1.2, I quickly run through some background about metaphysical possibility and sketch how conceptual possibility is supposed to guide us to metaphysical possibility. Section 1.3 covers Chalmers and Jackson’s version of a conceptual possibility view, and catalogs the extensive ways that the view depends on idealization. Section 1.4 turns to Ichikawa and Jarvis, and—in addition to idealization—pays special attention to how conceptual possibility views depend on our ability to detect incoherence in a set of propositions. I’ll argue in Sections 1.4 and 1.5 that while we are not very good at detecting this kind of incoherence by conceiving, we might be able to detect it using imagery. Hence imagery might be able to do what non-imagistic conceiving cannot do, and if it does, then imagery plays a distinctive and indispensable role in the epistemology of modality.

1.1 The Role of Metaphysical Possibility I will run through stage setting fairly quickly, since this story has been told elsewhere. 2 Why should philosophers care about metaphysical possibility? There are many reasons; here is one. 3 Philosophical thought experiments provide essential data for philosophical theorizing. Philosophers are interested in metaphysical possibility in part because knowing about the metaphysical possibilities helps us determine what is true in actuality. This works in a familiar way: Proponent holds a theory, T. T dictates that in circumstances C, proposition P is true. But lo, clever Objector comes up with a counterexample: a case in which C, yet P is false. Too bad for Proponent’s theory. We explain this back and forth between Proponent and Objector by appealing to modality. If theory T is a run-of-the-mill philosophical theory, it purports to express a necessary truth. Clever Objector’s thought experiment points to a possibility, a metaphysically possible situation, in which Proponent’s theory T delivers the wrong result. Hence theory T cannot be a necessary truth. Thus T isn’t true. Therefore, we learn about the actual from the metaphysically possible. According to this playbook, it’s important that we be in a position to form justifed beliefs about metaphysical possibilities. There is another playbook that I will not examine here—one that suggests the detour into metaphysical possibility is superfuous.4 Conceptual possibility is what matters, and not because it is a guide to metaphysical possibility. This playbook dictates that it’s conceptual possibility

Why We Need Something Like Imagery

25

that tells us, all by itself, what we need to know. The transplant case is supposed to pose a problem for consequentialism; Frankfurt cases are meant to make us question the principle of alternate possibilities.5 On this alternative playbook, it doesn’t matter whether the transplant case is metaphysically possible, or whether Frankfurt cases are metaphysically possible; the mere conceptual possibility of those cases shows there is something wrong with consequentialism and the principle of alternate possibilities, respectively. I won’t be addressing this view in this chapter.

1.2 Conceptual Possibility Here are some claims that, at one point in my life, my concepts did not rule out. i

There is a chocolate chip cookie that has the exact dimensions of the soccer pitch in Rio’s Maracanã stadium. ii Amy Kind is, unbeknownst to her, Julie Andrews’s secret love child. iii Cassius Clay defeated Muhammad Ali to make the 1960 Olympic team. iv There is a barber who shaves all those who do not shave themselves. v There is a highest prime number. The natural worry with conceptual possibility is that our concepts might fail to rule out a possibility simply because we often fail to be in the very best epistemic position. In grammar school, my concepts seemed to leave open the possibility of a highest prime because I had only a rudimentary grasp on the concept of a prime and I hadn’t seen the proof to the contrary: my conceiving was driven by ignorance. Similarly, my ignorance of an empirical fact meant I could not rule out the Kripkean a posteriori impossibility that Clay defeated Ali. But even if I am in a good epistemic position, I often fall short of the very best epistemic position. While I do have very good reason to believe (and maybe even know) that Amy Kind isn’t Julie Andrews’s daughter, I’m not absolutely certain of it, so my concepts don’t rule it out. In other cases, like the barber case, my concepts didn’t rule out a possibility because I wasn’t clever enough to spot an incoherence. Sometimes our concepts leave a matter open because (a) we are ignorant, or are in a less-than-ideal epistemic position, or because (b) we haven’t thought hard enough, or because (c) we have an imperfect grasp of our concepts. But other times our concepts leave a possibility open because (*) we do have a good grasp on our concepts, are fully informed, and have considered the matter thoroughly, and we can see there isn’t any conceptual absurdity lurking. In cases like (*), the story goes, conceiving counts as an epistemic credit and informs us about metaphysical possibility. For proponents of the conceptual possibility family of views,

26 Peter Kung the trick is to isolate an aspect of conceptual possibility that is an epistemic credit. The devil is in the details. Fortunately, there are views that delve deep into the details. Each view that we will examine has two parts: Part One, an account of conceptual possibility—what it takes to properly conceive a proposition to avoid issues like those in (a)–(c) above; and Part Two, an account linking conceptual possibility to metaphysical possibility. Our focus will be on Part One. The issue will be that these views describe conceptual possibility in idealized terms; we idealize to avoid problems like (a)–(c): the subject knows all relevant empirical facts, like Clay = Ali; the subject is smart enough to spot the incoherence in the barber case. I’ll be arguing that this idealization raises a problem that conceptual possibility views can’t solve. We will begin in Section 1.3 by working through David Chalmers and Frank Jackson’s version of the conceptual possibility view. Part Two of this view is complex and somewhat controversial. But Part One (while also somewhat controversial) is well known, worked out in painstaking detail, and explicit about how much idealization is required. In Section 1.4, we turn to Jonathan Ichikawa and Benjamin Jarvis’s version of the conceptual possibility view. They take on the burden of explaining why we should think that the kind of idealization needed to draw the in-principle connection between conceptual possibility and metaphysical possibility doesn’t render the view practically useless. Ichikawa and Jarvis argue that the way in which the rational capacities we use to discover conceptual possibilities are continuous with the rational capacities we use in ordinary reasoning about the actual world. In Section 1.5, I argue their view fails in an instructive way. Their view and others like it rely on our ability to detect incoherence that I argue we have little reason to think we do very well. I suggest that imagery might be our best hope for flling the gap.

1.3 Chalmers and Jackson In brief, here is the Chalmers and Jackson view.6 Part One: a proposition P is a conceptual possibility when it is not ruled out by a priori refection. Part Two: every conceptually possible scenario—where a scenario, roughly, is a fully described world in which P is true—is also a metaphysically possible scenario. We will spend most of our time on Part One. For the sake of completeness, let’s briefy cover Part Two. 1.3.1 Part Two: The Conceptual Possibility-Metaphysical Possibility Link The rough idea here is that an idealized subject can use a priori refection to consider a complete description of a world (not necessarily the

Why We Need Something Like Imagery

27

actual world, just a world). Chalmers and Jackson believe that idealized a priori refection will reveal every incoherence; hence any scenario that survives this idealized a priori scrutiny is a way the world might be for all the subject knows a priori. Call any such scenario a conceptual possibility, or an epistemic possibility.7 According to modal monism, which Chalmers and Jackson accept, every epistemically possible world is also a metaphysically possible world. There is one space of worlds, which can be used to explain logical, conceptual/epistemic, and metaphysical possibility. On this model, every conceptual/epistemic possibility is a centered metaphysically possible world.8 For example, allegedly there is no a priori incoherence in a complete description of the world where XYZ, not H 2O, plays the water role (fows in the lakes and streams, rains from the sky, and so on). Modal monism means that the XYZ world is metaphysically possible. Modal monism goes hand in hand with two-dimensional (2D) semantics. Roughly, on the 2D semantics picture, sentences have two meanings: the primary intension is the a priori slice of meaning, whereas the secondary includes a posteriori elements of meaning. ‘Water is H 2O’ has a false primary intension at the XYZ world (because it’s XYZ, not H 2O, that’s playing the water role), but a true secondary intension (because the secondary intension respects the a posteriori fact that in the actual world, H 2O plays the water role). Both modal monism and 2D semantics have been explored at great length.9 But our focus will be elsewhere. In the frst paragraph of this subsection, I wrote, “idealized a priori refection will reveal every incoherence.” Our focus will be on how much idealization is required, and whether we have any reason to think that non-idealized refection is good enough at revealing incoherence. 1.3.2 Part One: Conceptual Possibility and A Priori Entailment Let’s return to Part One, and consider in some detail what it takes for a scenario to be conceptually possible. For Chalmers and Jackson, the conceptual possibility thesis is a crucial piece of their a priori entailment (APE) thesis. According to APE, every non-basic fact must be derivable, a priori, from more basic facts; those more basic facts must themselves be derivable a priori from still more basic facts. The regress ends with the most basic facts, the set of facts that a priori entails all other facts. The overarching idea is that for each non-basic truth, there must be, in principle, an explanation of how the basic facts make this the non-basic fact a fact.10 This explanation must be a conceptual one, for otherwise the dependence of the non-basic facts on the basic facts is utterly mysterious. This applies not just to the sorts of “bachelor is an unmarried male” cases familiar in philosophy; it is a general claim about explanation that

28

Peter Kung

covers scientifc cases as well. If the basic facts are microphysical, and everything else supervenes on the microphysical facts, then there must be an a priori explanatory link that reveals how facts about baldness, water, and peach cobbler follow from the microphysical facts. Chalmers and Jackson are especially concerned with deriving non-basic truths from basic microphysical truths. I’ll present their view without that focus. Let’s work through an example to see how this view works. Consider an ordinary macro truth, like “that is a trashcan.” Chalmers and Jackson are quite explicit that there need not be any kind of analysis of being a trashcan in simpler terms; a fortiori, it is not the claim that there is some analysis of being a trashcan in microphysical terms.11 What does follow from APE is that, even lacking an analysis, a subject possessing a concept C should, with non-trivial empirical information, be able to identify a priori whether a particular item falls under C’s extension. Suppose in Shayan’s den there is a certain object. Given enough (non-trivial) information about the world in general and this object in particular, Shayan should be able to say, from his armchair, based solely on the (non-trivial) information he is given, whether or not this object is a trashcan.12 In particular, Shayan should give one of the following four answers to the question “Is this object a trashcan?”: (Y) (N) (V) (I)

The object is (clearly) a trashcan. The object is (clearly) not a trashcan. It is vague whether the object is a trashcan. It is (clearly) indeterminate whether the object is a trashcan.

Because vagueness or indeterminacy may be ineliminable features of our concepts (or perhaps of the objects themselves), in borderline cases, (V) or (I) may be the correct way for Shayan to respond.13 Chalmers and Jackson contend that Shayan’s answer is a priori. If S is the correct answer, and W is the empirical information we give him, then Shayan is able to affrm W ⊃ S from his armchair, without doing any empirical investigation and without relying on additional empirical knowledge. So if it is plausible that Shayan is always able to answer (Y), (N), (V), or (I), then APE is vindicated. Can Shayan do this? There is one answer that is not consistent with APE: (~K)

I do not know which of (Y), (N), (V), or (I) is correct.

If we give Shayan empirical information W, and it is true that W ⊃ S, but he does not know which answer is correct, then he does not know W ⊃ S a priori. Let us see how Chalmers and Jackson rule out (~K) as a possible answer for Shayan. Suppose we give Shayan the following empirical information about the object: it is hollow metallic cylinder 21 inches in length and 14 inches in diameter. One end of the cylinder is removed. The open

Why We Need Something Like Imagery

29

end contains several discarded crumpled papers. When asked, “Is this object a trashcan?” he may reply with (~K). Every time he says he does not know, we ask him why. There are three sorts of answers he can give to the why question. First, he can protest that we have not given him enough information. The object is not a trashcan if it weighs as much as a star, and we have not told him how much it weighs. If the object was put there for the purpose of collecting dirty laundry, it may not be a trashcan, and we haven’t told him anything about the intentions of the person who put it there. If ignorance of some important fact is the reason Shayan does not know, then we can remedy the situation by telling him the relevant fact. Include in S whatever fact Shayan claims he needs to know to give a determinate answer. We repeat this as many times as necessary: every time Shayan replies with (~K) because he is ignorant of some fact, we provide that fact. Of course, Shayan may not be able to tell us what fact he needs to know, or, at least, we do not want to require that he be aware of where, exactly, his ignorance lies. The strategy, however, remains the same. This picture of repeatedly asking Shayan why he doesn’t know is a way of making vivid the underlying point, that we can alleviate this reason— ignorance—for Shayan to give the “don’t know” answer if we pack all the facts there are to know (at a level of description) into the antecedent. If every fact (at this level of description) is in the antecedent, including the assurance, “that’s all,” then Shayan cannot say he does not know whether the object is a trashcan because he does not have enough information. Second, Shayan may answer “Don’t know” because of cognitive limitations. For instance, we’ve just packed a considerable amount of information into the antecedent description W, more than any human being could remember. Perhaps he is also not smart enough to fgure out that if the object is 21 by 14 inches, and the solar system is seven billion miles in diameter, then the object is smaller than the solar system. To avoid this reason for answering (~K), we idealize away from Shayan’s cognitive limitations; we expand his memory to allow him to entertain the huge description W, we increase his intelligence so he can make the required inferences. Third, Shayan may defer on matters involving trashcans his community. To avoid worries about division of linguistic labor, we can take idealized Shayan to be an expert possessor of all of his concepts. This turns out to be a fairly dramatic idealization, as discussion of a worry raised by Block and Stalnaker (1999)14 will make plain. Block and Stalnaker worry that Shayan may not know which of (Y), (N), (V), and (I) is correct because he does not know the verdict of our best scientifc theories on key issues. For example, the object is not a trashcan if it is a gas, and he cannot know a priori what our best scientifc theories say about gases, because the results of science are a posteriori. When Shayan

30

Peter Kung

considers the antecedent of one of these conditionals, won’t he need to defer to what our best science would tell us in such a situation? Bear in mind that sometimes the consequent of the conditional Shayan is considering will touch on an area where science is relevant; the consequent might be something like the window glass is a supercooled liquid, not a solid. Before Shayan asserts that such consequents are true, he needs to know the verdict of the best science. We idealize away such worries by either noting that the antecedent will include metalinguistic information about how a term is used by members of the community, including scientists, and/or allowing idealized Shayan by himself to conduct the best scientifc theorizing about information presented in the antecedent.15 To sum up, the idealized thinker is an expert possessor of all his concepts; has additional concepts that we ordinary thinkers do not (how many more is not clear); has ideal powers of memory, attention, and reasoning; has access to the complete and accurate underlying description of the world; is an ideal evaluator of scientifc theories for the virtues of simplicity, elegance, explanatory completeness, and so on. Chalmers and Jackson rhetorically ask whether, given the idealizations allowed and all the information we have packed into description W, Shayan could still reasonably respond to the question “Is the object a trashcan?” with “I don’t know.” Bearing in mind that “it is vague” and “it is indeterminate” count as acceptable answers, what, wonder Chalmers and Jackson, could possibly keep Shayan from knowing a priori whether or not his concept applies in the described situation? It cannot be lack of empirical facts, because we have included all such facts in the antecedent. It cannot be because he is cognitively incapable in some way; we’ve idealized away all such limitations. No one is in a better position to know how the concept applies, and unless some further argument is supplied, why should we accept that one of (Y), (N), (V), or (I) is correct yet unknowable a priori? To deny that Shayan is able to affrm a priori W ⊃ S seems tantamount to claiming that although S is the correct answer, it is impossible to know this. If Shayan does not know it, who could? In short, any limitation that would prevent giving a defnitive a priori answer to the material conditional “if W is actual, then S” is idealized away. Chalmers (2012, ch.3) introduces a tool to help the thinker marshal all the information packing into the antecedent W: the Cosmoscope.16 A Cosmoscope is a device that stores all the information in [the antecedent] and makes it usable. In particular it contains (i) a supercomputer to store the information and to perform any necessary calculations; (ii) tools that use [the antecedent] to zoom in on arbitrary regions of the world, and to deliver information about the distribution of matter in those regions; (iii) a virtual reality device to produce direct knowledge of any phenomenal states described in

Why We Need Something Like Imagery

31

[the antecedent]; (iv) a ‘you are here’ marker to convey [indexical] information in [the antecedent]; and (v) simulation devices that deliver information about counterfactuals, exhibiting the physical and phenomenal states that will be produced under various counterfactual circumstances specifed in [the antecedent]. (p. 114) The Cosmoscope is intended to make vivid that there isn’t any relevant question that an idealized thinker like Shayan could not answer. Why is it necessary? Because the information packed into the antecedent W is so vast, the task of making sense of that information is monumental, even for an idealized cognizer. The Cosmoscope helps make plausible how an idealized cognizer could process all the information in the antecedent. Note that the Cosmoscope combines two kinds of cognizing that in the introduction I said I wanted to keep distinct. Cosmoscope items (i) and (ii) are most easily understood as facilitating non-imagistic reasoning; whereas (iii) and (v) are like VR devices that augment or replace the mental imagery. I think it is revealing that the Cosmoscope affords a prominent role to VR and simulation. The distinction between nonimagistic reasoning and imagistic imagining will be critical in Section 1.5. An obvious concern is whether the Chalmers and Jackson picture of conceptual possibility is of any use to non-ideal cognizers. As we have repeatedly seen, the amount of information packed into the antecedent of “if W is actual, then S” is vast. How can a non-ideal cognizer determine whether the vast amount of information in W contains some incoherence? And of course actual non-ideal cognizers don’t have anything like the Cosmoscope to help them. Chalmers and Jackson do not have as much to say about this as you might expect. Chalmers (2002) asserts that, with more limited antecedents falling far short of a complete description of a world, examples of hidden incoherence are hard to fnd. In (Chalmers, 2012), he remarks: Still, my own view is that cases where ideal reasoning makes a difference of this magnitude are relatively rare. In most cases, if one has a good human reasoner making inferences from partially specifed scenarios and the like, their inferences will be good ones and their results will not differ greatly from those of ideal reasoning. This is brought out by the fact that we are reasonably good at determining the truth of sentences given information about the external world. (pp. 70–1) Ichikawa and Jarvis are a bit more explicit. They purport to have an explanation as to how we know whether the antecedent W is coherent. Let’s run through their account, and then explain their view about how we can know that a description in the antecedent is coherent.

32

Peter Kung

1.4 Ichikawa and Jarvis Ichikawa and Jarvis focus on “the rational imagination.”17 We can quibble about whether the capacity they have in mind deserves the label “imagination.” For our purposes, they seem to have in mind something similar to Chalmers and Jackson’s conceivability.18 For the sake of consistency with the above discussion of Chalmers and Jackson, I will continue to use the word “conceive.” Ichikawa and Jarvis draw attention to our rational capacity to draw inferences about things we believe, and argue that same capacity is at work when we conceive.19 We conceive a set of propositions from which we draw inferences, in the same way that we draw inferences from the things we believe. Once they isolate the notion of “inference in imagination,” Ichikawa and Jarvis use that notion to build a notion of conceptual entailment. Similar to Chalmers and Jackson, they argue that if you suitably constrain conceptual entailments, they will reveal metaphysical possibilities. While Ichikawa and Jarvis’s general approach to epistemology of modality is similar to Chalmers and Jackson, there are important differences. Ichikawa and Jarvis are not as concerned whether the inferences in question are a priori. And they admit that we can conceive the impossible. Their strategy is to articulate a notion of conceptual entailment, defne conceptual possibility in terms of conceptual entailment, and then develop a formula for when conceptual possibility entails metaphysical possibility. As before, we will principally be concerned with Part One of their account, about what it means for a proposition to be conceptually possible. 1.4.1 Conceptual Entailment Ichikawa and Jarvis start with our capacity to draw inferences in actuality. We infer from It is raining to The ground is wet, or from Lots of people around here are sheltering in place to The store is out of toilet paper. We use that same capacity to draw inferences when we conceive. The inferences we would make in conceiving parallel inferences we would make in actuality. Bernie Sanders is a human being, so I infer that he has a belly button and one of the usual blood types. Ichikawa and Jarvis believe that we make such inferences in interpreting fction. When I read The Prisoner of Azkaban, I conceive that Harry Potter is a human being (even if a special one!), so I would infer (if the issue came up) in my conceiving that Harry has a belly button and one of the usual blood types. Sometimes learning more information should block an inference. The new information defeats the inference. If Rowling had written that wizard’s blood is special, and wizards don’t have one of the usual human blood types, then that should block me from inferring from “Harry is a human being” to “Harry has one of the usual blood types.”

Why We Need Something Like Imagery

33

Like Chalmers and Jackson, Ichikawa and Jarvis take special note of a way that inferences can also be defeated: a subject’s rational limitations, or evidence of a subject’s current rational limitations. Rational limitations are of the same sort we identifed above in reviewing Chalmers and Jackson’s view: limitations concerning the processing of evidence, and might include things like limited computational capacity, limited memory, performance errors, limitations in conceptual repertoire, and so on. Conceptual entailment is defned as rational commitment to infer that can be defeated only by a subject’s rational limitations, or evidence regarding a subject’s current rational limitations. Ichikawa and Jarvis defne propositions “that do not conceptually entail absurdities […] [as] conceptually possible propositions” (p. 121). If you can coherently conceive a scenario—the scenario does not conceptually entail an absurdity—then the scenario is conceptually possible. 1.4.2 Conceptual Possibility to Metaphysical Possibility To get from conceptual possibility to metaphysical possibility, Ichikawa and Jarvis have to deal with a posteriori necessities. It is conceptually possible that Hesperus is closer to Earth than Phosphorus; it is conceptually possible that my dad is Bruce Lee. Neither of those propositions conceptually entails an absurdity. But they agree with Kripke that neither of these propositions is metaphysically possible. Their solution is to notice that Kripke-Putnam a posteriori necessity cases rest on tacit background assumptions about the actual world. In their view, when we imagine that water is XYZ, it conceptually entails that In the actual world, samples of water we interact with are not composed of H 2O. The trick, then, is to exclude cases where conceiving a proposition conceptually entails something false about the actual world. Let’s run through an example, to see how this solution works. Is it metaphysically possible for me, Peter Kung, to be the son of Bruce Lee? Kripke says no. But it certainly seems conceptually possible that I am. We can imagine that my parents are not Marcia Smiley and Edward Kung, but instead Marcia Smiley and Bruce Lee. According to the doctrine of the necessity of origins, Bruce Lee being Peter Kung’s father in conception conceptually entails that Bruce Lee is Peter Kung’s father in actuality. And my father is not Bruce Lee; he is Edward Kung. So even though we can conceive that Bruce Lee is my dad, so conceiving conceptually entails something false about the actual world. And Ichikawa and Jarvis rule cases like that out. Putting it all together, when you coherently conceive a proposition P, and conceiving that proposition does not conceptually entail that some Q is true in the actual world, but Q is actually false in the actual world, then P is metaphysically possible.

34 Peter Kung 1.4.3 Detecting Incoherence We fnally come to the crux of the matter, as far as this chapter is concerned. Ichikawa and Jarvis’s notion of conceptual possibility is constructed around an ideal reasoner, similar to Chalmers and Jackson’s. Ichikawa and Jarvis take on the burden of arguing that even though we are not ideal reasoners, our beliefs about conceptual possibility can nevertheless be justifed. They explain that ordinary human reasoning is similar enough to ideal reasoning. 20 Let’s take a closer look at this explanation. Ichikawa and Jarvis acknowledge that frequently we non-ideal reasoners try to establish the conceptual possibility of a proposition by conceiving a scenario in which the proposition is true. Because we are not ideal reasoners, when we construct these scenarios in conception, there is the danger that we may construct a scenario that is incoherent. They refer to this as the danger of overstipulation. Here is their reply to that danger: We can avoid overstipulation by being sensitive to potential conceptual entailment relations. If we take care to stipulate only matters conceptually independent from what is already stipulated, then our mental facsimile will represent coherent scenarios. Taking such care is not unduly diffcult, nor does it require antecedent knowledge of what exactly is conceptually impossible and what isn’t. One can take care not to overstipulate a represented person’s gender when one has already stipulated he’s a bachelor just by being aware that there is some conceptual link between something having to do with gender and bachelorhood. One can take care not to overstipulate that a represented object is orange when one has already stipulated it is green just by being aware that there are conceptual necessities concerning color exclusion. This sort of care can plausibly be taken a priori. Being sensitive to which features of a scenario are conceptually independent so as to restrict stipulation to those features that are thus independent is the sort of ability we have in virtue of rational competences. This is not to say, of course, that we’re infallible at avoiding pitfalls. Sometimes, we overstipulate without realizing it. Accidental overstipulation, however, is the exception rather than the rule. More to the point, it can be recognized and abandoned by a priori means; we can derive a reductio by inferring in accordance with conceptual entailments. (Fleshing a scenario out even further can help us do so.) (Ichikawa & Jarvis, 2012, p. 152)21 Why do they think that overstipulation is the exception, rather than the rule?

Why We Need Something Like Imagery

35

We do not expect completely to mollify the skeptic as to how knowledge of conceptual possibility is possible. But whatever diffculties remain, discovering that [conceivings] are coherent can’t be any more diffcult than discovering that beliefs are coherent, and most of us are pretty confdent that most of the beliefs that we have, even if they are false, are not incoherent. (Ichikawa & Jarvis, 2012, p. 153) If this thought were right, it would be an important way to defend the whole family of conceptual possibility views. When we conceive a scenario, and then deem it to be coherent, we are using the very same capacity that we use to judge ordinary sets of beliefs to be coherent. Denying that we can detect incoherence in our conceived scenarios would mean denying that we can detect incoherence in our beliefs, and that would to put us several miles down the road toward an intolerable skepticism about our actual world beliefs. Our justifcation for believing that scenarios are conceptually possible stands and falls with our justifcation for our actual world beliefs. This thought is reassuring. But I think it is wrong in an important way. We are pretty confdent that our beliefs are not incoherent, and we are often pretty confdent that a subset of our beliefs are not incoherent. However, that is not because we are particularly good at recognizing the kind of incoherence that is most relevant to conceptual possibility views. The reason that we can be justifably confdent that our beliefs are not incoherent in the relevant way is that we are justifably confdent that each of our beliefs is true. 22 We know that the actual world is consistent. 23 The actual world does not contain contradictions. (That’s not to say the actual world can’t be perplexing!) Take a set of non-modal beliefs that we have about the world. As long as we are justifably confdent that each member of the set is true, we will be justifably confdent that the whole set is coherent, that it doesn’t contain hidden contradictions or hidden absurdities. Our justifcation for thinking that our beliefs are coherent in this sense can come mostly from our individual justifcations for thinking that each of our beliefs is true. I’ve just said our justifcation for coherence can come mostly from our justifcation for individual beliefs. Are there reasons to think that this is how it actually is? I think so. I think we learn this lesson from a natural extension of the preface paradox. The preface paradox gets its name from expressions of fallibility that a humble author might include in the preface of her book: “Undoubtedly I’ve made some mistakes in this book.” The author believes each assertion in her book. By agglomeration, she should believe the conjunction of all the assertions in her book. Yet she also believes that she is fallible, meaning she accepts that it is likely there is at least one false claim in the book. So she believes the

36

Peter Kung

conjunction of all the assertions in her book is likely false. This is the preface paradox: the author is apparently committed to a contradiction. Now let’s fll in more of the details. The self-acknowledged fallible author has little idea which assertions in her book are false—otherwise she’d remove them from the book. Books are not collections of unrelated assertions. The assertions in her book form a web with all sorts of the usual coherence relationships between them. There will be conceptual entailments (in Ichikawa and Jarvis’s sense), relationships that refect empirical generalizations, relationships of evidential and explanatory support, and so on. The author must be positioned to recognize some coherence relations, in order to recognize those supporting relationships between the various propositions that she asserts in her book. Recognizing those sorts of coherence relationships among a set of propositions is often regarded as providing additional reason to think some of the propositions are true. Suppose the inspector antecedently believes the butler did it. Then the inspector comes to believe a few other things. The butler was absent during the cheese course. There was poison on the victim’s spoon. The butler was seen taking the cutlery out of the pantry. Empty poison vials were found on the ground outside the butler’s bedroom window. All that coheres very well with the butler did it! That the set of propositions the inspector already believes forms a coherent story is an additional reason to believe the butler did it. But whatever additional justifcation comes from a judgment of coherence rests on the inspector having justifcation for each of the beliefs individually. A coherent story that the disgruntled son-in-law did it, absent justifcation for believing any of the individual propositions that compose the story, does not increase our justifcation for believing that the son-in-law did it. That would be rank conspiracy theorizing. With actual world beliefs, our judgment of coherence goes hand in hand with our justifcation for each individual belief. In “Can coherence generate warrant ex nihilo?” James Van Cleve (2011) runs through a mathematical analysis of this point using jurors as an example. Does the fact that all the jurors agree on a guilty verdict provide evidence for guilt? The answer is no, unless the jurors are independently reliable at individually determining guilt or innocence. Van Cleve believes this point generalizes. Coherence by itself does not generate warrant. I think that our capacity to detect coherence is designed to detect explanatory and narrative support relations: do the propositions ft together to form a coherent narrative or explanation? We can detect incoherence when it comes in the form of a proposition that clearly doesn’t ft well, either explanatorily or narratively. The butler loved the victim. Why would he want her dead? But that is quite different from a capacity to ferret out any and every incoherence (an absurdity, in Ichikawa and Jarvis’s terms). If it doesn’t matter to the explanation of the murder, do we think that we have the capacity to detect minor tensions in, e.g., the

Why We Need Something Like Imagery

37

way the butler dresses versus the way the frst footman dresses? With our actual world beliefs, we don’t need that latter capacity, because we have independent justifcation for each belief individually. As for narrative fction, that kind of attention to details seems to miss the narrative point. If the major plot lines of the Harry Potter series are narratively satisfying and compelling, does it really matter that whether the narrative is consistent in who exits a tunnel in what order?24 You need legions of obsessive fans to uncover these minor tensions and minor inconsistencies. 25 That seems to me to be the lesson of the preface paradox. In the preface the author is tacitly conceding that she is not good at ferreting out any and every incoherence. She is justifably confdent that each assertion is true, and she is justifably confdent that there is no major explanatory or narrative incoherence. But there are, she recognizes, probably some unrecognized tensions among the book’s assertions. Using her ordinary capacity to detect coherence and incoherence, she just has no idea where those tensions are. The upshot here is that there is a crucial difference between our ordinary actual world beliefs and propositions in a conceived scenario. With actual world beliefs, we can be justifably confdent that the set of beliefs contain no hidden incoherences because we can be justifably confdent that each belief is true. We have no such assurance for the propositions in a conceived scenario. With a conceived scenario, we rely exclusively on our ordinary capacity to detect coherence and incoherence. That capacity, I suggest, is not up to the job that Ichikawa and Jarvis are assigning to it.

1.5 Imagery to the Rescue? While I didn’t realize it at the time, I’ve made basically this point in other work. Thought experiments are usually coherent in the frst sense I described above. They “hang together” in a way that is explanatorily and narratively satisfying. They make sense as stories. But it’s much less obvious that they are coherent in the second sense I described above: they may contain hidden inconsistencies or hidden incoherences that ought to make us wonder whether the scenario is a genuine metaphysical possibility. For example, ethics thought experiments frequently feature forced choices with fxed outcomes. Well-written ethical thought experiments are narratively compelling; they present the moral issue at play starkly. But it is a lot less clear—or so I argue—that forced choices with fxed outcomes, cases that are designed to present moral challenges without risk, are genuine metaphysical possibilities. 26 It’s here that imagery can step into the breach. The idea is that imagery does “at a glance” what conceiving, and thinking through the implications of our conceivings, does only with diffculty. Thanks to Neil Van Leeuwen for pointing out that Aristotle puts this point quite succinctly.

38

Peter Kung In constructing the plot and working it out with the proper diction, the poet should place the scene, as far as possible, before his eyes. In this way, seeing everything with the utmost vividness, as if he were a spectator of the action, he will discover what is in keeping with it, and be most unlikely to overlook inconsistencies. (Poetics, Part 17)

Mental images are not the world and so they are not guaranteed to be consistent. But I believe that they are similar enough to the way that we perceive the world that it is reasonable to think that, enough of the time, they contain no hidden absurdities. 27 Our capacity for imagistic imagining can play the indispensable role that conceiving cannot. It is now widely accepted in the imagination literature that imagination must be constrained for it to play an epistemic role. When imagination runs free, our power to imagine is virtually limitless. We can imagine unicorns, talking teapots, sentient numbers, and so on. To put imagination to serious epistemic work, this power needs to be constrained, and constrained in a way that aligns with the epistemic purpose. 28 Imagery provides one type of constraint, and it is a constraint that is tailor-made to provide justifcation that we highlighted the need for above, justifcation for thinking that there are no hidden inconsistencies. In the visual case, for example, imagery “lays out” for inspection one way that space could be flled. It’s a constraint that no space can be both orange and green at the same time, or that an object has to have (at least fuzzy) boundaries. It’s a constraint that certain shapes are symmetrical. Our visual imagery tracks and respects these constraints, at least to some extent. I think it is no accident that compelling examples of people like Temple Grandin and Nicola Tesla putting imagination to use involve particularly vivid imagery. 29 This idea, that imagination operates under constraints, isn’t new. Hume makes this point, as this explanation of his Copy Principle makes plain. Although nothing seems freer than the power of thought, which isn’t “restrained within the limits of nature and reality” (EHU 2.4/18), Hume insists that our imagination is in fact “confned within very narrow limits.” We can separate and combine our ideas in new and even bizarre ways, imagining creatures we’ve never seen or faraway galaxies, but all the materials of thinking are ultimately derived from our impressions. Since “all our ideas or more feeble perceptions are copies of our impressions or more lively ones”; we are restricted to “compounding, transporting, augmenting, or diminishing the materials afforded us by the senses and experience” (EHU 2.5/19). (Morris & Brown, 2020)

Why We Need Something Like Imagery

39

This Humean insight is the engine that drives the combinatorial views of metaphysical possibility like Armstrong (1989). But it’s important to notice that Humean imagination adds constraints beyond the Copy Principle. Here are three Humean ideas: the idea of a trumpet playing middle C, the idea of an oboe playing middle C, and the idea of the aroma of freshly brewed espresso. Even though these are distinct ideas, we cannot imagine all the permutations in the same way: a person with a well-developed musical imagination can imagine the combined sound of the trumpet and the oboe. But imagination produces no similar musical combination of trumpet and espresso aroma. 30 That is a constraint in the way that imagination works. One promising approach to spelling out these constraints begins with the similarity between imagery and perceptual experience. Magdalena Balcerak Jackson (2018) argues for one version of this approach. She focuses on imagination as recreative simulation: imagistic imagining of the situation is perspectival, and because of the similarities between imagistic imagination and perception, it justifes a belief about the way that things could look. From there, she thinks it is plausible that that in turn justifes a belief about the way things could veridically look. And that leads to a justifed belief in metaphysical possibility. Another version of this approach has been developed quite elegantly in detail by Peter Langland-Hassan (2016), who argues that imagination borrows pieces of our perceptual cognitive architecture that have been trained by past perceptual experience, and that because those pieces constrain how imagination unfolds, imagination can inherit the epistemic credentials of past perceptual experience. Another perhaps complementary position taken by Jonathan Ichikawa (2016) and Tim Williamson (2016) is that our imaginative faculties evolved to inform us about metaphysical possibilities. My own view is that the route is more direct than those two views suggest. The thought would be that perceptual experience, even when misleading, presents us in a direct and immediate way with a perspective on the world that is at least metaphysically possible. There’s something about the nature of imagistic content, whether perceptual or imaginative, that directly justifes belief in possibility. 31 Regardless of which approach we adopt, imagistic imagination promises to supply what conceptual possibility views need but cannot supply (without overreliance on idealization): a way to guarantee that a scenario is free from hidden contradictions. It guarantees more than narrative and explanatory coherence; it plays the role that the truth of beliefs plays for inferences from beliefs. The ability to picture (in the visual case) allows us to stage for ourselves a scenario, and inspect it in a way that, as Aristotle says, makes us “most unlikely to overlook inconsistencies.” The challenge for views that start at this point, noticing that imagistic imagination has constraints (so that not just anything can be imagistically imagined) and that those constraints align with metaphysical

40

Peter Kung

possibility, is how far that alignment stretches. The alignment is at its most plausible when we confne ourselves to narrow range of perceptual cases. Yes, it is metaphysically possible to have a sound that combines trumpet and oboe; yes, it is metaphysically possible that a cube could be orange on one side and green on the opposite side; yes, it is metaphysically impossible to construct a triangle out of non-overlapping squares. One’s reason for thinking so is that one can imagistically imagine the frst two, and that one cannot imagistically imagine the third. But does that really help when we use imagistic imagination to imagine much richer and more complex scenarios, like the outcomes of elections or split hemisphere brain transplants? That is a lot less obvious. Explaining why the epistemic credential of imagistic imagination extend from the narrow toy cases to rich, real-life cases that can do more philosophical work is, in my view, the principal challenge for champions of imagistic imagination. In my work, I’ve emphasized the way the non-imagistic elements of imagination lead to imagining metaphysical impossibilities, and tried to construct a recipe for how to incorporate non-imagistic elements into an imagistic imagining in a way that allows us to recognize when we are imagining impossibilities. But there is more work to be done, to extend accounts like those of Kind, Balcerak Jackson, and Langland-Hassan. I foresee that this is where the action will and should be in future discussions about imagistic imagination. There will be more of an emphasis on the role of imagery in constructing models.32 I suspect that some of this discussion will mirror discussions in philosophy of science about the role of pictures and diagrams in science. In that debate, the issue was whether pictures and diagrams were epistemically reducible to propositional statements, or whether they have some important, epistemically distinctive role to play. 33 In our case, it will be to explain in detail how our capacity to work with imagery lends new epistemic credit to models that we construct. Regardless, there is a critical and indispensable epistemic role for imagery to play, a role that cannot be flled by our ordinary non-imagistic rational capacities. 34

Notes 1 2 3 4 5

This is similar to Balcerak Jackson’s (2016) notion of supposing. See, e.g., Gendler and Hawthorne (2002) and Kung (2016a). For other reasons, see Kind and Kung (2016) and Mallozzi (2019). Leon and Tognazzini (2010). For transplant, see Thomson (1976); Frankfurt cases appear in Frankfurt (1969). 6 See Chalmers and Jackson (2001), Jackson (1994), and Chalmers (2012). 7 For technical clarifcations on the notion of epistemic possibility, see Chalmers (2012, Tenth Excursus). 8 For criticism of modal monism, see Soames (2007), Vaidya (2008), and Mallozzi (2019).

Why We Need Something Like Imagery

41

9 See in particular Chalmers (2012). 10 There are various of caveats that will not concern us here about which non-basic facts that APE applies to. See Chalmers (2012) for extensive discussion. 11 Possessing a concept does not guarantee that we will be able to come up with even a rough analysis of it. It does endow us with the ability to examine a particular case and decide whether the concept applies does not apply (Chalmers & Jackson, 2001, 320). 12 Chalmers and Jackson (2001, 323) require the “non-triviality” clause in order to ensure that the empirical information does not render trivial the identifcation of the concept’s extension. In this case, were the empirical information to include the fact that the object is a garbage bin, or an ashcan, it would be a trivial task for Shayan to identify the object as a trashcan. Triviality may be diffcult to describe precisely, but it is, as Chalmers and Jackson say, intuitive. 13 I do not mean to prejudge the issue of whether vagueness and indeterminacy are distinct phenomena. On several theories, vagueness is a species of indeterminacy. On other theories, indeterminacy is impossible. 14 And Ned Block, in personal communication many years ago. 15 For more on deferential terms in particular, see Chalmers (2012, §6.9); for idealization generally, see Chalmers (2012, §§2.7–2.8). 16 Thanks to Antonella Mallozzi for suggesting discussion of the Cosmoscope. 17 See Ichikawa and Jarvis (2012) and Ichikawa and Jarvis (2013). 18 For discussion of these taxonomic issues, see Balcerak Jackson (2016). As I noted in the introduction, on Balcerak Jackson’s taxonomy, Ichikawa and Jarvis are talking about supposition. 19 See also Nichols and Stich (2000) and Nichols (2004). Thanks to Christopher Badura for noting the connection. 20 See also Chalmers (2012, 69–71). 21 Compare to Chalmers’s (2002) discussion of prima and secunda facie conceivability vs. ideal conceivability. 22 As Christopher Badura pointed out to me, if our beliefs are not justifed, then that undermines much of our justifcation for thinking the beliefs are coherent. 23 Though as Christopher Badura noted, dialetheists like Graham Priest disagree. 24 https://harrypotter.fandom.com/wiki/Mistakes_in_the_Harry_Potter_ books notes that in the US edition of The Prisoner of Azkaban, when exiting the tunnel that lead out of the Whomping Willow, the group was in the following order: Crookshanks, Lupin, Pettigrew, Ron, Snape, Sirius, Harry, and Hermione. Later, when Hermione uses the Time-Turner, Harry and Hermione see themselves coming out in the order: Lupin, Ron, Pettigrew, Hermione, Snape, Harry, and Sirius. 25 As someone pointed out to me at the 2020 COVID conference, there is a sense in which we collectively might have the ability to detect low level incoherence. The legions of fans do what the author herself cannot. That doesn’t mean that each of us individually has this ability. If that is right, then philosophers should individually be a lot less confdent than they actually are in their own judgments that a scenario is coherent, and a lot more receptive than they actually are to criticisms that is it not coherent. 26 See my “Thought experiments in ethics,” Kung (2016a). 27 As Antonella Mallozzi pointed out, this may be an empirical question that cognitive scientists can help answer.

42

Peter Kung

28 See Balcerak Jackson (2018), Balcerak Jackson and Balcerak Jackson (2013), Kind (2016), Kind (2018), Kind and Kung (2016a), Kung (2010), Mallozzi (2020), and, in this volume, chapters by Badura, Mallozzi, and Myers. 29 Grandin and Tesla play a starring role in Kind (2016). The fip side of the Grandin-Tesla coin is that those with impoverished imagery might not be as well equipped to form justifed beliefs about metaphysical possibilities. 30 See Kung (2016b) for further discussion of Hume. 31 See Hart’s (1988) underappreciated monograph for discussion. 32 See, e.g., the work of Aronowitz and Lombrozo, such as their (2020). 33 See Laura Perini’s work, such as her (2005a, 2005b). 34 Many thanks to Ned Block, Christopher Badura, Amy Kind, Masahiro Yamada, students in my Spring 2020 imagination seminar at ASU, and attendees at the online COVID conference in May 2020 for help in making this chapter better. Special thanks to Antonella Mallozzi for detailed and instructive comments.

References Armstrong, D. (1989). A combinatorial theory of possibility. Cambridge: Cambridge University Press. Aronowitz, S. & Lombrozo, T. (2020). Learning through simulation. Philosopher’s Imprint, 20(1), 1–18. Balcerak Jackson, M. (2016). On the epistemic value of imagining, supposing, and conceiving. In A. Kung & P. Kung (Eds.), Knowledge through imagination. Oxford: Oxford University Press. Balcerak Jackson, M. (2018). Justifcation by imagination. In F. Dorsch & F. Macpherson, (Eds.), Perceptual imagination and perceptual memory. Oxford: Oxford University Press. Balcerak Jackson, M. & Balcerak Jackson, B. (2013). Reasoning as a source of justifcation. Philosophical Studies, 164(1), 113–26. Block, N. & Stalnaker, R. (1999). Conceptual analysis, dualism, and the explanatory gap. Philosophical Review, 108(1), 1–46. Chalmers, D. (2002). Does conceivability entail possibility? In Gendler & Hawthorne (Eds.), Conceivability and possibility. Oxford: Oxford University Press. Chalmers, D. (2012). Constructing the world. Oxford: Oxford University Press. Chalmers, D. & Jackson, F. (2001). Conceptual analysis and reductive explanation. Philosophical Review, 110(3), 315–60. Frankfurt, H. (1969). Alternate possibilities and moral responsibility. Journal of Philosophy, 66(23), 829–39. Gendler, T.S. & Hawthorne, J. (Eds.). (2002). Conceivability and possibility. New York: Oxford University Press. Hart, W.D. (1988). The engines of the soul. Cambridge: Cambridge University Press. Ichikawa, J. (2016). Modals and modal epistemology. In A. Kung & P. Kung (Eds.), Knowledge through imagination. Oxford: Oxford University Press. Ichikawa, J. & Jarvis, B. (2012). Rational imagination and modal knowledge. Noûs, 46, 127–58. Ichikawa, J. & Jarvis, B. (2013). The rules of thought. Oxford: Oxford University Press.

Why We Need Something Like Imagery

43

Jackson, F. (1994). Finding the mind in the natural world. In R. Casati, B. Smith, & G. White (Eds.), Philosophy and the cognitive sciences: Proceedings of the 16th international Wittgenstein symposium. Vienna: Verlag Holder-PichlerTempsky. Reprinted in Block, Flanagan, & Güzeldere (Eds.) (1997), 483–91. Kind, A. (2016). Imagination under constraints. In A. Kung & P. Kung (Eds.), Knowledge through imagination. Oxford: Oxford University Press. Kind, A. (2018). How imagination gives rise to knowledge. In F. Dorsch & F. Macpherson (Eds.), Perceptual imagination and perceptual memory. Oxford: Oxford University Press. Kind, A. & Kung, P. (2016). Introduction: The puzzle of imaginative use. In A. Kung & P. Kung (Eds.), Knowledge through imagination. Oxford: Oxford University Press. Kung, P. (2010). Imagination as guide to possibility. Philosophy and Phenomenological Research, 81(3), 620–63. Kung, P. (2016a). Thought experiments in ethics. In A. Kung & P. Kung (Eds.), Knowledge through imagination. Oxford: Oxford University Press. Kung, P. (2016b). Personal identity without too much science fction. In B. Fischer & F. Leon (Eds.), Modal epistemology after rationalism, Synthese Library. Cham: Springer. Langland-Hassan, P. (2016). On choosing what to imagine? In A. Kung & P. Kung (Eds.), Knowledge through imagination. Oxford: Oxford University Press. Leon, F., & Tognazzini, N. (2010). Why Frankfurt-examples don’t need to succeed to succeed. Philosophy and Phenomenological Research, 80(3), 551–65. Mallozzi, A. (2019). Special issue of synthese on new directions in the epistemology of modality: introduction. Synthese, online. https://doi.org/10.1007/ s11229-019-02358-8. Mallozzi, A. (2020). Superexplanations for counterfactual knowledge. Philosophical Studies, published online. https://doi.org/10.1007/s11098-020-01477-0 Morris, W.E. & Brown, C.R. (2020) “David Hume”. In Edward N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2020 Edition). https:// plato.stanford.edu/archives/sum2020/entries/hume/. Nichols, S. (2004). Imagining and believing: The promise of a single code. Journal of Aesthetics and Art Criticism, 62(2), 129–39. Nichols, S. & Stich, S. (2000). A cognitive theory of pretense. Cognition, 74, 115–47. Perini, L. (2005a). The truth in pictures. Philosophy of Science, 72, 262–85. Perini, L. (2005b). Visual representations and their role in confrmation. Philosophy of Science, 72, 913–26. Soames, S. (2007). Reference and description. Princeton: Princeton University Press. Thomson, J.J. (1976). Killing, letting die, and the trolley problem. The Monist, 59(2), 204–17. Vaidya, A. (2008). Modal rationalism and modal monism. Erkenntnis, 68, 191– 212. https://doi.org/10.1007/s10670-007-9093-7. Van Cleve, J. (2011). Can coherence generate warrant ex nihilo? Probability and the logic of concurring witnesses. Philosophy and Phenomenological Research, 82, 337–80. Williamson, T. (2016). Knowing by imagining. In A. Kung & P. Kung (Eds.), Knowledge through imagination. Oxford: Oxford University Press.

2

An Imaginative Person’s Guide to Objective Modality Derek Lam

2.1 Our Modal North Star Chance is an objective feature of reality. But it’s a puzzling feature. It isn’t even immediately clear how we might begin to theorize sensibly about chance. We need some kind of methodological anchor to set us on a proper path and tell us what to look for. Lewis has a suggestion. In “A Subjectivist’s Guide to Objective Chance” (1980), he approaches this question by articulating a rational norm governing the relation between credence and chance, a norm known as the Principal Principle (PP). It’s a principle that tells us how a rational person’s credences should correlate with her beliefs about objective chances. Roughly put, it says that our credences should track our beliefs about objective chances. We will get into the detail of PP shortly. But the methodological point for now is that, even if we don’t know what objective chance really is, we clearly have a rudimentary conception of objective chance already; otherwise, we wouldn’t even be asking questions about the nature of objective chance. The notion has already entered our intellectual life somehow, playing a certain cognitive role. Whatever metaphysics of objective chance we come up with, it shouldn’t deviate signifcantly from this rudimentary grasp; otherwise, we risk simply changing the subject. Lewis argues that PP is important because it captures our rudimentary grasp of objective chance. If we have a handle on the idea of objective chance at all, it stems from our thoughts about how certain we should be about things. A good metaphysics of objective chance must therefore be able to explain why chance and ideal credence correlate in the way PP says they do. PP serves as an Archimedean point that guides our search for a theory of chance. This chapter isn’t primarily about objective chance (though it will play a role). My main goal is to show what we can do with the epistemic connection between imagination and metaphysical modality by mimicking what Lewis did with the relation between credence and chance. Assuming that there is objective, metaphysical modality, it’s hard to wrap our heads around what modal reality is supposed to be. Drastically different hypotheses have been offered, from views that appeal

An Imaginative Person’s Guide

45

to disconnected spacetime regions (Lewisian possible worlds) to views that appeal to powers of actual entities (e.g., Jacobs (2010)). The logical space of admissible options is so vast that it can be intellectually crippling. Fortunately, like the concept of objective chance, the concept of objective modality isn’t a theoretical invention. The idea of possibility and necessity has already entered our intellectual life, playing a cognitive role, a role related to our imagination. Is a talking donkey metaphysically possible? I try to imagine a donkey talking and succeed. I thereby conclude that it is possible. Is it metaphysically possible for something to be all green and all red? I try but cannot imagine a surface colored that way. I thereby conclude that it is metaphysically impossible. It’s fairly common to move from what one can imagine to what is possible this way. It might be controversial to say that imaginability entails possibility (and that unimaginability entails impossibility). But it’s relatively safe to assume that imagination provides defeasible evidence for possibility (and that unimaginability provides defeasible evidence for impossibility). In this chapter, I’m going to assume that imagining something provides justifcation for believing in its metaphysical possibility and a failed attempt to imagine something provides justifcation for believing in its metaphysical impossibility.1 Now I want to say something slightly more controversial. Whereas imagination is not our only epistemic route to modality, the epistemic connection between imagination and modality is special. This connection is how the idea of objective modality enters our intellectual lives in the frst place. It captures our rudimentary grasp of objective modality in the same way PP captures our rudimentary grasp of objective chance. My contention is that if the idea of objective possibility and necessity has any intellectual traction on us to begin with, it stems from what we can or cannot imagine. This modal epistemic connection can serve as an intellectual anchor that guides our search for a theory of modal reality in the same way PP sets us on the right path in search for a theory of objective chance. So, a proper theory of objective modality should be able to explain the way our imagination provides evidence for modal claims intuitively. By “the way our imagination provides evidence for modal claims,” I mean not only how our imagination succeeds in justifying modal claims but also how imagination fails to justify modal claims in certain cases (i.e., how imagination-based justifcation may be epistemically defeated). This chapter is particularly interested in the latter. 2 The goal of this chapter is to argue that using the imagination-modality connection as our Archimedean point and attending particularly to how imagination-based modal justifcation gets defeated, we have good reason to think that the proper logical structure of modal reality is S4 instead of S5.

46

Derek Lam

2.2 The Curious Case of Etiological Defeat Let’s focus on the justifcation for possibilities and set aside justifcation for impossibilities for the moment. Imagination-based modal justifcation may be presented as epistemically analogous to perceptual justifcation, as Yablo puts it: Without suggesting that Hume would go quite so far, I take the idea to be that conceiving is in a certain way analogous to perceiving. Just as someone who perceives that p enjoys the appearance that p is true, whoever fnds p conceivable enjoys something worth describing as the appearance that it is possible. In slogan form: conceiving involves the appearance of possibility. (1993, 5)3 And Geirsson, for example, appeals to this analogy in response to those who are skeptical of imaginability’s relevance for modal justifcation because of its fallibility: That I can imagine a scenario where something appears true while it cannot be true does not mean that conceivability does not carry with it justifcation of beliefs of what is possible. It is a general feature of justifcation that it does not guarantee truth, and that one can be justifed in believing what is false. For example, perceptual beliefs are typically justifed although they do not guarantee truth. (2005, 296) If I perceive a piece of broccoli, I have prima facie justifcation for believing that there is a piece of broccoli. In the same way, my imagining a talking donkey gives me prima facie justifcation for believing that a talking donkey is possible. Imagination is like perception of the modal space. Although perception can be used as an analogy to illustrate the sense in which imagination provides prima facie modal justifcation, there are signifcant differences. For our purpose in particular, epistemic defeat for perceptual justifcation and imagination-based modal justifcation don’t seem to work the same way. My perceiving a spider gives me prima facie justifcation to believe that there is one. Such justifcation is defeasible. Here is one way my perceptual justifcation may be defeated: being told that I just consumed some hallucinogenic mushrooms. The information that I have consumed a hallucinogenic substance undermines the support that my perceptual experience provides to my belief that there is a spider. This kind of epistemic defeater is standardly called an undermining or undercutting defeater. (This is in contrast to rebutting defeaters, which are evidence

An Imaginative Person’s Guide

47

against the belief at stake directly, not the epistemic force of the prima facie justifcation involved.)4 More specifcally, the kind of undermining defeater involved in this scenario is etiological. The justifcatory force of my perception is undermined by information about the causal origin of the perception. This is worth mentioning because not all undermining defeaters are etiological. For example, in the well-known fake barn case, if I’m told about the existence of the numerous fake barns in the area, my perceptual justifcation for thinking that I’m standing in front of a barn is undermined, yet there is nothing wrong about the causal origin of my perceptual experience. By contrast, unlike perceptual justifcation, it seems that etiological undermining defeaters don’t work for imagination-based modal justifcation for beliefs about possibility. Suppose I’m justifed to believe that a talking donkey is a metaphysical possibility because I imagined one. Say I’m told that I started imagining and fantasizing about talking donkeys because someone slipped something into my drink. Without the drug, being an uptight person (or perhaps I suffer from some kind of aphantasia), I wouldn’t be able to imagine such nonsense even if I tried. Does this information about the causal origin of my imagining undermine the modal justifcation my imagining provides? I’m inclined to say no. It would have undermined my perceptual justifcation for believing that there is an actual talking donkey if the substance caused me to see talking donkeys. But as long as I’m really imagining talking donkeys, I don’t see how that etiological information undermines my justifcation for believing that talking donkeys are possible in light of my imagination. The same is true even if the source of my imagination is more invasive. Perhaps my imagining a talking donkey is caused by a neurosurgeon stimulating certain areas of my brain. Whereas this kind of information would undermine perceptual justifcation about what is actual, it doesn’t undermine the modal justifcation provided by my imagination. Assuming that imagination provides modal justifcation, it provides modal justifcation regardless of the imagining’s causal origin.5 This disanalogy between imagination-based modal justifcation and perceptual justifcation alone is interesting and calls for an explanation, but the puzzle runs deeper. Although it seems that there is no etiological undermining defeater to modal justifcation for possibility based on our imagination, modal justifcation for believing that something is impossible based on a failed attempt to imagine something can be undermined etiologically. For example, if I fnd myself incapable of imagining something, I have prima facie justifcation for believing that that thing is impossible. But if I’m then told that my imagination’s failure is due to a brain injury, that information would undermine my modal justifcation for believing that that thing is impossible.6 Recognizing this asymmetry in etiological defeat can play an important role in modal epistemological discussions. For example, Kung (2010)

48 Derek Lam states that non-sensory imagination – unlike sensory imaginations – is far too liberal to be of value for modal justifcation. How does he argue that non-sensory imagination is too liberal? Kung’s strategy is to frst identify what constrains our capacity to imagine something (i.e., what stops us from being able to imagine something) and then argue that such constraints are not relevant to objective modality. Since the only constraints on imagination are not relevant to what is objectively possible, he concludes that our power of imagination isn’t suffciently constrained to provide evidence for objective possibilities. Notice that learning about what constrains our capacity to imagine is simply to obtain information about the source of our inability to imagine something. This kind of information can indeed undermine justifcation for believing that something is impossible based on what we cannot imagine. If Kung has succeeded in showing that the factors that prevent us from imagining something – in a non-sensory way – are irrelevant to objective modality, he may have offered a general reason to undermine the legitimacy of using our inability to non-sensorily imagine as evidence for the impossibility. However, even if we grant him that, given what we have said about the asymmetry that etiological defeaters work for undermining justifcation for impossibility based on our inability to imagine but do not work for undermining justifcation for possibility based on our imagination, it remains unclear that we should jump to any conclusion about justifcation for possibility based on what we can and do imagine.7 There is certainly much more to be said about the application of this epistemological asymmetry. One might even question the reality of this asymmetry. In this chapter, I take this apparent asymmetry at face value as a starting point. My goal is to show that assuming as a premise that this asymmetry in the imagination-modality connection obtains, and assuming as a premise that the imagination-modality connection is an Archimedean point for theorizing about modality, this asymmetry presents a substantive constraint upon our view about the modal reality.

2.3 Lewisian Chance-Credence Norm To explain the asymmetry regarding epistemic defeat for modal justifcation, we need to explain two facts: [Etiological Immunity] modal justifcation about possibility based on imagination is not defeasible etiologically; and [Etiological Vulnerability] modal justifcation about impossibility based on inability to imagine is defeasible etiologically. There is a ready explanation for [Etiological Immunity] based on Lewis’s PP. So, let’s take a closer look at this chance-credence norm frst.

An Imaginative Person’s Guide

49

Credence and chance are different things. Yet, if one’s credence about something is supposed to be refected in one’s rational betting behavior, then it’s plausible to think that there is some rational connection between the two. After all, our rational betting behaviors should be somehow sensitive to what the chances are. PP says that how certain we are about a proposition (i.e., our credence) should match the objective chance of that proposition. Or, in technical terms that some philosophers somehow prefer: a subject S’s credence for a proposition Q conditional on the proposition that the objective chance of Q is x should be equal to x. So, if I believe that there is 0.3 chance that it will rain, my credence for the fact that it will rain should also be 0.3. This is of course a simplifcation of the matter. Lewis argues that there is what he calls inadmissible evidence that, when present, breaks the rational connection between credence and objective chance. To put it in terms of notation introduced earlier, Lewis explains inadmissible evidence about Q as a piece of evidence that is directly about Q – not by being a piece of evidence about chance. Here’s an example to illustrate the idea. Say a coin, which I believe was a fair coin, was fipped in my absence. A friend tells me that it landed heads. Notice that my friend’s testimony is evidence for the proposition that the coin landed on head. It isn’t evidence about the chance of it landing heads. The testimony is therefore inadmissible evidence. (By contrast, information about the past frequency of the coin landing heads would be evidence about the chance of it landing heads. Such information would be admissible evidence.) In this case, despite my believing that the coin was fair and had a 0.5 chance of landing heads, I am not thereby rationally required to have 0.5 credence that the coin was going to land heads. Given the testimonial justifcation for believing that it has landed heads, my rational credence for that should be higher than 0.5 (see Lewis 1980, 265). Having inadmissible evidence (i.e., the testimony from my friend) breaks the chance-credence connection. So, what PP says isn’t that our credence should track what we believe to be the chance of things no matter what. The thought is that credence should track objective chance given any admissible evidence and admissible evidence alone. It’s contentious how “admissibility” is to be analyzed more rigorously and, accordingly, what exactly is inadmissible. On one occasion, Lewis argues that information directly about the future is all inadmissible (e.g., a crystal ball that reveals the future directly) and information about the past and probabilistic laws of nature is all admissible. But the example about testimonial evidence indicates that some information about the past can be inadmissible evidence that breaks the chance-credence norm. Different philosophers have since then offered different ways to formalize or standardize the idea.8 I’m not going to engage in those debates in this chapter because, for those who accept that imagination provides modal justifcation, imagination

50

Derek Lam

is admissible evidence. This might be surprising to some. One might be tempted to think that since perceptual evidence is not evidence about chance and therefore inadmissible, 9 if we model imagination-based justifcation about possibility with perceptual justifcation about actuality, perhaps we should think that imagination-based modal evidence is inadmissible evidence as well. But notice the fact that a state of affairs p is possible just is the fact that there is a non-zero chance that p. Evidence about possibility is evidence about chance, unlike perceptual justifcation. Therefore, evidence for possibility is admissible. Since modal evidence is admissible and doesn’t break the chance-credence normative connection, the exact analysis of admissibility doesn’t make signifcant difference to what I’m going to say.

2.4 Explaining Etiological Immunity Let’s take a step back from modal justifcation and think about evidence more generally. To receive information as evidence for p is to accept information that, all else being equal, should boost one’s credence for p. If a piece of information doesn’t make it rational to be more certain about p, that piece of information isn’t evidence for p. A rational person’s credence at a time should satisfy the axioms of probability calculus. So, a rational person’s credence updating should be governed by Bayes’ theorem (assuming P(B) > 0): P(A|B) = P(B|A) P(A)/P(B). This dictates that if P(A) = 0, there is no B such that P(A|B) is larger than zero. In other words, if a rational person is absolutely certain that A isn’t the case so that Cr(A) = 0 (where Cr stands for the person’s credence function at a time), then no information should be able to boost her credence over zero. Since evidence boosts credence, a rational person should never think that there is evidence for A if she begins with Cr(A) = 0. On the fip side, if she thinks that there may be evidence for A, her credence for A should be non-zero. Now let us consider the following state of affairs: Z : it is possible that ˜. If a rational person thinks that there is any evidence for Z at all, her credence for Z must be non-zero (due to Bayes’ theorem). Then the chance-credence norm PP dictates that she ought to believe that Z has a non-zero chance. It would have been irrational of her to believe that there is no chance for Z yet have some degree of belief that Z is the case. As I have stated in the previous section, to say that something is possible just is to say that that thing has a non-zero chance. In other words, this rational person, who ought to believe that there is a non-zero chance that Z, ought to believe that it is possible that Z; otherwise, she shouldn’t believe that there can be any evidence for Z at all.

An Imaginative Person’s Guide

51

Here is a modal logic principle:

[ Possibly Possible ] If it is possibly possible that p, then it is possible that p ( ˙˙p ˆ ˙p ) . To think that it’s possible that Z is to think that it is possibly possible that ψ. The principle [Possibly Possible] requires that if a rational person believes that it is possible that Z (hence ◊◊ ψ), she ought to also believe that it is possible that ψ. So, based on Lewis’s PP and [Possibly Possible], a rational person who thinks that there is evidence that it is possible that ψ ought logically to believe that it is possible that ψ. Notice that this is a very substantive claim. That there is evidence for p doesn’t normally entail that p even for an ideal rational agent. When a rational person hears her cat vomiting, that auditory experience is evidence for the fact that her cat is vomiting. That doesn’t, however, entail that her cat is vomiting. The perceptual evidence is non-conclusive. What we have shown based on PP and [Possibly Possible] is that evidence for statements about possibility works differently. That there is evidence for a statement about something’s being possible entails that the statement is true. In other words, there is only conclusive evidence for possibility. If we think that imagination provides evidence for possibility at all, we ought to think that it’s conclusive evidence. This explains [Etiological Immunity]. If I managed to imagine a talking donkey, I ought to believe that it’s possible for there to be talking donkeys. Since entailment is monotonic, no extra information can undermine the evidence for the possibility of a talking donkey. That is why it should not change anything to learn that my imagination is triggered by a surgeon stimulating my brain.

2.5 Defeating an Appeal to Conclusive Evidence My explanation of [Etiological Immunity] is based on my demonstrating that one’s imagining something entails its possibility in virtue of PP and [Possibly Possible]. I reckon that most would fnd the entailment claim intuitively questionable. Unlike Hume, most philosophers who endorse imagination-based modal justifcation fnd it too strong to claim that imagination entails possibility. Even Chalmers (2002), who argues that conceivability entails possibility, only defends the view that ideal conceivability entails possibility. One can see where the hesitation comes from. We appear to imagine things that are impossible. For example, here’s a mathematical statement: there is at least one prime number between 155921 and 360653. Given how many numbers there are between 155921 and 360653, one might think that one can imagine fnding a prime number between them. The truth is, there are no prime numbers between them. Assuming that

52

Derek Lam

mathematical statements have their truth-value necessarily, it’s necessary that there are no prime numbers between 155921 and 360653. So, as a source of evidence for possibility, our imagination seems obviously fallible. And that seems to undercut my explanation for [Etiological Immunity]. Actually, my explanation of [Etiological Immunity] is compatible with our appeal to imagination for modal justifcation being fallible. To illustrate why, let’s consider a mathematical proof. A proof is conclusive evidence for a mathematical statement. If a person has a proof for a mathematical statement p, that is the end of the matter. One doesn’t go on to weigh the strength of various evidence for p after that. But this doesn’t mean that when she presents her justifcation for p by constructing a proof, she is infallible. There can still be an epistemic defeater for her justifcation for accepting p. If information becomes available which indicates that what she presented was in fact not a proof for p at all (perhaps there was a mistake in her attempt to construct a proof), her justifcation would be nullifed. This information shows that she has misidentifed something as a proof and hence as evidence for p. Mathematical proofs are conclusive but our ability to identify proofs is fallible. As Peacocke (2003) rightly points out: [We should] distinguish two kinds of defeasibility, which I will call defeasibility of identifcation and defeasibility of grounds. A ground for accepting a proposition can be conclusive even though our entitlement to believe that we have identifed such a ground is defeasible. Identifying something as a conclusive ground is one thing; its being a conclusive ground is another. (30) Let us call this an Identifcation Defeater. To think that evidence for possibility, focusing on imagination for our purposes, must be conclusive is compatible with our fallibility because our ability to identify relevant imaginings is fallible. For example, I could think that I was imagining fnding a prime number between 155921 and 360653, but all I was in fact doing was imagining my reaction to fnding such a number instead of imagining there being such a number. (For example, I might react by thinking out loud, “Huh, so there is a prime number between 155921 and 360653. I wonder whether there is a second one.”) So, perhaps, our imaginings are modal epistemologically fallible not because imagination is inconclusive, just like mathematicians are fallible not because mathematical proofs are inconclusive. This echoes much of what we see in debates about metaphysical modality. Consider the claim that light is a stream of photons, which Kripke considers necessary a posteriori if true. There is an anti-essentialist concern that light might have turned out to be something else. We could

An Imaginative Person’s Guide

53

imagine a scenario where there are photons fring yet we see no light. Kripke’s response is: Imagine a situation in which human beings were blind or their eyes didn’t work. They were unaffected by light. Would that have been a situation in which light did not exist? It seems to me that it would not. It would have been a situation in which our eyes were not sensitive to light. […] Even if all people had had awful vestigial growths and just couldn’t see a thing, the light might have been around; but it would not have been able to affect people’s eyes in the proper way. (1980, 129–30) Kripke’s point is that the anti-essentialists are mistaken about the content of their imagining. They are not, as they thought they are, imagining a scenario with photons but no light. Instead, they are imagining a scenario where there is light but people have no ability to detect light. The question is: why go this route? If imagination is inconclusive modal evidence, Kripke could have simply said that the anti-essentialists have indeed imagined a scenario in which there is a stream of photon and no light, but maintained that imagination is defeasible and this is a case in which light remains essentially a stream of photon despite what we imagine. Note that it is not at all clear that this is a weaker defense than saying that the anti-essentialists are wrong about the content of their own imagining: they aren’t really imagining light being something other than streams of photons, Hesperus not being identical to Phosphorus, water being something other than H 2O, etc. The same applies to the Kripkean defense of dualism about phenomenal qualities. Kripke states that it is conceivable for us to have C-fbers fring without the feeling of pain. Central to Kripke’s defense is the argument that since we couldn’t have been mistakenly imagining something else when we appeared to be imagining ourselves to have C-fbers fring without the feeling of pain (due to our direct reference to both), and since identity is necessary, we should conclude that pain is not C-fbers fring. A similar question arises: what is the point of focusing on arguing whether one could have been mistaken about the content of one’s imagination? The question wouldn’t have been even half as pressing if Kripke alone chooses to go down this road. On the contrary, a huge portion of the literature – from both dualists and physicalists – on philosophical zombies has been devoted to pinning down a proper characterization of the content of the relevant imaginings (e.g., the analysis of phenomenal concepts). This is puzzling. If imagination is inconclusive evidence for possibility, a physicalist can simply concede that we can imagine pain and C-fbers fring coming apart yet note that this is just one of those cases in which our imagination gets modality wrong. It’s unclear how

54

Derek Lam

much extra dialectical mileage, if any, dualists can gain from cornering physicalists into conceding that we couldn’t have been imagining something else when we imagine zombies.10 All this makes sense, however, with what we said about [Etiological Immunity]. If imagination is evidence for possibility, imagination entails possibility; and thus the only way to defeat an appeal to imagination as evidence for possibility is by showing that the alleged imagining in fact didn’t happen (i.e., presenting an Identifcation Defeater). That’s why essentialists need to argue that the anti-essentialists aren’t really imagining light and photon streams coming apart. And that’s why it’s a challenge physicalists need to take seriously if dualists are right that there is no plausible way to reinterpret what dualists claim to be an imagining about zombies into an imagining about something else (or not as an instance of the relevant kind of imagining at all). The result is what Kung (2016) calls the Error Theory about imagination, which says that we cannot imagine what is impossible, and if we thought we have imagined something impossible, it is an error and something else must have happened. Kung objects that if we set our modal epistemology aside and try to just honestly describe our imagination, it is far more intuitive to acknowledge that we can imagine the impossible.11 To reinterpret and restrict intuitive cases of imaginings as the Error Theory suggests is, according to Kung, to adopt a “telescopic view of imagining” that uses “imagine” as a success term like “see” (2016, 91). One cannot see what does not exist because there is nothing to see; similarly, one cannot imagine what is impossible because there is nothing to imagine. What one ends up with is a “modal-conclusion-frst reasoning” (2016, 104): one attributes imaginings (to oneself and others) according to one’s prior modal beliefs. Kung thinks that doing so snuffs out any hope of using imagination as a guide to possibility. My appeal to Identifcation Defeaters is to embrace the Error Theory. One point that makes Kung suspicious of the Error Theory is that philosophers like Kripke offer no clear motivation for the view that fies in the face of our ordinary self-ascriptions of imaginings. And it isn’t immediately clear that we need to use “imagine” as a success term to treat imagination as a guide to possibility. Now we know better. There is an independent reason to accept the Error Theory. It follows from two highly intuitive principles: Lewis’s PP and [Possibly Possible]. We obtain a neat explanation for [Etiological Immunity] as a result. Note that even those who reject the Error Theory should accept that our self-ascriptions of imaginings are fallible. Attitude ascriptions aren’t immune to the same cost-beneft analysis that all our other beliefs are subjected to. One may refuse to revise our ascriptions of imagining as the Error Theory requires. The cost is having to reject either PP or [Possibly

An Imaginative Person’s Guide

55

Possible] and leave [Etiological Immunity] unexplained. People may value the trade differently. But the Error Theory is not obviously getting a bad bargain. Surely, it opens the gates to some “modal-conclusion-frst reasonings” that revise our imagining ascriptions based on modal facts external to a theory of imagination. That doesn’t mean we can no longer rely on imagination as a guide to possibility. It just means using imagination to reach possibility is a holistic enterprise. One person’s modus ponens is another person’s modus tollens. And I don’t see why taking a person’s self-report of imaginings seriously implies, as Kind (2016, 347) seems to suggest, that revisions to our imagining ascriptions must be based entirely on factors internal to a theory of imagination.

2.6 Explaining Etiological Vulnerability Suppose we are satisfed with the explanation for [Etiological Immunity]. What is possible and what is impossible are closely tied logically. It’s natural to wonder whether what I argued for about evidence for possibility would logically lead to something similar about evidence for impossibility. But that would be bad news given that we want to explain [Etiological Vulnerability]. Here’s a parallel argument for the conclusiveness of evidence for impossibility. To have evidence that a state of affairs is impossible is to have evidence that the negation of that state of affairs is necessary. Let’s consider the following state of affair: T : it is necessary that ˜. Again, a rational person’s credence is governed by Bayes’ theorem. So, for a rational person to have evidence for T, her credence for T, i.e., Cr(T), has to be non-zero. The Principal Principle then says that she ought to believe that there is a non-zero chance that T is the case. The fact that T has a non-zero chance just is the fact that T is possible. Hence, a rational person ought to think that it’s possible that T, i.e., that it is possibly necessary that ψ, if she thinks that there is evidence for T. Here is a modal logic principle:

[ Necessarily Possible ] If it is possible that p, then it is necessarily possible that p ( ˙p ˆ˙p ) . [Necessarily Possible] is logically equivalent to its contrapositive: ~□◊p → ~◊p, which is, in turn, equivalent to ◊□~p → □~p and, hence, ◊□p → □p (by substitution). Since PP requires a rational person who believes that there is evidence for T to believe that it’s possibly necessary that ψ, it requires her to believe that it’s necessary that ψ. As I mentioned, every statement about necessity just is a statement about something being impossible. Therefore,

56

Derek Lam

a rational person is required to think that if there is any evidence that something is impossible at all, that evidence is conclusive. If a rational agent thinks that a failed attempt to imagine that p is evidence that p is impossible, she ought to think that her failed attempt is conclusive evidence that p is indeed impossible. But that cannot be if we want to think about objective modality in a way that is in line with the modal epistemic behavior of imagination, which includes the way imagination-based modal justifcation can/cannot be undermined by defeaters. Specifcally, although one’s inability to imagine that p is evidence that p is impossible, that evidence can be undermined by information about the source of such inability – [Etiological Vulnerability]. Learning that your inability to imagine that p is due to a brain injury defeats your inconceivability-based modal justifcation for thinking that it is impossible that p. To do so, this defeater doesn’t need to suggest that you aren’t really incapable of imagining p; that is, evidence for impossibility can be defeated by something other than an Identifcation Defeater. Evidence for possibility ought to be conclusive; evidence for impossibility need not be. To preserve this epistemic asymmetry, something in this parallel argument must go. Here is my hypothesis. The best way to explain [Etiological Immunity] without sacrifcing [Etiological Vulnerability] is to accept the principle [Possibly Possible], which is key to my explanation of [Etiological Immunity], while giving up the principle [Necessarily Possible], which is key to the argument for the conclusiveness of evidence for impossibility. Doing so has substantive implications about the shape of the objective modal landscape. [Possibly Possible] is a theorem in the modal logic system S4 and [Necessarily Possible] is a theorem in the system S5 but not S4. In other words, to think of modal reality in a way that is in sync with the modal epistemic behavior of imagination, we better think of objective modality as being governed by S4 instead of S5. In S4, the accessibility relation among possible worlds is not symmetric. That means, w1 can be possible with respect to w2 without w2 being possible with respect to w1. But S4’s accessibility relation is refexive so that w2 is possible with respect to itself. As a result, it isn’t logically guaranteed that what is possible with respect to one possible world must also be possible with respect to all other possible worlds. This is a signifcant result. Philosophers have long been interested in the true logic for objective modality with S5 being the most popular system. But discussions on this topic often have the fates of S4 and S5 bundled together. Whereas there is a heated debate about whether S4 can be the true logic for metaphysical modality, the arguments involved are typically applicable to both S4 and S5. For example, Chandler (1976) and Salmon (1989) challenge S4 by arguing against [Possibly Possible] based on the assumption that although it’s possible for a material object to have a slightly different material constitution, it’s impossible for it to

An Imaginative Person’s Guide

57

have a massively different material constitution. There are ways to resist their objection. The details do not concern us here.12 The relevant point is that [Possibly Possible] is a theorem of both S4 and S5. The fates of S4 and S5 are therefore bundled together in this debate. By studying the modal epistemic behavior of imagination, we obtain an argument in favor of S4 against S5; and we do so without treating this conclusion simply as a result of our prior commitment to some substantive metaphysics of modality.13

2.7 Overkill? My explanation of the modal epistemic behaviors of imagination specifcally – [Etiological Immunity] and [Etiological Vulnerability] – is based on my argument about the way modal evidence in general behaves (or should behave). In this section, I’ll discuss the potential worry that my approach overkills: while capturing the epistemic behaviors of imagination, my explanation implies the same for all sources of modal justifcation; and that may not be a desirable consequence. Even those who acknowledge that imagination is a source of modal justifcation would never go so far as to say that it’s the only source. There is no way I can cover every possible source of modal justifcation.14 What I’ll do instead is to focus on testimonial justifcation for modal claims as a case study. My hope is that by showing how my explanation’s implication for testimonial modal justifcation isn’t as objectionable as it frst seems, I can offer some assurance that it isn’t implausible for my explanation to imply something similar for all forms of modal justifcation. I’m justifed in believing that there is a non-zero chance for a closed system’s entropy to decrease over a limited time span. That is, it’s possible that a closed system’s entropy decreases over a limited amount of time. As someone who has no advanced knowledge in physics, my justifcation is based on textbooks: a textbook case of justifcation by testimony. (Pun intended!) If we grant that there is testimonial justifcation for modal claims, and if I’m right that evidence for possibility must be conclusive, testimonial justifcation for claims about possibility must be conclusive, too. So, when I received testimony about the possibility of a decrease in entropy, I have conclusive evidence that it’s possible. As much as we think testimony can provide justifcation, we think that testimony is fallible, no matter what the testimony is about. We might, for example, receive information that the person delivering the testimony has been conducting her research with malfunctioning equipment. Similarly, on modal matters, we might learn that a person who testifes about chances and possibilities has been using a faulty calculator or doing modal logic wrong. In cases like this, it seems clear that

58

Derek Lam

whatever testimonial justifcation one gets is defeated. So, my explanation for [Etiological Immunity] seems to imply too much by implying that all modal evidence about possibility is conclusive. My explanation is derived from PP and [Possibly Possible]. If one doesn’t like the implication, one has to give up one of them. Either way, it’s going to be a big theoretical sacrifce; and, on top of that, one will need to fnd another explanation for [Etiological Immunity]. The good news is, one doesn’t have to do so. Saying that testimony about possibility is conclusive is not to say that appealing to testimony about possibility is indefeasible. Again, we should distinguish the conclusiveness of a piece of evidence and the defeasibility of one’s appeal to evidence. To have testimonial justifcation is to be justifed in believing something on the basis of taking another person’s testimony at face value. To understand testimonial justifcation, we need to know what constitutes a piece of testimony. All that is uttered is not testimony. To be a piece of testimony, the speaker must have the intention that the hearer believes what is said. For example, overhearing someone mumbling to herself doesn’t count as a case of testimony in the relevant sense because she doesn’t even intend to be heard. Whatever epistemic signifcance overhearing others’ self-talk has, it’s epistemologically unhelpful to lump it together with testimonial justifcation. Furthermore, I want to argue that offering testimony is an inherently social phenomenon. What counts as testimony doesn’t only depend on the testifer, but also depends on the hearer. Say I move to a community where no one speaks English. I cannot offer testimony by speaking English even if I sincerely intend to do so. It isn’t that I produce testimony in English but the people fail to pick it up. What I say simply doesn’t count as testimony because no one even acknowledges that I’m telling them something.15 As we said earlier, one can receive information indicating that what we thought was a mathematical proof is in fact not a mathematical proof. Similarly, testimonial justifcation is susceptible to Identifcation Defeaters. But, due to testimony’s inherently social nature, Identifcation Defeaters for testimonial justifcation can happen in two ways. First of all, you might receive information that indicates that what you received from the speaker was not intended to be a piece of testimony. Second, you might receive information that makes you outright refuse to even acknowledge that the speaker said anything meaningful to the conversation, shunning the speaker’s contribution. Since testimony is partly constituted by the recipient’s recognition, this effectively makes what the speaker said not even count as a piece of testimony. Testimonial justifcation is fallible no matter what the testimony is about. This conficts with my view’s implication that testimonial justifcation for possibility must be conclusive, so the objection goes. I agree with the frst claim but reject the second one. Just like the fallibility of

An Imaginative Person’s Guide

59

imagination-based evidence for possibility, the fallibility of testimonybased evidence for possibility can be explained by Identifcation Defeater. Hence, the fallibility of testimonial justifcation for possibility is compatible with my view, which implies that testimonial justifcation for possibility is conclusive. Let me illustrate with an example. Say someone told me that certain subatomic process L has a non-zero chance of happening. I thereby have testimonial justifcation for believing that L is possible. But then I learn that that person makes things up all the time just to look knowledgeable. According to my view, the testimonial justifcation is nullifed not because the new information undermines the justifcatory force of the testimony; instead, it’s because the new information gives me reason to simply ignore whatever that person said. With no recognition at all, what she says fails to be testimony and I therefore no longer have testimonial justifcation at all – not just a weakened or undermined one – that L is possible. One might wonder: why should we describe defeaters for testimonial modal justifcation in the way I suggest? I’m not saying that we should. All I’m saying is that this is a coherent way to think about my explanation of [Etiological Immunity] and the fallibility of testimonial justifcation. The mere fact that testimonial justifcation for modal claims is fallible is therefore not an objection to my explanation.16 I focus on testimonial justifcation because it’s a kind of justifcation that is most obviously fallible. But what I said applies to all sources of justifcation for beliefs about possibility. If it works for testimonial justifcation, it should work for other sources of evidence for possibility. For example, some argue that we have modal intuitions for what is possible. If that is the case, such intuitions should be conclusive evidence for what is possible. Having modal intuitions should be factive. Of course, we are fallible when we appeal to our modal intuitions (assuming that we have such a cognitive faculty). But like our appeal to imagination, that is because we are fallible in identifying instances of modal intuition.

2.8 Conclusion Modal notions come into our intellectual life via the operation of our imagination. The epistemic connection between imagination and modality serves as a constraint upon our attempt to theorize about modality. Attempts to integrate the epistemology and metaphysics of modality are not exactly new. My treatment of the subject is novel in its emphasis on observing the way epistemic defeaters work for imagination-based modal justifcation. I bring attention to an epistemic asymmetry: whereas information about the causal origin of our imagination cannot defeat imagination-based modal justifcation about possibility, information about the causal origin of our inability to imagine can defeat modal

60

Derek Lam

justifcation about impossibility. We should think about modal reality in a way that can account for this modal epistemic asymmetry. I argue that the best explanation for the asymmetry is by accepting Lewis’s Principal Principle, and accepting that the objective modal landscape is governed by S4, not S5. This explanation implies that evidence about possibility must be conclusive. Although this seems to go against the fact that our modal reasoning is fallible, such fallibility is compatible with our modal justifcation being conclusive as long as we recognize that an appeal to conclusive evidence can be defeated by Identifcation Defeaters.

Notes 1 Of course, there are skeptics, e.g., Van Inwagen (1998). But the goal of this chapter isn’t to address the skeptics. I also limit myself to current imaginings. I set aside whether having been able to imagine something in the past counts as evidence for possibility. 2 The emphasis on the epistemic defeat is what sets my project apart from Peacocke’s (1999), which seeks to integrate the epistemology and metaphysics of modality. 3 Yablo and Geirsson don’t distinguish conceivability and imaginability. 4 The distinction between undermining and rebutting defeaters is due to Pollock (1987). See also Sturgeon (2012) and Melis (2014) for further analysis of the distinction. 5 One might argue that perceptual justifcation is based on a perceptual experience’s causal connection with the perceived in normal circumstances. That isn’t the case for imagination-based justifcation about possibility. Hence, etiological information may defeat perceptual justifcation but not imagination-based justifcation about possibility. I don’t know that this is a good explanation. I’m skeptical of this view about perceptual justifcation because of the New Demon Problem: although perceptual experience in the Matrix doesn’t provide knowledge, it should still provide justifcation. 6 Information about the origin of one’s unimaginability can undermine modal justifcation for impossibility even if it doesn’t involve brain injury. See Levin (2011) for a discussion of the epistemic signifcance of imaginative resistance as a source of unimaginability. 7 See Lam (2018) for a further discussion. 8 For example, see Hall (1994), Thau (1994), Vranas (2004), Ismael (2008), and Pettigrew (2015). 9 This is unless one thinks of our minds as predictive systems through and through; see Clark (2013). 10 For more examples of debates that appear to assume that imaginability implies possibility, see Kind (2016). 11 See also Kind (2016) and Priest (2016: 2658–9). Priest proposes a logic for imagination that allows impossible worlds to be imagined. Embracing the Error Theory, I clearly don’t accept this as the logic for imagination. Yet, none of this prevents us from accepting a semantics that includes impossible worlds for modeling thoughts about modality itself. For simplicity’s sake, however, I’ll focus on normal modal logic for that is the standard treatment for objective modality.

An Imaginative Person’s Guide

61

12 See Yagisawa (2017), Murray and Wilson (2012), and Roca-Royes (2006, 2011). 13 For example, if one is committed to Armstrong’s (1989) combinatorial metaphysics of modality, one would accept S4 and reject S5. 14 There are those who base their view on conceivability or imaginability (e.g., Yablo 1993, Geirsson 2005), there are those who rest their view on postulating the faculty of modal intuition (e.g., Bealer 2002; Fiocco 2007), there are those who rely on our implicit knowledge of the principle of admissible interpretations (e.g., Peacock 2003), and there are those who appeal to our power of forming supposition (e.g., Ichikawa & Jarvis 2012). 15 Instead of saying that testimony requires both the speaker’s intention and the hearer’s recognition, Lackey argues that we should distinguish two concepts of testimony: speaker testimony and hearer testimony (2006, 187). Since we’re interested in testimony as a source of (modal) belief for the hearer, if we accept Lackey’s distinction, we should focus on hearer testimony. What someone says doesn’t constitute a piece of hearer testimony if the hearer doesn’t reasonably recognize what the speaker says as conveying relevant information. 16 Does my proposal make our appeal to testimony for modal justifcation circular? No, it just means working with testimony is a holistic business.

References Armstrong, David. 1989. A Combinatorial Theory of Possibility. New York: Cambridge University Press. Bealer, George. 2002. “Modal Epistemology and the Rationalist Renaissance”. In Conceivability and Possibility. Edited by Tamar Gendler & John Hawthorne. 71–126. Oxford: Oxford University Press. Chalmers, David. 2002. “Does Conceivability Entail Possibility?” In Conceivability and Possibility. Edited by Tamar Gendler & John Hawthorne. 145– 200. Oxford: Oxford University Press. Chandler, Hugh S. 1976. “Plantinga and the Contingently Possible”. Analysis 36(2): 106–9. Clark, Andy. 2013. “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science”. Behavioral and Brain Sciences 36: 181–253. Fiocco, Marcello Oreste. 2007. “Conceivability, Imagination, and Modal Knowledge”. Philosophy and Phenomenological Research 74(2): 364–80. Geirsson, Heimir. 2005. “Conceivability and Defeasible Modal Justifcation”. Philosophical Studies 122: 279–304. Hall, Ned. 1994. “Correcting the Guide to Objective Chance”. Mind 103(412): 505–18. Ichikawa & Jarvis 2012. “Rational Imagination and Modal Knowledge”. Noûs 46(1): 127–58. Ismael, Jenann. 2008. “Raid! Dissolving the Big, Bad Bug”. Noûs 42(2): 292–307. Jacobs, Jonathan D. 2010. “A Power Theory of Modality: Or, How I Learn to Stop Worrying and Reject Possible Worlds”. Philosophical Studies 151: 227–48. Kind, Amy. 2016. “The Snowman’s Imagination”. American Philosophical Quarterly 53(4): 341–8.

62

Derek Lam

Kripke, Saul. 1980. Naming and Necessity. Cambridge: Harvard University Press. Kung, Peter. 2010. “Imagining as a Guide to Possibility”. Philosophy and Phenomenological Research 81(3): 620–63. ———. 2016. “You Really Do Imagine It: Against Error Theories of Imagination”. Noûs 50(1): 90–120. Lackey, Jennifer. 2006. “The Nature of Testimony”. Pacifc Philosophical Quarterly 87: 177–97. Lam, Derek. 2018. “Is Imagination too Liberal for Modal Epistemology?” Synthese 195: 2155–74. Levin, Janet. 2011. “Imaginability, Possibility, and the Puzzle of Imaginative Resistance”. Canadian Journal of Philosophy 41(3): 391–421. Lewis, David. 1980. “A Subjectivist’s Guide to Objective Chance”. In Studies in Inductive Logic and Probability. Edited by Richard C. Jeffrey. Volume II: 263–93. Berkeley: University of California Press. Melis, Giacomo. 2014. “Understanding Undermining Defeat”. Philosophical Studies 170: 433–42. Murray, Adam & Wilson, Jessica. 2012. “Relativized Metaphysical Modality”. In Oxford Studies in Metaphysics. Edited by Karen Bennett & Dean Zimmerman. 189–226. Oxford: Oxford University Press. Peacocke, Christopher. 1999. Being Known. Oxford: Oxford University Press. ———. 2003. The Realm of Reason. Oxford: Oxford University Press. Pettigrew, Richard. 2015. “What Chance-Credence Norms Should Not Be”. Noûs 49(1): 177–96. Pollock, John. 1987. Contemporary Theories of Knowledge. Lanham: Rowman-Littlefeld. Priest, Graham. 2016. “Thinking the Impossible”. Philosophical Studies 173: 2649–62. Roca-Royes, Sonia. 2006. “Peacocke’s Principle-Based Account of Modality: ‘Flexibility of Origins’ Plus S4”. Erkenntnis 65(3): 405–26. ———. 2011. “Essentialism vis-a-vis Possibilia, Modal Logic, and Necessitism”. Philosophy Compass 6(1): 54–64. Salmon, Nathan. 1989. “The Logic of What Might Have Been”. The Philosophical Review 98(1): 3–34. Sturgeon, Scott. 2012. “Pollock on Defeasible Reasons”. Philosophical Studies 169: 105–18. Thau, Michael. 1994. “Undermining and Admissibility”. Mind 103(412): 491–503. Van Inwagen, Peter. 1998. “Modal Epistemology”. Philosophical Studies 92: 67–84. Vranas, Peter. 2004. “Have Your Cake and Eat It Too: The Old Principal Principle Reconciled with the New”. Philosophy and Phenomenological Research 69(2): 368–82. Yablo, Stephen. 1993. “Is Conceivability a Guide to Possibility?” Philosophy and Phenomenological Research 53(1): 1–42. Yagisawa, Takashi. 2017. “S4 to S5”. Argumenta 2(2): 241–61.

3

Crossing Rivers Imagination and Real Possibilities Rebecca Hanrahan

3.1 Crossing Rivers: Imagination and Real Possibilities An analogy often cited within the literature on modal epistemology is that imagination is to the possible as perception is to the actual.1 There is a lot to say about this claim, but what to me is worthy of note is the implied division of labor. Perception’s job is to tell us what is actual, not what is possible. There is another mental process, specifcally, the imagination, whose job is to tell us what is possible. Of course, what is actual is possible, and assuming our senses reveal what is actual, our senses can tell us about what is possible. Vision informs you that these letters are before you right now, and from this fact, you can conclude that it is possible for these letters to be before you right now. But this inference is not ampliative, it is merely explicative. It doesn’t tell us anything about a world that we didn’t already know.2 For many philosophers, this inference marks the limit of perception’s role within modal epistemology. It is assumed that perception can’t tell us about “non-actual possibilities,” nor can it tell us about necessities (Strohminger 2015, 363). In this chapter, I want to defend this assumption by looking at an account that rejects it. In “Perceptual Knowledge of Nonactual Possibilities,” Margot Strohminger challenges this assumption. She argues that she can just see that the mug across from her is reachable. This knowledge is not a product of inference but rather a product of perception. She contends that her perceptual account of modal knowledge makes evolutionary sense and is more psychologically realistic than accounts that depend on inferences. I will present Strohminger’s view and argue that while what we perceive might prompt us to form psychologically immediate beliefs about what is possible or necessary, on her account these beliefs rise to the level of knowledge only because we have reaffrmed certain arguments. Hence, I argue, they shouldn’t be considered perceptual knowledge. I will next consider a theory of perceptual modal knowledge that doesn’t depend on such repeated affrmations. But I will argue that because this disposition to form certain modal beliefs in response to what we perceive is supposedly a product of evolutionary forces, we will at times

64

Rebecca Hanrahan

be led astray. We will be prompted to form modal beliefs that we won’t be justifed in holding, though our ancestors would have been so justifed. Thus, we will need another method of acquiring modal knowledge, such as the imagination, to correct for these disparities. Let’s begin with Strohminger’s theory of perceptual modal epistemology.

3.2 Perceptual Knowledge of Nonactual Possibilities Strohminger’s goal is not to convince the modal skeptic. She isn’t trying to establish that we have modal knowledge. She assumes we do. Her goal is to establish that some of our modal knowledge is perceptual knowledge. Strohminger begins by defning perceptual knowledge. For a belief to count as perceptual knowledge, it must be psychologically immediate, and it must be a product of our perceptual faculties and not a product of inference. When I convey this kind of knowledge, my claims can be modifed both by “gaze-related directional prepositions” and with “the modal auxiliary ‘can’” (Strohminger 2015, 365, 366). Say I am hiking, and I see a herd of elk. If challenged about what I see, I might reply “I can see them through my binoculars sitting down in the tall grass.” The modal auxiliary “can” indicates that I am exercising an ability, specifcally, the ability to take in visual information. And the prepositional phrase “through my binoculars” describes how I am employing my visual faculties. Thus, my knowledge of the presence of these elks counts as perceptual knowledge. Strohminger next considers whether we can have perceptual knowledge of at least some non-actual possibilities. And she offers two arguments in support of this position. The frst concerns our knowledge of our own abilities. Say there is salt and pepper on the kitchen table. Say as well that I know that I can reach these shakers. (Again, Strohminger is not trying to answer the modal skeptic; she assumes we have modal knowledge, and this is an instance of that knowledge.) Strohminger contends that this knowledge should be considered perceptual. Merely by looking at these shakers, I form the belief that I can reach them. Thus, this belief is psychologically immediate. The process that produces it involves no inferences. Importantly, this process is “very reliable” (Strohminger 2015, 367). I can often reach that which I deem reachable. And, fnally, this belief can take the necessary modifers that mark it as perceptual knowledge. I can see “through my glasses” that the salt and pepper shakers are reachable (Strohminger 2015, 367). Here then is an instance of my having perceptual knowledge of my own ability. Knowledge of one’s abilities entails knowledge of what is metaphysically possible. “[A] claim of the form, ‘S can ϕ’ (‘I can reach the mug’), entails the possibility claim, ‘It is metaphysically possible that S ϕ’ (It is metaphysically possible that I reach the mug’)” (Strohminger 2015, 367). So, I know that it is possible for me to reach the salt and pepper. Strohminger’s second argument comes from an analogy with bird watching. When I am just learning how to identify a particular species,

Crossing Rivers  65 I will refer to my bird books and attend to an individual bird’s markings, size, and habits, and then infer its identity from what I see and have learned about this bird. From its red patch and mottled back, I infer that the bird before me is a flicker. But after correctly identifying enough flickers, I no longer need to consult the description of the bird, matching what I see to what I read. Such inferences are no longer necessary. I can just see that a bird is a flicker. Say for good reasons we accept arguments of the following form.3 The house is black. The house could have been some color other than its actual color. Thus, Possibly, the house is not black. (Strohminger 2015, 368) Assume that some state of affairs prompts me via my senses to form a belief in the first premise. If so, Strohminger argues, we can train ourselves to form a belief in the conclusion when prompted by that very same state of affairs. In the presence of a black house, the belief that “the house is black” is psychologically immediate. But I can work to develop the habit of believing that the house could have been yellow in the presence of that same house and if successful, that belief will in time become psychologically immediate. And in so far as the above argument is valid and in so far as this psychologically immediate belief can be modified by a modal auxiliary, this belief then counts as perceptual knowledge.4 Thus, just as I no longer need to make an argument that justifies my identification of a bird as a flicker and I can instead just see it as a flicker, Strohminger argues, after some training, I can forgo rehearsing the above argument and I can instead just see that the black house before me could have been yellow.5 In neither argument is Strohminger claiming that possibilities appear before us to be perceived (Strohminger, p. 371). Nor is she claiming that there is modal content to our sensory experiences. She instead strives to be neutral on such matters. That said, she does hold that “In general, what we can objectually see or otherwise perceive does not neatly align [with] what we can propositionally see or otherwise perceive” (Strohminger, p. 371). Here we see in Strohminger’s position a debt to Allan Millar. What is “perceptually manifest” to us, while lacking in modal content, can prompt me to form beliefs with modal content (Millar 2000, 74). The look of the shakers arrayed before me prompts the belief that they are reachable and from that it follows that it is possible for me to reach the salt. The look of the house prompts me to believe that it could have been painted something other than black. In both cases, I see these possibilities, but in both cases what it is to see is to form a psychologically immediate belief in these possibilities. Finally, while the above gives us reason to think that some modal knowledge can be perceptual knowledge in the stipulated sense, Strohminger is not claiming that perception can immediately justify the

66

Rebecca Hanrahan

modal belief at issue. Here again Strohminger would like to remain neutral. “If you know something by perception, then you are (at least on standard views) justifed in believing it to be true. But it does not follow that you are immediately justifed in believing it to be true” (Strohminger 2015, 372). For justifcation to be immediate, it can’t rest on any other knowledge or justifed belief you hold. So, I take her neutrality to mean the following: What I perceive will at times prompt me to form psychologically immediate modal beliefs. What justifes these beliefs might be other justifed beliefs I hold, in which case the justifcation for these beliefs will be mediated. But it might be that the justifcation for these beliefs doesn’t rest on other justifed beliefs, in which case these beliefs will be immediately justifed. But, importantly, in both cases, these beliefs count as perceptual knowledge. In my objections to Strohminger’s account, I will focus on this relationship between perceptual knowledge and immediate/mediate justifcation. Strohminger sees two benefts to her account. The frst is that it provides us with a way to respond to those who argue that evolutionary forces would not select for modal knowledge. If an individual is being hunted by a crocodile, she greatly increases her chances of survival if she can accurately assess whether the tree before her is climbable. Given this evolutionary advantage for those who live in crocodile-infested territories, there will be selective pressure in favor of those who accurately make this assessment. Moreover, what abilities an individual has are relative to her environment. Trees aren’t in and of themselves climbable; some trees are climbable for some individuals under some conditions. So, this individual needs to assess which of the trees before her she can climb and she best do so quickly, for crocodiles are fast. Thus, it is to our evolutionary advantage that we just see what we can do within our environment. And in so far as modal claims are entailed by claims about abilities, similar sorts of evolutionary arguments can be made about the capacity to just see what possibilities are before us. The second beneft of this account is that, in Strohminger’s assessment, it is more psychologically realistic than other modal epistemologies. Some hold that the imagination is a guide to possibility,6 yet my judgment that it is possible for me to reach the salt isn’t preceded by my calling forth images of my extending my arm. In looking at the salt, I judge that it is reachable and hence possible for me to reach. Similarly, some hold that it’s via following out the implications of a given statement that I determine its modal status.7 But I don’t entertain the proposition, “This ball is kickable,” and then follow out the implications of its truth in search of a contradiction, prior to concluding that that ball is kickable. I just see that it is kickable. As Strohminger readily admits, this theory of modal epistemology doesn’t account for all of our modal knowledge. While I can see that it is possible for me to reach that salt, I can’t see that it is possible for the

Crossing Rivers

67

mind to be separate from the body. Especially in so far as it is supported by evolutionary considerations, this modal epistemology is best suited to account for our knowledge of a restricted species of what Barbara Vetter refers to as “real or objective possibility” as opposed to doxastic or epistemic possibilities (Vetter 2018, 3). These possibilities are keyed to variations in our current environment and they allow for “real options for acting in the world” such that knowledge of them can be to our evolutionary advantage (Vetter 2018, 3). The dead tree limb hanging over my house is swaying gently in the wind, but I see that it is a real possibility for it to break off in a bad storm, so I best get the tree doctor out to the house. Having now worked through the basics of Strohminger’s theory, I want to return to her arguments for considering such modal knowledge to be perceptual. I will challenge Strohminger’s position that a belief counts as perceptual knowledge, regardless of whether its justifcation is mediated or immediate. I will argue that since the process she describes relies on our endorsing arguments and forming habits, the justifcation for the beliefs at issue should be considered mediated and hence, contra Strohminger, the knowledge we acquire shouldn’t be considered perceptual. Moreover, I question how psychologically realistic this account is. I will then consider an account of perceptual modal knowledge that jettisons Strohminger’s talk of endorsing arguments and forming habits, and instead opts for a more fully externalist account of such modal knowledge. On this account, we are prompted by what we perceive to form beliefs that are both psychologically and justifcatorily immediate. While such beliefs might best be considered perceptual knowledge, as we will see, this theory faces its own problems, requiring us to reaffrm the imagination’s role in justifying these modal beliefs.

3.3 Immediate versus Mediate Modal Perceptual Knowledge and the Psychologically Realistic Consider another example Strohminger offers: I can sometimes see that two objects in my visual feld are distinct. I can see that the glass is not the mug. If I accept the Principle of the Necessity of Distinctness, I might reason to the conclusion that it is necessary that the glass is not the mug. This reasoning can become unconscious or automatic by habit. Whether the reasoning is conscious or unconscious, it can yield knowledge. Eventually I can just see that it is necessary that the glass is not the mug. (Strohminger 2015, 369) As I read this quotation, Strohminger is describing a process whereby non-perceptual knowledge becomes perceptual knowledge. Prior to my

68

Rebecca Hanrahan

reasoning becoming “unconscious or automatic,” the belief I form in the necessary distinctness of the glass from the mug is not perceptual. For I must reason to the conclusion that the two are necessarily distinct. At this point, this belief seems to be neither psychologically nor justifcatorily immediate. But after this reasoning becomes unconscious or habitual, then eventually I just “see” this necessity. Psychologically, I believe immediately that the two are necessarily distinct. But what role does my reasoning play now that this belief has become psychologically immediate? Does it render my justifcation for this belief immediate or mediated? Oddly, Strohminger does not speak to these questions. But in so far as my habit here is a consequence of my repeated rehearsal of my reasons, including my endorsement of the Principle of the Necessity of Distinctness, it would seem as if the justifcation for this belief should be considered mediated. Moreover, if it were not, then it would seem as if the only times my justifcation would be considered mediated is when I explicitly articulate my reasonings for adopting the belief at issue. But this is just too narrow a conception of mediated justifcation. Thus, after I form the appropriate habit, my belief that the mug is necessarily distinct from the cup is a psychologically immediate but justifcatorily mediated belief. But why should such a belief be considered perceptual knowledge? Consider two individuals, one who has developed the appropriate habit and the other who has not. If Strohminger’s position is correct, the frst has perceptual knowledge, while the second does not, even though neither would possess knowledge if their reasoning was faulty and hence the argument at play was invalid. But if this is the case, then that argument is doing important justifcatory work for both individuals. Admittedly, in the second case, the person comes to hold this belief because he has just worked through the argument at issue. And the frst comes to hold this belief because she is prompted to form this belief by what she perceives. But she is so prompted only because she has reaffrmed this argument so often. Why would the fact that she has generated this habit change the nature of the knowledge she acquires? If the second person does not have perceptual knowledge then neither should the frst, especially given that both depend on the same argument and the habit at issue arose only because of the repeated endorsement of that argument.8 Moreover, even if the frst person does in fact have perceptual knowledge in this instance, the process of acquiring modal knowledge that Strohminger describes can’t be considered psychologically realistic. On Strohminger’s process, an individual must endorse the Principle of the Necessity of Distinctness, rehearse the argument which establishes that two objects must necessarily be distinct from each other until it becomes a habit, see a mug and a glass, and then and only then is her claim that “the mug can’t possibly be the glass” considered perceptual knowledge. If this is what is required, only a few philosophy majors can see that

Crossing Rivers

69

the mug is necessarily not the glass. When forming our modal beliefs, Strohminger claims, people seldom employ the processes described by modal epistemologists. These beliefs aren’t often preceded by our employing our imagination or tracing out implications. Whether or not this is true, we certainly don’t engage in the process Strohminger describes. As even Strohminger admits, this process would be a “strange training regime” for a person to adopt but would nonetheless be possible.9, 10 So, if psychological realism is a priority, why not drop talk of habit and endorsing arguments and/or principles? Why not focus on whether perception can produce in us beliefs that are not only psychologically immediate, but as well justifcatorily immediate? My friend just painted her house an ugly brown and when I look at it, I think she should have chosen a different color. Thus, in my mind, it is possible for the house to be painted a vibrant blue or a soft yellow. This belief was psychologically immediate. It was spontaneously produced in me by what I saw. If it is as well justifcatorily immediate, if I am also justifed in believing it as a result of my having perceived this ugly brown house, then this would be a clear case of my having perceptual knowledge of this possibility. Return to the assumption that perception is not a source of modal knowledge. I suspect that, of those philosophers who accept this assumption, few would object to the notion that perception could cause us to spontaneously form modal beliefs. What they reject is that upon fnding ourselves with such beliefs as a consequence of what we see, these beliefs would be in no need of further justifcation to be considered knowledge. But in describing a process of acquiring modal knowledge that depends on our endorsing arguments and forming habits, it would seem as if Strohminger has offered up the very justifcation that philosophers who accept this assumption require. So at the outset of her paper, it seems as if Strohminger was going to argue against this assumption, but the process she ends up describing is seemingly in line with this assumption. But by dropping references to habits, arguments, and hence mediated justifcation, we will be considering a theory that holds that perception alone can provide us with modal knowledge, a theory that holds that via perception I can form modal beliefs that are both psychologically and justifcatorily immediate. Only that sort of position undermines the assumption under consideration. Such a theory is externalist in nature. In outline, it would take the following form: Evolution has instilled in us the disposition to see some of what is possible and some of what is necessary in our current environment. That is, without engaging in any inferential process, I spontaneously form beliefs via my perceptual faculties about what is possible and what is necessary with respect to my surroundings. The perceptual process that produces these beliefs is reliable, such that much of the time when, for example, I see that it is possible to reach the salt and I act on that belief, I fnd the shaker in my grasp. The arguments that justify

70 Rebecca Hanrahan these beliefs might depend on or include principles like the Necessity of Distinctness or the relationship between abilities and metaphysical possibility, but they will probably depend on the fact that these beliefs are consistently true. But I need not have any beliefs about or knowledge of these arguments. Whatever justifes these modal beliefs need not be accessible to me. But because such arguments hold, I am justifed in holding these beliefs. Moreover, since the beliefs in question are psychologically immediate products of perceptions, these beliefs count as perceptual modal knowledge. This theory of modal knowledge seems to at once capture the most salient aspect of Strohminger’s view, such as the emphasis on perception, reliability, and evolution, while avoiding what I contend is both philosophically problematic and psychologically unrealistic, i.e., that these beliefs count as perceptual knowledge even though their justifcation is mediated by arguments we have repeatedly endorsed. But since it is externalist in nature, it also faces problems. The process it describes will at times result in us believing in possibilities that aren’t possibilities for us. And it is to this problem that I turn next.

3.4 Evolution and My Utter Failure to Climb Trees I am an utter failure when it comes to climbing trees. What is worse, I am very bad at assessing my ability to climb trees. That is, I have seen many trees that I thought I could climb and when I tried to climb them, I have failed, horribly, miserably, and repeatedly. It can therefore be said that the perceptual process that compels me to form beliefs about the possibility of my climbing a tree is not reliable. I look at a tree and spontaneously form the belief that it is possible for me to climb that tree. But that I hold this belief is no indication that it is in fact possible for me to climb that tree. I shouldn’t take my psychologically immediate belief that I can climb that tree (which has arisen in me due to my looking at that tree) as reason to think that I can climb that tree. In accordance with the theory I outlined above, let us say that evolution has gifted us with the disposition to form modal beliefs in response to some of what we see. These beliefs are psychologically immediate. Sometimes these beliefs consistently match what is possible, as is the case with saltshakers. But sometimes these beliefs don’t match what is possible, as is the case with my beliefs about trees. What can be said about this disparity? Here I want to focus on my weird confdence that it is possible for me to climb trees that I can’t actually climb. If it is to my evolutionary advantage to know what it is possible for me to do within my current environment, then why aren’t I markedly better at assessing whether it is possible for me to get up in that tree? Why do my beliefs lead me to consistently conclude that I can do something when I can’t, which is surely to my evolutionary disadvantage?

Crossing Rivers

71

The abilities and dispositions evolution gifted us with are the ones that enabled our ancestors and our ancestors’ ancestors to survive and reproduce. In most cases, but not all, they will enable us now to survive and reproduce. There are, though, evolutionary leftovers, vestigial traits, and dispositions that were once conducive to survival but are no longer so. When we were covered with hair, goosebumps allowed us to better trap our body heat on cold days. Without the hair, our goosebumps have lost their evolutionary function. At the cognitive level, our fear of the dark is also considered a vestigial trait. When we were not at the top of the food chain, our predators hunted us at night. Fear of the dark was to our beneft. With the invention of reliable light sources and fortifed structures, this fear produces a lot of unwanted anxiety. Thus, my disposition to believe that it is possible for me to climb the trees around me might be cognitive residue from a time when my ancestors needed to and could climb trees. As I noted earlier, in a treerich, crocodile-infested environment, it is to an individual’s advantage to be able to quickly assess whether it is possible to get up a nearby tree. Say this was the environment my way-back ancestors found themselves in. And hence over generations, they got better and better at making just this kind of assessment. What they saw prompted them to form the psychologically immediate beliefs that the various trees around them were climbable and these beliefs were consistently true. Hence, given the theory at issue, our ancestors had perceptual modal knowledge about whether or not it was possible for them to climb the trees around them. But let’s say in time our ancestors moved out of this environment; after all, there were just too many crocodiles in the neighborhood. Because they no longer lived around crocodiles, they no longer needed to climb so many trees, so the ability to climb trees was no longer selected for. But the evolutionary forces at play here need not have simultaneously worked on their cognitive dispositions. Since I and my more recent kin don’t need to climb as many trees, our mistaken beliefs won’t be selected against. I can wrongly identify a tree as one I could climb and pay no evolutionary cost.11, 12 So, if the disposition to form beliefs about what is possible within my current environment is instilled in us by evolutionary forces, we can see how and why those beliefs might not match what is possible. The mental capacities evolution provided us with might not be aligned with our current environment or our current abilities. Some might challenge this analysis, specifcally my claim that I made a mistake in my assessment of those trees. I saw a tree and I formed the psychologically immediate belief that it was possible to climb that tree that I then failed to climb it. But it could still be possible to climb that tree, even if I didn’t or couldn’t at that moment. Thus, we have no reason yet to say I made a mistake. In response to this, it is worth reminding ourselves of the kind of possibility at play here. This is a restricted species of “real or objective possibilities” (Vetter 2018, 3). These possibilities are

72

Rebecca Hanrahan

such that to have knowledge of them is to our evolutionary advantage. It does me no evolutionary good to know that it is possible for me to climb that tree in some world in which I am markedly stronger, had prehensile toes, and gravity’s pull was lessened. The possibilities I need to know about are those that allow me to act for my advantage in the world that I fnd myself in. Evolution might have instilled in me the disposition to believe that something is really possible, but given the changes in me or in my environment, I am no longer justifed in believing in these possibilities. These beliefs are consistently wrong, for the possibility at issue is not a real possibility for me. Thus, I would be mistaken to take my belief that the tree is climbable to indicate that I can in fact climb it. But this disposition, acting in my evolutionary ancestors, would have produced in her the same beliefs (under the appropriate conditions) and she would be justifed in believing it was possible for her to climb the trees she saw. For her beliefs in these cases do match the real possibilities available to her. But if this is so, then in some sense the objection raised above is right. Based on what I saw, I formed the belief that it was possible to climb the tree before me. And it is possible to climb that tree. It is just not a possibility for me, but rather it is a possibility for my evolutionary kin. Here it might help to bring in the notion of affordances. Affordances are collections of properties or relations that provide opportunities for those that can perceive them. “The affordances of the environment are what it offers animals, what it provides or furnishes, for good or ill” (Gibson 1986, 68). The apple affords me the possibility of a nutritious, energy-rich meal. The hill affords me a perch where it will be possible to scan the land for predators or prey. What affordances are available to me are relative to me and my environment. Coming to a river’s edge, the rocks afford me a path across, but they do so because of the length of my stride as well as how the rocks are spaced. For someone shorter than me, these rocks might not afford them the same opportunity to cross. It is to my and my evolutionary kin’s advantage to identify what possibilities our environment affords us. But I am arguing that because of how evolution works, the affordances I can identify in my environment might not always match the real possibilities available to me. Evolution might have allowed me to identify affordances that would have been relevant to my kin’s survival, though they might be a detriment to mine. So, I will have correctly identifed an affordance. It is possible to climb that tree. But it is not an affordance for me. It isn’t a real possibility that I climb that tree. But this is where the imagination comes in.

3.5 Crossing Rivers and the Imagination I once spent three days trying to cross a river. I did eventually make it across, but only because I and my hiking companions were very careful.

Crossing Rivers

73

Over the years, I have learned how to safely cross a river. Can you hear rocks being tumbled? Well, it is best to look for another place to cross. Is there a way to zig zag up and across a stream by looking for bands of rocks that run the length of a river section? Is there a place where it braids or fattens? With rivers of certain sizes and fow it is best to scout out your route, considering and assessing alternatives. But this all involves acts of the imagination. Hiking up and down that river for three days, we were imagining various ways across and looking for dangers, deep spots, or tree snags. Let’s assume that what I see prompts me to form modal beliefs. But from the examples of trees and saltshakers, we learn that while some of those beliefs are consistently true, some are consistently false (though they would have been consistently true, if it had been my evolutionary kin in my shoes). And of course, between these extremes will be beliefs about which there is no defnable pattern and that’s what brings us to the river. So, at the side of the river, what I see might prompt me to believe that it is possible for me to cross this river. But I have learned to set aside those initial beliefs. I need to do the work of the imagination to chart out various ways across. Maybe I use my sensory imagination and try to picture myself making my way across and/or I describe my route across, trying to indicate where to be careful and where things might go wrong. What my imagination provides me with are the “real options for acting in the world,” that is possible routes that will get me across the river, safely, and it is these options that I choose between when I step in that river (Vetter 2018, 3). Relatedly, when I am challenged on my choice of routes, it is my ability to describe and hence picture that route that will silence my detractors or force me to reconsider my route. I might not have adequately considered the diffculties of getting up the bank or my companions might not have seen how by looking upstream they can connect various steps together to get across. So, my initial beliefs in what was possible, though prompted by what I saw, were not enough to justify my plunging into that river. It was my imagination that revealed to me the real possibilities before me. It is also my imagination that allows me to justify to others my course of action. And here we can see this account dovetailing with my theory of modal epistemology.13 On my theory, the imagination provides us with a guide to possible variations in our current environment. Specifcally, I argue that if I can imagine some variation in my current environment that I can account for as a veridical perception via an inference to my best explanation, then I have reason to take what I have imagined as justifying my belief in the possibility of that variation. Sitting at my desk, I wonder if it is possible for my puppy to escape from my home. And I call forth images of him pushing through my front door. I argue that because I have called forth these images, I have reason not only to think that these very images could have come to my epistemic twin, but that she would with

74

Rebecca Hanrahan

good reason take these images to be veridical perceptions. My epistemic twin is an individual who exists in some possible world who from the frst-person perspective has in most every respect the same mental life as me. For the most part, she shares all of my beliefs, experiences, desires, and justifcatory practices. So, via her best explanation (which is as well my best explanation), my twin can account for these images with reference to a door that poorly latches and a new puppy who desperately wants to meet an old cat. Thus, my epistemic twin has reason to believe that her puppy has escaped. Moreover, since she is my epistemic twin, I know she would reason as I would, and I know that I can’t doubt her reliability without putting my own into question. So, I can trust her judgments, which means, I have reason to believe it is possible for my puppy to escape.14 What is worthy of note for our purposes here is how well suited this theory is to tell us about the real possibilities at issue here. On my theory, I am to imagine variations to my current environment, and I am to account for them as veridical perceptions via my best explanation of how this world works. If I succeed at accounting for these variations, I am justifed in believing in the possibility in question. The scope of this theory is clearly limited to worlds that are quite similar to the actual world, because of its dependence on inferences to the best explanation and because of its focus on variations to our current environment. But as a consequence, it can’t help but pick out those possibilities that are relevant to actions I may or may not take that can affect my well-being. So, at the river’s edge, say I see a line of rocks and boulders. I note that I am at the bend of the river. The water is thigh deep and at the inside of the curve there is a tree snag. While I can chart out a potential route across, given my understanding of currents, I also imagine falling in and getting swept into that tree snag. So, given that what I imagine is what my twin will experience (at least from the frst-person perspective), and given what she is experiencing is something I want to avoid, that is potentially drowning, I have reason to opt for another way across. So, it is the imagination that can enable me to identify which route I shouldn’t take and hence eventually which route I should take. Of course, this process won’t guarantee my safe passage. But there is reason to believe that it does increase my chance of survival. Over and over, it is the hasty decision that leads to problems in the backcountry.15 With all this in mind, let’s return to the question of whether perception on this theory provides us with modal knowledge. On the one hand, it could be argued that when it comes to the possibility of reaching such things as salt shakers, the beliefs we form should count as perceptual modal knowledge, for they are consistently true and prompted by what we see. But in cases of the climbability of trees and crossability of rivers, we don’t have modal knowledge merely as a consequence of being prompted to form a modal belief. For these beliefs aren’t consistently

Crossing Rivers

75

true. On the other hand, with respect to the modal beliefs themselves, there is nothing that marks one set of beliefs as the product of a reliable process and the other not. Experience might reveal which kinds of modal beliefs are reliable and which are not, but it need not. I might never have been tempted to climb a tree and hence have never learned that I am godawful at assessing their climbability. Worse yet, if the disposition to form consistently true modal beliefs is selected for, that means some of our ancestors were good at making these assessments and some were bad, and those who were bad were selected against. With respect to any particular kind of belief, we don’t know where we are in this journey to modal reliability, a journey that can reverse itself as we saw with the case of trees. So while evolution might have gifted us with a disposition to form certain modal beliefs, the above gives us reason to question whether we can say that we are justifed in holding these beliefs as a consequence. Admittedly, on an externalist account, we need not have access to that which justifes our beliefs. But I am questioning whether the process that produces these dispositions makes them such that even from an externalist perspective we can consider the beliefs they produce justifed. These dispositions are a product of forces working on my evolutionary ancestors living in a particular environment, forces that might not be acting on me now. Moreover, they are produced by a process that can reverse itself and must fail some of the time for some of us. Finally, the usual causal relationship that acts as a corrective when it comes to perceiving what is actual is not in play when it comes to what is possible. Whatever relationship I do or don’t have to that possibly black house is not the same relationship I have to that yellow house and this matters as to how and when I can be corrected in my modal belief. All this gives us reason to reject these immediately formed perceptual modal beliefs as justifed. And given that we want a modal epistemology that will tell us about real possibilities in our current environment, it also gives us reason to turn to a modal epistemology, such as mine, that is based on what we currently perceive and believe, that depends on the imagination, and that justifes our beliefs as to what is possible by way of an inference to our best explanation.

3.6 Conclusion Strohminger has argued that some of our modal knowledge should be considered perceptual knowledge. After we endorse certain arguments and develop the right habits, we can just see that the house could be yellow and that a tree is climbable. I have argued that in so far as this knowledge does rest on arguments and habits, then this knowledge shouldn’t be considered perceptual knowledge. And even if it should, such a theory is far from psychologically realistic.

76 Rebecca Hanrahan I then turned to a theory of perceptual modal knowledge that doesn’t locate the justifcation of our modal beliefs in arguments we affrm and habits we form. On this theory, evolution has instilled in us the disposition to form modal beliefs that are conducive to our survival when prompted by what we perceive, though what justifes these beliefs need not be accessible to us. I have argued against this theory. Because the dispositions to form these modal beliefs are instilled in us by evolution, the beliefs they produce need not track what is really possible for us. Thus, I  have argued, we shouldn’t consider the modal beliefs produced by these dispositions justifed merely be the fact that we are prompted to believe them by what we see. Instead, we should turn to our imagination to tell us what in our current situation is possible. At the river’s edge, I might be prompted to believe that it is possible to cross. But instead of stepping in, I consider that river and I imagine what would happen if I take one route over another. In arguing for the imagination and against our having perceptual modal knowledge, I am reaffrming the division of labor noted at the outset of the chapter. Perception doesn’t tell us what is possible; the imagination does. It tells us what is possible, even – and especially – when the kind of possibility at issue is real possibility.

Notes 1 Hart (1988, 10), Peacocke (1985, 31), and Yablo (1993, 7). 2 Hanrahan (2017). 3 Strohminger does not explain why we are justifed in accepting this argument nor does she explain why we are justifed in inferring from an ability claim to a possibility claim. But let’s assume with her that we are so justifed. 4 Strohminger notes that such conclusions can be modifed by modal auxiliaries. Just as I can see that the house is black, “I can also see that it could have been yellow” (Strohminger 2015, 368). But can such conclusions be modifed by “gaze-related directional prepositions” (Strohminger 2015, 365)? It is odd to say that “I saw through my window that the house could have been black.” Or that “I saw through my binoculars that the elk could have been standing.” If my linguistic intuitions are right, this creates a problem for Strohminger’s theory. For Strohminger, the fact that a claim can be modifed by gaze-related directional prepositions is not suffcient to establish that that claim counts as perceptual knowledge. But she does use the fact that my ability claims can be so modifed to support her position that these claims are perceptual. Moreover, including gaze-related directional prepositions enables us to exclude certain knowledge from being perceptual that we should want to exclude. My logician friends “can just see” whether certain proofs work. I take it as a given that we would not want to include such knowledge as perceptual and we wouldn’t have to if modifcation by gaze-directional prepositions was a requirement for perceptual knowledge. For it would be wrong to say that these logicians see through their glasses that these proofs work. Thus, we have a reason to think that being modifed by such prepositions is a necessary condition on Strohminger’s theory. But if this is correct,

Crossing Rivers

5

6 7 8

9 10

11

12

13

77

then the fact that we don’t say “I saw through my binoculars that the elk could have been standing” is reason to not take my knowledge of what the elk could be doing as perceptual knowledge. Just to be clear, Strohminger isn’t arguing that how we come to identify birds is anything like how we come to make color judgments. Instead she is using the bird example to illustrate how with training we no longer refer to or consider the argument that justifes a belief we might hold. Instead, with the right perceptual input we jump straight to the belief. Such beliefs in her estimation count as perceptual knowledge. See Hanrahan (2007, 2009) and Kung (2010). See Chalmers (2002). Strohminger’s two arguments both depend on an individual endorsing some argument or making an inference that becomes habitual. She explicitly presents this argument in her discussion of how we can see that the black house could have been yellow. In regard to the saltshaker, once I see that it is reachable, there is a further inference that grounds my belief that “It is possible for me to reach the saltshaker.” And it is an inference, as Strohminger acknowledges, that rests on the principle that ability claims entail claims about what is metaphysically possible. Thus, I have perceptual knowledge of this possibility only after I endorsed this principle and developed the relevant habit. This comment was made in a referee report for this paper. An epistemological theory should describe a process of acquiring knowledge that is human-sized. That is, if we have knowledge, what it takes to acquire that knowledge should in some sense be compatible with human psychology. But that doesn’t mean that the process described must occur with or before a person endorses a modal claim. Nor does it mean that the only acts of the mind referred to within a theory must be those that occur with or before a person makes a modal claim. Strohminger is right that most of my modal beliefs don’t arise in me through an act of the imagination or inference. But that doesn’t make a theory of modal epistemology grounded in the imagination or inference psychologically unrealistic. Such theories are psychological unrealistic when at a minimum they describe what we can’t do or what we don’t ever do. And this can’t be said of theories of modal epistemology that are grounded in either the imagination or inference. Evolution weeds out quicker those traits that are no longer useful and are costly to maintain (and/or are easy genetically to weed out). It would seemingly be costly to maintain the muscle mass to climb trees, but not so costly to maintain the disposition to form certain beliefs. See https://www. sciencedaily.com/releases/2009/09/090908103904.htm. I have concentrated on an example of my believing that something is possible for me when it is not. But I could instead fail to form beliefs about possibilities before me, even when recognizing those possibilities would be to my evolutionary advantage. Koalas eat eucalyptus leaves that they themselves strip from branches. If koalas are given leaves separated from the branch, they may not eat them (https://koalainfo.com/does-koalas-lackintellectualabilities). One way of explaining this behavior is that while the koalas still see these leaves, they don’t recognize that these leaves could provide them with a good meal. Here I am trying to illustrate how evolution could instill in us the disposition to believe what we shouldn’t (as is the case with the trees) or it could fail to instill in us the disposition to believe what we should. We can also see this account dovetailing with Greg Currie’s theory of the imagination. Currie holds that the imagination is a kind of “internal

78

Rebecca Hanrahan

simulator” that allows us to safely test various strategies to achieve our ends (Currie 1995, 157). 14 See Hanrahan (2007, 2009). 15 The leading cause of death in national parks in the United States is drowning and most of these deaths are the result of bad judgment. See Ingraham (2015).

References Chalmers, David. 2002. “Does Conceivability Entail Possibility?” In Conceivability and Possibility. Edited by T. Gender & J. Hawtthorne, 145–200. Oxford: Clarendon. Currie, Gregory. 1995. “Imagination and Simulation: Aesthetics Meets Cognitive Science.” In Mental Simulation: Evaluations and Applications. Edited by Martin Davies and Tony Stone, 151–84. Cambridge: Wiley Blackwell. Gibson, James. 1986. The Ecological Approach to Visual Perception. New York: Psychology Press. Hanrahan, Rebecca. 2007. “Imagination and Possibility.” The Philosophical Forum 38(2): 125–46. Hanrahan, Rebecca. 2009. “Consciousness and Modal Empiricism.” Philosophia 37: 281–306. Hanrahan, Rebecca. 2017. “The Actual and the Possible.” Journal of Philosophical Research 42: 223–42. Hart, W.D. 1988. Engines of the Soul, 10. Cambridge: Cambridge University Press. Ingraham, Christopher. 2015. “Forget Bears: Here’s What Really Kills People in National Parks.” The Washington Post, August 12. Kung, Peter. 2010. “Imagining as a Guide to Possibility.” Philosophy and Phenomenological Research 81(3): 620–63. Millar, Allan. 2000. “The Scope of Perceptual Knowledge.” Philosophy 1: 73–88. National Evolutionary Synthesis Center (NESCent). “Evolutionary Fate of ‘Useless’ Traits; Why Some Traits Break Down Quickly While Others Persist Over Time.” Science Daily. www.sciencedaily.com/releases/2009/ 09/090908103904.htm Peacocke, Christopher. 1985. “Imagination, Experience, and Possibility: A Berkeleian View Defended.” In Essays on Berkeley. Edited by L. Foster and H. Robinson, 19–35. Oxford: Oxford University Press. Strohminger, Margot. 2015. “Perceptual Knowledge of Nonactual Possibilities.” Philosophical Perspectives 29: 363–75. Vetter, Barbara. 2018. “Perceiving Potentiality: A Metaphysics of Affordances.” Topoi 39(5): 1177–91. Yablo, Stephen. 1993. “Is Conceivability a Gide to Possibility?” Philosophy and Phenomenological Research 53(1): 1–42.

4

Imagination, Metaphysical Modality, and Modal Psychology Michael Omoge

4.1 Introduction We use imagination in many activities: pretense, thought experiments, counterfactual reasoning, modal judgments, and so on. However, in modal judgments notably, there is considerable skepticism about whether imagination can play the required epistemic role. Even here, using imagination to reach conclusions about practical modality (i.e., the kind that features in everyday reasoning) doesn’t elicit the same vivacious skepticism as using it to reach conclusions about metaphysical modality (i.e., the kind that features in sophisticated philosophical reasoning) does. For instance, if I say I can free climb the Yosemite’s El Capitan because I have imagined it and it is possible, you probably won’t object too much. But if I say something can be water without being H 2O because I have imagined it and it is possible, you are most likely going to object. In short, using imagination to reach conclusions about metaphysical modality elicits strong skepticism, and most of the views (e.g., Quine (1969); Blackburn (1993); Hill (1997); Van Inwagen (1998); Nichols (2006)) that express skepticism about whether imagination can deliver modal judgments target metaphysical not practical modality. Metaphysical modal judgments, or ‘metaphysical modalizing’ for short, is the bugbear, and this chapter aims to address this skepticism about it.1 Standardly, the approach taken to address the skepticism is often metaphysical 2 in that the focus is on whether imagination can enable agents to modalize (e.g., Yablo (1993); Chalmers (2002); Gregory (2004); McGinn (2004); Kung (2010)).3 The idea is to connect imagination with modality through some conceptual means, e.g., the possible worlds framework. This way, the offered epistemological explanations are often rationalistic in that they are independent of the empirical sciences that study imagination. And rationalism about modalizing seems to be one of the factors that underwrite skepticism about it in the frst place. This is because metaphysical modal facts are often (but not always) taken to be mind-independent, and so, it becomes unclear how a mind-dependent

80

Michael Omoge

cognitive faculty like imagination enables agents to modalize (see, e.g., Roca-Royes (2019)). Instead of this whether-response then, we might give a how-response, i.e., we can focus on how imagination might enable agents to modalize. This sort of response would be psychological in that it focuses on the process underlying our imaginative capacities when used to modalize. Shaun Nichols calls this approach ‘modal psychology’: In addition to metaphysical and epistemic questions about modality, there is another question, which is sometimes treated as a shameful relative of the others: What is involved, psychologically, in making modal judgments? That is, in addition to modal metaphysics and modal epistemology, one might study modal psychology. (2006, 237, original italics) Thus, the method of modal psychology is naturalistic, in that it is informed by the sciences that study imagination. Since the widespread skepticism about the usage of imagination to modalize is underwritten in part by rationalistic methods, and since modal epistemology is typically rationalistic, it would help if we step away from modal epistemology, as it is typically done, to modal psychology in our attempt to address this skepticism. That will be how I proceed in what follows. Specifcally, I intend to explain what is involved psychologically when we use imagination to modalize. In doing so, I set aside epistemological questions about whether we are justifed in relying on imagination in modalizing, or whether imagination has the required epistemic status to be so used. But though Nichols classifes modal psychology as a different project from modal epistemology, Daniel Nolan suggests that it is a project in ‘naturalized modal epistemology’4: If our discovery of modal facts is a process not dissimilar to our discovery of other, better understood, ranges of facts, then the investigation of ourselves and our capacities to respond to the world might indirectly shed some light on what we are responding to when we get the modal facts correct. (2017, 19, my italics) But why does modal psychology potentially resolve modal epistemological problems ‘indirectly’? Because [m]erely detecting what prompts modal judgments would not settle epistemological questions, but on the assumption that ordinary users of modality are doing a good job […] investigating how they in

Imagination, Metaphysical Modality

81

fact form their judgments should provide us defeasible accounts of what it takes to correctly form those judgments. (Nolan 2017, 19) Thus, besides explaining what is involved psychologically when we use imagination to modalize, I want to use the psychological explanation I will develop as part of an effort to give a defeasible account of what it takes for an agent to correctly use their imagination to modalize. In short, I consider my psychological explanation and the defeasible account it affords as offering a way to address the widespread skepticism about the usage of imagination in modalizing. In this way, following Nolan, my project in modal psychology is also a project in naturalized modal epistemology: it seeks to address a modal epistemological question naturalistically. I should also clarify that although psychologizing a realm of study often implies anti-realism about that realm, I will be agnostic about the metaphysics of metaphysical modality here, and so, my modal psychology shouldn’t be read as endorsing modal anti-realism. I would like to think that the psychological processes underlying realists’ and antirealists’ usages of imagination in modalizing are exactly the same. Describing those processes is my target here. To be safe, therefore, my task here can be taken as this conditional: if imagination does indeed enable agents to metaphysically modalize, then the view I will describe here provides the best way of accounting for this naturalistically. To give this naturalistic account, I will rely on Nichols and Stich’s (2003) cognitive theory of imagination, using it to explain what is involved psychologically when we use imagination to modalize (Section  4.2). Since their theory is speculative, my account too must be speculative, and, hence, might seem overly optimistic. But this isn’t just a problem for me; any attempt to psychologize imagination’s epistemic role will be speculative—imagination is one of our less understood cognitive faculties, compared to perception, say. Nonetheless, I will address this worry by psychologizing philosophers’ disagreements about what is metaphysically possible (Section 4.3). I will argue that by scrutinizing the psychological processes underlying their usages of imagination in modalizing, we can tell who among them is correctly using their imagination to modalize. This way, the account also addresses whatever skepticism might arise from such disagreements. Thereafter, I will respond to Nichols’ (2006) view that rather than providing an answer to the skeptics, modal psychology instead ends up promoting skepticism about the usage of imagination in modalizing, at least in the metaphysical domain5 (Section 4.4). I will show how his argument is faulty, and using my arguments in Sections 4.2 and 4.3 as additional premises, I will argue that modal psychology dispels rather than supports such skepticism.

82

Michael Omoge

4.2 The Psychology of Imagination and Metaphysical Modality The cognitive theory of imagination offered by Nichols and Stich focuses on propositional imagination (i.e., imagination as a propositional attitude that’s not necessarily imagistic), and I will here follow their lead. They put forth the theory in a bid to explain some practical usages of imagination, like mindreading and pretense. So, their examples involve everyday cases, like a child imagining that a banana is a telephone handset. But since my goal is to explain the metaphysical usage of imagination, I will apply what they say to examples of modalizing cases, like the possibility of phenomenal zombies. Nichols and Stich build their theory on a standard model of the basic architecture of the mind. The perception/belief/desire architecture is widely accepted among philosophers of mind as a model for almost all animals that possess some sort of central nervous system. For Nichols and Stich, building the architecture of imagination on this basic model requires three additional boxes: an ‘imagination box’, an ‘UpDater’, and a ‘script elaborator’. Talk of ‘boxes’ here, the employment of which is often called ‘boxology’, doesn’t imply that representation tokens in a box share a spatial location in the mind, but that they share an important cluster of causal properties which other representation types don’t share. Though some philosophers have raised concerns about the fruitfulness of boxology in theorizing about imagination (see, e.g., Stock (2011)), I will follow the boxology approach here. According to Nichols and Stich, the imagination box is a “workspace in which our cognitive system builds and temporarily stores representations of one or another possible [scenario]” (2003, 28). As they explain it, imagination episodes begin with an imagination premise, an arbitrary explicit input that kick-starts imaginative acts. Take the case of phenomenal zombies. The debate is about whether they are conceptually, and, therefore, metaphysically possible. David Chalmers gives the description of a (phenomenal) zombie this way: The creature is molecule for molecule identical to me, and identical in all the low-level properties postulated by a completed physics, but he lacks conscious experience entirely […] he is physically identical to me [but] none of this functioning will be accompanied by any real conscious experience. There will be no phenomenal feel. There is nothing it is like to be a zombie. (1996, 94–5) For simplicity’s sake, let us take the representation token, there are creatures that are physically identical to us but not conscious, as the imagination premise that kick-starts imaginative episodes about the possibility of zombies.

Imagination, Metaphysical Modality

83

Following Nichols and Stich, since the imagination box is the workspace where imaginings are built, this imagination premise would be put inside it. But the imagination premise won’t be suffcient to yield the possibility of zombies: depending on one’s pre-existing beliefs, many things or nothing could follow from it. Thus, Nichols and Stich say that “in addition to the [imagination] premise, the cognitive system puts [copies of the] entire contents of the belief box into the [imagination] box” (2003, 29). This, they say, ensures that there will be enough premises to yield the target imagining. The challenge, however, is that some pre-existing beliefs will most certainly be incompatible with the imagination premise, and so, they may block the target imagining from forming. Nichols and Stich say that the UpDater ensures that this doesn’t happen. The UpDater’s job, according to them, is to flter out pre-existing beliefs that are incompatible with the imagination premise. For instance, getting to the possibility of zombies from the combination of the imagination premise (there are creatures that are physically identical to us but not conscious) and one’s pre-existing beliefs would require revising some nomological beliefs about rationality. Perhaps one might have to be fexible about rationality, and that would require revising the belief that ‘consciousness is indispensable to rationality’. Likewise, if one wants to imagine physically impossible but metaphysically possible objects, many pre-existing beliefs about mass, gravity, and so on might have to be revised—impossible objects may just lack any known physical property altogether. Without the UpDater then, our inference mechanisms won’t have the relevant and appropriate premises (the imagination premise and the UpDater-fltered beliefs) to work with, in which case, our target imaginings mightn’t form. But often, we imagine very rich, detailed, and fantastical things for which it would be farfetched to say our pre-existing beliefs are suffcient as the required premises. For instance, I might imagine a zombie world alongside imagining the possibility of zombies, i.e., I might not just imagine individuals who lack consciousness in worlds like ours, but also imagine worlds very much unlike ours by virtue of lacking consciousness altogether.6 And it is very much possible that my pre-existing beliefs might not be able to get me all the way to imagining zombie worlds, even though they got me to imagining zombie individuals—modalizing involves intuitions that cannot be explained solely in terms of pre-existing beliefs. The natural question that comes to mind then is where the non-inferable details come from. Nichols and Stich’s answer is that they are supplied by the script elaborator. The script elaborator works to fll “in those details of [an imagination episode] that can’t be inferred from the [imagination] premise, the UpDater-fltered contents of the belief box and the [imaginer’s] knowledge of what has happened earlier on in the [imagination]” (2003, 35).

84 Michael Omoge But how does the script elaborator achieve this task? Here, Nichols and Stich bring in the notion of scripts from cognitive psychology (Schank and Abelson (1977); Abelson (1981)). According to Schank and Abelson, “a script is a structure that describes appropriate sequences of events in a particular context” (1977, 41). A script, they say, is generated for an event on account of the event’s repeatedness; e.g., by repeatedly going to restaurants, a restaurant script is generated.7 Initially, Schank and Abelson take scripts to explain how we act in a given scenario: your restaurant script is at work when you entered a restaurant, went through the routine, without much thought. But later, Abelson adds that there are scripts that guide how we understand a given scenario: “Scripts play a double role in psychology. There are scripts in understanding and scripts in behavior” (1981, 719). If I frequent Moroccan restaurants but you don’t, and so, I am acquainted with Moroccan eating etiquette, then I might be able to make sense of this short story and you mightn’t: “Seun went to a Moroccan restaurant, and after washing his hands, he left”. For you might ask: “Why didn’t he eat, why did he just wash his hands and leave?” Simply, your imagination is limited because your restaurant script isn’t as detailed as mine. It is part of Moroccan etiquette to eat with hands, and just by hearing that Seun left after washing his hands, I can fll in the gaps—that he already ate. Thus, a restaurant script can be a behavior script or an understanding script, depending on the imaginer’s goal. Another way to explain this story is to say that due to my frequenting Moroccan restaurants, I now have a restaurant script specifcally for Moroccan restaurants—a Moroccan-restaurant script, say—but you just have a (generic) restaurant script. The Moroccan-restaurant script is, therefore, a more detailed restaurant script. Notice, however, that even though the Moroccan-restaurant script is an understanding script in the story, a behavior script is implicit in it: were there no script for how people behave in Moroccan restaurants, even I would not be able to understand the story. Since the goal in modalizing is to understand modal scenarios, there are understanding scripts that detail how modal scenarios typically unfold. Let’s call them ‘modalizing scripts’.8 For example, to imagine whether zombies are possible, the relevant modalizing script (call it, a zombie script) is that which details how thoughts involving ‘consciousness’ typically unfold. Thus, just as the Moroccan-restaurant script is a more detailed restaurant script, the zombie script is a more detailed modalizing script, one that’s generated when an agent repeatedly uses their modalizing script to make sense of whether zombies are possible. So, Chalmers, say, would have a zombie script, whereas a freshman philosophy student would have only a modalizing script, i.e., where Chalmers can skip many steps when modalizing about zombies, the freshman

Imagination, Metaphysical Modality

85

mightn’t be able to. In addition, the zombie case (and arguably many others) is like the above story, i.e., a behavior script is implicit in it: were there no script for how humans, a fortiori, zombies behave, no modalizing script would work for understanding whether zombies are possible. More on this in Section 4.3. In my view, scripts play a much larger role in imagining, a fortiori in modalizing, than Nichols and Stich reckon. Not only can their operation explain where details that aren’t inferable from pre-existing beliefs and the imagination premise come from, as Nichols and Stich say, scripts can also inform prescribing the correct usage of imagination. More on this also in Section 4.3. What then do Nichols and Stich say about how scripts enable the script elaborator to supply those non-inferable details? They argue, frst, that scripts supply some of the details by supplying intuitions that aren’t produced by explicit reasoning in imagination. They say ‘some’ because scripts are non-restrictive, in that we can deviate from them and imagine what they couldn’t inform. I could also imagine that Seun stood while eating, and this wouldn’t be informed by my Moroccan-restaurant script, since it is part of Moroccan etiquette to eat while sitting on mats. I can imagine whatever I want. So, Nichols and Stich argue, second, that there must be a mechanism, which teases deviations out from scripts, and they take these deviations to be the details that are left which scripts do not supply. Hence, the script elaborator. Thus, with my zombie script, my script elaborator will be able to tease out any other details, which are neither supplied by scripts nor inferable from the combination of the imagination premise and my pre-existing beliefs. This, it seems to me, is a plausible description of how imagination yields the modalizing episode that zombies and zombie worlds are possible. Simply, we can explain how imagination enables agents to have metaphysical modal judgments (i.e., modalize) through modal psychology. As things stand, why a philosopher and a physicist, say, might have fundamentally different imaginative contents about the possibility of zombies seems clear enough. Given the same imagination premise (there are creatures that are physically identical to us but not conscious), the philosopher might get to the possibility of ‘phenomenal’ zombies, and the physicist, to the kind of zombies depicted in Hollywood movies, which Chalmers (1996, 95) calls ‘psychological’ zombies. Simply, the philosopher’s zombie script has more details than the physicist’s, if they are even using the same modalizing script. But the current analysis won’t work for why two philosophers disagree about the possibility of (phenomenal) zombies, since they presumably have equally detailed zombie scripts. And this seems to be a general fact about modalizing: philosophers often have equally detailed modalizing scripts and still disagree about whether the given modal scenario is possible. What then?

86

Michael Omoge

4.3 The Psychology of Imagination and Disagreements about Metaphysical Modality Talk of having more detailed scripts implies that what is imaginable is a function of domain-specifc knowledge. If I have more domain-specifc knowledge than you in a given area, then the content of my imagination in the area will be richer than yours. I could but you couldn’t imagine that Seun had eaten because my domain-specifc knowledge about Moroccan etiquette is more than yours. Mutatis mutandis for the philosopher and the physicist with regard to the zombie script. But two disagreeing philosophers presumably have equal domain-specifc knowledge about a modal scenario. Shoemaker (1999), who denies the possibility of zombies, presumably has equal domain-specifc knowledge in metaphysics and logic as Chalmers (1996), who countenances their possibility. Why then do they disagree if their zombie scripts are presumably equally detailed? Do such disagreements not give weight to the widespread skepticism about the usage of imagination in modalizing? One way of explaining the disagreements that ensue from the usage of imagination in modalizing comes to us from the independent discussions of Hawthorne (2002), Stalnaker (2002), and Braddon-Mitchell (2003) on the mind-body debate, which seem to suggest that much about the disagreements is semantic. The coherence of this semantic view is debatable, but it nonetheless represents what is, so far, a useful way of thinking about disagreements in modalizing. I will discuss Stalnaker’s version here.9 According to him, the disagreement between Chalmers and Shoemaker about the possibility of zombies arises because they disagree about what kind of world we live in. Chalmers says we live in a world where some properties, namely, phenomenal consciousness properties don’t supervene on the physical: for him, consciousness is irreducibly nonphysical. Shoemaker disagrees: for him, consciousness is physical/behavioral/ functional, and so, whatever phenomenal consciousness properties there could be in our world are already part of the subvenients. Stalnaker’s move (or, as Lycan (2007) explains him) then is this. We are phenomenally conscious, but we can’t tell in which world we are so: Chalmers’ or Shoemaker’s. So, Chalmers isn’t entitled to say ‘zombies are imaginable’, since that turns on us being phenomenally conscious in a Chalmersian world, i.e., in a world where phenomenal consciousness properties fail to supervene on the physical. Since Shoemaker denies this, he isn’t wrong then on semantic grounds that ‘zombies are unimaginable’. Thus, Stalnaker concludes: The issue concerns how theoretically loaded the idea of phenomenal consciousness is […] it seems clear that [Chalmers is] building some

Imagination, Metaphysical Modality

87

theoretical content into the concept of phenomenal consciousness [Shoemaker] uses the word ‘consciousness’ in a contrasting way, but a way that may also be theoretically loaded. (2002, 391 & 393–4) If so, then we can generalize: much about disagreements in modalizing is semantic, because the disagreements turn on how theoretically loaded relevant key concepts are. It might be said, however, that if this is true, then modalizing itself becomes an untrustworthy process, since philosophers have always been engaging in this sort of disagreements through modalizing, thereby treating the disagreements as substantive not merely semantic. I think this challenge works only if the disagreements are entirely semantic, which they aren’t. Lycan clarifes: “Stalnaker does not hold that the entire mind-body issue is semantic, because the dualist and the materialist disagree over [what kind of world] we live in” (2007, 475). If so, then it doesn’t automatically follow that modalizing becomes untrustworthy, for the non-semantic, a fortiori substantive, aspect of the disagreements might be underlaying philosophers’ engagements in them via modalizing all these years. What substantive aspect? Stalnaker is clear. The mind-body issue divides into two subdisagreements: whether (a) we live in a zombie world, and (b) zombie worlds are possible. Where (a) is factual—neuroscience can settle it— and so substantive, (b) is semantic—we can’t take an intergalactic space shuttle to zombie worlds to settle it. Since (a) is substantive, the relevant question, then, is whether it has always been underlaying philosophers’ engagements in the mind-body debate via modalizing, such that it gives some legitimacy to modalizing. I think it does. It is our attempt to answer the question of what kind of world we live in that led philosophers to postulate zombie worlds through modalizing— this, I take it, is why some of the best modal epistemologies were given in a bid to clarify issues in metaphysics and philosophy of mind (see Section 4.1, esp. n. 2). Simply, modalizing is inevitable if metaphysics is to deliver on its promise of unraveling the fundamental structure of the world (see, e.g., Lowe (2011)). With this epistemological challenge out of the way, let us return to our psychological analysis. As we have seen, Stalnaker’s reason for saying that much about disagreements in modalizing is semantic is that the disagreements are fueled by the theoretical baggage with which philosophers modalize. This gives us a promising link to the picture I have been painting about the psychological process underlying the usage of imagination in modalizing. Precisely, we have a link that connects us to scripts. Scripts, recall, are psychological paradigms that detail how repeated events typically unfold. What is relevant about them now is how they are activated.

88

Michael Omoge

Schank and Abelson say: To defne when a script should be called into play, script headers are necessary. The headers for the restaurant script are concepts having to do with hunger, restaurants, and so on in the context of a plan for action for getting fed […] The rules for activating a script are dependent on certain key concepts or conceptualizations when found in certain context. (1977, 46 & 48, my italics) Scripts are activated conceptually. Hence, I could imagine that Seun had eaten just by hearing ‘Moroccan restaurant’ in the relevant context. Schank and Abelson also note that “a script must be written from one’s particular role’s point of view. A customer sees a restaurant one way, and a cook sees it another way” (1977, 42, my italics). In other words, different restaurant scripts are activated for the cook and the customer, although the scripts share common headers/key concepts. I explained earlier (Section 4.2) that Schank and Abelson initially conceive scripts as behavior scripts; hence, their limiting what informs the cook’s and the customer’s points of view to the role they play in restaurant scenarios. I then explained that Abelson later adds that there are also understanding scripts. So, when the restaurant scripts are understanding scripts, what is informing the cook’s and the customer’s points of view would be their domain-specifc knowledge—the customer might be a food blogger, and so, her perspective on issues about restaurants will differ from the cook’s. In short, while sociological differences usually lead to different behavior scripts, theoretical differences usually lead to different understanding scripts. Since modalizing scripts are understanding scripts, and since disagreements in modalizing turn on the theoretical load of key concepts, different modalizing scripts will be activated for the disagreeing philosophers. For their points of view will be different since they are building different theoretical contents into, or applying different theoretical interpretations to, the same concepts. We shall see what this amounts to, shortly. For now, it suffces that Chalmers’ zombie script will be different from Shoemaker’s, although sharing key concepts. Since scripts guide the unfolding of imaginative events, different events will then be unfolded for them: for Chalmers, the possibility of zombies, but for Shoemaker, their impossibility.10 Thus, to answer the frst question we began this section with: even though their zombie scripts are presumably equally detailed, philosophers disagree about the possibility of zombies because their zombie scripts are different, thanks to their different theoretical baggage with which they modalize. This doesn’t mean that they just recycle their existing views when they modalize, however. For not only is imagination a form of reasoning (see Myers, this volume), it is clear from Section 4.2 that many inferences occur in modalizing, and so, new views are reached.

Imagination, Metaphysical Modality

89

What about the second question? Given that disagreements in modalizing boil down to the disagreeing philosophers using different but presumably equally detailed modalizing scripts, doesn’t this add to why we should question the reliability of imagination when so used? No. I will extend the ongoing analysis to offer one way of resolving this residual skepticism. Earlier I said scripts generate on account of events’ repeatedness (Section 4.2), but that was an oversimplifcation. Actually, what Schank and Abelson say is that for scripts to generate, events must be repeated in a (or rather, the same) sequence. You have a restaurant script because you repeatedly went to restaurants where you always ordered before you could eat. If so, then we can say more about the generation of scripts if we know what determines events’ sequences. Since events can be physical or mental, candidates would include perception and introspection: by repeatedly going to restaurants, a restaurant script is generated; by repeatedly thinking about zombies, a zombie script is generated. This then means that scripts can be compositional, i.e., two perceptually and/or introspectively formed scripts might be activated concurrently and sometimes (but not always) compose a new one. From my Moroccan-restaurant script and my script for singing karaoke, my cognitive system can generate a script for singing karaoke at Moroccan restaurants. It stands to reason that all modalizing scripts are compositional, and that some of them have behavior scripts as part of their components. Hence, I said earlier (Section 4.2) that behavior scripts are implicit in some modalizing scripts, and I gave the zombie script as an example. Let me now explain further. Chalmers thinks human actions are decomposable into phenomenal and functional descriptions, such that only the latter applies to zombies whereas both apply to us. Shoemaker disagrees: for him, human actions are inextricably both phenomenal and functional, and so, there can’t be zombies because that would mean we are zombies too, which we aren’t. Both Chalmers’ and Shoemaker’s zombie scripts thus have, as components, the behavior script for phenomenal actions and the behavior script for functional actions. Hence, as I said in Section 4.2, were there no scripts for how humans behave, there won’t be zombie scripts.11 If so, then we can explain how zombies are seen as possible for Chalmers but impossible for Shoemaker. Given the manner in which their zombie scripts are composed, which divides into at least two behavior scripts for Chalmers but one for Shoemaker, the imaginative event unfolds in two ways for Chalmers, such that one leads to the possibility of zombies; whereas, it unfolds in only one way for Shoemaker, and that way rules out the possibility of zombies. Simply, for compositional scripts, the manner of composition plays a big role. Another such role is informing a prescription of the correct usage of imagination.

90 Michael Omoge When scripts are compositional, there are often interferences, i.e., “states or actions which prevent the normal continuation of a script” (Schank and Abelson 1977, 52). Since interferences surface during the unfolding/elaboration of scripts, possible candidates are few. For instance, instructions like ‘suppose you were in a restaurant but there was no food’, which conficts with the normal unfolding of your restaurant script, drop out; they are rather the imagination premises on such occasions, since they are invitations for you to imagine. A good example of interferences, at least for understanding scripts, are propositional states that do not directly mesh with the agent’s domainspecifc knowledge. Importantly, Schank and Abelson use ‘interference’ normatively, in that they take interfering with (imaginative) goals as something bad. I will follow suit here. With interferences thus understood, we can give some prescriptions about the usage of imagination in modalizing. First, we have seen that stages in scripts are sequential, but that’s not all. Abelson adds that what matters is that the stages be connected causally. It is true that ordering a meal precedes eating it, but what matters is that in order to eat a meal in a restaurant, one must have ordered it. The former causally enables the latter to unfold: “The distinctive aspect of scripts is the relevance of learned associations between prior and consequent events. These associations are usually meaningful rather than rote because of causal ‘enablements’ between script events” (1981, 717). Now, suppose that during the composition of their zombie scripts, there were interferences for Chalmers but not Shoemaker. Further suppose that the interferences blocked the causal enablements between the stages of Chalmers’ phenomenal-action and functional-action scripts. Perhaps they blocked how phenomenal consciousness properties fail to supervene on the physical, such that he reached the conclusion that we live in a zombie world illicitly. Since such causal enablements are supposed to ground the composition of his zombie script in the frst place, the imaginative event that was unfolded for him would be problematic. The details his script elaborator supplied would be defective, and so, what it takes for him to correctly use imagination to modalize is hampered. Simply, he would be wrongly using imagination to modalize. Mutatis mutandis for Shoemaker. Second, Abelson adds: [E]vents in scripts differ in their centrality to the action fow; some events are indispensable to the script and summarize scenes consisting of lower level actions. ‘Ordering the meal’ and ‘eating the meal’, for example, are more central than are, say, ‘discussing the menu’ and ‘lifting the fork’. (1981, 718)

Imagination, Metaphysical Modality

91

Given what Stalnaker (2002, 388, n. 1) says, “whether the actual world is a zombie world” is more central to the zombie script than “whether there could be two discernible possible worlds ftting the description of a zombie world”, or “whether there could be physical laws that permit the emergence of nonphysical phenomena”. Now suppose that both Chalmers’ and Shoemaker’s zombie scripts have interferences, but that while the interferences are blocking the latter less central stages of Chalmers’ zombie script, they are blocking the former more central stage in Shoemaker’s. It would follow that the supplied script-elaborated details for Shoemaker would be much more defective than those supplied for Chalmers, and so, Chalmers would be correctly using imagination to modalize more than Stalnaker. Again, this works the other way around. Nonetheless, interferences are typically easily correctable. A waiter who brings you a burger after you ordered chicken wings, though interfering with the unfolding of your restaurant script, can easily be reminded about the correct order, and your restaurant script will continue to unfold normally. But not so for modalizing scripts. Due to the peculiar theoretical load of their key concepts, interferences often go unnoticed, and so, they aren’t easily correctable. This is what the theoretical load amounts to, and it explains why the ensuing disagreements run very deep. I, for one, am in Chalmers’ camp, but I could be wrongly using imagination to modalize, in that some interferences might be preventing the normal unfolding of my zombie script. Even now that I can see how this could be the case, I am no less convinced that my imagination leads me to the possibility of zombies. My view about the possibility of zombies and my usage of imagination to reach the view come apart. After all, if I am not using imagination to modalize but, say, I am using counterfactual reasoning,12 the ongoing psychological analysis mightn’t work—what is involved psychologically in imagination and counterfactual reasoning are most likely nonidentical. To be clear then, the analysis here isn’t about my or Chalmers’ or Shoemaker’s or any philosopher’s view on the mind-body (or any other metaphysical modal) debate, but about the usage of imagination to reach such views. Although ‘whether the views are correct’ and ‘whether imagination is correctly used to reach them’ are both epistemological questions, the former requires a metaphysical account, whereas the latter requires a psychological one (see Section 4.1). So, to reemphasize, ‘correctly use’, as I’ve been using it, qualifes imagining not modalizing.13 The crucial point, then, is that by taking a closer look at the psychology of imagination, one way of addressing whatever skepticism about the usage of imagination in modalizing that might arise from disagreements in modalizing opens up. If we shift our focus to scripts, we can tell who, between the disagreeing philosophers, is correctly using their imagination (i.e., in accordance with an interference-free modalizing

92

Michael Omoge

script) to modalize. If so, then the psychological account here offered sheds any exaggerated optimism it could be said to have. For it no longer just explains how imaginability yields possibility as with Chalmers, it also explains how it might yield impossibility as with Shoemaker, why it can go either or both ways, and how to tell who is correctly using their imagination to modalize when it does go both ways. Thus, even though the account is speculative, it caters for some of the necessary aspects of modalizing,14 and so, it is a step in the right direction, at least relative to naturalistic concerns. Simply, much of the skepticism about the usage of imagination in modalizing can be addressed through modal psychology.

4.4 Can We Really Psychologize Metaphysical Modality? Nichols (2006) raises one potential problem for this (and any) sort of imagination-based psychological analysis of metaphysical modality. According to him, specifying the architecture of imagination makes it clear that imagination is reliable only when used for practical not metaphysical modalizing: [I]f the modal psychology sketched here is right, it might contribute to a skeptical view about the epistemic status of imagination-driven intuitions of absolute modality. Given certain widely shared assumptions, like the assumption that the future will resemble the past, it’s plausible that the imagination provides a good guide to risk and opportunity. That’s in part what it’s designed to do. However, when this imaginative capacity is appropriated in the service of judgments of absolute possibility and necessity, it’s less clear that we can trust the verdicts. For here the psychological systems are being used outside their natural domain. Hence, there’s less reason to think that they will be successful guides in this foreign terrain of absolute modality. (2006, 253) As Sections 4.2 and 4.3 show, I began from the same psychological perspective as Nichols—he too relies on Nichols and Stich (2003). But unlike him, I arrived at a view that does not further promote skepticism about the usage of imagination to metaphysically modalize but rather helps to dispel it. What accounts for this difference? I will argue that his view that modal psychology supports skepticism relies on a faulty premise: (P) The reliability of a cognitive capacity, when used for a certain purpose, depends on the functions the capacity evolved to perform.15

Imagination, Metaphysical Modality

93

The goal is for the debunking argument I will give here to corroborate the vindicating one I’ve given in Sections 4.2 and 4.3. In what follows, the argumentative tone will differ from what has come before—still psychological, but more evolutionary than cognitive. The truth is this: we use many cognitive capacities outside their natural domain, and we don’t start distrusting their verdicts when they are used in those unnatural domains. Consider geometric reasoning. It is generally regarded as the product of some basic spatial cognitive capacity, like the one that facilitates our ducking to avoid bumping our heads, reaching for objects at a distance, and so on. Mateusz Hohol and Marcin Miłkowski put it better: an account of geometric reasoning “should recognize ‘hardwired,’ or evolutionarily early, cognitive abilities that are necessary to engage in Euclidean geometry, such as natural sensitivity to distance (proximal–distal), sense (left–right) and angle, and thus explain where geometric cognition comes from” (2019, 4). Even though geometry is an elaborated development of a function of spatial capacity, and even though it evidently wasn’t what the capacity evolved to perform, that doesn’t entitle us to start distrusting it. Geometry has been, is, and will continue to be very important to our understanding of the universe. In fact, some other sophisticated aspects of our life, like architecture and space-traveling, would become unworkable without it. It is just absurd to say that because spatial capacity evolved only to stop us from banging our heads on the tops of caves, we should have distrusted its verdicts when we started to use it for geometric reasoning. Our spatial capacity already got us into the geometry game, and even though the elementary geometry-ish things we were able to do with it, like cutting the distance between three L-shaped points by walking diagonally from one end of the L to the other, weren’t guaranteed to be right, we have disciplined them to such a degree that we can now do things that spatial reasoning could never do. From walking to cut distances between three L-shaped points, we have constructed the idea of triangles, and from there to quadratic equations, trigonometry, Pythagoras’ theorem, and so on. And it is highly doubtful that we would have arrived here had we not begun to use spatial capacity for geometric reasoning in addition to spatial reasoning. The point here is that the close relationship between practical and metaphysical modalizing as the natural and unnatural usages of imagination can be akin to the one between spatial and geometric reasoning as the natural and unnatural usages of spatial capacity. Imagination already got us to practically oriented modalizing, such that getting from there to metaphysically modalizing didn’t require evolving a separate capacity—this, I take it, is why it is common practice to begin discussions of metaphysical modality from practical modality (see n. 3). If so, then we shouldn’t distrust metaphysical modalizing just as we

94  Michael Omoge didn’t distrust geometry, since they are both unnatural usages of their respective capacities. Nichols might respond, however, that before we trust a given capacity in a new employment, we need some positive reason to think it is trustworthy. For instance, demonstrated success, which geometry has but metaphysical modalizing lacks. Also, evolutionary usefulness, which favors practical but not metaphysical modalizing. The idea here is that if we evolved a particular capacity, then presumably the functions it performed that got it selected was useful, or at the least, its selection was a function of demonstrated success. What then? Consider our capacity for sugar detection, which subserves detecting which food is rich in fructose and its ilk, and which food lacks them completely. While the former is basic, the latter is an add-on. Evidence abounds that we only initially evolved to detect sugar; detecting sugarless foods evolved much later when the former starts to be maladaptive as diabetes surfaced (e.g., Watve and Yajnik (2007)). Detecting sugarless food is a positive not negative function, and so, it requires action not inaction from the sugar detector (cf. Millikan 1984, esp. ch. 14). After all, there is a whole lot of biological connections that went into evolving detecting sugar-rich foods, which informs the thought that detecting sugarless foods would also require its own biological connections, or at least some modifications of existing ones, and so, it evolved later than detecting sugar-rich foods.16 Nonetheless, at the current stage in our evolutionary history, it is clear that if the sugar detector fails to deliver the whole package, we die off: sometimes, what we need isn’t more sugar but less and even no sugar at all. Now, geometry and metaphysical modalizing aren’t like this: we certainly won’t die if we couldn’t reason geometrically or metaphysically modalize. They are add-ons, which spatial capacity and imagination, respectively, need not deliver. Thus, the fact that geometry but not metaphysical modalizing has demonstrated success counts for less in judging whether we should trust either or both of them. What matters is that there was no selective pressure for the relevant capacity to deliver them, and so, they stand or fall together. Demonstrated success, therefore, fails to explain why we evolved using spatial capacity for geometric ­reasoning and imagination for metaphysical modalizing. In both cases, demonstrated success is post facto. If so, then how to determine the usefulness of a capacity’s functions becomes unclear. No doubt, practical modalizing (like geometry) has empirical usefulness, which differentiates it from metaphysical modalizing. But this can no longer be the reason metaphysical modalizing is untrustworthy. What we need is a reason why practical modalizing’s empirical usefulness renders it trustworthy, and metaphysical modalizing’s theoretical usefulness renders it untrustworthy. So, pointing to

Imagination, Metaphysical Modality

95

empiricality as the reason practical modalizing but not metaphysical modalizing is trustworthy becomes question-begging. Moreover, it isn’t even entirely true that metaphysical modalizing is only theoretically useful. Matteo Morganti puts it this way: “metaphysical conjectures and theories can turn into empirically relevant theses—at least in the sense that they become indirectly testable, i.e., relevant for the interpretation of science, at specifc junctures in the history of science (and philosophy)” (2015, 62). Thus, in some sense, metaphysical modalizing is also empirically useful, such that even if there is a way in which it isn’t question-begging and empiricality is the determinant factor for the usefulness of a capacity’s functions, then trusting practical modalizing but not metaphysical modalizing on the basis of empirical usefulness becomes arbitrary. In short, the original evolutionary functions of a capacity aren’t as signifcant in appraising the capacity’s verdicts when used for a newly evolved function, as Nichols might think. Certainly, our capacity to use imagination to metaphysically modalize evolved much later, perhaps with the evolution of logical reasoning,17 and we have already begun to use imagination to practically modalize by then, but that doesn’t mean imagination becomes unreliable when we now start to use it to metaphysically modalize. Timothy Williamson puts it summarily: The cognitive view of the imagination does not predict that it will be cognitively reliable only for tasks just like those it evolved to serve […] Whatever the function or evolutionary origin of our capacities, we are not forbidden to use them for other ends. (2016, 121) Although, like Williamson, I haven’t shown that reliability is sustained in this co-optation,18 it suffces that reliability isn’t also lost, which is what Nichols is saying. Simply, (P) is false, and so, Nichols’ view that builds on it falls apart. Modal psychology doesn’t “contribute to a skeptical view about the epistemic status of imagination-driven intuitions of [metaphysical] modality” (Nichols 2006, 253); nothing stands in the way of addressing such skepticism. I have offered one way that modal psychology does this in Sections 4.2 and 4.3. What we have here, then, is both vindicating and debunking arguments that even though modal psychology “does hold promise for partly constraining the theoretical space” (Nichols 2006, 253), it does so in a way that dispels rather than support skepticism about the usage of imagination in metaphysically modalizing. In so doing, my modal psychology responds to the demand for naturalizing how imagination enables agents to metaphysically modalize, which its rationalistic counterparts fail to apprehend as one important

96

Michael Omoge

factor that underwrites skepticism about using imagination in such a way. It does so by appropriating what was originally designed for practical modality for metaphysical modality, thereby removing any charge of exceptionalism. The psychological process underlying our usage of imagination is, by and large, the same for both practical and metaphysical modalizing. Modal psychology bears fruit in the philosophy of metaphysical modality.19

Notes 1 Unless otherwise stated, when I just say ‘modalizing’ hereafter, I mean metaphysical modalizing. 2 Hence, Fischer and Leon (2017) point out that, until recently, modal epistemology was done concurrently with metaphysics and philosophy of mind. 3 The approach taken here is usually to show, frst, that modalizing in general is a trustworthy process through practical modalizing, and then, second, that metaphysical modalizing is a standard application thereof. But, as I’ve said, only a few modal skeptics are against practical modalizing and/or modalizing in general; their skepticism often targets metaphysical modalizing specifcally. At any rate, while this approach is a viable option, I will isolate metaphysical modalizing here. 4 He gives two other lines of inquiry: the study of (i) the behavior of modal expressions in natural languages, and (ii) the roles modal claims/commitments play in our best scientifc theories. Thus, naturalized modal epistemology isn’t always psychological. 5 Although Nichols talks in terms of ‘absolute modality’, I take it that at least some metaphysical modalities (e.g., arithmetical ones) are absolute (cf. Hale 2012). Since his examples are arithmetical cases, I am therefore licensed to talk about him in terms of metaphysical modality. 6 Explicating what imagining zombie worlds amounts to exactly is, however, a diffcult and general problem. For it seems to require that one’s imaginative perspective is not part of the imagined world, which then makes it diffcult to see how one can have such imagining in the frst place. 7 It might be said, however, that scripts are unnecessary even for repeated events, since we are almost always engaging in active inferences (Kintsch and van Dijk 1975). See Abelson (1981) for a response to this challenge. 8 Not to be confused with scripts for modalizing, which are behavior scripts. Scripts for modalizing would encode information like “when people are modalizing, they often close their eyes or look off into the distance; they furrow their brow if they are having trouble with something; and so on.” They tell us how people behave when they modalize. 9 Although Stalnaker doesn’t name Chalmers and Shoemaker explicitly—he uses fctional characters—I will speak in terms of the real-life philosophers. 10 Here, I am taking it for granted that ‘the possibility of a scenario’ is a thought, and that thoughts are mental events (see Sellars 1981). 11 It might be said that the component scripts aren’t behavior but understanding scripts. Fine, but that just pushes my point one step back. For were there no scripts for how humans behave, there wouldn’t be any script for understanding the phenomenal and functional aspects of human behaviors, and ultimately, no zombie scripts. 12 See, e.g., Williamson (2007) for a rationalistic account of how counterfactual reasoning is a guide to possibility.

Imagination, Metaphysical Modality

97

13 Although one can infer correct or incorrect modalizing from correct or incorrect imagining, I am not making that inference here. To make such an inference, a process-reliabilist account of imaginative modal justifcation is required, which I have no space for now. See n. 14. 14 There are other important aspects I haven’t addressed here, but which can be potentially addressed through modal psychology as well; e.g., how imagination provides justifcation for metaphysical modal beliefs. 15 It might be said that I have set up a straw with (P), since it reminds us of ‘exaptation’, i.e., the process by which a trait previously shaped by natural selection for a particular function is co-opted for another use (Gould and Vrba 1982), which Nichols (see, e.g., page 64 of his 2003 work) is aware of. But given his statement that’s quoted above, there seems to be no better way to parse his talk of “designed to do” and “being appropriated in service of” than (P). Thus, to be fair to him, my arguments against (P) won’t rely on the seemingly uncontroversial truth of exaptation, even though they can be shortened by doing so. 16 Thanks to Peter Godfrey-Smith here. 17 See Woleński (2016) for an account of the evolution of logical reasoning. 18 Again (as in ns. 13 and 14), to show this, a process-reliabilist account of imaginative modal justifcation is required. 19 An earlier version of this chapter was presented at the Philosophical S ociety of Southern Africa Conference in Durban, South Africa. A signifcantly improved version was presented at the Fiction, Imagination, and Epistemology Conference in Bochum, Germany. The feedbacks from those presentations helped to clarify many parts of the chapter. I thank David Spurrett, Monique Whitaker, Adriano Palma, and Jacek Brzozowski for reading earlier drafts and providing numerous valuable comments. I specially thank Christopher Badura, Amy Kind, and an anonymous referee, whose comments and suggestions are so insightful that they have often been directly incorporated into the text.

References Abelson, Robert. 1981. “Psychological Status of the Script Concept.” American Psychologist 36(7): 715–29. Blackburn, Simon. 1993. Essays in Quasi-Realism. Oxford: Oxford University Press. Braddon-Mitchell, David. 2003. “Qualia and Analytical Conditionals.” Journal of Philosophy 100: 111–35. Chalmers, David. 1996. The Conscious Mind. Oxford: Oxford University Press. ———. 2002. “Does Conceivability Entail Possibility?” In Conceivability and Possibility, edited by Tamar Gendler and John Hawthorne, 145–200. Oxford: Oxford University Press. Fischer, Bob, and Felipe Leon. 2017. “Introduction.” In Modal Epistemology after Rationalism, edited by Bob Fischer and Felipe Leon, 1–6. Cham: Springer. Gould, Stephen, and Elisabeth Vrba. 1982. “Exaptation—A Missing Term in the Science of Form.” Paleobiology 8(1): 4–15. Gregory, Dominic. 2004. “Imagining Possibilities.” Philosophy and Phenomenological Research 69(2): 327–48. Hale, Bob. 2012. “What is Absolute Necessity?” Philosophia Scientiæ 16(2): 117–48.

98

Michael Omoge

Hawthorne, John. 2002. “Advice for Physicalists.” Philosophical Studies 108: 17–52. Hill, Christopher. 1997. “Imaginability, Conceivability, Possibility, and the Mind-Body Problem.” Philosophical Studies 84: 61–85. Hohol, Mateusz, and Marcin Miłkowski. 2019. “Cognitive Artifacts for Geometric Reasoning.” Foundation Science 24: 657–80. Kintsch, Walter, and Teun van Dijk. 1975. “Toward a Model of Text Comprehension and Production.” Psychological Review 85: 363–94. Kung, Peter. 2010. “Imagining as a Guide to Possibility.” Philosophy and Phenomenological Research 81: 620–63. Lowe, Jonathan. 2011. “The Rationality of Metaphysics.” Synthese 178: 99–109. Lycan, William. 2007. “Stalnaker on Zombies.” Philosophical Studies 133(3): 473–9. McGinn, Colin. 2004. Mindsight: Image, Dream, Meaning. Cambridge: Harvard University Press. Millikan, Ruth. 1984. Language, Thoughts, and Other Biological Concepts. Cambridge: MIT Press. Morganti, Matteo. 2015. “Science-based Metaphysics: On Some Recent Anti-metaphysical Claims”. Philosophia Scientiæ 19(1): 57–70. Nichols, Shaun. 2006. “Imaginative Blocks and Impossibility: An Essay in Modal Psychology.” In The Architecture of Imagination, edited by Shaun Nichols, 237–55. Oxford: Oxford University Press. Nichols, Shaun, and Stephen Stich. 2003. Mindreading. Oxford: Oxford University Press. Nolan, Daniel. 2017. “Naturalized Modal Epistemology.” In Modal Epistemology after Rationalism, edited by Bob Fischer and Felipe Leon, 7–28. Cham: Springer. Quine, Willard. 1969. Ontological Relativity and Other Essays. New York: Columbia University Press. Roca-Royes, Sonia. 2019. “The Integration Challenge.” In The Routledge Handbook of Modality, edited by Otavio Bueno and Scott Shalkowski. Oxford: Oxford University Press. Schank, Roger, and Robert Abelson. 1977. Scripts, Plans, Goals, and Understanding. Mahwah: L. Erlbaum Associates. Sellars, Wilfred. 1981. “Mental Events.” Philosophical Studies 39(4): 325–45. Shoemaker, Sydney. 1999. “On David Chalmers’ The Conscious Mind.” Philosophy and Phenomenological Research 59(2): 439–44. Stalnaker, Robert. 2002. “What is it Like to be a Zombie?” In Conceivability and Possibility, edited by Tamar Gendler and John Hawthorne, 385–400. Oxford: Oxford University Press. Stock, Kathleen. 2011. “Unpacking the Boxes.” In The Aesthetic Mind, edited by Elisabeth Schellekens and Peter Goldie, 268–82. New York: Oxford University Press. Van Inwagen, Peter. 1998. “Modal Epistemology.” Philosophical Studies 92: 67–84. Watve, Milind, and Chittaranjan Yajnik. 2007. “Evolutionary Origins of Insulin Resistance: A Behavioral Switch Hypothesis.” BMC Evolutionary Biology 7(61): https://doi.org/10.1186/1471-2148-7-61

Imagination, Metaphysical Modality

99

Williamson, Timothy. 2007. “Philosophical Knowledge and Knowledge of Counterfactuals.” Grazer Philosophiche Studien 74: 89–123. ———. 2016. “Knowing by Imagining.” In Knowledge through Imagining, edited by Amy Kind and Peter Kung, 113–23. Oxford: Oxford University Press. Woleński, Jan. 2016. “Logic in the Light of Cognitive Science.” Studies in Logic, Grammar and Rhetoric 48(61). 87–101. Yablo, Stephen. 1993. “Is Conceivability a Guide to Possibility?” Philosophy and Phenomenological Research 53: 1–42.

Section II

Reasoning

5

Reasoning with Imagination Joshua Myers

5.1 Introduction The imagination is epistemically useful. When used in the right way, it can justify beliefs about what the world is like. But how should we understand the epistemic role and structure of the imagination? In this chapter, I will motivate and defend the thesis that epistemic uses of the imagination are a kind of theoretical reasoning. This thesis has important implications for the epistemology of the imagination. On one hand, it is often claimed that imagination justifes beliefs in a way similar to perception. However, if epistemic uses of the imagination are best understood as a kind of reasoning, then this analogy between perception and imagination is misleading: while perception allows us to take in new evidence about the world, imagination allows us to reason out what follows from evidence that we already have. On the other hand, although some philosophers refer to certain imaginative episodes as a kind of reasoning, little has been done to precisify this idea beyond a mere fgure of speech.1 In this chapter, I intend to do just that by articulating the epistemic structure of imaginative reasoning. I will proceed as follows. I begin in Section 5.2 by giving important background on epistemic uses of the imagination. In Section 5.3, I argue that instances of theoretical reasoning instantiate a certain epistemic structure. Most centrally, in Section 5.4 I argue that epistemic uses of the imagination instantiate that same epistemic structure and thus are a kind of reasoning. In Section 5.5, I argue that reasoning with imagination is not reducible to reasoning with beliefs. I conclude that understanding epistemic uses of the imagination as a sui generis kind of reasoning provides us with a promising framework through which to understand the epistemology of the imagination.

5.2 Epistemic Uses of the Imagination People put their imagination to many uses. Some of those uses are purely recreational, as when I daydream. Some of those uses are aesthetic, as when I imaginatively engage with fction. Some of those uses are practical,

104

Joshua Myers

as when I use my imagination to rehearse a speech. Importantly, some uses of the imagination are epistemic. We can and often do use our imagination to learn new things. Consider the following example: Jenga: Jessica is playing a game of Jenga, the classic party game in which players take turns trying to remove blocks from a tower without causing it to topple over. Jessica is unsure whether removing a particular block will cause the tower to fall. In order to fnd out, she imagines reaching out her hand towards the tower and gently removing the block in question. As she imagines herself pulling the block away from the tower, she imagines that the tower does not fall over. On the basis of this imagining, she forms the belief if she removes the block, the tower will not fall. This example has fve important features. First, Jessica’s imagining is an instance of sensory imagination. Sensory imagination involves mental imagery, which represents perceptible properties in a perceptual format. It is plausibly a kind of offine perceptual representation in the absence of any sensory stimulation. In what follows, I will set aside what is sometimes called propositional imagination, which does not involve mental imagery, and will restrict my focus to sensory imagination. Second, Jessica’s belief is intuitively justifed on the basis of her imagining. The thesis that imaginings can justify beliefs has found many adherents in recent years (Balcerak Jackson (2018), Balcerak Jackson & Balcerak Jackson (2013), Dorsch (2016), Kind (2016, 2018), Kind & Kung (2016), Langland-Hassan (2016), Williamson (2016)). 2 I am not assuming that imaginative justifcation is indefeasible or even particularly strong. I only claim that the imagination is capable of conferring some amount of defeasible justifcation. Third, in Jenga the content of the belief Jessica goes on to form is empirical; it is a contingent proposition about the external world. Her belief concerns the actual Jenga tower that is in front of her. Even as she imagines the nonactual scenario in which she reaches out her hand to remove the block, she continues to imagine the tower behaving as it would in the actual world. Thus, the epistemic uses of the imagination that I will focus on are importantly different from the widely discussed role imagination plays in justifying beliefs about metaphysical possibilities (see Chalmers (2002), Kung (2010)). Fourth, Jenga involves an imaginative episode which is constituted by a sequence of imaginative states. Jessica’s initial imaginative state represents the tower in its current state and, as her imagining develops over time, each successive state represents her hand moving toward the tower and gradually pulling away one of the blocks. I will refer to both imaginative episodes and states as imaginings, but my primary focus will be on imaginative episodes.

Reasoning with Imagination

105

Finally, the content of Jessica’s imagining is at least partially determined by her beliefs, memories, and perceptions about the tower of blocks that is in front of her. For example, the state that she imagines the tower to be in is informed by her current perceptual experience of the tower. If she is already deep into the game and the tower has become a highly irregular shape from having many blocks removed already, she reconstructs that shape in her imagination on the basis of her perception of the tower. Similarly, when she imagines removing one of the blocks from the tower, the content of this imagining is infuenced by her beliefs about the size, weight, and material of the block. Finally, the content of her imagining may be infuenced by her memories of previous Jenga games. Jenga is an example of constrained imagining (see Kind (2016, 2018) and Kind & Kung (2016)). An imagining is constrained when its content is (partially) determined by the content of other mental states which act as constrainers.3 There are many potential constrainers on imaginings in addition to perceptions, beliefs, and memories. For example, Jessica’s intention to imagine a Jenga tower constrains her imagining insofar as it partially determines its content. It is also plausible that modular perceptual information constrains how Jessica’s imagining unfolds. Her perceptual system may encode regularities in her environment, such as how towers tend to fall, which infuence how she imagines the Jenga tower falling. Different types of constrainers will be useful for different tasks. For example, imaginatively engaging with fction typically involves different constrainers than imaginatively daydreaming, which in turn typically involves different constrainers than imagining how I take the actual world to be. It is worth noting that, as I am conceiving of it, the constraining relation can be either personal or subpersonal. Sometimes I consciously and deliberately impose a constraint on my imagination, as when I form a conscious intention to imagine a Jenga tower. Other times my imagination is constrained subpersonally, as when my sensory imaginings are constrained by modular perceptual information that I am not consciously aware of. Constraints are often invoked as a way of explaining the epistemic power of the imagination (Balcerak Jackson (2018), Kind (2016, 2018), Kind & Kung (2016), Langland-Hassan (2016), Williamson (2016)). The rough idea is that by properly constraining our imagination, we can make it more sensitive to the way that things are, and thus (if all goes well) learn about the way things are. I think this proposal is correct, but incomplete. Given the many different ways that the imagination can be constrained, more needs to be said about the conditions on properly constraining the imagination such that it has the power to justify beliefs. I will try to fll in some of these important details in what follows.

106

Joshua Myers

5.3 The Epistemic Structure of Reasoning I am going to argue that we can reason with imagination. In order to do that, I had better say something about what it means to reason with some state. At its core, theoretical reasoning involves transitioning from some previously held attitudes to a new doxastic state. One obvious sense of “reasoning with” a certain mental state is for that mental state to be one of the previously held attitudes that enters into the reasoning process. I will call these the input states of reasoning. Analogously, I will call the attitude that one forms at the end of the reasoning process the output state of reasoning.4 Thus, one way of precisifying the question of which states we can reason with is to ask what kind of states the inputs or outputs of reasoning can be. Since I am focusing on theoretical reasoning, which by defnition results in a doxastic state as its output, the interesting question is which states can be the inputs of reasoning. Do input and output states exhaust the states involved in reasoning? Following Staffel (2019) and others, we can distinguish between input states and mediating states. Input states are initial states which enter into the reasoning process. Mediating states are states which help to generate  the output state on the basis of the input states. They mediate between the input and the output. In a sense, this is where the reasoning actually happens; the mediating states are constitutive of the reasoning itself. In reasoning, one does not merely have some initial states and then form a doxastic state out of the blue. One reasons one’s way to this doxastic state from the input states via some mediating states, and the mediating states are what constitute that process of reasoning. One illustrative, although controversial, example of a mediating state is a taking state. According to some views of reasoning, one must take the output state to be supported by the input states (Boghossian (2014)). This taking state cannot be an additional premise that is an input to the reasoning, due to the regress worries raised by Carroll (1895). One can think of it as a background condition that allows the output to be properly based on the input. Other putative mediating states may be quite different in their epistemic structure. Thus, the question of which states one can reason with is the question of which states can act as input or mediating states to reasoning. One methodological issue that faces us when we ask about which states we can reason with is that the nature of reasoning is a matter of widespread dispute. Arguing for a particular theory of reasoning is beyond the scope of this chapter and simply assuming a particular theory of reasoning would signifcantly diminish the dialectical force and philosophical interest of my argument. I’m going to circumvent this methodological worry by identifying some commonly accepted platitudes about reasoning and using them to explicate some markers that are good evidence for reasoning regardless of the theory of reasoning one accepts.

Reasoning with Imagination

107

These platitudes pick out an epistemic structure that is distinctive of reasoning. In the next section, I’ll go on to argue that epistemic uses of the imagination instantiate this epistemic structure. Here is one platitude about reasoning: reasoning involves operations on content. Note that in reasoning the output states are different in content from the input states (Boghossian (2003)). If they were the same in content, then you would not need to do any reasoning. You could simply transfer the content from input to output without any intervening deliberation or inference. The way reasoning bridges that gap in content between the inputs and outputs is by operating on the contents of the inputs (Broome (2013)). In paradigmatic cases of reasoning, these operations involve rules of inference. More generally, it is plausible that these operations are facilitated by states that mediate between the inputs and outputs of reasoning. This frst platitude is closely related to the commonplace idea that reasoning allows us to infer new information (the output state) from information that we already possess (the input states). If the new information were simply identical to the information that we already possess, then we would not need to do any reasoning. Here is a second platitude about reasoning: garbage in, garbage out. More perspicuously, for an output of reasoning to be justifed, the inputs to reasoning need to have justifcatory force. By justifcatory force, I mean the property of being able to justify beliefs. Reasoning involves arriving at a belief on the basis of some evidence. If you start with bad evidence, then the output belief will not be justifed no matter how well you reason from it. Only epistemically good inputs can generate epistemically good outputs. A third platitude: in order for the output belief to be justifed, you need to reason well. In other words, it is not enough to simply have epistemically good inputs. You also need to reason in a way that is epistemically good. This has to do with how one operates on the content of the input states. For example, reasoning your way to some conclusion via an invalid inference rule results in an unjustifed output, no matter how epistemically justifed the input beliefs are. The frst, second, and third platitudes are complementary. Together, they say that in reasoning, the epistemic status of the output belief depends on the epistemic status of the input states and the epistemic status of how one operates on their content. One fnal platitude: you can be held epistemically responsible for how you reason. Reasoning poorly can incur personal-level epistemic praise and criticism. This is not merely an epistemic evaluation of the belief that one forms on the basis of reasoning, but of the agent who performed that reasoning. If an agent reasons with some logical fallacy then not only is the belief that is formed on the basis of that reasoning unjustifed, but it is also the case that the agent reasoned poorly and that they ought to have reasoned differently.

108

Joshua Myers

To summarize, (1) reasoning involves operations on contents, (2) justifcation by reasoning depends on the justifcatory force of the inputs, (3) justifcation by reasoning depends on how well one reasons, and (4) reasoning implicates epistemic responsibility. Each of these is a marker of reasoning. Taken together, these markers delineate an epistemic structure that is distinctive of reasoning. In order to further motivate this picture of reasoning beyond mere platitudes, consider how a paradigmatic case of reasoning such as deduction exhibits each of these markers. First, in deduction, the content of the conclusion is different from the content of the premises. One bridges this gap by operating on the content of the premises according to certain rules of inference in order to derive the conclusion. Second, justifcation by deduction depends on the justifcatory force of the input beliefs. This is motivated by the intuitive idea that the conclusion of some deductive inference will only be justifed if the input beliefs are justifed as well. I cannot form a justifed belief that q on the basis of unjustifed beliefs that p and that if p then q. Third, justifcation by deduction depends on how well one reasons. In the case of deduction, this involves reasoning according to certain logically valid rules of inference. If one reasons according to a rule that exemplifes some logical fallacy, then the output of that inference will be unjustifed. Finally, one can be held epistemically responsible for one’s deductive reasoning. This is a level of normative appraisal over and above evaluating whether the output belief of the reasoning is justifed, or whether the input states really entail the output state. If someone reasons poorly according to invalid rules of inference, but by coincidence happens to arrive at a conclusion that is actually entailed by their premises, there is still a sense in which they reasoned poorly and are epistemically blameworthy as a result. We can further motivate these markers as distinctive of reasoning by considering perception as a foil to reasoning. It is generally uncontroversial that the process of forming beliefs solely on the basis of perception is not a form of reasoning (although see Siegel (2017)). The markers I have identifed substantiate this claim. First, perception does not involve operations on contents. On a popular and plausible view of perceptual justifcation, perception (prima facie) justifes a belief that p only if your perceptual experience represents that p (Pryor (2000), Tucker (2010)). There is no gap in content between these two states, and thus no operations on the content are needed to transition from the perception to the belief. Perception provides one with new information and does not involve operating on the content of information that one already has. Second, perceptual justifcation does not depend in any substantive sense on the justifcatory force of the inputs. Perceptual states are immediate justifers. That is, they confer prima facie justifcation to beliefs

Reasoning with Imagination

109

in a way that does not depend on the justifcatory status of one’s other beliefs. So, the justifcatory force of perception does not depend on the epistemic status of one’s previously held attitudes. As a result, there is no substantive possibility of having “garbage in and garbage out” in perception as there is in reasoning. Third, perceptual justifcation does not depend on how well you perceive. Recall that how well one reasons is related to how one operates on the content of the input states. Insofar as perception does not involve operating on content, then it trivially fails to meet this platitude. Nevertheless, perhaps there is some loose sense in which one can perceive well or poorly. For example, perhaps being the subject of a visual hallucination or illusion is an example of perceiving “poorly.” But, even if one grants this, it is not plausible that visual hallucinations or illusions defeat one’s perceptual justifcation. One still possesses prima facie perceptual justifcation when one is hallucinating. If someone hallucinates an apple in front of them (without knowing that they are hallucinating), this justifes them in believing that there is an apple in front of them. Possessing other evidence that there is not an apple in front of them, or evidence that they are hallucinating defeats this perceptual justifcation, but the mere fact that one is hallucinating does not. Finally, and relatedly, you generally cannot be held epistemically responsible for how you perceive in the way that you can be held responsible for how you reason. Faulty perception is not epistemically evaluable in the way that faulty reasoning is. This can be partly motivated by the fact that perception is passive. Agents cannot be held responsible for how they perceive because they cannot choose how they perceive. This can also be motivated by the aforementioned fact that perceiving poorly does not undermine perceptual justifcation. If upon viewing the Müller-Lyer illusion you form the belief that the two lines are unequal in size, you are not epistemically blameworthy for doing so (unless, of course, you were aware of the illusion and thus have evidence which acts as a defeater). The fact that these four features cleanly distinguish the epistemic structure of reasoning from the epistemic structure of perception is some evidence that these four features carve epistemic nature at its joints and demarcate an important epistemic kind. This is further supported by the fact that debates over whether perceptual states can enter into or be the product of a reasoning process often hinge on these very features (see Siegel (2017)). Ultimately, whether the epistemic kind picked out by these features is particularly interesting or important will be decided by how theoretically fruitful its applications are. In what follows, I argue that it has theoretically fruitful applications for thinking about the epistemology of imagination and thus provide some evidence for its importance.

110 Joshua Myers

5.4 Imaginative Reasoning Sensory imagination is often called “perception-like” with respect to its functional roles, phenomenology, and representational properties (for an infuential example, see Currie and Ravenscroft (2002)). But is imagination also perception-like in its epistemic structure? Analogies between perception and imagination are widespread throughout the literature on the epistemology of the imagination. For example, Hyde states that “the imagination can be put to good epistemic use, justifying certain beliefs based on its content in a way comparable with perceptual experience” (2019, p. 32). Similarly, Balcerak Jackson suggests that by focusing on “the similarities, rather than the differences, between imaginings and perceptual states […] we can explain why imaginings have epistemic value, and what that value is” (2018, p. 209). Finally, Kind argues that “perceiving a state of affairs S justifes (or contributes to the justifcation of) the belief that P [and] in at least some cases, imagining a state of affairs S can have the same justifcatory power” (2018, p. 228). Other theorists draw similar analogies between the epistemic roles of imagination and perception (see, for example, Dorsch (2016)). In the preceding quotations, the perceptual analogy is invoked in order to claim that imagination and perception have similar epistemic structures; that they justify beliefs in broadly the same way. There are a number of reasons one might fnd this model of imaginative justifcation to be plausible. First, perception and imagination are phenomenally and representationally similar; both states involve similar kinds of experiences and represent similar kinds of properties. One might take these similarities to suggest epistemic similarities as well. Second, perception and imagination can justify similar beliefs. For example, one can form a justifed belief that one’s sofa will ft through a doorway either by seeing the sofa ft through that doorway or by imagining the sofa ft through that doorway. Nevertheless, in this section I will argue that epistemic uses of the imagination (or, at least, an important class of them exemplifed by Jenga) are best understood as a kind of reasoning. This, in turn, suggests that the perceptual analogy is misleading. Although sensory imaginings may be perception-like in various respects, they do not justify beliefs in a perception-like way. Instead, imaginative justifcation is best understood as a species of justifcation by reasoning. Recall that in epistemic uses of the imagination, one bases a belief on an imagining which is constrained in various ways by one’s other mental states. I will be arguing that the constrainers act as the input states to imaginative reasoning, the imaginings which they constrain act as the mediating states of imaginative reasoning, and the belief formed on the basis of the imagining is the output state of imaginative reasoning. 5 As

Reasoning with Imagination

111

we will see, this framework is useful for understanding the different epistemic roles that constrainers and imaginings play. Let us now consider whether epistemic uses of the imagination exhibit the markers of reasoning identifed in the previous section. Epistemic uses of the imagination meet the frst platitude about reasoning: they involve operations on content.6 Consider Jenga. At the outset of Jenga, Jessica has no beliefs about whether or not the tower will fall. Neither does she have any imaginative states that represent the state of the tower after she has removed the block at the outset of the imaginative episode. Thus, there is a gap in content between the output belief and any of the input states. In order to bridge this gap, Jessica allows her imaginative episode to unfold, thereby performing various operations on the content of her initial imaginative state, before arriving at an imaginative state which represents the tower after she has removed the block. As her imaginative episode unfolds, it takes the content of the constrainers regarding the physical structure of the tower and the motor control of her hand and develops that content in various ways by representing her hand as moving toward the tower and then removing a block so as to modify the structure of the tower. Only by operating on the content in this way can Jessica arrive at an imaginative state which explicitly represents the tower after the block has been removed and thus arrive at a justifed belief about whether the tower will fall. If Jessica simply began by imagining the tower falling over, then her belief would not be justifed on its basis because her imagining would not mediate between her prior attitudes and the resulting belief in an epistemically appropriate way. Epistemic uses of the imagination also meet the second platitude about reasoning: imaginative justifcation depends on the justifcatory force of the constrainers. This is clear from considering two variations on Jenga. The frst is as follows: Unjustifed Jenga: Jessica uses her imagination to learn whether removing a Jenga piece from the tower will result in the tower collapsing. This imagining is partially constrained by her beliefs about the mass and material of the blocks and her memories of previous Jenga games. However, her memories are hazy and her beliefs are based on poor evidence. On the basis of this imaginative episode, Jessica forms the belief the tower will not fall if she removes a certain block. Intuitively, the output belief is unjustifed. Indeed, it seems like it is unjustifed because it was based on an imagining which was constrained by states which themselves lack justifcatory force. Because the inputs to the imagining are epistemically bad, so too is the belief that is based

112

Joshua Myers

on it. This hypothesis is further borne out when we consider a contrasting case: Justifed Jenga: Jessica uses her imagination to learn whether removing a Jenga piece from the tower will result in the tower collapsing. This imagining is partially constrained by her beliefs about the mass and material of the blocks and her memories of previous Jenga games. Her memories are vivid and precise, and her beliefs are based on excellent evidence. On the basis of this imaginative episode, Jessica forms the belief the tower will not fall if she removes a certain block. In this case, the output belief is intuitively justifed. The only difference between Unjustifed Jenga and Justifed Jenga is the justifcatory force of the constrainers. We can even stipulate that the imaginative content and phenomenology is held constant between the two cases. Thus, imaginative justifcation depends on the justifcatory force of the constrainers.7 No matter how careful Jessica is in letting her imagination unfold, it will not result in a justifed belief if she starts with constrainers that are not capable of justifying beliefs.8 As in paradigmatic cases of reasoning, merely having epistemically good inputs is not enough. The third platitude about reasoning also applies: imaginative justifcation depends on how well one imagines. In order to generate a justifed output, you not only need good constrainers, but also need to properly operate on those constrainers in your imagination. One way of imagining poorly might be failing to construct a stable imaginative representation. It is diffcult to construct a stable imaginative representation of suffciently complex scenes, regardless of whether one’s constrainers have a positive epistemic status or not. For example, in Jenga, Jessica might fail to hold the structure of the tower constant in her imagination. Another way of imagining poorly may be failing to imagine what is entailed by its constrainers. For example, Jessica might correctly imagine the structure of the tower after removing the block but fail to imagine the fact that the tower will fall as a direct consequence of removing the block, perhaps because she fails to recognize the extent to which the tower’s structure is unstable. Another way of imagining poorly may be failing to constrain one’s imagining with a reason that one ought to be sensitive to. For example, Jessica might fail to imaginatively simulate the effects of friction on the dynamics of the tower. While it remains an open question exactly what demarcates epistemically appropriate imagining from epistemically inappropriate imagining, it does at least seem that there is such a distinction to be drawn. Holding the epistemic status of the constrainers constant, there are ways of imagining that confer justifcation onto the output belief and ways of imagining that do not. Thus, how one develops the content of their imagining is epistemically relevant. Finally, epistemic uses of the imagination meet the fnal platitude about reasoning: epistemic uses of the imagination implicate epistemic

Reasoning with Imagination

113

responsibility. Consider the following variation on Jenga (adapted from a similar case in Balcerak Jackson & Balcerak Jackson (2013)): Wishful Thinking Jenga: Jessica really wants to win the game of Jenga. Her desire to win, coupled with her irrational optimism about how good she is at Jenga, cause her to imagine cleanly removing a piece from the tower without it falling, and thereby form the belief that if she removes the piece the tower will not fall. It would be quite natural to hold Jessica epistemically responsible for her imagining. Intuitively, she is blameworthy for imagining in the way that she did. One might rightly criticize her for letting irrational factors infuence the content of her imagination. It is intuitive to say that she ought to have constrained their imagination in light of her evidence about the Jenga tower and ought not to have constrained it in light of her irrational optimism. These criticisms are natural to make even if her imagination was veridical and the Jenga tower really will remain standing. Although the belief she forms is intuitively unjustifed, these are appraisals of Jessica, the agent who did the imagining, and not merely of the belief she formed. This suggests that agents can be held epistemically responsible for the imaginative projects they engage in. Where does this leave us? I began this section by noting that theorists often invoke an analogy with perception when discussing the epistemic role of the imagination. Understood loosely, this analogy is harmless. After all, it is true that both perception and imagination have the capacity to justify beliefs. However, when understood more strictly as a claim about the epistemic structure of the imagination, the analogy fails. Imagination and perception do not play the same epistemic roles. Rather than acquainting one with new evidence about what the world is like, epistemic uses of the imagination allow one to appropriately form new beliefs on the basis of evidence which one already possesses. More precisely, epistemic uses of the imagination involve operations on contents, implicate epistemic responsibility, and only justify beliefs when they both have epistemically good inputs and when one properly operates on the content of one’s imagining. Thus, epistemic uses of the imagination are best understood as a kind of reasoning.

5.5 Reasoning with Imagination Is Not Reducible to Reasoning with Beliefs So far, I have argued that some epistemic uses of the imagination instantiate an epistemic structure that is distinctive of reasoning. One might concede that epistemic uses of the imagination such as Jenga are in fact cases of reasoning, but object that they are reducible to paradigmatic cases of reasoning with beliefs. On this view, epistemic uses of

114 Joshua Myers the imagination involve reasoning, but do not involve reasoning with imaginings. Instead, they involve reasoning with one’s beliefs. The imaginative episode that accompanies the reasoning is epistemically epiphenomenal; it does not play any essential epistemic role. The imagination may contribute to the causal story of how the subject arrives at a particular belief, but it does not contribute to the epistemic story of whether and how that belief is justifed. This analysis has the advantage of explaining why epistemic uses of the imagination exhibit each of the markers of reasoning without introducing a sui generis and heretofore mostly unrecognized type of reasoning. It might therefore be thought of as a more parsimonious way of understanding epistemic uses of the imagination. There is a fat-footed response to this way of formulating the objection. Namely, it is implausible in a case like Jenga that beliefs alone could account for the subject’s justifcation. Jenga is an extremely complicated case in which Jessica must track many physical properties of the tower while also regulating motor imagery of her hand removing the block. If beliefs alone were doing the epistemic work here, this would require her to believe a theory of physics. Empirical work on mentally simulating the stability of towers of blocks has suggested that a model which obeys roughly Newtonian mechanics best matches human performance (Battaglia et al. (2013), Hamrick et al. (2016)). But it is very implausible that humans all believe Newtonian mechanics. It is more likely that roughly Newtonian mechanical laws are implicitly built into the functioning of the system involved in imaginatively simulating physical systems, or, at the very least, are represented sub-doxastically. This best accounts for the fact that conceptually unsophisticated animals and young children can perform physical reasoning tasks using their imagination. The more general point is that our imaginative capacities cannot be fully explained by our beliefs. Sensory imagination constitutively involves offine perceptual processing, and thus our imaginative capacities are tightly linked to our perceptual capacities. Imagination is sensitive to regularities in our environment that have been registered by perception. One particularly striking demonstration of this is the fact that perceptual learning, the process by which one becomes more perceptually sensitive to certain features by being repeatedly exposed to them, can occur within the imagination. That is, merely imagining some feature repeatedly causes one to be able to better perceptually discriminate that feature (Tartaglia et  al. (2009, 2012)). Similarly, we can get better at imagining in a way that does not seem to depend on gaining any new beliefs. Studies suggest that people can increase their skill at easily generating vivid mental imagery simply through practice (Rodgers et al. (1991), Calmels et al. (2004)). Finally, it is plausible that we can constrain our imaginings with non-doxastic states such as memories or occurrent perceptions. All of this points to the fact that what we imagine cannot be reduced merely to what we believe. In epistemic uses of the imagination,

Reasoning with Imagination

115

beliefs may partly infuence what is imagined, but they do not exhaust what is imagined. Thus, it is likely that imaginings contribute something epistemically over and above beliefs. Reasoning with imagination cannot be reduced to reasoning with beliefs. In response, one might concede that imaginings have content that cannot be reduced to the content of one’s beliefs. But this does not yet show that imaginings play an epistemic role over and above one’s other cognitive capacities.9 Let the constrainers be any kind of state you would like, belief or otherwise, the point is that the imagining does not play a role in epistemically mediating between those constrainers and the output belief. The constrainers on the imagining are what justify the output belief, not the imagining itself. At best, the imagining plays a kind of heuristic role in generating the output belief, but not an epistemic role in justifying it. As Shannon Spaulding puts it, “Imagination generates ideas, but other cognitive capacities must be employed to evaluate these ideas in order for them to count as knowledge” (2016, p. 207). Once we expand the potential constrainers from only beliefs to non-doxastic states as well, this objection seems more plausible. Nevertheless, there are at least two epistemically relevant roles that the imagination plays: imaginings integrate and develop the content of their constrainers in a distinctive way. What do I mean to say that imagination integrates the content of constrainers? I mean that sensory imagination is a central workspace that takes inputs from many different modules and subsystems such as beliefs, memories, and perceptions and not only translates them into a single format but can also integrate them into a single imaginative episode. For example, in Jenga Jessica’s imagining integrates her beliefs, memories, and perceptions about the Jenga tower into a single continuous imaginative episode. Indeed, we have seen that the imagination, in virtue of being housed in the perceptual system, plausibly has access to modular information that other systems cannot as a matter of principle access. The epistemic upshot is that imagination often has more information available to it than other cognitive capacities on their own. Intuitively, Jessica’s imaginative episode in Jenga is much more informationally rich than her beliefs about Jenga towers on their own, or her perceptual experience of the Jenga tower on its own. Thus, imagination is often in the position to justify more beliefs than other cognitive capacities on their own in virtue of being a central workspace that integrates representations from these other capacities. Once imagination has integrated the information encoded by the constrainers, it goes onto develop that information in various ways. What does it mean to say that imagination develops the content of the constrainers? I just mean the fact, explored in the previous section, that imagination operates on the content of the constrainers to arrive at a fnal imaginative state that is different in content from the initial imaginative state. For example, in Jenga Jessica’s initial imaginative state

116 Joshua Myers represents the Jenga tower as it currently is, and her fnal imaginative state represents the Jenga tower after a piece has been removed. The intermediate imaginative states represent her hand reaching toward the tower and removing one of the blocks in a particular way. Together, these states constitute a single imaginative episode. The fact that imaginative episodes develop their content is epistemically relevant. The process of operating on the content of the imagination can make certain inferential transitions epistemically appropriate that otherwise would not be. In general, merely having some evidence that supports some belief is not suffcient for that belief to be doxastically justifed. Some gaps between evidence and belief are too large to cross in an epistemically appropriate manner, even when the evidence actually does support the belief. For example, consider some extremely complicated mathematical theorem that follows from some basic axioms. Even though the axioms really do support the theorem, if I simply form a belief that the theorem is true on the basis of the axioms, I will have done something epistemically wrong. Simply basing my belief that the theorem is true on the axioms without any intervening inferential steps would result in an unjustifed belief. Instead, I need to go through a number of smaller inferential steps which constitute a proof of the theorem from the axioms in order for it to be epistemically appropriate to base my belief that the theorem is true on the axioms. Intuitively, these smaller steps help me to recognize exactly why the theorem follows from the axioms. Thus, developing content in certain ways is an epistemically relevant feature of belief-forming processes. An analogous process occurs in epistemic uses of the imagination. It is plausible that imagination can play the role of developing the content of input states such that one can derive the conclusion in an epistemically appropriate manner. That is what the imagination is doing in Jenga. Jessica starts with a collection of beliefs and memories and perceptions about the Jenga tower. These are the input states to her reasoning. However, it would be too great a leap for her to immediately infer that the tower will fall on the basis of these input states. She needs to do some reasoning to bridge the gap in content between these inputs and the output belief that the tower will fall. Intuitively, engaging in an imaginative episode that takes in these input states and traces out their entailments is one way to bridge this gap in an epistemically appropriate manner. Indeed, this is exactly the epistemic role that mediating states play in paradigmatic cases of reasoning. In some cases, there may be multiple ways to bridge the gap between input and output in an epistemically appropriate manner. Perhaps there are cases where one could either reason with imagination or reason with beliefs to reach the relevant conclusion. But this is no threat to the epistemic relevance of the imagination, as long as imagination is one way to bridge the gap in an epistemically appropriate manner.

Reasoning with Imagination

117

Nevertheless, there are many cases where imaginative reasoning is the only epistemically appropriate way to develop the content in order to reach the conclusion. To see this, consider how Jessica might reason her way to the conclusion that the tower will not fall in Jenga without using her imagination. It would have to involve some extremely complex chain of mathematical reasoning about physics that would simply be too complicated for anyone to carry out with their beliefs. This demonstrates the fact that while reasoning with beliefs is well suited for developing content in certain ways, reasoning with imagination is well suited for developing content in other ways. For example, deductive reasoning with beliefs is well suited for tracing out the logical entailments of the input states. However, reasoning with imagination is well suited for tracing out the spatial, or physical, or causal entailments of the input states. Since imaginings develop, at least in part, according to stored perceptual information, and since that information is particularly relevant to how spatially extended physical objects interact with each other, then imagination will be particularly well suited to fguring out how certain physical systems will change over time, among other things. The imagistic format of imagination plausibly also allows for certain distinctive transitions between imaginative states. These features go some way toward explaining the distinctive epistemic value that imagination has in a case like Jenga. Of course, reasoning with imagination is not limited to physical reasoning. For example, many theorists have argued for the importance of the imagination in reasoning about other people’s mental states (Currie  & Ravenscroft (2002), Goldman (2006)). In this case, one can tell a similar story about how imagination is particularly well suited to simulating mental states such that it can integrate and develop your information about the mental states of another person, and so on for other kinds of epistemic uses of the imagination. Thus, not only are imaginings epistemically relevant, but they are sometimes distinctively epistemically relevant; not only is the imagination a way of forming justifed beliefs, but it is sometimes the best or only way to form justifed beliefs.

5.6 Conclusion In this chapter, I have argued that we can reason with imagination. In particular, I have argued for a framework for understanding epistemic uses of the imagination in which constrainers on an imagining act as input states to imaginative reasoning; the imagining itself is a mediating state which operates on the content of those input states and thereby allows one to form a justifed belief as the output state of the imaginative reasoning. I argued for this structure by way of arguing that epistemic uses of the imagination involve operations on contents, implicate epistemic responsibility, and only justify beliefs when they both have

118

Joshua Myers

epistemically good inputs and when one develops their imagining appropriately. I also argued that reasoning with imagination is not reducible to other, more familiar kinds of reasoning with doxastic states. Thus, epistemic uses of the imagination are a sui generis form of reasoning. This thesis has wide-ranging implications for theorizing about reasoning and the imagination. First, there is a tendency within the literature on reasoning to exclusively focus on reasoning with beliefs (e.g., Boghossian (2003), Broome (2013), Staffel (2019)). If the arguments of this chapter are on the right track, then theories of reasoning should also accommodate reasoning with imagination. Second, although it is widely accepted that constraints on the imagination are epistemically important, the thesis that we can reason with imagination clarifes the particular epistemic role they play as inputs to imaginative reasoning. Third, although it is often claimed that perception and imagination are epistemically similar, the arguments of this chapter suggest that they play fundamentally different epistemic roles. While perception takes in new evidence about the world, imagination allows us to reason out what follows from evidence that we already have. Finally, although I have focused on theoretical reasoning, the framework developed in this chapter has the potential to shed light on imagination’s role in practical reasoning as well. Understanding epistemic uses of the imagination as a form of reasoning offers a fruitful way forward for theorizing about the epistemology of the imagination.10

Notes 1 See Balcerak Jackson and Balcerak Jackson (2013) who explicitly discuss and argue for the idea that one can reason with their imagination. See also Williamson (2016), who argues that epistemic uses of the imagination are a kind of conditional reasoning. 2 There are also notable dissenters (e.g., O’Shaughnessy 2000, Spaulding 2016, Egeland 2019). I will address some of their objections in Section 5.5. For now, let me address an initial skeptical worry about the justifcatory force of the imagination: perhaps imagination is simply not very reliable. While there are no doubt many domains about which imagination is unreliable, empirical evidence suggests that subjects are reliable at using imagination to make physical inferences of the kind exemplifed by Jenga. One study found that not only are subjects who use their imagination reliable at predicting whether a tower of blocks will fall or not, but they are also reliable at determining which direction the tower will fall in, and that this reliability is maintained even when the tower consists of blocks of different masses (Battaglia et al. 2013). Other studies that demonstrate the reliability of the imagination in physical inferences include the classic mental rotation paradigm (Shepard & Metzler 1971), as well as studies implicating imagination in making predictions on the basis of mechanical diagrams (Hegarty 1992). 3 Plausibly, there are also constraints which result from the architecture of the imagination, rather than from other contentful mental states. I set these aside in what follows.

Reasoning with Imagination

119

4 Input states are often referred to as premise attitudes and output states are often referred to as conclusion attitudes in the literature on reasoning. 5 Plausibly, imaginings can also act as input states to reasoning. For example, Jessica’s initial imaginative state represents her hand reaching toward the Jenga tower despite the fact that she does not believe this to be the case. This suggests that Jessica’s imagining is a kind of conditional reasoning, in which she reasons with a merely imagined premise as an input. Nevertheless, not all epistemic uses of the imagination are best understood as conditional reasoning which takes imaginings as inputs. Consider a subject in a psychological experiment who is asked to perform a mental rotation in order to fgure out whether two objects are identical (Shepard & Metzler 1971). The subject’s initial imaginative state does not represent a scenario which is contrary to what they believe (it simply represents the objects as they appear to the subject), and they do not go on to form a belief in a conditional (they simply form the belief that the two objects are identical). Thus, I will understand imaginative reasoning as distinguished by its imaginative mediating states, rather than imaginative inputs. 6 I am indebted to the discussion of this point in Balcerak Jackson and Balcerak Jackson (2013). 7 In other work, I argue that this condition on imaginative justifcation is best explained by the thesis that imaginings are themselves capable of being epistemically justifed or unjustifed (Myers 2021). While nothing I say here depends on this thesis, it does provide further support for the view that imaginings play the role of mediating states in reasoning. If imaginings can have a justifcatory status and this justifcatory status is what allows them to justify beliefs, then epistemic uses of the imagination involve justifcation being transferred from inputs to imaginings to beliefs in much the same way that justifcation is transferred from premise to conclusion in typical cases of deductive reasoning. 8 This feature explains why non-epistemic uses of the imagination cannot justify beliefs about the actual world. Imaginatively engaging with a fantasy novel, for example, cannot justify beliefs about the actual world precisely because fantasy novels are not typically good sources of evidence about what the actual world is like and thus lack justifcatory force with respect to those beliefs. 9 This more general form of the objection is argued for forcefully by Spaulding (2016) and Egeland (2019). 10 Special thanks to Paul Boghossian for many helpful discussions and comments on multiple drafts of this chapter. Thanks also to participants at the NYC PoPRocks and the 2020 meeting of the Southern Society for Philosophy and Psychology for helpful questions, comments, and discussion.

References Balcerak Jackson, Magdalena, & Balcerak Jackson, Brendan (2013). Reasoning as a source of justifcation. Philosophical Studies 164(1): 113–26. Balcerak Jackson, Magdalena (2018). Justifcation by imagination. In Fiona Macpherson & Fabian Dorsch (eds.), Perceptual Imagination and Perceptual Memory. Oxford: Oxford University Press, 209–26. Battaglia P.W., Hamrick J.B., & Tenenbaum J.B. (2013). Simulation as an engine of physical scene understanding. Proceedings of the National Academy of Sciences 110(45): 18327–32.

120  Joshua Myers Boghossian, Paul (2003). Blind reasoning. Aristotelian Society Supplementary Volume 77(1): 225–48. Boghossian, Paul (2014). What is inference? Philosophical Studies 169(1): 1–18. Broome, John (2013). Rationality through Reasoning. Chichester: Wiley-Blackwell. Calmels, C., Holmes, P., Berthoumieux, C., & Singer, R.N. (2004). The development of movement imagery vividness through a structured intervention in softball. Journal of Sport Behaviour 27: 307–22. Carroll, Lewis (1895). What the Tortoise Said to Achilles. Mind 104: 691–3. Chalmers, David J. (2002). Does conceivability entail possibility? In Tamar S. Gendler & John Hawthorne (eds.), Conceivability and Possibility. Oxford: Oxford University Press, 145–200. Currie, Gregory & Ravenscroft, Ian (2002). Recreative Minds: Imagination in Philosophy and Psychology. Oxford: Oxford University Press. Dorsch, Fabian (2016). Knowledge by imagination – How imaginative experience can ground knowledge. Teorema: International Journal of Philosophy 35(3): 87–116. Egeland, Jonathan (2019). Imagination cannot justify empirical belief. Episteme 17(1): 1–7. Goldman, Alvin L. (2006). Simulating Minds: The Philosophy, Psychology, and Neuroscience of Mindreading. Oxford: Oxford University Press. Hamrick J.B., Battaglia P.W., Griffiths T.L., & Tenenbaum J.B. (2016). Inferring mass in complex scenes by mental simulation. Cognition 157: 61–76. Hegarty, M. (1992). Mental animation: Inferring motion from static displays of mechanical systems. Journal of Experimental Psychology: Learning, Memory, and Cognition 18: 1084–102. Hyde, Madeleine. (2019). Keeping the imagination epistemically useful in the face of bias. Journal of Cognition and Neuroethics 6(1): 31–55. Kind, Amy (2016). Imagining under constraints. In Amy Kind & Peter Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press, 145–59. Kind, Amy (2018). How imagination gives rise to knowledge. In F. Dorsch & F. Macpherson (eds.), Perceptual Memory and Perceptual Imagination. New York: Oxford University Press. Kind, Amy & Kung, Peter (2016). Introduction: The puzzle of imaginative use. In Amy Kind & Peter Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press, 1–37. Kung, Peter (2010). Imagining as a guide to possibility. Philosophy and Phenomenological Research 81: 620–33. Langland-Hassan, Peter (2016). On choosing what to imagine. In A. Kind & P. Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press, 61–84. O’Shaughnessy, Brian (2000). Consciousness and the World. Oxford: Clarendon Press. Myers, Joshua (2021). The epistemic status of the imagination. Philosophical Studies. https://doi.org/10.1007/s11098-020-01600-1. Pryor, James (2000). The skeptic and the dogmatist. Noûs 34(4): 517–49.

Reasoning with Imagination

121

Rodgers, W., Hall, C., & Buckolz, E. (1991). The effect of an imagery training program on imagery ability, imagery use, and fgure skating performance. Journal of Applied Sport Psychology 3: 109–25. Shepard, R.N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science 171(3972): 701–3. Siegel, Susanna (2017). The Rationality of Perception. Oxford: Oxford University Press. Spaulding, Shannon (2016). Imagination through knowledge. In Amy Kind & Peter Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press, 207–26. Staffel, Julia (2019). Attitudes in active reasoning. In Magdalena Balcerak Jackson & Brendan Balcerak Jackson (eds.), Reasoning: New Essays on Theoretical and Practical Thinking. Oxford: Oxford University Press, 44–68. Tartaglia, E.M., Bamert, L., Mast, F.W., & Herzog, M.H. (2009). Human perceptual learning by mental imagery. Current Biology 19: 2081–85. Tartaglia, E.M., Bamert, L., Herzog, M.H., & Mast, F.W. (2012). Perceptual learning of motion discrimination by mental imagery. Journal of Vision 12: 1–10. Tucker, Chris (2010). Why open-minded people should endorse dogmatism. Philosophical Perspectives 24(1): 529–45. Williamson, Timothy (2016). Knowing by imagining. In Amy Kind & Peter Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press, 113–23.

6

Equivalence in Imagination Francesco Berto

6.1 Imagination as ROMS In their introduction to the beautiful Knowledge Through Imagination, Amy Kind and Peter Kung state what they call ‘the puzzle of imaginative use’ (Kind and Kung 2016, 1): imagination seems to be arbitrary escape from reality; how can it give knowledge of reality? The short version of their answer, shared by a number of authors in that collection (Kind (2016), Langland-Hassan (2016), Williamson (2016)), is: imagination, of the kind that can give us knowledge, is constrained. It deviates from reality, but in a regimented way. This chapter is about logical regimentation. Our question will be: given that one imagines that A, what Bs is one to imagine, as a matter of logical necessity? As ‘imagining’ is highly ambiguous (daydreaming, hallucinating, mental wandering, entertaining), we need to pin down the sort of imagination whose logic we are after. For our question to make sense, it must be propositional: one imagines that one jumps a stream, that John is tall and thin, that there are functional zombies devoid of phenomenal consciousness, that Stauffenberg puts the bomb on the other side of the table. A number of authors (e.g. Currie and Ravenscroft (2002), Goldman (2006)) agree on a simulationist account. Imagination in general recreates counterpart non-imaginative mental states. In perceptual imagination, one simulates perception: one hears the riff of Smoke on the Water in one’s head and it’s relevantly like the real thing, except that one’s auditive apparatus is left offine. Propositional imagination, the kind thereof which seems to have a chance, if any does, at helping with knowledge, is the one in which one simulates belief (Arcangeli 2019 calls it ‘cognitive imagination’). On the other hand, imagination is often contrasted with belief: the former is voluntary in a way the latter isn’t, at least according to most authors (Dorsch (2012), Gendler (2000), Mulligan (1999), Walton (1990)): one can imagine that all of Edinburgh has been painted yellow but, having overwhelming evidence of Edinburgh’s greyness, one cannot make oneself believe it. What’s the point of a simulation that differs from what it simulates in such a core feature?

Equivalence in Imagination

123

The link between imagination as mental simulation and belief is, I think, that in mental simulation we imagine in order to assess what will happen if something is the case, or what would have happened if something had been the case. What if Stauffenberg had put the bomb on the other side of the table? How would the course of WWII have changed? What if I jump the stream? Will I make it to the other side, or will I fall in the water and drawn? The kind of belief mental simulation typically connects to for epistemic purposes, then, must be conditional belief (which may be quite different from belief in the corresponding conditional: more on this below). I don’t really believe, say, that Brexit will cause a recession. I explore the consequences: what if Brexit causes a recession? If I conclude, given the supposition that Brexit causes a recession and what I know or believe, that I will lose my job in the hypothetical scenario, I form the relevant conditional belief, representing a subjective conditional probability. This gives us guidance on the kind of logical form we should be after, when we ask about how logical closure works for imagination as mental simulation. The form must be related to conditionals – ‘Given input A, one imagines that B’; or, less tersely: ‘In an exercise of mental simulation starting from suppositional input A, one imagines that B’. This obviously relates to the Ramsey test for conditionals: one evaluates a conditional ‘If A, then B’ by supposing the antecedent, A, minimally adjusting one’s belief or knowledge system, and seeing whether the consequent B turns out in the imagined scenario. The difference between supposing that A in an exercise of mental simulation, and supposing that one believes that A, is big. And it’s the frst one that goes on, in general, in suppositional thinking, as per the Ramsey test. (Of course, there will be special cases in which one supposes that one believes something.) Supposing I am prime minister of the UK, I may conclude, e.g., that I will hire a bunch of friends in key governmental roles. Supposing I believe I am prime minister of the UK, I may conclude very different things, e.g., that I need to be hospitalized for I am seriously deluded. The Ramsey test ties our evaluation of conditionals to our updating our priors in the light of new information. It’s around noon and one sees that John has moved into the kitchen. One then minimally revises one’s beliefs compatibly with the news. In the updated belief system, one will have dropped, say, the previously held belief that John was in the living room; on the other hand, that John is cooking will now look likely enough for one to come to believe it. What is typical of mental simulation is that one doesn’t really get the news online, e.g., by seeing that John is in the kitchen; one just pretends, in an offine mode (Williamson (2016)), that John is there, and checks what would follow. If in the hypothetical situation it turns out that John is cooking, one does not believe that John is cooking; but one can acquire a conditional belief: one believes, so to speak, that John is cooking conditional on John being in the kitchen.

124 Francesco Berto (Does one come to believe the corresponding conditional, too, ‘If John is in the kitchen, then he’s cooking’? We know from Lewis’ (1976) triviality results that, however tight the connection between conditional beliefs and beliefs in the corresponding conditionals, it won’t be identity. Philosophers and psychologists have tried to circumvent the problem: some keep the core of the Ramsey test by stating that conditionals don’t express propositions and cannot generally be embedded, e.g., Edgington (1995), Bennett (2003). Psychologists often content themselves with a robust, empirically corroborated correspondence, short of identity, e.g., Evans and Over (2004), Oaksford and Chater (2010). On this whole issue, see the masterful Douven (2016).) As stressed e.g. by Williamson (2016), Langland-Hassan (2016), Canavotto et al. (2020), not everything is voluntary in such an exercise. The offine suppositional input is: we are free to suppose whatever A we like (Arcangeli (2019) makes a forceful case). Once the supposition is in, however, what Bs will come out true in the imagined scenario is largely not up to us: it has to do with how our prior belief system is disposed to adjust itself conditional on A, and how this works is largely involuntary, as beliefs are. As many have acknowledged (e.g. Byrne (2005), Kind (2016), Williamson (2016)), the adjustment is governed by principles of relevance and minimal alteration. One who mentally simulates jumping the stream to predict whether if one jumped, one would make it won’t imagine growing wings on one’s back allowing one to fy to the other side. One keeps reality, as one knows or believes it to be, unaltered as much as possible, compatibly with one’s jumping. Imagination is reality-oriented. And so we have a label for what we’re after: the logic of Reality-Oriented Mental Simulation (ROMS). The epistemological importance of such an activity is hardly over-estimated; if we use ROMS to gain knowledge of metaphysical modality (Williamson (2016)), there are good arguments to the effect that we use it to handle modalities of much more mundane kinds, too (Strohminger and Yli-Vakkuri (2019)). And so we may want logics to reason about ROMS.

6.2 The Logic of ROMS In a couple of papers on ROMS (Berto (2017, 2018)), I have proposed to use the simple notation ‘[A]B’ for ‘Given input A, one imagines that B’, or ‘In an exercise of mental simulation starting from suppositional input A, one imagines that B’. Logical tradition suggests that it be interpreted as some modality. We have learned since Hintikka (1962) that we can treat intentional notions like knows, believes, is informed that, using modal logic. We take them as normal modals: quantifers over possible worlds, restricted by an epistemic accessibility relation. What one knows is what holds at all worlds one looks at, which represent open epistemic possibilities.

Equivalence in Imagination

125

The Hintikkan approach notoriously gives logical omniscience: if X is the relevant mental state, given that the agent Xs that A, the agent Xs any B that follows logically from A (A is true at all worlds one looks at; they all make true B as well: one Xs that B). As a special case, one Xs all logical truths. Now we ordinary humans are not like that: we sometimes fail to see even obvious logical consequences of what we know or believe, because we are tired, cognitively impaired, or otherwise busy. For various Xs, logical omniscience is implausible even for perfect reasoners. It has been questioned for knowledge, even for ideal knowers, on the grounds of sceptical paradoxes (Hawke (2016), Nozick (1981)) and the presence of defeaters (Brown (2018)). While these stances are controversial (see e.g. Hawthorne (2004) for criticisms), other propositional attitudes obviously fail to be closed under logical consequence: one desires that one’s headache goes away, and this implies that one has a headache; but one doesn’t desire that one has a headache, even as a perfect reasoner. What about imagination as ROMS? Even if it’s not subject to full logical closure, it should be closed under some logical operations. Try to imagine that John is tall and thin without imagining at the same time that he’s thin and tall, and without imagining that he’s tall; or that Mary is an expert chess player, without imagining that she’s a chess player. ROMS must have some conjunctive structure: one can’t imagine that A ∧ B without simultaneously imagining that B ∧ A, and without imagining that A. Imagining that John is tall and thin without imagining that he is tall would be a bit like imagining that John is tall without imagining that John is tall, wouldn’t it? And it might also be that one cannot imagine that A and that B, without imagining that A ∧ B. You can’t imagine that John is tall and that John is thin without imagining that John is tall and thin, can you? (This may sound more controversial; I’ll get back to it.) Besides, the aforementioned reality-oriented nature of ROMS must induce plausibility and relevance constraints for our logic. When the suppositional input is that Stauffenberg puts the bomb on the other side of the table, we don’t imagine that Hitler suddenly grows an armour protecting him from the explosion: this would be an implausible departure from reality. We imagine historical reality largely as we know or believe it to be, save that Stauffenberg puts the bomb on the other side. As for relevance, in ROMS we focus on what is on-topic given an input: when the input is that Stauffenberg puts the bomb on the other side of the table and we conclude that, then, Hitler gets killed, we don’t imagine that either Hitler gets killed or there’s life on Kepler-442b, although A or B is a logical consequence of A. And we don’t imagine each irrelevant logical or necessary truth or consequence given that suppositional input, although logical truths are implied, in classical logic, by any input, and necessary truths hold at all possible worlds, thus at all worlds, where the input is true.

126

Francesco Berto

To comply with the plausibility requirement, we should be looking at what goes on at worlds where the suppositional input, A, is true, and which are, for the rest, the most plausible worlds given one’s knowledge or beliefs. Readers familiar with conditional logics will hear the workings of the Lewis (1973) semantics for counterfactuals. One evaluates ‘If it were/had been the case that A, then it would be/have been the case that B’ by looking at the closest worlds where A is true. Similarly, our [A] will be a sententially indexed universal modality (Chellas (1975)), a restricted universal quantifer over worlds, where the restriction is determined by the A at issue: the suppositional input. Relevance tells us that [A] cannot be a normal modality, i.e. just a restricted quantifer over possible worlds. Any possible world where B is true will also be one where B ∨ C is, but we don’t want our agents to automatically imagine the irrelevant either Hitler gets killed or there’s life on Kepler-442b because they imagine that Hitler gets killed. Nor do we want our agents busy in working out what would have been the case, had Stauffenberg put the bomb on the other side of the table, to imagine unrelated necessary truths, e.g., that either Melbourne is in Sweden or not, or that Fermat’s Last Theorem is true. In Berto (2017, 2018) I have explored two semantic approaches that make of the [A]-operators non-normal modals: one – Semantics 1, let’s say – sticks to the traditional possible worlds apparatus but adds a topicality constraint that flters out irrelevant logical or necessary truths and consequences. Another one, Semantics 2, drops ‘possible’ and takes [A] as a restricted quantifer over worlds, some of which may be non-normal or impossible: worlds where logical or necessary truths may fail (Berto and Jago (2019)). One reason for considering both here is that they can embed an idea which is, fnally, the main topic of this chapter – and that they do it in slightly but interestingly different ways. This is the idea of equivalence in imagination.

6.3 Equivalence in Imagination The logically untamed nature of intentional states shows up in the fact that the operators expressing them seem to be hyperintensional: they can fail substitution salva veritate of intensional (necessary or logical) equivalents in their scope. One Xs (knows, believes, hopes, etc.) that A, but not that B, although A and B are such equivalents – in standard possible worlds semantics: true at the same possible worlds (of all the relevant models). This may happen because one is computationally limited (A is an obvious tautology, say p → p; B is a long and complicated logical truth), or because one is not on top of some concept (A is equivalent to A ∨ (A ∧ B) in classical and many non-classical logics; but one has an attitude only towards the former, for one lacks B-involved concepts); or because of the defeasibility of knowledge; etc.

Equivalence in Imagination

127

ROMS is no exception. Sometimes [A]C will be true while [B]C won’t, for equivalent A and B: one may be subject to framing effects (Kahneman et al. (1982)), whereby one imagines different things starting from differently presented, necessarily equivalent suppositions (‘If you apply for the job, you have 60% chances of making it’ vs ‘… 40% chances of failing’). One may imagine something given A, nothing given an equivalent B, for one lacks some B-involved concept but no A-involved concept. Importantly, not all aspects of logical de-idealization are represented by the frameworks below. The suppositional agents are conceptually nonidealized: they may lack some concept needed to grasp a proposition; but they are syntactically idealized, in that they cannot fail to parse the syntax of certain sentences in the correct way. For example, whenever, given suppositional input A, they imagine that B ∧ C, they also imagine that C ∧ B when ‘∧’ is Boolean, order-insensitive conjunction. We know that real humans are, instead, susceptible to order-of-presentation effects in their attitudes. The anarchy of imagination may be heuristically fruitful (see Stuart (2020)). But an operator failing substitution of intensional equivalents is likely to have a rather weak logic. When we fnd operators of the form ‘Given A, C’ (e.g. in probability logic – ‘C is likely given A’: Adams (1998); in dynamic belief revision – ‘After revision of one’s beliefs by A, C is the case’: van Benthem (2011); in conditional logic: ‘If A is/was/ had been the case, then C is/would be/would have been the case’), they nearly always satisfy the principle whereby ‘Given A, C’ entails ‘Given B, C’ when A and B are intensional equivalents. This is so even when such operators are non-monotonic, i.e. ‘Given A, C’ does not entail ‘Given A ∧ D, C’ (adding information in the antecedent or premise is not guaranteed to preserve the consequent or conclusion: given that Tweety is a bird, it fies. Given that it’s a penguin bird, it doesn’t). For example, we fnd the principle that ‘Given A, C’ entails ‘Given B, C’ when A and B are intensional equivalents, under the label of Left Logical Equivalence, in the mainstream system P, often taken as a set of minimal principles any non-monotonic logic should obey (Kraus et al. (1990)). However, ROMS may satisfy some restricted substitutivity principle. This is the idea of equivalence in imagination: although it is not the case that, given two intensionally equivalent suppositional inputs, A and B, one will imagine the same things, this will happen when A and B are at least equivalent in one’s imagination. What is equivalence in imagination? We can understand it as cognitive equivalence or synonymy, which is different from synonymy toutcourt. A and B are cognitive equivalents or cognitive synonyms when they play the same role in one’s cognitive life (Hornischer (2017)): whatever one concludes – deductively, abductively, inductively – supposing either, one does, supposing the other. Whatever one understands when either is uttered, one does, when the other is. One cannot take either as

128

Francesco Berto

true, without taking the other as true. Whatever one thinks about given either, one does, given the other. Cognitive synonymy is a respectable notion in linguistics. Absolute synonymy, understood as substitutivity salva veritate in all contexts, has notoriously raised eyebrows among philosophers. Linguists sometimes take it as a purely theoretical, limit notion, for it cannot be empirically tested (Cruse (2000), Stanojecvić (2009)). However, cognitive equivalence or synonymy is the working concept for a number of works in linguistics (Lyons (1996), Murphy (2003)). Cognitive equivalence should be relative to the thinker’s available knowledge base and storage of concepts – thus, what’s in this or that specifc agent’s knowledge base, and thus what is cognitively equivalent for this or that agent, is not for a general logic of suppositional thinking to settle. John is a bachelor and John is an unmarried man are equivalent for nearly any competent speaker of English: if one takes either as true but the other as false, this is likely to generate suspicion on the level of English profciency of the speaker. Ex falso is classically valid and Pseudo-Scotus’ Law is classically valid are equivalent for most logicians. Groundhogs are rodent and Woodchucks are rodent for most zoologists. One way to represent such cognitive interchangeability in AI is via pairs of defeasible, non-monotonic conditionals, ‘If A, then B’, ‘If B then A’, stored in the agent’s knowledge base, e.g., in Logic Programming (Stenning and van Lambalgen (2008)). Our formalism will be in the same ballpark.

6.4 Getting Formal We use a simple propositional language L with an indefnitely large set L AT of atoms p, q, r (p1, p2 ,…), negation ¬, conjunction ∧, disjunction ∨, a strict conditional , square brackets [, ] for our imaginative modals, and round parentheses (,) as auxiliary symbols. The well-formed formulas of L are the items in L AT, and if A and B are formulas: ¬A|(A ∧ B)|(A ∨ B)|(A  B)|[A]B (We normally omit outermost brackets.) Read ‘[A]B’ as ‘In an act of ROMS with suppositional input A, one imagines that B’, or ‘Given input A, one imagines B’, or so. 6.4.1 Semantics 1 Semantics 1 starts with a frame F = (W, {R A|A ∈ L}, T, +, t) with fve items: 1 W is the usual set of possible worlds; 2 {R A|A ∈ L} is a set of binary accessibilities between worlds: each formula A of L has its own R A ⊆ W × W;

Equivalence in Imagination

129

3 T is a set of possible topics or subject matters: what the formulas of L can be about (more on these in a minute); 4 + is an idempotent, commutative, associative operation on T: unrestricted topic fusion; it takes two topics and returns their sum or fusion; topic parthood can be defned from it the usual way: topic x is part of topic y (x ≤ y, a partial ordering) when x + y = y, the fusion of x and y is just y. 5 is a function assigning topics to all formulas of L via the following recursion: it frst assigns topics to all items in L AT; then if At(A) = {p1,…, pn} is the set of atoms in A, t(A) is the fusion over the atoms of A, +At(A) = t(p1) +… + t(pn); intuitively: what a formula is about is what its atoms, taken together, are about. Items (1) and (2) need little comment. Items (3)–(5) do: what are topics? Why should they allow fusion into further topics, and stand in parthood relations? Why are topics assigned to formulas in that way? Topics or subject matters are what meaningful items are about (Fine (2016), Humberstone (2008), Yablo (2014)). Aboutness is usually treated as a feature of linguistic representations, but of course mental representations bear aboutness, too: we call it intentionality. When one thinks that John is tall, one is thinking about John’s height. Topics are sometimes linked to questions. Lewis (1988) took them as partitions of the set of worlds: when the topic is the number of stars, the associated question is, ‘How many stars are there?’. Worlds end up in the same cell when they agree on the answer: all zero-star worlds, all one-star worlds, etc. Others (Fine (2016)) take topics as sets or fusions of truthmakers understood as situations à la Barwise and Perry (1983). We don’t need to take a stance on the nature of topics. For our logical purposes, we only need them to obey three constraints, widely agreed upon in subject matter semantics: i

Intensionally equivalent sentences A and B can differ in content when they are about different things. In Yablo’s (2014)’s version, the content of a sentence (in context) is not specifed just by the set of worlds in which it is true, but also by what it’s about. Subject matter semantics is, thus, hyperintensional: ‘2 + 2 = 4’ and ‘Either Jupiter is a planet, or not’ differ in content in spite of being true at the same worlds (all of them), for they say different things: only one is about the number 2. ii The space of topics has a mereological structure (Fine (2016), Yablo (2014)), hence our talk of fusions and parthood in (4) above. Topics can have proper parts; distinct topics may have common parts. Mathematics includes arithmetic. Mathematics and philosophy overlap, having (certain parts of) logic as a common part. The topic of A ∧ B includes that of A as a (proper) part; if Mary says ‘John is tall and thin’ and Ann says ‘John is tall’, what Ann said has already

130 Francesco Berto been said by Mary – who also said more: Ann has addressed a topic, say John’s height, which is a proper part of the larger topic addressed by Mary, the height and looks of John. iii The Boolean operators should add no subject matter of their own: they are ‘topic-transparent’ (Fine (2016), Perry (1989), Yablo (2014)). The topic of ¬A is the same as that of A (‘John isn’t tall’ is exactly about what ‘John is tall’ is about: John’s height; it certainly is not about not). Conjunction and disjunction merge topics (‘John is tall and thin’ and ‘John is tall or thin’ are about the same topic: the height and looks of John). In our L above, this reduces the topic of a formula to the fusion of the topics of its atoms. Hence our recursion in (5) above. A model M = (W, {R A|A∈L}, T, +, t, v) is a frame with a valuation v, assigning to each atom p in L AT a truth set v(p)⊆W, the set of worlds where p is true. This is extended to the whole L as follows. Read ‘w  A’ as saying that A is true at world w; we omit reference to the model as this is understood: (Sat) w  p iff w ∈ v(p) (S¬) w  ¬A iff w  A (S∧) w  A ∧ B iff w  A and w  B (S∨) w  A ∨ B iff w  A or w  B (S) w  A  B iff for all w1: if w1  A, then w1  B (S[A]) w  [A]B iff (i) for all w1: if wR Aw1, then w1  B and (ii) t(B) ≤ t(A) Negation, conjunction, and disjunction work as usual and the fshhook is a strict conditional. As for the last item: read ‘wR Aw1’ as saying that w1 is one of the worlds one looks at, in an act of ROMS (carried out at w) whose initial suppositional input is A. (What worlds does one look at? The intuition behind the minimal alteration principle would be that these are the most plausible worlds given one’s belief or knowledge state. This, however, is not really represented in the semantics. To do it, one can impose a total ordering of worlds by plausibility, mimicking what is done in the Lewis (1973) sphere semantics and in various epistemic logics for belief revision – see e.g. van Benthem (2011). I haven’t done it here, or in Berto (2017, 2018), for Semantics 1 or 2. But I have done it, for Semantics 1, in Berto (2019). One reason not to do it in the present context is that a semantics with the plausibility ordering automatically satisfes the condition representing equivalence in imagination, which I will present below. I want to discuss, instead, the opportunity of adding it manually.) One can equivalently formulate (S[A]) using set-selection functions à la Lewis (1973): each formula A of L has its own function, fA , taking as input the world where the act of mental simulation starting from A takes

Equivalence in Imagination

131

place, and outputting the corresponding bunch of accessible worlds, fA(w) = {w1 ∈ W|wR Aw1}. If |A| = {w ∈ W|w  A}, the truth set for A, one can rephrase the clause for [A] as the more compact: (S[A]) w ∈ [A]B iff (i) fA(w) ⊆ |B| and (ii) t(B) ≤ t(A) The two formulations are equivalent, as wR Aw1 iff w1 ∈ fA(w). But either is at times handier than the other to talk about the semantics. For [A]B to be true, we ask for two things: (i) is the truth-conditional component making of [A] a sententially indexed modal: B must be true throughout the worlds one looks at, given input A. (ii) is the topicality component: B must be fully on-topic with respect to what A is about. One natural Basic Constraint on the semantics is that, for all A ∈ L and w ∈ W: (BC) fA(w) ⊆ |A| This says that all the worlds one looks at, when one starts by supposing that A, will be worlds where A is true (plausibly enough: when we wonder what happens if Brexit causes a recession, we consider situations where Brexit causes a recession to begin with). From now on, we will only consider models satisfying BC. We defne logical consequence as truth preservation at all worlds of all models: where S is a set of formulas, S  A when in all models M = (W, {R A|A ∈ L}, T, +, t, v) and for all worlds w in W if w  B for all B in S, then w  A. As a special case, A is a logical truth,  A, when true at all worlds of all models. Our frst logical validity is secured via BC (for proofs of all the validities and invalidities listed from now on for Semantics 1, see Berto (2018)):  [A]A The suppositional input is always imagined. The next validities, taken together, characterize ROMS as ‘fully conjunctive’: (Simplifcation) [A](B ∧ C)  [A]B [A](B ∧ C)  [A]C (Adjunction) [A]B, [A]C  [A](B ∧ C) The insight is that imagined scenarios have some sort of mereological structure. Simplifcation has it that one who imagines the whole of a situation imagines the parts. You can’t imagine that John is tall and thin without imagining that John is tall because by imagining the former – the whole – you have already imagined the latter – the part. Adjunction has it that one who imagines the parts imagines the whole. You can’t

132 Francesco Berto imagine that John is tall and John is thin without imagining that John is tall and thin because, by imagining the parts all together, you have already imagined the whole. One may have more doubts about Adjunction than about Simplifcation. I discussed the point in Berto (2017, 2018), resorting to an old chestnut by Quine (1960) concerning same-antecedent counterfactuals allowing for conjunction of their consequents. The suppositional input in [A] has Caesar being in command of the US troops in the Korean war. We can imagine him using atomic bombs, B, if in our ROMS we keep fxed the weapons available in the Korean war, or we can imagine him using catapults, C, if we keep fxed the military apparatus available to Caesar. One would not thereby infer [A](B ∧ C), that Caesar employs both bombs and catapults. One can imagine that, too, if one wills, but it will spoil the reality-oriented nature of the exercise. It should not come as an automatic entailment mandated by the logic of mental simulation. In those works, I tried to argue that Adjunction should be maintained, and that doubts on it may have to do with the formalism not capturing the contextual character and temporal development of ROMS episodes. These are indexed only to formulas (the suppositional inputs), but it seems clear that the same input can trigger different imaginings in different contexts, and at different times, as it unfolds. The Quinean example embeds a shift. Once one sticks to a single context and time, Adjunction will work: you can’t imagine in one go, that is, in the same act of ROMS and at the same time, that John is tall and that John is thin without imagining that John is tall and thin, can you? Another line of argument may start by assuming that imagination (ROMS included) essentially involves mental imagery (see Kind (2001)). Unlike propositional or language-like mental representations, pictorial mental representations, also called mental imagery, have quasi-spatial features (Paivio (1986)). This may point at some mereological structure represented in the mind: when we visually imagine our bedroom, we can zoom into one part of the scenario – where the bedside table is – then zoom into one further sub-part – the book on the bedside table – then move upwards, etc. Classic empirical work in psychology (Block (1983), Shephard and Metzler (1971)) showed that the time taken to scan between two points of a mental image is generally proportional to their subjective distance; that larger objects fll the imagined scenario sooner than smaller ones; etc. Maybe the conjunctive nature of imagination is supported by such considerations on the intuitive mereology of mental imagery. It is, however, controversial (Gregory (2016), Van Leeuwen (2013), Williamson (2016)) that imagination essentially involves mental imagery, even if one admits that some mental representations represent pictorially, which is itself a controversial claim (see the so-called imagery debate, e.g., Pylyshyn (2002)). In mental simulation, we sometimes imagine

Equivalence in Imagination

133

scenarios involving only abstract objects (think of a mathematician mentally going through a proof, without making use of quasi-spatial structures like graphs or so); and sometimes we imagine complex situations whose representation needn’t involve mental imagery (think of yourself trying to predict how the markets worldwide will react to the next economic downturn). More logical reasons for dismissing such conjunctive hunches will come from considerations linked to equivalence in imagination, to which we get below. Lots of things come out invalid in the semantics, making the logic of our imaginative operators correspondingly weak – and rightly so. Four examples. First, their variable strictness models a key feature of ROMS: imagination is non-monotonic. [A]B  [A ∧ C]B Supposing Tweety is a bird, you imagine that it can fy. Supposing Tweety is a penguin bird, you don’t imagine that. Second, imagination under-determines its contents. This is guaranteed by A-accessibilities allowing one to look at a plurality of worlds: [A](B ∨ C)  [A]B ∨ [A]C When you imagine, given inputs provided by the page of The Lord of the Rings you are reading, that Boromir is either left-handed or righthanded (or ambidextrous) – he’s a normally endowed human being, after all – you don’t thereby imagine that he is left-handed, and you don’t thereby imagine that he is right-handed (LOTR tells you nothing about Boromir’s dominant hand). There’ll be worlds compatible with what you imagine where he’s left-handed, and compatible worlds where he’s right-handed. Third, imagination is not additive. This is guaranteed by topic-sensitivity: [A]B  [A](B ∨ C) When, starting from the supposition that Stauffenberg puts the bomb on the other side of the table, you imagine that Hitler gets killed, you don’t thereby imagine that either Hitler gets killed or there’s life on Kepler-442b. Fourth, imagination is hyperintensional. Even when A strictly or necessarily implies B, that is, there is just no possible way for A to be true without B being true, one needn’t imagine B starting from suppositional input A, when B is off-topic: A  B  [A]B

134

Francesco Berto

For a detailed discussion of the workings of Semantics 1, see Berto (2018). Now let’s look at the alternative. 6.4.2 Semantics 2 Semantics 2 starts with a frame F = (W, N, {R A|A ∈ L}) with three items: 1 W is a set of worlds; 2 N ⊆ W is the subset of normal worlds (think possible worlds); the items in W – N are the non-normal worlds (think of them as logically impossible worlds); 3 {R A|A ∈ L} is as in Semantics 1. There is no structure of topics. The truth conditions go as follows. A model M = (W, N, {R A|A ∈ L}, r) is a frame with a relation r, relating, for each world w, each atom p in L AT to truth, ‘rw p1’, falsity, ‘rw p0’, both, or neither. Unlike the usual valuations, a relation can connect a formula to more than one truth value at a world: formulas can be both true and false, truth value gluts. Formulas can also be truth value gaps, being related neither to truth, nor to falsity. Readers familiar with nonclassical logics will recognize one way of presenting the semantics for First Degree Entailment (FDE) logic (Belnap (1977), Dunn (1976), Priest (2008)), where one drops the classical presupposition that truth and falsity be exhaustive and exclusive. This is extended to the whole L via the following truth-and-falsity conditions at normal worlds (we need to give falsity conditions, too, as the semantics is not bivalent): (S1¬) rw(¬A)1 iff rwA0 (S2¬) rw(¬A)0 iff rwA1 (S1∧) rw(A ∧ B)1 iff rwA1 and rwB1 (S2∧) rw(A ∧ B)0 iff rwA0 or rwB0 (S1∨) rw(A ∨ B)1 iff rwA1 or rwB1 (S2∨) rw(A ∨ B)0 iff rwA0 and rwB0 (S1) rw(A  B)1 iff for all w1 ∈ N: if rw1A1, then rw1B1 (S2) rw(A  B)0 iff for some w1 ∈ N: rw1A1 and rw1B0 (S1[A]) rw([A]B)1 iff for all w1 ∈ W such that wR Aw1, rw1B1 (S2[A]) rw([A]B)0 iff for some w1 ∈ W such that wR Aw1, rw1B0 Negation, conjunction, and disjunction get the relational semantics for FDE. The fshhook is, again, a strict conditional: true at a normal world when all normal worlds where the antecedent is true also make the consequent true. [A] works, again, with accessibilities indexed by formulas. We can equivalently rephrase its truth and falsity conditions in a more compact way using set-selection functions:

Equivalence in Imagination

135

(S1[A]) rw([A]B)1 iff fA(w) ⊆ |B| (S2[A]) rw([A]B)0 iff fA(w) ∩ |¬B|≠ ∅ At points in W – N, r relates complex formulas to truth values directly, irrespective of their syntax. At non-normal worlds, a disjunction can be true although both disjuncts are false, etc. – this is the logically anarchic nature of such worlds. (The trick of assigning arbitrary truth values to complex formulas at non-normal worlds is due to Rantala (1982). Priest (2005, 2008) has put it to work in logic and metaphysics; I’ve done the same in Berto and Jago (2019).) The Basic Constraint goes thus for all A ∈ L: (BC) If w ∈ N, then fA(w) ⊆ |A| (We restrict to normal worlds, as non-normal worlds can do what they like.) Logical consequence is truth preservation at all normal worlds of all models: S  A when in all models M = (W, N, {R A|A ∈ L}, r)) and for all worlds w in N, if rwB1 for all B in S, then rwA1. Logical truth,  A, is truth at all normal worlds of all models. Non-normal worlds are taken as logically impossible worlds, where logic can fail. In the defnitions of logical truth and consequence, we only look at worlds where logic does not fail. Like Semantics 1, Semantics 2 makes of [A]A a validity, thanks to the BC. Unlike Semantics 1, Semantics 2 doesn’t make ROMS operators adjunctive as such: one can imagine, given input A, that B ∧ C is the case without imagining that B is the case and that C is the case, and vice versa. If we want to rule this out, we need the following two conditions. For Simplifcation: (C1) For all w ∈ N: if wR Aw1, and rw1(B ∧ C)1, then rw1B 1 and rw1C1 For Adjunction: (C2) For all w ∈ N: if wR Aw1, rw1B 1 and rw1C1, then rw1(B ∧ C)1 (For proofs that C1 and C2 validate, respectively, Simplifcation and Adjunction, and for proofs of all the validities and invalidities listed from now on for Semantics 2, see Berto (2017).) One may want to retain C1 and Simplifcation but make imagination non-adjunctive by not having C2, e.g., if one is persuaded by the quasi-Quinean worries discussed above, and unpersuaded by my replies. Semantics 2 wins in fexibility over Semantics 1. One may say that it loses in naturalness, as C1 and C2 may look contrived, or ad hoc. Otherwise, Semantics 2 proceeds hand in hand with Semantics 1 on a number of invalidities. It agrees with the latter in modelling the

136

Francesco Berto

non-monotonic nature of ROMS ([A]B does not entail [A ∧ C]B) and its under-determinacy ([A](B ∨ C) does not entail [A]B ∨ [A]C), thanks to [A]’s being a variably strict modal and to the accessibility of a plurality of worlds. But it can also mimic hyperintensionality (A  B does not entail [A]B) and the failure of Addition ([A]B does not entail [A](B ∨ C)), with no need for a topicality flter, thanks to the accessibility of non-normal worlds which can break necessary truths and consequences. The resulting logic is rather weak in both cases, but we can make it stronger by adding a constraint that captures the idea of equivalence in imagination.

6.5 Adding Equivalence in Imagination The constraint – let’s call it the Principle of Imaginative Equivalents (PIE) – is the same for both kinds of semantics. For all A, B in L: (PIE) If fA(w) ⊆ |B| and f B(w) ⊆ |A|, then fA(w) = f B(w) If all the selected A-worlds make B true and vice versa, then A and B are equivalent in imagination. Intuitively, when we take either as our suppositional input in an act of ROMS, we look at the same circumstances. PIE validates, both in Semantics 1 and in Semantics 2, a Substitutivity principle that greatly strengthens the logic: (Substitutivity) [A]B, [B]A, [A]C  [B]C Substitutivity says that equivalents in imagination, A and B, can be replaced salva veritate as modal indexes inside [ ]. It may be the case that we imagine different things in exercises of ROMS that start with intensionally equivalent suppositional inputs. But when we start from inputs that play the same role in our cognitive life, we will imagine the same things. (Again, which As and Bs count as cognitively equivalent depends on the specifc imaginative agent, and it’s not an issue for the logic to settle: x is a woodchuck and x is a groundhog may be equivalent for your fellow zoologist but not for you, etc.) Given the naturalness of the notion of imaginative equivalence – given how it connects to the plausible ideas of cognitive equivalence and cognitive synonymy – Substitutivity looks like a good principle to have in a logic of imagination as ROMS. Say that bachelor and unmarried man are cognitively equivalent for you qua competent speaker of English: when you suppose that John is unmarried, you imagine that he’s a bachelor ([A]B) and vice versa ([B]A). Supposing that John is unmarried, you imagine that he has no marriage allowance, [A]C. Then the same happens when you suppose that John is a bachelor, [B]C. Your imagination won’t work that smoothly with Groundhogs are rodent and Woodchucks are rodent – unless you’re a zoologist; in this case, it probably will.

Equivalence in Imagination

137

Now for the glitch. In Semantics 1, PIE validates a further inference we may call, following literature on non-monotonic logics, Special or Restricted Transitivity: (RT) [A]B, [A ∧ B]C  [A]C Is RT good? It’s easy to give plausible instances. The issue has to do with cases where C easily pops to mind given B alone, but is only dimly related to A. Then RT, acting a bit like a Cut rule in a logical calculus, washes the bridging B away in the conclusion. Here’s a situation suggested by Claudio Calosi, which may work as a counterexample. [A]B: supposing that I am wearing a red shirt in Pamplona, I imagine that I am being chased by bulls. [A ∧ B]C: supposing that I am being chased by bulls on the streets of Pamplona while wearing a red shirt, I imagine that I die on the street. But it’s not the case that [A]C: supposing that I am wearing a red shirt in Pamplona, I don’t imagine that I die on its streets. We here mark a crucial difference with Semantics 2. There, the proof that RT is valid given PIE needs to make use of the Simplifcation- and Adjunction-validating conditions C1 and C2 above. It goes thus: suppose normal world w verifes (i) [A]B and (ii) [A ∧ B]C. Given the BC, w also verifes [A]A and so, by (i) and C2, it verifes [A](A ∧ B). From BC again, w also verifes [A ∧ B](A ∧ B) and hence, by C1, it verifes [A ∧ B A. Then by (S1[A]), fA(w) ⊆ |A ∧ B| and fA∧B(w) ⊆ |A| and so, by PIE, fA(w) = fA∧B(w). Since (ii) w also verifes [A ∧ B]C, by (S1[A]) again we have fA∧B(w) ⊆ |C|, hence fA(w) ⊆ |C|, and so w verifes [A]C. Now both C1 and C2 are added to Semantics 2 manually: if one drops either, one can retain Substitutivity as desired, without having RT. For the proof of Substitutivity doesn’t need them. It goes thus: suppose normal world w verifes [A]B, [B]A, [A]C. By (S1[A]), we have fA(w) ⊆ |B|, f B(w) ⊆ |A|, fA(w) ⊆ |C|. PIE then gives us fA(w) = f B(w), and hence f B(w) ⊆ |C|. So by (S1[A]), w verifes [B]C. Semantics 1 cannot afford such fexibility: it is inherently adjunctive, for there is no addition of alien topic in the move from a conjunction to its conjuncts or vice versa. It may be that intuitive counterexamples to RT are forceful enough to have us reconsider the idea that ROMS is fully conjunctive after all! Here’s another glitch involving PIE. Pierre Saint-Germier (forthcoming) has recently pointed out that if A and A′ are logically equivalent and the topic of A′ is included in that of A, then [A]B and [A′]B turn out to be logically equivalent. The problem is generated by the equivalence of instances like [p]q and [p ∨ (p ∧ r)]q. I reply that the only case in which the difference between p and p ∨ (p ∧ r) matters, for they should not be cognitively equivalent qua suppositional inputs, is when the agent does not grasp the proposition expressed by r, due to some conceptual defcit. But I grant that it is indeed a problem for my approach as it is now, that it doesn’t represent the difference. I should think about what to do!

138

Francesco Berto

6.6 Summary I have proposed to understand one kind of epistemically useful imagination as reality-oriented mental simulation (ROMS): we suppose that A takes on board relevant background beliefs and knowledge, and we wonder whether B turns out in the imagined scenario. The scenario we focus on is constrained by its relevant connections with the suppositional input A and governed by a principle of minimal alteration. ROMS allows us to enlarge our stock of conditional beliefs: if B turns out in the imagined scenario, we add to our belief stock the belief in B conditional on A. I have then discussed the idea that, in spite of its hyperintensional anarchy, ROMS obeys a principle of equivalence in imagination: when A and B play interchangeable roles in one’s cognitive life, they play interchangeable roles, in particular, as suppositional inputs in ROMS. I have proposed to use the notation ‘[A]B’ for ‘Given input A, one imagines that B’, or ‘In an exercise of mental simulation starting from supposition A, one imagines that B’. I have interpreted [A] as a variably strict, non-normal modal and given two alternative semantics for it: Semantics 1 uses possible worlds plus an added structure of topics. Semantics 2 uses non-normal or impossible worlds. I have shown how a constraint representing equivalence in imagination can be added to both semantics, and how it works within them in slightly but interestingly different ways. Specifcally, Semantics 2 has the advantage of validating Substitutivity, which seems to capture something important about likeness in cognitive role, without validating Special/Restricted Transitivity, which may have clear counterexamples. But it does so via manually added constraints that may look artifcial, or ad hoc. Semantics 1 cannot keep Substitutivity and Special/Restricted transitivity apart. It is, however, based on adding a topicality flter to an otherwise very classical, normal modal logical framework; this may make Semantics 1 more appealing to conservative logicians.1

Note 1 This research is published within the Project ‘The Logic of Conceivability’, which is funded by the European Research Council (ERC CoG) under Grant Number 681404. Thanks to Amy Kind, Chris Badura, and an anonymous reviewer for tremendously helpful comments which allowed me to greatly improve the initial draft. The two formal semantics presented below rely on previously published works: Berto (2017, 2018, 2019). A number of people are thanked there already.

References Adams, E. 1998. A Primer of Probability Logic. Stanford, CA: CSLI. Arcangeli, M. 2019. Supposition and the Imaginative Realm. London and New York: Routledge.

Equivalence in Imagination

139

Barwise, J. and J. Perry. 1983. Situations and Attitudes. Cambridge, MA: MIT Press. Belnap, N. 1977. “A Useful Four-valued Logic”. In Modern Uses of Multiple-valued Logic, ed. by J. Dunn and G. Epstein. Dordrecht: Reidel. Bennett, J. 2003. A Philosophical Guide to Conditionals. Oxford: Oxford UP. van Benthem, J. 2011. Logical Dynamics of Information and Interaction. Cambridge: Cambridge UP. Berto, F. 2017. “Impossible Worlds and the Logic of Imagination”. Erkenntnis 82: 1277–97. Berto, F. 2018. “Aboutness in Imagination”. Philosophical Studies 175: 1871–86. Berto, F. 2019. “Taming the Runabout Imagination Ticket”. Synthese, On Line First https://doi.org/10.1007/s11229-018-1751-6. Berto, F. and M. Jago. 2019. Impossible Worlds. Oxford: Oxford UP. Block, N. 1983. “Mental Pictures and Cognitive Science”. The Philosophical Review 92: 499–541. Brown, J. 2018. Fallibilism: Evidence and Knowledge. Oxford: Oxford UP. Byrne, R. 2005. The Rational Imagination. Cambridge, MA: MIT Press. Canavotto, I., F. Berto and A. Giordani. 2020. “Voluntary Imagination: A FineGrained Analysis”. Review of Symbolic Logic, forthcoming. Chellas, B. 1975. “Basic Conditional Logic”. Journal of Philosophical Logic 4: 133–53. Cruse, D.A. 2000. Meaning in Language. Oxford: Oxford UP. Currie, G. and I. Ravenscroft. 2002. Recreative Minds. Oxford: Oxford UP. Dorsch, F. 2012. The Unity of Imagining. Frankfurt: Ontos Verlag. Douven, I. 2016. The Epistemology of Indicative Conditionals: Formal and Empirical Approaches. Cambridge: Cambridge UP. Dunn, J.M. 1976. “Intuitive Semantics for First-Degree Entailments and Coupled Trees”. Philosophical Studies 29: 149–68. Edgington, D. 1995. “On Conditionals”. Mind 104: 235–329. Evans, J. and D. Over. 2004. If. Oxford: Oxford UP. Fine, K. 2016. “Angellic Content”. Journal of Philosophical Logic 45: 199–226. Gendler, T.S. 2000. “The Puzzle of Imaginative Resistance”. Journal of Philosophy 97: 55–81. Goldman, A. 2006. Simulating Minds. Oxford: Oxford UP. Gregory, D. 2016. “Imagination and Mental Imagery”. In The Routledge Handbook of Philosophy of Imagination, ed. by A. Kind. London; New York: Routledge. Hawke, P. 2016. “Questions, Topics, and Restricted Closure”. Philosophical Studies 173: 2759–84. Hawthorne, J. 2004. Knowledge and Lotteries. Oxford: Oxford UP. Hintikka, J. 1962. Knowledge and Belief. Ithaca, NY: Cornell UP. Hornischer, L. 2017. Hyperintensionality and Synonymy. MSc Master of Logic, University of Amsterdam. Humberstone, L. 2008. “Parts and Partitions”. Theoria 66: 41–82. Kahneman, D., P. Slovic and A. Tversky. 1982. Judgment under Uncertainty. Cambridge: Cambridge UP. Kind, A. 2001. “Putting the Image Back in Imagination”. Philosophy and Phenomenological Research 62: 85–109. Kind, A. 2016. “Imagining under Constraints”. In Knowledge Through Imagination, ed. by A. Kind and P. Kung. Oxford: Oxford UP: 145–59.

140 Francesco Berto Kind, A. and P. Kung. 2016. Knowledge Through Imagination. Oxford: Oxford UP. Kraus, S., D. Lehmann and M. Magidor. 1990. “Nonmonotonic Reasoning, Preferential Models, and Cumulative Logics”. Artifcial Intelligence 44: 167–207. Langland-Hassan, P. 2016. “On Choosing What to Imagine”. In Knowledge Through Imagination, ed. by A. Kind and P. Kung. Oxford: Oxford UP: 61–84. Lewis, D. 1973. Counterfactuals. Oxford: Blackwell. Lewis, D. 1976. “Probabilities of Conditionals and Conditional Probabilities”. Philosophical Review 95: 581–9. Lewis, D. 1988. “Relevant Implication”. Theoria 54: 161–74. Lyons, J. 1996. Linguistic Semantics. Cambridge: Cambridge UP. Mulligan, K. 1999. “La varietà e l’unità dell’immaginazione”. Rivista di estetica 11: 53–67. Murphy, M.L. 2003. Semantic Relations and the Lexicon. Cambridge: Cambridge UP. Nozick, R. 1981. Philosophical explanations. Cambridge, MA: Harvard UP. Oaksford, M. and N. Chater (eds.). 2010. Cognition and Conditionals. Oxford: Oxford UP. Paivio, A. 1986. Mental Representation. Oxford: Oxford UP. Perry, J. 1989. “Possible Worlds and Subject Matter”. In The Problem of the Essential Indexical and Other Essays, ed. by J. Perry. Palo Alto, CA: CSLI: 145–60. Priest, G. 2005. Towards Non-Being. Oxford: Oxford UP. Priest, G. 2008. An Introduction to Non-Classical Logic. Cambridge: Cambridge UP. Pylyshyn, Z. 2002. “Mental Imagery: In Search of a Theory”. Behavioral and Brain Sciences 25: 157–82. Quine, W.V.O. 1960. Word and Object. Cambridge, MA: MIT Press. Rantala, V. 1982. “Impossible World Semantics and Logical Omniscience”. Acta Philosophica Fennica 35: 106–15. Saint-Germier, P. forthcoming. “Hyperintensionality in Imagination”. In Logic in High Defnition, Trends in Logical Semantics, ed. by A. Giordani and J. Malinowski. Dordrecht: Springer: 79–117. Shephard, R. and J. Metzler. 1971. “Mental Rotation of Three-Dimensional Objects”. Science 171: 701–3. Stanojecvić, M. 2009. “Cognitive Synonymy: A General Overview”. Facta Universitatis 7: 193–200. Stenning, K. and M. van Lambalgen. 2008. Human Reasoning and Cognitive Science. Cambridge, MA: MIT Press. Strohminger, M. and J. Yli-Vakkuri. 2019. “Knowledge of Objective Modality”. Philosophical Studies 176: 1155–75. Stuart, M. 2020. “The Productive Anarchy of Scientifc Imagination”. Philosophy of Science, forthcoming http://philsci-archive.pitt.edu/16749/. Van Leeuwen, N. 2013. “The Meanings of ‘Imagine’ Part I: Constructive Imagination”. Philosophy Compass 8: 220–30. Walton, K. 1990. Mimesis as Make-Believe. Cambridge, MA: Harvard UP. Williamson, T. 2016. “Knowing by Imagining”. In Knowledge Through Imagination, ed. by A. Kind and P. Kung. Oxford: Oxford UP: 113–23. Yablo, S. 2014. Aboutness. Princeton, NJ: Princeton UP.

7

How Imagination Can Justify Christopher Badura

7.1 Introduction According to Amy Kind, some imaginative episodes are epistemically useful: they prompt and justify beliefs (Kind, 2018). To defend this claim, Kind discusses examples of highly skilled imaginers whose imaginative episodes were crucial to the development of their inventions (Kind, 2016, 2018). Not only extraordinarily gifted imaginers are able to engage in epistemically useful imaginative episodes. Especially in decisionmaking, even those of us who are not especially gifted at imagination employ our imagination. Consider the following (simplifed) example: Mountain Climbing (MC) Anika contemplates whether to go mountain climbing in the Alps. She has never been mountain climbing and doesn’t know what it is like to go mountain climbing, in the Alps or anywhere else. She sets out to imagine that she is mountain climbing in the Alps. Of course, it is important for her to approximate in her imagination what it is like to be mountain climbing. For if she doesn’t think she’ll like what it’s like, she won’t go. Anika has previously tried out indoor climbing. So, she knows what it’s like to go indoor climbing. She has beliefs about this experience and beliefs about indoor climbing: she believes that indoor climbing exhausts her. Since she believes that this experience is similar to what it’s like to go mountain climbing, she believes that going mountain climbing in the Alps will exhaust her. Given her desire to not be exhausted, she decides not to go mountain climbing. Seemingly, Anika’s imaginative episode justifes her belief that going mountain climbing in the Alps will exhaust her (BE). What enables this imaginative episode to be justifcatory? Various authors have suggested that such imaginings unfold under suitable constraints; see, e.g., Langland-Hassan (2016); Williamson (2016); Berto (2018). Amy Kind suggests that epistemically useful imaginative episodes are characterized

142 Christopher Badura by obeying, or at least trying to obey, the Reality Constraint and the Change Constraint; cf. Kind (2016, p. 151): (RC) The world is imagined as it is; (CC) If a change to the world as it is believed to be is imagined, then one’s further imagining is guided by the logical consequences of this change. Kind points out that these are not necessary conditions for epistemically useful imaginings and offers “relevantized” versions as an alternative; cf. Kind (2016, p. 153): (rRC) The world is imagined as it is in all relevant respects; (rCC) If a change to the world as it is believed to be is imagined, then one is guided by the relevant consequences of this change. Whether (rRC) and (rCC) are necessary conditions for epistemically useful imaginings, Kind notes, depends on the notions of relevance and relevant consequence. The latter and hence (rCC) seem to capture the idea that one’s imagination unfolds in a suitable manner. The frst aim (Aim 1) of this chapter is to provide a rigorous understanding of the unfolding process that obeys (rRC) and (rCC) and that delivers the relevant consequences of an imagined change. The second aim (Aim 2) is to explain how an imaginative episode unfolding like this can be justifcatory. The chapter approaches these two aims from the perspective of epistemology and philosophy of mind, although it also features some considerations and concepts from epistemic-doxastic logic and formal epistemology. The proposal is inspired by Peter Langland-Hassan’s discussion of what he calls “guided chosen imaginings”, which are epistemically useful imaginings.1 He points out that a characteristic of such imaginings is that they are ‘belief-like’ in that they unfold more or less in accord with the norms that govern belief. […] If true, it would explain why most imaginings do not develop in a completely arbitrary manner, but instead often match what we would come to believe if we believed the initiating premise of the imagining. […] We can then understand the lateral constraints on propositional imagination to the extent – and only to the extent – that we understand such constraints with respect to belief. (Langland-Hassan, 2016, pp. 68–9) In formal epistemology and doxastic-epistemic logic, conditional beliefs encode the belief change behavior of an agent and thus are a suitable

How Imagination Can Justify

143

way of understanding “what we would come to believe” if we believed something else. Conditional beliefs satisfy certain norms/constraints/ postulates that govern how we change our beliefs. Taking the quote from Langland-Hassan seriously, conditional beliefs thus seem to be a suitable candidate to feature in epistemically useful imaginative episodes and to provide the relevant consequences to an imagined change. Since they also come with a notion of justifcation (Leitgeb, 2007), they also help in achieving my second aim. The plan of the chapter is as follows. I take a conditional belief in C given A as a disposition to believe C if one believed A. I explain how, in principle, their involvement in imagination provides us with relevant consequences and how they transmit justifcation, addressing Aim 1 and Aim 2. I then raise a problem, namely that we might imagine some content A for which we do not have any conditional belief such that we are disposed to believe (some) C if we believed A. To solve this problem, I appeal to the notion of “imaginative scaffolding” introduced by Kind (2020a), which I argue can be generalized from being involved in experiential imagination to being involved in propositional imagination. Adding this provides us with a more rigorous understanding of the overall process that is epistemically useful imagination, pursuing Aim 1. Thereafter, I discuss why I appeal to conditional beliefs and not beliefs in conditionals. In the last section, by way of conclusion, I summarize my proposal and point toward two anticipated objections. A note on terminology: from here on, I speak of “imaginative episodes” only but always mean “epistemically useful imaginative episodes” or “imaginative episodes put to epistemic use”. The latter are imaginative episodes which are meant to provide justifcation for a certain belief, although they might be unsuccessful in doing so (that is roughly the instructive use of imagination, as introduced by Kind & Kung (2016)). In the context of justifcation, I will be concerned with propositional imagination, which I express by “agent a imagines that p”. However, experiential imaginings, which I express by “agent a imagines what it’s like to G” or “a imagines being G”, will also feature when I talk about imaginative scaffolding. These two can combine into imaginings expressed by “a imagines that this is what it’s like to G”. This, I assume, is a propositional imagining accompanied by an experiential imagining. Occasionally, I address sensory imaginings, which I express by “a imagines b” or “a imagines a/an/the F”. I leave open the exact relation between the three types of imagination; see, e.g., Kind (2001); Balcerak Jackson (2016).

7.2 Conditional Beliefs Anika sets out to imagine that she is mountain climbing. The example is set up in such a way that she doesn’t really have any beliefs about

144

Christopher Badura

mountain climbing that she can appeal to directly. She needs to make a detour through her beliefs about indoor climbing, and specifcally, that indoor climbing exhausts her. This detour, it seems, is associative rather than by logical consequence. Yet, it is not entirely arbitrary. There are two questions I will answer at this stage: frst, how do her beliefs concerning indoor climbing provide justifcation for (BE)? Second, how does she “select” these beliefs? To answer the frst question, I say more about what I take to be the kind of beliefs involved, namely conditional beliefs. To the second question, I turn in the next section. According to Hannes Leitgeb, a conditional belief in C given A – notation: B(C|A)—is a disposition to believe C if one believed A ceteris paribus (Leitgeb, 2007). 2 (From here on I will omit the “ceteris paribus” for brevity.) This disposition can also be expressed by a counterfactual, although I am not claiming that this is the correct analysis of the disposition: if one believed A (ceteris paribus), then one would believe C. I call the belief in the antecedent A the “antecedent belief” and the belief in the consequent C the “consequent belief”. This means that the stimulus condition of the conditional belief B(C|A) is the antecedent belief, and the manifestation condition is the consequent belief. For now, grant me that conditional beliefs are different from beliefs in conditionals like “Anika believes that if she goes/went mountain climbing, she will/would be exhausted”—notation: B(If A, C). I will address the relation between the two in Section 7.4 and explain why I rely on conditional beliefs rather beliefs in conditionals. Anika has set out to imagine that she is mountain climbing in the Alps. Somehow (to be explained in the next section) that has made her think about indoor climbing. I assume that Anika is disposed to believe that she will be exhausted if she believed that she went indoor climbing, i.e., Anika has the conditional belief B(I will be exhausted|I go indoor climbing). A disposition is manifested just in case the stimulus condition occurs (and there are no fnks or antidotes); see Choi & Fara (2018) for an overview of dispositions. When Anika imagines or thinks about that she goes indoor climbing, usually she doesn’t believe that she goes indoor climbing. Neither does Anika believe that she feels or will feel exhausted, and hence the manifestation condition does not occur either in the imaginative episode. So, how can the conditional belief play any role in the imaginative episode if neither its stimulus condition nor its manifestation occurs? Some infuential accounts of imagination understand imagination as the simulation, or recreation, of other mental states that are run “offine” (Currie & Ravenscroft, 2012; Arcangeli, this volume). My suggestion is along similar lines, although I am not going as far as to claiming that Anika recreates some other mental state(s).3 Anika “offine-manifests” her disposition to believe that she will be exhausted if she believed she went indoor climbing. Let me explain how I would like to understand this offine-manifestation.

How Imagination Can Justify

145

Following Currie and Ravenscroft, one might suggest that Anika is simply recreating believing that she is indoor climbing. On their account, this would mean that she is imagining (or conceiving or supposing) that she is indoor climbing. But this isn’t quite right, it seems. Anika certainly has some affrmative attitude toward the proposition that she is indoor climbing, but she is not imagining that she is indoor climbing, rather she imagines that she is mountain climbing. If we cannot take the offine-stimulus condition for her disposition B(I will be exhausted|I go indoor climbing) to be that she imagines that she is indoor climbing, what, then, is the offine-analogue of the stimulus condition that she believes that she is indoor climbing? It seems plausible that Anika is thinking of a situation in which she is/was indoor climbing. She might be thinking of a particular occasion when she was indoor climbing (that is, she might be remembering episodically), but she might also just think of some general representative of an indoor climbing situation. Is “thinking of” suitable to explain the offine-manifestation of the disposition and offer an offine-analogue of the online stimulus condition? It does not seem so. The disposition requires the occurrence of a propositional attitude as a stimulus condition, namely that Anika believes that she is indoor climbing. The expression “Anika is thinking of that Anika is indoor climbing” is ungrammatical. What about “thinking that”? The expression “Anika thinks that she is indoor climbing” seems equivalent to “Anika believes that she is mountain climbing”. So, the pickle is this: while Anika seems to be thinking of a situation, her disposition requires some propositional attitude as an offine-analogue of the stimulus condition for the disposition to be offine-manifested. So, her thinking of the situation cannot offine-manifest the disposition in a straightforward way. I suggest the technical term “acknowledgment” to account for the affrmative attitude Anika has toward the situation she is thinking of and toward the proposition that she is indoor climbing. That is, Anika acknowledges a situation in which she is indoor climbing, and she acknowledges that she is indoor climbing. Hence, acknowledgment is a non-propositional but also a propositional attitude. Moreover, I assume that if one imagines that A, then one acknowledges that A (but not vice versa). (So the story I am about to tell transfers to the case in which Anika does have conditional beliefs that involve as antecedent what she imagines.) It is this affrmative attitude that plays the role of the offine-stimulus condition analogous to the online stimulus condition that Anika believes that she is indoor climbing. That is, Anika’s disposition to believe C if she believed A is offine-manifested if Anika acknowledges that A. The manifestation condition that Anika believes that C is offine-manifested if she acknowledges that C (as a consequence of the occurrence of the offine-stimulus condition). In our example, Anika acknowledges that she will be exhausted.

146 Christopher Badura So far, I have explained what it is for a disposition to believe C if one believed A to be offine-manifested. This was necessary because in (MC) Anika doesn’t end up believing that she is exhausted, and hence her disposition is not online-manifested. Moreover, she does not imagine that she is indoor climbing when imagining that she is mountain climbing. Rather, she has some general affrmative attitude, what I have called “acknowledgment”, toward the proposition that she is indoor climbing. This alone, however, doesn’t explain why Anika is now justifed in believing that mountain climbing will exhaust her. To this, I turn next. Part of the fnal answer of how an imaginative episode is justifcatory will have to wait until the next section, however, for I have assumed that there is some way in which Anika gets from imagining that she is mountain climbing to acknowledging that she is indoor climbing. The justifcatory force of an imaginative episode will also depend on the justifcatory force of this “blackbox”, which I address only in the next section. Here, to address the issue concerning the justifcatory force of an imaginative episode, I show how conditional beliefs add some justifcatory force to one’s imaginative episodes. Hannes Leitgeb offers three ways to understand what it is for a conditional belief to be justifed, depending on one’s favorite account of justifcation: [W]e are led to the following process-reliabilist account of justifcation for conditional beliefs: (a) a conditional belief is justifed if and only if it was produced by a reliable process; (b) consider any process that only produces conditional beliefs: such a process is reliable if and only if most of the conditional beliefs it produces, or the great majority thereof, do not lead from truths to falsehoods. […] In the case of foundationalist internalism, justifcation is supposed to be transferred from basic to non-basic beliefs. Whatever the notion of deductive or inductive justifcation-transfer – the reason-for relation – looks like that is used to explain in what sense beliefs can be epistemically founded on, and argued for, on the basis of more primitive beliefs, it must automatically yield a notion of justifcation for conditional beliefs: call a conditional belief in [C] given A justifed if and only if the transition from the belief in A to the belief in [C] is justifcation-transmitting, or A is a reason for [C]. On the coherentist picture, a conditional belief will be justifed (with respect to a system S of beliefs) if and only if (a) it belongs to S, (b) S is coherent, and (c) the coherence of S is suffciently accessible to the agent to enable her to put forward arguments in favour of it on the basis of the coherence of S. The coherence of a belief system can be explained in terms of rationality constraints on beliefs and on the way these beliefs get updated (the axioms for subjective probability spaces and those for belief revision operators are standard

How Imagination Can Justify

147

examples); the constraints themselves are justifed on the basis of Dutch book arguments. (Leitgeb, 2007, p. 130, notation adjusted) In the example, Anika acknowledges that she is indoor climbing. Given her conditional belief B(I will be exhausted|I go indoor climbing) is justifed in the process-reliabilist sense, this means that if her belief that she is indoor climbing was true, her belief that she will be exhausted would also be true. So, if Anika was indoor climbing, it would be true that she will be exhausted. As I have argued above, Anika also acknowledges that she will be exhausted if she acknowledges that she is indoor climbing. Given these two acknowledged propositions and her conditional belief and nothing else, it follows that if the acknowledged proposition that she is indoor climbing was true, then also the acknowledged proposition that she will be exhausted would be true. Hence, justifcation in the process-reliabilist sense is preserved. What about the case of foundationalist internalism? Suppose Anika’s conditional belief B(I will be exhausted|I go indoor climbing) is internalistically justifed. It means that the transition from Anika’s believing that she goes indoor climbing to her believing that she will be exhausted is justifcation-transmitting, or that her belief that she is indoor climbing is a reason for her belief that she will be exhausted. This entails that if Anika has a reason for her belief that she is indoor climbing, then she has a reason for her belief that she will be exhausted. Does it follow that the transition from Anika acknowledging that she is indoor climbing to acknowledging that she will be exhausted is justifcation-transmitting; or that Anika’s acknowledging that she is indoor climbing is a reason for her acknowledging that she will be exhausted? Recall that it is plausible that Anika acknowledges not only that she is indoor climbing, but that she also acknowledges a situation in which she goes indoor climbing. This situation is, all things being equal, such that Anika would be justifed in believing that she goes indoor climbing (if she were in that situation).4 Together with her justifed conditional belief B(I will be exhausted|I go indoor climbing), this entails, all things being equal, that in such a situation, Anika would be justifed in believing that she will be exhausted. So, the transition from Anika acknowledging that she is indoor climbing to her acknowledging that she will be exhausted is justifcation-transmitting if her conditional belief is justifed. Finally, let me address the coherentist account. Both the probabilistic and the belief revision account are formal accounts that come with a formal notion of conditional belief. Assuming a naive probabilistic account, Anika’s belief system is given by her subjective probability function Pr (also: degree of belief), which obeys the Kolmogorov axioms of probability. Anika’s conditional belief B(I will be exhausted|I go indoor climbing) is justifed with respect to her belief system if it is identical to

148 Christopher Badura the subjective conditional probability Pr(I will be exhausted|I go indoor climbing), which is defned in terms of her subjective probability function.5 In such a framework, one can derive Jeffrey’s principle of conditionalization, which states that one’s degree of belief in C after “learning” A should be identical to the conditional degree of belief Pr(C|A) times one’s degree of belief Pr(A) before learning A, plus the conditional degree of belief Pr(C|not-A) times one’s degree of belief Pr(not-A) before learning A.6 The detailed workings of the principle don’t matter for my purposes. For simplicity, let’s assume when Anika acknowledges that she goes indoor climbing, she temporarily sets Pr(I go indoor climbing)=1.7 Suppose Anika’s conditional belief is justifed. Let us also take on board Langland-Hassan’s assumption that our imagination is constrained by the same principles or postulates that govern our beliefs. Then it is reasonable that Anika’s “degree of acknowledgment” in “I am exhausted” should be equal to the conditional degree of belief Pr(I will be exhausted|I go indoor climbing) times her degree of belief Pr(I go indoor climbing) before acknowledging that she goes indoor climbing, plus the conditional degree of belief Pr(I will be exhausted|I do not go indoor climbing) times her degree of belief Pr(I do not go indoor climbing) before acknowledging that she goes indoor climbing. The fact that Anika is not really updating her degree of belief in C after learning A but rather offine-updates her belief in C (i.e., assigns a degree of acknowledgment) is captured by the fact that she considers her degrees of beliefs assigned prior to acknowledging A, rather than the one prior to learning A. Of course, it is likely that her degree of belief Pr(I go indoor climbing) before learning A is equal to her degree of belief Pr(I go indoor climbing) before acknowledging A. Since her degree of acknowledgment in “I will be exhausted” is governed by her justifed conditional beliefs and Jeffrey conditionalization (which is one of the standard rationality postulates governing updating), her degree of acknowledgment in “I will be exhausted” will be coherent with her other beliefs and acknowledgments during the imaginative episode. There are some issues lurking here because traditional probabilistic and belief revision accounts come with a formal concept of conditional belief that makes believers logically omniscient (see endnote 5). For example, from the formal perspective, in standard Bayesian epistemology, an agent has conditional beliefs B(C|A) for any C and A, where C is a tautology and A has non-zero probability. I won’t discuss these issues here because they are problems for the formal notion of conditional belief. That is, whatever the solution to these formal problems is, it carries over to the dispositions to believe as psychological entities. Of course, which of the three accounts of justifcation one employs might depend on one’s overall philosophical theory. I am not advocating any of the three. As I have argued, all three provide sensible explanations of how imaginative episodes are justifcatory.

How Imagination Can Justify

149

So far, I have only been concerned with Anika’s conditional belief B(I will be exhausted|I go indoor climbing). I have argued that this disposition is offine-manifested. The disposition is offine-manifested, I  assumed, because Anika has an affrmative attitude toward “I go climbing”. I introduced the technical term “acknowledgment” for this analogue to the stimulus condition that Anika believes that she is mountain climbing. The offine-stimulus condition is thus that Anika acknowledges that she is indoor climbing. The manifestation condition is that Anika acknowledges that she will be exhausted. I have also explained how, if Anika acknowledges that she is indoor climbing and she acknowledges that she will be exhausted, she is justifed in her latter acknowledgment if the initial conditional belief itself is justifed. What remains to be shown is how Anika’s imagining that she is mountain climbing gives rise to her acknowledging that she is indoor climbing in a way that is, in principle, justifcatory. This is the subject of the next section.

7.3 Imaginative Episodes, Subject Matter, Topic, Scaffolding The following two problems are to be addressed: frst, how does Anika get from imagining that she is mountain climbing to acknowledging that she is indoor climbing? Second, how does all this provide justifcation for (BE)? To answer the frst question, Amy Kind’s notion of imaginative scaffolding comes in handy; see Kind (2020a, 2020b). In her discussion of L.A. Paul’s transformative experiences, Kind argues that it seems at least possible to know what a certain transformative experience is like. For example, Paul argues that one cannot know what it is like to be a parent unless one experiences being a parent. Kind argues against this that by adding, subtracting, and modifying experiences we had had, we can imagine a new experience that approximates that of being a parent. Kind calls this process of addition, subtraction, and modifcation of one’s previous experiences in one’s imagination “imaginative scaffolding”. I do not mean to defend Kind’s claim against Paul here. I am not even claiming that we are especially good at imaginative scaffolding. Rather, I want to put the notion of imaginative scaffolding to use in explaining how Anika gets from imagining that she is mountain climbing to acknowledging that she is indoor climbing. The justifcatory force of an imaginative episode will then crucially depend on how well one imaginatively scaffolds. Kind introduced imaginative scaffolding in the context of experiential imagination but the part of (MC) relevant for justifcation is relating Anika propositionally imagining that she is mountain climbing with acknowledging that she is indoor climbing. Although I have set up the

150 Christopher Badura example in such a way that Anika imagines what it’s like to go mountain climbing, one might argue that the whole example could be phrased in terms of propositional imagination. In such a case, it is not clear how imaginative scaffolding is meant to apply. So, I will frst explain how imaginative scaffolding helps in the example as it is set up. Then I proceed to generalize it to propositional imagination. Anika imagines that she is mountain climbing. This, according to (MC), involves her to imagine what it’s like to go mountain climbing. Since she has no experience of what it is like to go mountain climbing, she needs to draw on experiences that she has had already. She knows what it is like to go indoor climbing: exhausting. Since mountain climbing is a kind of climbing and indoor climbing is too, she has good reason to believe that this will involve similar experiences. Let me generalize this to the case of propositional imagination. Kind focuses on the “synthesizing” aspect of imaginative scaffolding, which combines previous experiences into something new. What seems to be a key component and a pre-condition of the scaffolding is to frst select the relevant experiences. But this selection cannot be based on the phenomenal content of Anika’s experience of mountain climbing for she has none prior to the scaffolding.8 That is, how can Anika know which experiences are relevant to that of mountain climbing if she has no clue on what the experience she is trying to imagine is like? The point here is this: if one doesn’t know what X is like, how does one know that what Y is like is relevant for what X is like? I take it that, usually, one knows what one’s cognitive attitudes are about.9 I assume that just like sentences are about certain topics (or subject matters), our cognitive attitudes are about topics or subject matters, too.10 Anika’s imagining is about the topic mountain climbing and is partly about the topic climbing (in general) (I use capitals to refer to topics and subject matters).11 These two topics are related to each other: the frst topic is more informative than the second because the frst conceptually entails the second. Note that this relation is determined by reality and thus obeys (rRC). Similarly, for the case of indoor climbing and climbing. But what is the relation between mountain climbing and indoor climbing? Their relation is somehow mediated by both being related to the topic climbing.12 Intuitively, Anika’s imagining that she is mountain climbing and her acknowledging that she is indoor climbing thus have a common topic, namely climbing.13 So, here is what I think is going on (conceptually) when Anika imagines that she is mountain climbing and then gets to acknowledge that she is indoor climbing. Anika frst attends to a topic her imagining is about: mountain climbing.14 She then attends to a more general topic, namely climbing, and then acknowledges a proposition that is also about climbing, namely that she is indoor climbing. Of course, Anika is not proceeding by classical logical entailment (or even entailment of

How Imagination Can Justify

151

some relevant logic) from “I am mountain climbing” to “I am indoor climbing” but there is a relation of relevance between the two, which is spelled out in terms of the two having a common topic. So, this provides a way of explicating a new kind of relevant consequences that guide our imaginative episodes, thereby obeying (rCC).15 This is the (conceptually) frst step of imaginative scaffolding: selecting some relevant experience one had had, or, in the case of propositional imagination, selecting a relevant topic. This resolves the question of how Anika gets to offine-manifest her conditional beliefs involving that she is indoor climbing: because that she is mountain climbing and that she is indoor climbing have a common topic. What is crucial now for the justifcatory force of an imaginative episode is not so much the connection between imagining that she is mountain climbing and acknowledging that she is indoor climbing, but rather how her imagining that she is mountain climbing is a reason for her acknowledging that she is exhausted, although she is only justifed in her acknowledging that she will exhausted because she acknowledges that she is indoor climbing. Is the following principle acceptable? If A and A′ have a common topic, then if A′ justifes C′, A justifes C (where C is scaffolded out from C′) It is not. Consider A: “Trump is the 45th president of the U.S.” and A′: “The 44th U.S. president, Obama, has a J.D.”. They have a common topic: the U.S. president. Now, that Obama has a J.D. justifes (believing) that he went to law school (C′). But clearly, that Trump is the 45th president is no reason to believe that he (Trump) went to law school (C). The scaffolding of C from C′ here is taking care of “he” referring back to Donald Trump. Intuitively, the scaffolding “adjusts” the consequence C′ of the acknowledged A′ to a C matching the initial imagining A. What is important for the justifcation here seems to be that Obama has a J.D., and this is not part of the common topic. Consequently, when we consider A′ as a whole, then we might end up with justifcation for C′ but that being due to topics only A′ and not A is about. Thus, the corresponding C will be about topics A is not about. So, instead, I suggest the following: If A and A′ have a common topic T, then if the part of A′ that is about T justifes C′, then the part of A about T justifes C (where C is scaffolded out to from C′). This works out well in (MC): The part of “I go indoor climbing” that is about climbing justifes “I will be exhausted” for it is the climbing that exhausts Anika. By the principle, then, “I go mountain climbing” justifes “I will be exhausted”.

152 Christopher Badura In the case of Obama and Trump, the part of “The 44th U.S. president, Obama, has a J.D.” that is about the U.S. president (in general) does not justify “Obama went to law school” because just being U.S. president doesn’t justify that one went to law school. Summing up this section, I have answered the question of how one gets from imagining A (which one has no conditional beliefs about) to acknowledging some A′ one has conditional beliefs about by suggesting that one attends to a topic A is about, attends to a more general topic A is about, and then attends to a specifc topic A′ is about, such that A and A′ have the general topic in common. As for the question how this makes imaginative episodes justifcatory, I have proposed that if A and A′ have a common topic T and if the part of A′ about T justifes C′, then the part of A about T justifes C, where C is scaffolded out from C′. The notion of scaffolding has remained largely intuitive. Its two main features are, intuitively, a selection of relevant topics from an initial imagining, and relating back acknowledged consequences to the initial imagining. Let me also sum up what I have been proposing overall in the conceptually chronological order. First, one sets out to imagine some A. If one has conditional beliefs B(C|A) that involve an antecedent belief in A, some of these offine-manifest qua one’s acknowledging A. Hence, one acknowledges C. By considering some general topic A is about, one might also acknowledge some A′ that has a common topic with A, and corresponding conditional beliefs B(C′|A′) are manifested offine. This is especially the case if one imagines some A one doesn’t have any or just very few conditional beliefs about. I have explained how three different notions of justifcation for conditional beliefs provide justifcatory force imaginative episodes.

7.4 Conditional Beliefs vs Beliefs in Conditionals Why do we need the detour through conditional beliefs? Couldn’t one alternatively offer an explanation in terms of beliefs in conditionals that rely on logical reasoning alone? After all, Anika could just have the beliefs “If I am mountain climbing, then I am climbing” and “If I am climbing, then I am exhausted”, from which she could deduce “If I am mountain climbing, then I am exhausted”. Note that one of the questions I have tried to answer was how Anika gets from imagining that she is mountain climbing to acknowledging that she is indoor climbing. I am not concerned with the question whether there might be an alternative way for Anika to get from imagining that she is mountain climbing to (BE), for example, via a belief that climbing in general is exhausting. Since Anika believes that mountain climbing is a kind of climbing, she would reason her way to mountain climbing being exhausting. The example, as I have set it up, is one where she specifically relies on her previous experience of indoor climbing and uses that

How Imagination Can Justify

153

to scaffold out what it is like to go mountain climbing. Or, to not rely on the talk of experiences, she fgures out a relevant consequence of her imagining that she is mountain climbing, while not relying on (classical) logical reasoning alone. Perhaps in some other cases this would be one way how things could go. I do not contest that. But even if we suppose that things can be worked out along the suggested lines in terms of beliefs in conditionals, there are good reasons to consider my proposal as a serious alternative. It might be just a part of the whole story. That is fne. My goal here is to show that my story is among the plausible ones for explaining the epistemic usefulness of (MC) and, moreover, that the explanation it offers has some important advantages. First, psychological research suggests that the probability humans assign to a conditional approximates the corresponding conditional probability; see Douven (2015) and references therein. So, if one understands an assigned conditional probability as a subjective conditional probability, and takes this to be the conditional degree of belief of a reasonably rational agent, then it seems that conditional degrees of belief are the explanatory basis for the degree of belief one has in a conditional for conditional probabilities are a well-defned entity (if the probability of the antecedent is non-zero). If our dispositions to believe are what is modeled by the conditional degrees of belief/subjective conditional probabilities, then it seems that our dispositions to believe are the explanatory basis for our beliefs in conditionals. So, while one might be able to tell a story in terms of beliefs in conditionals and initial suppositions, like, e.g., Nichols & Stich (2003; Williamson (2016), my proposal can reduce such a story even further to an explanation solely in terms of conditional beliefs. Although the approximation of subjective conditional probabilities and subjective probabilities of conditionals is warranted by psychological research, there are some concerns from the semantics and formal epistemology regarding this approximation. As mentioned in the introduction, conditional beliefs are familiar from formal epistemology, doxastic logic, semantics of conditionals, etc.  – long story short, they are mostly a formal concept. In the early days of formal semantics, it was proposed that the probability of “If  A,  C” just is the conditional probability of C given A. If one then understands probability as subjective probability and accepts Lewis’s Principal Principle, see (Lam, this volume), one obtains the result that the credence in a conditional equals one’s conditional credence. Unfortunately, David Lewis has shown that this identifcation of the probability of the conditional with conditional probability leads to triviality given certain seemingly plausible assumptions (Lewis, 1976). Ever since there have been attempts to save the equation; see, e.g., Douven (2015). The details do not matter. What matters is that in the whole probabilistic tradition dealing with conditionals, conditional probabilities/credences/

154 Christopher Badura beliefs have always played an explanatory role for probability/credence/ belief of a conditional.16 So, from a formal perspective, it is not straightforward to identify conditional beliefs and beliefs in conditionals. The standard accounts take conditional beliefs as basic and then sometimes add beliefs in conditionals as an additional primitive. The object of study when it comes to belief change behavior remains conditional beliefs, though. Hence, in two of the major approaches to formal epistemology at least, conditional beliefs are the beliefs of interest – not beliefs in conditionals. In both cases, conditional beliefs turn out to be explanatorily prior to beliefs in conditionals. Conditional beliefs are meant to defne beliefs in conditionals. Thus, an account of the epistemic usefulness of imaginative episodes in terms of conditional beliefs – like the one I proposed  – promises to be explanatorily more basic than an account in terms of beliefs in conditionals.

7.5 Conclusion This chapter pursued two aims. First providing a more rigorous understanding of the notion of unfolding. Second explaining how imaginative episodes (like MC) are justifcatory. I have spelled out the unfolding process as follows. The imaginer starts out by imagining that A. Then she acknowledges some A′ such that A and A′ have a common topic T. This is part of the imaginative scaffolding process described by Amy Kind. Whether they have the topic in common is determined by (linguistic) reality, obeying (rRC). (The relevance is taken care of because the two are compared in terms of their topic.) Of course, the imaginer might make a mistake here and select A′ such that it does not have a common topic with A. The imaginer has certain conditional beliefs B(C′|A′) which may or may not be justifed. These are offine-manifested because the imaginer acknowledges that A′. This results in the imaginer acknowledging C′. I  have suggested that if A and A′ are about the common topic T and the part of A′ about T justifes C′, then the part of A about T justifes C (where C is adjusted for the respective pronouns and indexicals). The latter adjustment is also involved in imaginative scaffolding. The justifcation of C′ by the part of A′ about T is explained in terms of justifcation of conditional beliefs. All this elaborates on (rCC). Why, and to what degree are these imaginative episodes justifcatory? As should become clear from the described process, imaginative episodes rely on our ability to scaffold out from the “right” experiences. Or, more generally, to acknowledge the right propositions that have a common topic with what we imagine. Moreover, we need to have other conditional beliefs that are independently justifed. Finally, we need to connect the initial imagining with the offine-manifested acknowledgment. This

How Imagination Can Justify

155

can be understood as a second “scaffolding” step. So, the justifcatory force of an imaginative episode depends on the justifcatory force of our conditional beliefs and also on how well we scaffold. So, overall, I have fulflled the two aims. By way of conclusion, let me anticipate a family of objections: what exactly makes the process imaginative; isn’t the process rather hypothetical reasoning; if the process works like this, doesn’t this reduce imaginative episodes to (conditional) beliefs, and so forth. Following the orthodoxy (Nichols & Stich, 2003; Byrne, 2005; Langland-Hassan, 2016; Williamson, 2016), I have assumed that the process involves something like an initial imagining, or an initial supposition. If one grants that the initial state of this process is distinctively imaginative and that the unfolding of this process is offine, then this is suffcient for me to call the described process imaginative. But isn’t this just hypothetical reasoning if the initial state of one’s imaginative episode is just suppositional? There is something to this concern when we are concerned with propositional imagination only. However, as Magdalena Balcerak Jackson has recently emphasized, a great deal of our imaginings are experiential (Balcerak Jackson, 2016). In particular, if we grant that in (MC), part of what is going on is that Anika wants to fgure out what it is like to go mountain climbing, she accesses the phenomenal character of previous experiences. This phenomenal character is necessary for her belief that this is what mountain climbing is like to be justifed (via her conditional belief). Hypothetical reasoning doesn’t obviously have this phenomenological component. Does my account reduce imaginative episodes to some complex process involving only beliefs? Further investigation into my proposal might reveal that it does. But I don’t think it is obvious. In particular, if we grant that the initial state of the whole episode is an imaginative state and have reason to assume that such states are not reducible to beliefs, then there is something distinctively non-reducibly imaginative about imaginative episodes as I have described them. Specifcally, consider Amy Kind’s overall account of imagination, which holds that mental imagery is necessary for imagination. This means that for a mental process to count as imaginative, there must be a quasi-sensory experience present without a corresponding stimulus. Moreover, on her account, the content of this experience determines “the object of one’s imagination” (Kind, 2001), that is, what one’s imagining is about. How to spell this out precisely is beyond the scope of this chapter but in general, on my story, it would mean that the imagery is involved in determining the topic(s) or subject matter of one’s imagining. Kind’s view is certainly controversial, see, e.g., Gregory (2016), and Peter Langland-Hassan has recently put pressure on the orthodox view that imaginative episodes involve an initial imaginative or even suppositional state (Langland-Hassan, 2020). Thus, the proposed view has to be defended against these charges in

156 Christopher Badura more detail in future work. Yet, the work I have done here has set up a proposal that promises to be a serious alternative to views involving beliefs in conditionals because it is potentially explanatorily more basic. Furthermore, by applying ideas from formal epistemology and semantics to the epistemology of imagination, my account in terms of conditional beliefs is able to provide a more thorough and rigorous understanding of the relevant issues than the alternative in terms of beliefs in conditionals. Though I have here omitted the formal details of this account, the discussion here should show that they are worth investigating further. 17

Notes 1 Langland-Hassan accepts that there are some mental processes which we describe in terms of imagination but he aims to explain such processes, and hence imagination, solely in terms of beliefs and desires. 2 Leitgeb’s argument proceeds from the formal defnition of conditional beliefs in AGM belief revision theory by assuming that these can be expressed by a counterfactual with truth conditions à la David Lewis, and then assumes the conditional analysis of dispositions. I assume that the conclusion of the argument that conditional beliefs are dispositions to believe is intuitively plausible, even if Leitgeb’s argument could be challenged. 3 One might be tempted to claim that Anika’s imagining is an offine belief revision for conditional beliefs encode online belief revisions in the sense that if the antecedent belief occurs (i.e., there is an online stimulus), the consequent belief would occur (i.e., there is an online manifestation). This seems to be in fact how, e.g., Franz Berto understands imagination as reality-oriented mental simulation; cf. (Berto, this volume, pp. 123–4). As I am trying to stay faithful to Kind’s own position, I do not commit to simulationism because Kind has worries about it; see Kind (2001) and personal communication. 4 The point here is that, all things being equal, if one acknowledges that A, then one usually acknowledges that there is a reason for A, too. This does not mean that one has to explicitly acknowledge such a reason. It also does not exclude that one might, on further refection, come to acknowledge that one has been acknowledging something contradictory, and, hence, that it lacks justifcation. One might also, on further refection, acknowledge underminers or defeaters for what one initially acknowledged. Acknowledgment (of a proposition) is, in this respect, very much like belief, except that one doesn’t accept A as true. If we believe something, we usually have a reason for this belief, although we might be believing something contradictory, or might come to learn (and hence believe) about defeaters or underminers. 5 As a reminder, Pr(C|A) = Pr(C & A)\Pr(A) if Pr(A)>0. This has some unintuitive consequences. For example, she will be justifed in the conditional belief B(C|A) for any A and C, where Pr(A)>0 and where C is a tautology. If we take seriously the idea that A must provide some evidence or reason for A, then to account for this, one might add that A must be probability-raising for C (and also Pr(C|A) exceeds a certain threshold); see, e.g., Douven (2015). 6 This faces several problems, like the problem of old evidence; see Talbott (2016). In our case, everything else that Anika knows already adds to her degree of belief in C if she learnt A. Addressing these issues doesn’t matter for the explanatory purposes here. Whatever the right story of Bayesian epistemology is, it transfers to the justifcatory force of an imaginative episode.

How Imagination Can Justify

157

7 Strictly speaking, in such a probabilistic account, this would amount to her temporarily believing (to degree 1, i.e., being certain) that she is indoor climbing. This is clearly not what acknowledging amounts to. But as I have mentioned, I have been concerned with the non-technical side of things, and this might show that it is not straightforward to account for imaginative episodes in a probabilistic framework. 8 I use temporal notions like “prior” here, although these things might happen simultaneously. The ordering is rather conceptual than temporal. 9 Unfortunately, I can’t defend this assumption here. But my view is that while we might know what our imagining is about, we might fail to represent it accurately. For example, while I might succeed in imagining what it is like to be a parent, i.e., I am successful in fxing what my imagining is about, I might be representing it incorrectly; see Kind (2020a). See also (Roelofs, this volume) for an elaboration of the difference between the target and the content of an imagining. 10 Recently, Stephen Yablo has offered a comprehensive account of aboutness (Yablo, 2014). For a comparison of his and Kit Fine’s view, see Fine (2020). 11 Topics/subject matters are usually defned as the entities that our linguistic expressions – most notably assertoric sentences – are about. However, it makes perfect sense to say that Anika imagines about mountain climbing. So, one topic she is imagining about is the topic mountain climbing. A subject matter she is imagining about is that she is mountain climbing. Topics are usually objectual, while subject matters are propositional. While the latter difference is noteworthy in general, it does not matter for the purposes of this chapter. 12 Since every situation in which one is mountain climbing (indoor climbing), one is climbing, the topic mountain climbing (indoor climbing) is a disjunctive part of climbing; see Fine (2017a) for a notion of disjunctive part and Jago (2020) for a differing view of disjunctive part. 13 This notion of common topic can be made precise within Kit Fine’s truthmaker semantics (Fine, 2017a, 2017b), but in this chapter I just appeal to an intuitive understanding of common topic. 14 While we usually associate to a sentence its (unique) subject matter, there are usually several topics a sentence is about. Similarly for one’s imaginings, it seems: imagining that one is mountain climbing in the Alps is about climbing but also about the Alps, while the subject matter is that one is mountain climbing in the alps. Of course, the subject matter that one is climbing is part of that one is mountain climbing in the alps. Consequently, it makes sense to also consider just a subject matter of a sentence or an imagining. 15 Note that (MC) can’t be dealt with in Berto’s framework; see Berto (2018); Berto (this volume) because climbing is not a part of mountain climbing. I have raised a similar problem for Berto’s account and provided a more fnegrained account of topic/subject matter that resolves this (Badura, 2020). 16 A similar case has been established by Peter Gärdenfors (1986) for the alternative paradigm of AGM belief revision – after Alchourrón et al. (1985). 17 For extensive comments and discussion of previous drafts of this chapter, I thank Amy Kind, Margot Strohminger, Heinrich Wansing, Francesco Berto, and Tom Schoonen. I also thank all participants of the conference “Fiction, Imagination, and Epistemology” for their comments. The conference and this chapter have been supported by the Ruhr University Bochum Research School PLUS, funded by Germany’s Excellence Initiative [DFG GSC 98/3].

158 Christopher Badura

References Alchourrón, C.E., Gärdenfors, P., & Makinson, D. (1985). “On the Logic of Theory Change: Partial Meet Contraction and Revision Functions.” Journal of Symbolic Logic, 50(2), 510–30. Badura, C. (2020). “More Aboutness in Imagination.” Journal of Philosophical Logic. doi:10.1007/s10992-020-09575-4 Balcerak Jackson, M. (2016). “On the Epistemic Value of Imagining, Supposing, and Conceiving.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 41–60). Oxford: Oxford University Press. Berto, F. (2018). “Aboutness in Imagination.” Philosophical Studies, 175(8), 1871–86. Byrne, R.M. (2005). The Rational Imagination. How People Create Alternatives to Reality. Cambridge, MA: Bradford Books/MIT Press. Choi, S., & Fara, M. (2018). “Dispositions” (E.N. Zalta, Ed.). Retrieved from Stanford Encyclopedia of Philosophy: https://plato.stanford.edu/archives/ fall2018/entries/dispositions/ Currie, G., & Ravenscroft, I. (2012). Recreative Minds: Imagination in Philosophy and Psychology. Oxford: Oxford University Press. Douven, I. (2015). The Epistemology of Indicative Conditionals. Cambridge: Cambridge University Press. Fine, K. (2017a). “A Theory of Truthmaker Content I: Conjunction, Disjunction, Negation.” Journal of Philosophical Logic, 46(6), 625–74. Fine, K. (2017b). “A Theory of Truthmaker Content II: Subject-Matter, Common Content, Remainder and Ground.” Journal of Philosophical Logic, 46(6), 675–702. Fine, K. (2020). “Yablo on Subject Matter.” Philosophical Studies, 1(177), 129–71. Gärdenfors, P. (1986). “Belief Revisions and the Ramsey Test for Conditionals.” Philosophical Review, 95(1), 81–93. Gregory, D. (2016). “Imagination and Mental Imagery.” In A. Kind (Ed.), The Routledge Handbook of Philosophy of Imagination (pp. 97–110). Abingdon; New York: Routledge. Jago, M. (2020). “Truthmaker Semantics for Relevant Logic.” Journal of Philosophical Logic, 49(4), 681–702. doi:10.1007/s10992-019-09533-9 Kind, A. (2001). “Putting the Image Back in Imagination.” Philosophy and Phenomenological Research, 62(1), 85–109. Kind, A. (2016). “Imagining under Constraints.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 145–59). Oxford: Oxford University Press. Kind, A. (2018). “How Imagination Gives Rise to Knowledge.” In F. MacPherson & F. Dorsch (Eds.), Perceptual Imagination & Perceptual Memory (pp. 227–47). Oxford: Oxford University Press. Kind, A. (2020a). “Imaginative Experience.” In U. Kriegel (Ed.), The Oxford Handbook of Philosophy of Consciousness (pp. 124–41). Oxford: Oxford University Press. Kind, A. (2020b). “What Imagination Teaches.” In J. Schwenkler & E. Lambert (Eds.), Becoming Someone New: Essays on Transformative Experience, Choice, and Change (pp. 133–46). Oxford: Oxford University Press.

How Imagination Can Justify

159

Kind, A., & Kung, P. (2016). “Introduction: The Puzzle of Imaginative Use.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 1–37). Oxford: Oxford University Press. Langland-Hassan, P. (2016). “On Choosing What to Imagine.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 61–84). Oxford: Oxford University Press. Langland-Hassan, P. (2020). Explaining Imagination. Oxford: Oxford University Press. Retrieved from http://www.langland-hassan.com/uploads/4/7/3/ 1/47312141/explaining_imagination_--_langland-hassan_2020.pdf Leitgeb, H. (2007). “Beliefs in Conditionals vs. Conditional Beliefs.” Topoi, 26(1), 115–32. Lewis, D. (1976). “Probabilities of conditionals and conditional probabilities.” Philosophical Review, 85(3), 297–315. Nichols, S., & Stich, S. (2003). Mindreading: An Integrated Account of Pretence, Self-Awareness, and Understanding Other Minds. Oxford: Oxford University Press. Talbott, W. (2016). Bayesian Epistemology (E.N. Zalta, Ed.). Retrieved from Stanford Encyclopedia of Philosophy: https://plato.stanford.edu/archives/ win2016/entries/epistemology-bayesian/ Williamson, T. (2016). “Knowing by Imagining.” In A. Kind & P. Kung (Eds.), Knowledge Through Imagination (pp. 113–23). Oxford: Oxford University Press. Yablo, S. (2014). Aboutness. Princeton, NJ: Princeton University Press.

8

Imagination, Inference, and Apriority Antonella Mallozzi

8.1 Introduction Is imagination a source of knowledge? Timothy Williamson has enthusiastically advocated the epistemic role of imagination for knowing a variety of matters (2007, 2013, 2016). Its positive uses span from helping us plan and make choices in our life, to evaluating counterfactuals for various practical as well as philosophical purposes, to guiding us through knowledge of complex matters like logic and set theory. In core examples, we run cognitive processes “offine” in imagination by deploying helpful mental simulations. Imagination seems so powerful that Williamson conjectures it might have given us an evolutionary advantage as a species, as opposed to being a mere accidental byproduct of our evolutionary history. This fts an overall anti-exceptionalist program in epistemology, which urges us to avoid positing special faculties and instead look for explanations that are continuous with our ordinary and scientifc knowledge of the world (2007). Williamson’s account of imagination also contributes to a critical task. He challenges the philosophical signifcance of the long-established distinction between a priori and a posteriori knowledge by appealing to a similar role that imagination seems to play in typical cases from each category. Williamson’s argument proceeds by looking at examples of a priori and a posteriori knowledge and comparing the sorts of cognitive processes we go through in each case, aiming to show that they are actually not that different. If that’s correct, the distinction fails to capture anything epistemologically signifcant. The primary aim of this chapter is to defend the a priori-a posteriori distinction from Williamson’s challenge by questioning the claim that imagination is an epistemic source. I distinguish two notions of imagination at play in Williamson’s account – sensory imagination and belief-like imagination – and show that both face empirical and normative issues. Sensory imagination seems in general neither necessary nor suffcient for knowledge; additionally, Williamson’s particular examples are unconvincing because they don’t generalize. Belief-like imagination, on the other hand, isn’t suffciently disentangled from inference. Overall,

Imagination, Inference, and Apriority  161 the a priori-a posteriori distinction is still safe. Imagination seems at best to play only a supportive role for knowledge.

8.2  Williamson’s Challenge Williamson argues that the difference between a priori vs. a posteriori knowledge is only superficial. If we look carefully at the epistemic processes we carry out in clear cases from each category, those are apparently neither purely empirical nor purely a priori according to traditional standards. In particular, in many cases for Williamson experience plays a role that is neither strictly evidential, nor merely enabling for knowledge (2013: 13–4). The distinction is used in the epistemology of the a priori to answer the following worry. Even in apparently clear-cut cases of a priori knowledge and justification, experience will be involved in enabling us to form the relevant judgments. Appeal to the distinction clarifies that experience is allowed to play this type of “enabling role” consistent with the knowledge in question remaining a priori, as long as it doesn’t play an evidential role. In particular, experience might be enabling in the sense that we might only need it to initially acquire ingredient concepts of truths that we may then come to know a priori; or in the sense that it might enable our cognitive access to certain content. For example, one needs experience in order to acquire the concepts “red” and “green”. However, that is merely enabling for one to come to know a priori that “An object cannot be red all over and green all over at the same time”. Analogously, one needs to see the passages of a long mathematical proof on a piece of paper, or to memorize them, in order to carry out the proof. But seeing those passages, or recalling them in memory, isn’t what justifies one’s belief in the truth of the proof. Rather, those perceptual or mnemonic experiences only enable one to perform the relevant calculations and access the relevant mathematical propositions. Accordingly, coming to know the proof is still a priori. By contrast, when experience plays an evidential role, one’s having certain experiences is what justifies the resulting judgments, so that those are a posteriori. For example, seeing or remembering the keys on the table constitutes evidential experience for knowing that the keys are on the table. Williamson disagrees. He argues that in both apparently clear-cut cases of the a priori and the a posteriori, respectively, experience plays a role that’s intermediate between evidential and enabling.1 As we will see in a moment, the epistemic processes in question are distinctively imaginative. But it’s important to stress that Williamson is not trying to reject the a priori as such or reduce all knowledge to the deliverances of our experience. Radical empiricism isn’t an option, since experience might well be less than strictly evidential for knowledge. Nor should we conclude that such cases are “borderline”; in fact, he concedes that we can still identify examples from each category. The point

162

Antonella Mallozzi

is rather that the a priori-a posteriori distinction is a shallow one, which “does not cut at the epistemological joints” (2013: 8). That’s a pragmatic moral: epistemologists would do better without these obsolete categories. On the other hand, Williamson invites us to adopt a different framework, where imagination is a positive source of knowledge. For him, imagining has a key epistemic function in that we generally gain knowledge of various practical and theoretical matters by running our cognitive capacities “offine” via simulation in imagination. More precisely, from Williamson’s discussion it appears that simulative imagination can be understood in two ways. The frst, let’s call it sensory imagination, is imagistic and involves simulation of our perceptual capacities. The second, let’s call it belief-like imagination, is generally independent of the occurrence of mental imagery and involves simulation of our reasoning and planning capacities. Williamson’s early examples against the a priori-a posteriori distinction (2007, 2013) feature the former “sensory” kind of imagination. We shall start with that. We will turn to belief-like imagination in Section 8.4. In core cases Williamson discusses, imaginative procedures are distinctively quasi-perceptual, namely they simulate regular perception (particularly, vision) offine. Since our judgments in the relevant cases are taken to result from such simulations in imagination, Williamson can then claim that the contribution of experience is neither strictly evidential, nor merely enabling for the resulting knowledge. The contribution of experience is rather intermediate, due to the quasi-perceptual nature of the imaginative processes. Hence, even if by bottom-up standards we might still identify supposedly clear cases of the a priori-a posteriori, it turns out that the epistemic role of experience isn’t straightforward. If Williamson is right, that directly undermines the signifcance of the a priori-a posteriori distinction. Let us look at Williamson’s core examples (2013): (CR) All crimson things are red (WW) All recent copies of Who’s Who are red Williamson takes (CR) and (WW) to be clear examples of truths that, according to the tradition, one may come to know a priori and a posteriori respectively. Regarding (CR), the familiar idea is that one can come to know it by conceptual, or linguistic, analysis. “Red” may be understood as part of the concept “crimson”; or, by defnition, “crimson” picks out a particular shade of red. On the other hand, (WW) is generally taken to be an example of a truth that we can only come to know through empirical means – plausibly, visual perception, or testimony. In order to know (WW), we need some direct or indirect experience with recent copies of Who’s Who. However, Williamson disagrees with the traditional picture

Imagination, Inference, and Apriority

163

and offers a different story which distinctively features imagination of how one may come to know both of these truths. As to (CR), he asks us to suppose that an agent, Norman, learns the words “red” and “crimson” […] independently of each other, by ostensive means. He learns ‘crimson’ by being shown samples to which it applies and samples to which it does not apply, and told which are which. He learns ‘red’ in a parallel but causally independent way. He is not taught any rule like [CR], connecting ‘crimson’ and ‘red’. Through practice and feedback, he becomes very skilful in judging by eye whether something is crimson, and whether something is red. (9) How does Norman come to know (CR)? Apparently, simply by imagining a shade of crimson and judging, within the imaginative supposition, that it is in fact red – without any further empirical support. The relevant epistemic process is described in the following passage, which is worth quoting at length: First, Norman uses his skill in making visual judgments with ‘crimson’ to visually imagine a sample of crimson. Then he uses his skill in making visual judgments with ‘red’ to judge, within the imaginative supposition, ‘It is red’. This involves a general human capacity to transpose ‘online’ cognitive skills originally developed in perception into corresponding ‘offine’ cognitive skills subsequently applied in imagination. […] No episodic memories of prior experiences, for example of crimson things, play any role. As a result of the process, Norman accepts [CR]. Since his performance was suffciently skillful, background conditions were normal, and so on, he thereby comes to know [CR]. (9–10) Williamson’s treatment of (WW) mirrors (CR). Norman has already learned “red”. Now he learns all the other ingredient terms in (WW), i.e., “recent”, “volume”, etc., in a similar way. Through practice, he becomes similarly skillful at judging whether a certain volume is a copy of Who’s Who. Then, without any further empirical investigation, but simply refecting on what he has learned, Norman comes to know (WW) like he learned (CR). He visually imagines a recent volume of Who’s Who, and “he uses his skill in making visual judgments with ‘red’ to judge, within the imaginative supposition, ‘It is red’” (11). Note that although those cases are thought experiments, Norman’s use of imagination in his epistemic performances isn’t meant to be merely hypothetical. Nor is Norman special compared to other subjects.

164

Antonella Mallozzi

Instead, Norman’s imaginative processes are supposedly representative of what epistemic subjects actually do. Being able to gain knowledge by running certain skills offine in imagination instantiates for Williamson a “general human capacity” (9). Williamson’s conclusion is straightforward. If typical cases of a priori and a posteriori knowledge involve similar cognitive processes or epistemic procedures, where experience contributes to an extent that is neither enabling nor evidential for the resulting knowledge, then the signifcance of the a priori-a posteriori distinction fades away.

8.3 Issues for Sensory Imagination Williamson’s challenge against the signifcance of the a priori-a posteriori distinction depends on making a convincing case that one may come to know truths from each category, such as (CR) and (WW), through similar imaginative processes. Specifcally, imagination is here supposed to yield knowledge by deploying offine simulation of perceptual skills. Furthermore, the epistemic procedures featuring in Williamson’s candidate examples shouldn’t be special but generalizable to other truths and ways of knowing that are standardly regarded as a priori-a posteriori. Let us distinguish these two components: Imagination Claim We may gain knowledge via sensory imagination by simulating perceptual cognitive skills offine. Generality Claim Norman’s epistemic procedures for knowing (CR) and (WW) are generalizable to standard cases of the a priori-a posteriori. In Section 8.3.1, I discuss in detail the following three objections to the Imagination Claim (IC). First, although (IC) is an empirical thesis, it lacks adequate empirical support. Second, Williamson’s treatment of (IC) suggests that simulative sensory imagination guarantees that Norman gains the relevant knowledge, based on considerations of reliability. However, not only is the reliability of sensory imagination quite speculative, but appealing to it also doesn’t seem suffcient to establish that imagination is a source of knowledge. Third, (IC) prompts a dilemma concerning simulation, both horns of which have undesirable theoretical consequences. By undermining (IC), these criticisms undermine Williamson’s case against the a priori-a posteriori distinction. In Section 8.3.2, I argue that the Generality Claim (GC) is false. Even if (IC) were true and Norman came to know (CR) and (WW) in the described way, I argue that those examples are ad hoc and don’t generalize. Specifcally, sensory imagination doesn’t seem necessary for our epistemic processes. I conclude that the a priori-a posteriori distinction remains safe, whereas sensory imagination seems at best to only play a supportive role for knowledge. Finally, in Section 8.4, I turn

Imagination, Inference, and Apriority

165

to Williamson’s more recent treatment of imagination as not essentially imagistic but rather “belief-like”. I raise a general problem of clarifying the distinctive role of this type of imagination as opposed to inferential reasoning and contend that the latter best explains Williamson’s further examples. 8.3.1 Against the Imagination Claim (a) An Empirical Matter Williamson’s (IC) is an empirical thesis. It concerns what epistemic subjects actually do, by pointing to the psychological processes that appear to be involved (according to Williamson’s own introspective experience, at least) in coming to know truths such as (CR) and (WW). Those processes centrally involve visualizing “offine” percepts of the relevant shades of colors. More generally, imagination simulates cognitive skills that depend on perception. Simulation has received considerable attention in the cognitive sciences as the main alternative against the received view for explaining certain cognitive capacities, especially practical ones. Roughly, the received view invokes an internally represented body of rules or propositions (possibly “tacit” or “sub-doxastic”), which guide a range of cognitive tasks. By contrast, simulation theory holds that certain mental mechanisms or components can be detached from their usual function and used to perform or support some other function offine (often via pretense). Simulation has been mainly thought to explain the capacity for “mind-reading” or attributing mental states and predicting other people’s behavior without drawing from some underlying body of knowledge. But some maintain that similar offine processes might contribute to explain grammaticality judgments, empathy, counterfactual reasoning, and, importantly for our purposes, mental imagery (see Nichols et al. 1996 for an early survey). For example, Gregory Currie (1995) argues that episodes of imagery are episodes of simulated vision, based on evidence that they share central processing (same areas of brain involved, same content structure, similar incapacities from brain impairments). In his early examples involving (CR) and (WW), Williamson seems to think along similar lines as Currie. Imagination here simulates perception in a quasi-sensory way, by exploiting the same cognitive skills offine. Norman comes to know (CR) and (WW) through visual judgments within the imaginative supposition, which features offine simulation of visual experience. Additionally, for Williamson this hypothesis is supposed to explain even more abstract cases like knowledge of set theory and basic logical theorems, which are usually thought to be unrelated to sensory experience (and thus a priori). Williamson conjectures that we (well, mathematicians!) form conceptions of sets based on imaginative exercises that simulate “our online skill in observing and

166 Antonella Mallozzi engaging in processes of physical creation” and remarks that perceiving such conceptions as “intuitively compelling” may rest on helpful mental “metaphors” and “pictures” that we develop in imagination (2013: 21–2). Similarly, in the case of knowledge of basic logical theorems, e.g., the refexivity of identity (∀x x=x), imagination is alleged to play a key epistemic role by drawing from the experiential process of “continually judging numerical identity or distinctness among objects perceived or remembered in a wide variety of guises” (26). Williamson (2007) gives yet another example to the same effect: “If two marks had been nine inches apart, they would have been at least nineteen centimeters apart”. This judgment also supposedly depends on a simulative capacity of running naked eye measurements “visually offine” in imagination (166).2 Since (IC) concerns what epistemic subjects supposedly actually do, we would expect Williamson to support it with plenty of empirical evidence, like data from psychological or neuroscientifc studies on actual subjects. It is thus surprising that he doesn’t provide any such evidence for his claim, thus leaving it largely speculative. We saw that Williamson mentions a “general human capacity” for transposing perceptual cognitive skills into corresponding offine cognitive skills applied in imagination. Although that’s broadly consistent with certain literature from the cognitive sciences, it is important to fag that simulation theory broadly construed is still contentious. Empirical data haven’t clearly indicated that subjects actually deploy simulation in the relevant circumstances. For example, Nichols et al. (1996) assess empirical results concerning various applications of the theory and conclude that they are “still extremely sceptical of the off-line simulation theory of behavior prediction” (45) as well as “not sure how simulation is supposed to illuminate the issue of mental imagery” (35). (See also e.g. Gallagher (2007) and Barlassina and Gordon (2017) for evidence against mindreading in particular.) Additionally, researchers disagree as to the nature of simulation itself and its alleged contribution to our cognitive tasks. For example, Jane Heal remarks that “we have very little idea of what would be involved, neurophysiologically or functionally, in taking a system ‘off-line’”. What’s worse, “we do not really know how to test simulationism, regarded as this empirical hypothesis” (1998: 89–90). Similarly, Shannon Spaulding warns us that the concept of simulation itself is problematic: “There is no consensus on what exactly simulation is in the cognitive sciences, not even in the area of mindreading” (2016b: 263). Insofar as the simulation hypothesis is still contentious and developing, (IC) rests on an uncertain basis. (Note that these considerations don’t affect just sensory imagination qua simulating our perceptual processes, but also belief-like imagination, which I discuss below.) Lacking support from the simulation hypothesis, Williamson’s psychological remarks won’t have their purported import. For it is not obvious that investigating the phenomenology of imagination per se would

Imagination, Inference, and Apriority

167

tell us anything relevant for epistemological purposes. The introspective appearance and contents of our conscious processes might tell us nothing much about their origin and justifcation. In particular, once divorced from the thesis that imagination and perception share central processing, the occurrence of mental imagery might be completely inessential to the cognitive tasks in question, or merely a contingent byproduct of the actual epistemic processes. I will return to the signifcance of this point. (b) Sensory Imagination and Reliability Appealing to simulation of perceptual capacities might help Williamson’s case by suggesting that the imaginative procedures in question are reliable. Through offine visual simulation in imagination, Norman comes to know (CR) and (WW). I take Williamson’s implicit assumption to be that since perception is a reliable source of knowledge, imagination qua offine perception must somehow “inherit” that reliability. By exploiting the same cognitive resources as perception, sensory imagination should be equally generally truth-conducive, to the effect that judgments reached within the imaginative supposition may constitute knowledge as well as in the corresponding online perceptual cases. Indeed, Norman seems guaranteed to reach the correct judgments based on the simulative processes’ reliability. Williamson’s claims presuppose an externalist conception of knowledge, where our epistemic states primarily depend on our relations with the external environment. Additionally, whether a subject knows (or is at least justifed in believing) a certain truth seems to be a matter of the reliability of the relevant belief-forming processes, rather than one’s reasons for believing that truth. Experiential processes like perceiving and remembering constitute sources of knowledge because they are generally truth-conducive not because they provide evidence for believing their content. The thesis that Norman gains knowledge of (CR) and (WW) in the described way heavily relies on such externalist and reliabilist assumptions (similarly, Boghossian: Ch. 9 in Boghossian and Williamson 2020). But the idea that sensory imagination guarantees knowledge is problematic. To start, note that appealing to simulation per se doesn’t ensure that this kind of imagination is generally truth-conducive. Simulation might clarify how sensory imagination operates at the cognitive level, namely by sharing central processing with perception. But unless our imaginative exercises are suitably constrained, such that they consistently yield correct judgments, simulative imagination might still lead us astray. The scenarios we imagine depart from reality in all sorts of ways. Even among equally realistic scenarios, nothing in the imaginative exercise per se indicates which ones are correct. Indeed, by contrast with perception, our imaginative exercises do not generally entitle us to form beliefs based on the imagined content, like we do based on percepts. Candidate accounts of the epistemic uses of imagination should clarify

168

Antonella Mallozzi

under what conditions our imaginative exercises yield the correct outputs (similarly Kind and Kung 2016; Langland-Hassan 2016; Nichols 2006; Spaulding 2016a.). Williamson seems to trust our imaginative capacities to be reliable somehow by default, because they simulate online skills that are themselves reliable. Norman might mistakenly visualize a peripheral shade of crimson, but “the relevant cognitive skills must include sensitivity to such issues” (2013: 10). Still, trusting imagination to be somehow reliable by default when deployed for certain cognitive tasks is far from establishing that it is a source of knowledge. Additionally, even granting that it is in fact reliable, that doesn’t seem suffcient for knowledge. Norman’s imaginative capacities might be perfectly reliable and still he may not have knowledge, since he might invariably be making merely lucky guesses. One should be able to provide reasons for taking the judgments resulting from our imaginative exercises to be correct. The issue is symptomatic of a broader concern that Williamson’s empirical hypothesis about the epistemic roles of imagination neglects central questions of normative epistemology.3 In order to establish that imagination is a source of knowledge, it is not suffcient to speculate about the psychological processes subjects go through when engaging in imaginative exercises and trust that these are reliable based on possibly sharing central processing. One should integrate an account of the specifc constraints that bind our imaginative exercises and explain why they consistently deliver the correct results, while also ruling out that they are merely lucky guesses. (c) Simulation and Cognitive Architecture The simulation hypothesis at play in core examples (CR) and (WW) seems to commit one to a precise cognitive architecture. If our imaginative exercises simulate perceptual capacities offine, imagination and perception at least share the same cognitive mechanisms. Indeed, there might be just one faculty or cognitive unit, which can work in both online and offine modes. We saw that this might provide partial support for Williamson’s thesis that imagination is a source of knowledge, at least granted that the offine operations of the unit are generally as truth-conducive as the online perceptual ones. However, on this supposition the alleged productive work of imagination remains unexplained. In Williamson’s examples, imagination seems to be doing more than just simulating perception offine. It does not merely reproduce one’s perceptual experience, but is also responsible for flling a gap, so to speak, between what we experienced and what we have not in reasoning aimed at generalization. Imagination seems responsible for productive transitions beyond observed instances of a certain phenomenon. In the described procedure for knowing (CR) and (WW), imagination yields the relevant generalizations by connecting and supplementing – in a

Imagination, Inference, and Apriority

169

Humean fashion, we might say – the limited data provided by perception.4 Through an imaginative procedure, Norman can somehow go beyond his repeated observations of shades of crimson as well as copies of Who’s Who and reach general judgments. More generally, these sorts of transitions seem possible because of imagination, whereas perception alone does not suffce to move beyond the observed cases. Imagination has a central epistemic role by deploying this synthetic or productive capacity. But note that there’s nothing especially sensory at play in this alleged productive capacity. Indeed, this interpretation seems incompatible with understanding imagination as simulating perception. In simulation, sensory imagination runs our perceptual processes vicariously, by reproducing them offine. But it’s not clear how such a cognitive unit can also yield generalizations like (CR) and (WW). If imagination does play that synthetic or productive role, imagination and perception might not share central processing after all; or, at least, they might not constitute one single faculty or cognitive unit but rather two distinct ones with different cognitive tasks. But this is an unwanted result for Williamson. If there is a separate faculty or cognitive unit of imagination, the thesis that our imaginative exercises ought to be explained in terms of perceptual simulation loses support. What is more, introducing such a special faculty or cognitive unit in order to explain the relevant ampliative work appears to clash with Williamson’s anti-exceptionalism in epistemology, a view that requires that one avoid postulating sui generis epistemic faculties and that one seek to explain various types of knowledge by appealing to the same general human cognitive capacities. Williamson seems to be facing a dilemma. Either (a), following the simulation hypothesis for sensory imagination, imagination and perception share the same cognitive mechanisms, which simplifes the cognitive architecture of imagination with all the annexed advantages that we saw. However, in this case we can’t accommodate the supposed productive work of imagination. Imagination merely reproduces or copies what we have already experienced. Or (b), imagination and perception are separate and imagination is distinctively productive. However, that seems to give up the advantages that come from treating imagination as quasi-perceptual, particularly against the a priori-a posteriori distinction. Moreover, it may lead to positing some potentially exotic separate source of knowledge. From Williamson’s discussion, it isn’t clear that he could avoid the dilemma. There seems to be no third option that allows one to account for both the quasi-perceptual and the productive capacities that imagination seems to have in his examples. To take stock, (IC) rests on shaky grounds. The claim lacks adequate empirical support, while it appears to assume that simulative sensory imagination somehow guarantees knowledge outputs. Additionally, the purported roles of sensory imagination should be clarifed to avoid

170  Antonella Mallozzi an unwanted dilemma for our cognitive architecture. Overall, more is needed to establish (IC) and thereby undermine the a priori-a posteriori distinction. 8.3.2  Against the Generality Claim There is a deeper difficulty for Williamson’s challenge against the a ­priori-a posteriori distinction: (GC) seems false. Even if we granted that (IC) were true and Norman came to know (CR) and (WW) in the described way, those epistemic procedures aren’t generalizable to standard cases of the a priori-a posteriori. Hence, they won’t manage to jeopardize the significance of the a priori-a posteriori distinction. 5 More generally, sensory imagination doesn’t seem necessary for knowledge. Williamson remarks that imagination need not in general involve mental imagery (more below, Section 8.4). However, note that mental imagery is essential for Norman’s coming to know (CR) and (WW). Norman needs to engage in imaginative processes that involve offline visual simulation in order to formulate the relevant judgments. That, in turn, is crucial for Williamson’s dialectical purposes. Since offline simulation in imagination is only quasi-perceptual, experience appears to contribute to the resulting knowledge merely to an intermediate extent. More precisely, experience is supposedly neither strictly evidential nor merely enabling for coming to know (CR) and (WW). Thus, eliciting a familiar feeling of “experiencing” quasi-perceptual contents in imagination is crucial for Williamson’s argument because that’s how we would find support for the claim that imaginative processes straddle the traditional a priori-a posteriori categories. Indeed, those examples might feel intuitive, since they fit well the phenomenology of imagining based on simple introspection. However, I shall argue that the examples are devised to have sensory imagination play a crucial role and misleadingly induce us to think that they would generalize to similar cases. Additionally, there is a problem with the fact that the examples characteristically feature universal statements. That misleadingly induces us to think that imagination (this time not necessarily imagistic) is required to explain the productive epistemic process through which Norman reaches the universalizations. That’s especially relevant in the case of (WW), so let’s start with that. Example (WW) seems a rather strange choice for an a posteriori truth. Why “All recent copies of Who’s Who are red”, rather than, simply, “This copy of Who’s Who [here, in front of me] is red”? Ostensive, “here and now” judgments would more naturally come to mind as clear-cut cases of the a posteriori, likely because of their direct perceptual character. On the other hand, (WW) is an empirical generalization. According

Imagination, Inference, and Apriority  171 to Williamson’s narrative, Norman sees a certain number of copies of Who’s Who and concludes based on his imaginative assessment that all copies look alike. Why choose this example, rather than a simply ostensive one? Well, universalizations call for an explanation of the epistemic process that allows one to go beyond particular observations, and one might find imagination to be a promising candidate. As we saw in the previous section, imagination might be thought of as productive. But then (WW) seems ad hoc and won’t generalize. We are induced to think that imagination is crucially involved in coming to know (WW) qua a standard case of the a posteriori; whereas the pull of the example lies in the imagination’s potential to justify knowledge of universal statements. Could someone complain that generalizations constitute nearly all of empirical science? What better example of the a posteriori? To my mind, the objection backfires. First, (WW) is not a scientific generalization. Furthermore, pushing this line requires showing how the proposed account applies to real generalizations of science – those that actually matter to us. The imaginative procedure carried out by Norman in the case of (WW) doesn’t seem to plausibly account for the way in which, say, a biologist may come to know that “African rhinos have two horns, while Indian rhinos have one horn”.6 On the contrary, imagination seems completely inessential for gaining that knowledge, which is done instead via empirical observation and inference. Thus, the objection highlights that example (WW) does not generalize but is rather ad hoc. Turning specifically to sensory imagination, both (CR) and (WW) generally serve Williamson’s argumentative purposes because they hinge on visualizing shades of red. But for (CR) that is crucial. For it suggests that even in standard cases of the a priori some experiential contribution that is more than merely enabling is required for knowledge. However, because of the key role played by color terms, this case seems ad hoc, as well. To illustrate, note that Williamson’s argument wouldn’t go through if the example involved terms from different domains. Consider a sentence that has the same logical structure as (CR), but whose ingredient terms won’t obviously prompt perceptual associations as they belong to chief a priori categories – say, mathematical, or moral terms. Consider: (EV) All even numbers are divisible by 2 Suppose that Norman has learned the ingredient terms in (EV), “number”, “division”, “2”, etc., in the usual way and has become similarly skillful at applying them. If one then asked Norman whether (EV) holds, it’s implausible that this time he would engage in any imaginative exercise to formulate the relevant judgment. In particular, sensory imagination seems completely inessential for gaining knowledge. Using color

172

Antonella Mallozzi

terms is devised to prompt corresponding mental imagery in the reader, aiming to generate a sense that even traditionally a priori ways of knowing are actually quasi-perceptual. But the example does not generalize: (CR) is ad hoc, as well. So we have a similar verdict with respect to (CR) as we reached with respect to (WW). The main takeaway is that both of these examples, by involving color terms and universal statements, are devised to induce us to think that imagination is crucially involved in coming to know those truths. But the examples don’t generalize, so they fail to jeopardize the signifcance of the a posteriori-a priori distinction. So far in this section we have granted that Norman is as described and comes to know (CR) and (WW) like Williamson says. We granted in particular that although Norman is “very skillful” with the relevant terms, he’s never connected them in the appropriate way. I’d like to conclude by fagging that Norman’s coming to know (CR) and (WW) by sensory imagination is quite implausible. Once certain odd assumptions are removed, sensory imagination no longer seems necessary even in those cases. Think of (CR). It seems implausible that someone who mastered the words “crimson” and “red” wouldn’t be able to know that shades of crimson are shades of red without adverting to imagination. Based on his semantic competence, Norman should be able to come to know (CR) simply by refecting on the ingredient terms and drawing the appropriate inference. In fact, (CR) seems a clear instance of an epistemic analytic truth (in Boghossian’s 1996 terminology), namely a truth that can be known to be true only by virtue of grasping its meaning.7 If Norman truly has mastered the relevant terms, he would come to know (CR) without any contribution from experience, even imaginative experience, besides what’s needed for acquiring them and enabling him to reason with them. Nor would he have to rely on visual simulations in imagination. While mental images of red might pop up in his mind as he assesses these matters, these images arguably won’t play an epistemic role for the resulting knowledge. At most they play merely a supporting role. For those mental images won’t add any further content to what Norman already knows from his semantic mastery. To add to this point, consider Aphantasiac Norman, who’s like Norman in all respects except he suffers from a congenital inability to experience mental imagery. It seems implausible that, given that he can’t deploy visual simulation in imagination, he wouldn’t be able to come to know (CR). Hence, Norman could come to know (CR) purely a priori by usual means. Next consider (WW). Besides conceptual mastery, Norman arguably needs to further rely on specifc memories as well as background empirical knowledge in order to know this truth. However, Williamson denies that episodic memory contributes to knowing (WW) and concludes

Imagination, Inference, and Apriority

173

that its status can’t be obviously a posteriori. But that seems odd. Norman plausibly remembers one or more copies of Who’s Who and puts those (episodic) memories to good use when drawing the generalization. Furthermore, Norman will need to draw from background empirical knowledge. Particularly, he’ll need some basic grasp of the type-token relation. Norman needs to know (possibly implicitly) that volumes of a certain publication are all printed in copies which come in the same size, color, number of pages, etc. That also contributes to inferring from the observation of particular samples of Who’s Who to the generalization (WW). Once the odd assumption that Norman doesn’t rely on memory or other information is removed, sensory imagination seems again inessential for the relevant epistemic process. Norman could come to know (WW) a posteriori by usual means.

8.4 Imagination vs. Inference Williamson’s core examples tied the epistemic powers of imagination to offine simulation and mental imagery. That is the frst notion of imagination at play in his account – what I called “sensory” imagination, as it involves simulating our perceptual capacities. However, Williamson has more recently stressed that our imaginative processes need not involve mental imagery but can be expanded to cover other cases. For example, “imagining that there is a golden mountain in Austria […] does not entail that one forms a mental image of a golden mountain in Austria” (2016: 15). Likewise, “a politician [who] is trying to work out what his core supporters would do at the next election if he voted for gun control […] imagines their reactions, but doing so need not involve mental imagery” (9). In other words, Williamson now cashes out his account in terms of so-called propositional imagination. It is common in the literature to distinguish between sensory imagination on the one hand and imagining-that or propositional imagination on the other hand. The frst involves consciously entertaining mental images. These are analogous to our percepts (not only visual) and have referential content, namely the objects they represent. Instead, imagining-that or propositional imagination is a propositional attitude with a content that can be evaluated as true or false, but possibly abstract and non-imagistic. Propositional imagination seems rather isolated from perceptual experience and more similar to conceptual thought. Several contemporary philosophers take imagination to be intrinsically quasi-perceptual (e.g., Balcerak Jackson (2016); Kind (2001); Kung (2010)); but others contend that imaginative processes can be independent of the occurrence of mental images, namely propositional (e.g., Chalmers (2002); McGinn (2004); Yablo (1993). Conversely, according to Arcangeli (2019), mental imagery might be independent of imaginative processes). Thoughtexperiments, and practical and counterfactual reasoning, for example,

174 Antonella Mallozzi are often thought to be carried out via propositional imagination (Kind and Kung (2016); Nichols (2006)). Importantly, if imagination is propositional and so need not involve mental imagery, the simulation hypothesis is no longer strictly tied to perception. Instead, the idea now seems to be that in imagination we may simulate online belief or inference, which need not be perceptual (possibly akin to Currie and Ravenscroft’s (2002) notion of “recreative” imagination, where imaginings are mental states that may mimic beliefs and decisions, inter alia). Leaving aside concerns tied to the empirical accuracy of this version of the simulation hypothesis, turning to propositional imagination might help expand the scope of Williamson’s account. But it also raises a number of philosophical concerns. First, some taxonomy would help understand better Williamson’s desired notion. Hypothetical attitudes such as supposing, or conceiving, while obviously in the vicinity of imagining in the propositional sense might be distinct from it and have different phenomenology and different cognitive goals (Balcerak Jackson (2016)). Elsewhere Williamson has argued against the epistemic powers of conceivability (particularly for modal knowledge: 2007: Ch.5). On the other hand, he often seems to use “imagining” and “supposing” interchangeably. So Williamson could offer us some useful clarifcation of those various notions. Second, there is a more fundamental worry concerning the role of (propositional) imagination within the broader architecture of our cognitive functions. One might wonder in what sense the capacity at stake is supposedly still imaginative, when it no longer deploys quasi-perceptual imagining but it’s rather akin to belief and reasoning. Simply put, how does the revised candidate account distinguish between imagination and inference? “Inference” is here meant to broadly denote reasoning with beliefs toward a certain judgment. More precisely, following a prominent characterization, inference is a transition or “movement of thought” from some beliefs to a conclusion, where the thinker takes the conclusion to be based on or supported by those beliefs (Boghossian (2014)). One might suggest that the difference has to do with the hypothetical nature of imaginative thinking, and possibly a functional difference between imaginative representation vs. belief. But inference can certainly involve suppositions as well as beliefs. The question simply remerges at a more fne-grained level. Namely, what characterizes distinctively imaginative hypothetical thinking as opposed to inferential hypothetical thinking? Once imaginative thinking has been emptied of its sensory imagistic content, the suspicion is that we might be dealing with the very same mental operation, namely one where the thinker proceeds from a certain supposition to its consequences. The issue is especially pressing if we consider Williamson’s recent remark that principles of deductive logic supposedly shape rigorous

Imagination, Inference, and Apriority

175

imaginative procedures (2016). In developing conditional suppositions in imagination, according to Williamson we often follow roughly the logical deductive procedures of tree tableaux. For example, in imagining that there is a golden mountain in Austria, one does something formally similar to applying the frst-order tableaux rule for the existential quantifer and developing the imaginative supposition via deduction (similarly for the rules for conjunction, disjunction, etc.). Williamson himself raises the worry that the analogy with tableaux procedures might not be strongest, since in imaginative exercises we do not normally look for contradictions. However, if correct, the general contribution of rigorous deductive logical reasoning may enhance the reliability of our imaginative procedures, given that logical rules are truth-preserving. But on the other hand, the analogy reinforces the worry that the procedures in question may no longer be imaginative in any clear sense. The general problem (surely not just for Williamson’s account) is that while a notion of propositional imagination may help us build a more inclusive theory, it also calls for clarifying the principled distinction between imagining vs. inferring as hypothetical attitudes that similarly feature suppositions. Furthermore, the theory should say why we need to resort to this distinctive notion of imagination in order to explain the relevant epistemic processes, as opposed to hypothetical inference. In general, a candidate account should identify the limitations of an inferentialist theory as well as the theoretical advantages of appealing to imagination. Part of the pressure to clarify the distinctive explanatory work of propositional imagination compared to hypothetical inferential reasoning stems again from Williamson’s own anti-exceptionalism. If hypothetical inferential reasoning effectively explains the relevant epistemic processes, we don’t need to invoke any “extra” faculty of (propositional) imagination. To illustrate, consider Williamson’s recent claim that imagination yields knowledge of “mundane, widespread matters of immediate practical relevance” (2016: 1). For example, he describes a group of distant ancestors about to enter an unknown forest. What sorts of epistemic and practical advantages might imagination yield? An “obvious answer” for him is that An imagination will alert them to various potential dangers and opportunities […] They imagine wolves in the forest; warned of the danger, they keep a sharper look-out for signs of wolves. They imagine edible berries in the forest; alerted to the potential opportunity, they look about for bushes of the right kind. In both cases, their imagination enables them to prepare for practically relevant possibilities, helping them avoid dangers and take advantage of opportunities. (3)

176

Antonella Mallozzi

But in what way is the episode described a case of “knowledge by imagination”? More precisely, how is imagination playing an epistemic role? And how does suddenly realizing that one might be surrounded by various dangers or opportunities require the work of imagination? In fact, nothing especially imaginative seems to be at play as those ancestors go explore that forest and make the relevant decisions. A more natural reading of Williamson’s story sees the ancestors assessing various possibilities concerning their environment, thus reasoning hypothetically based on certain premises. Those premises implicitly or explicitly convey information based on the ancestors’ own past experience (e.g., episodic memories of similar forests they may have previously explored) and/or analogous testimony from others. Such information is put to use in hypothetical reasoning, drawing a conditional conclusion. Something along the lines of: “If present unknown forest y is suffciently similar to previously experienced forest x, then there might be wolves and berries in forest y”. Clearly the inference need not be explicit. Appealing to simple inference seems to be all that is needed in order to explain how our characters may reach some (conditional) empirical knowledge; whereas it is not clear why one should invoke imagination. Furthermore, for Williamson this is an “involuntary” manifestation of the imagination, by “breaking into the stream of consciousness with reminders of dangers and opportunities” (4). He distinguishes that from a “voluntary” mode, where imagination can instead be “turned on” when needed for problem-solving purposes. However, Williamson also claims that imagination constitutes a reliable method for forming true beliefs, and hence yields knowledge, only in the voluntary mode not the involuntary one (7). In sum, it is unclear that imagination plays the required epistemic role for knowledge in such cases, that anything more than inferential reasoning is involved. Perhaps all Williamson wants imagination to contribute in the story is mental imagery – thus going back to the frst, sensory type of imagination. The ancestors might be visualizing wolves and berries as they assess those matters. If that’s the point, however, he should clarify why the occurrence of such images should play such a crucial epistemic role, as opposed to only being an accidental byproduct of inferring based on background knowledge. Plausibly, our ancestors would be able to reach the same conclusions regarding their surroundings without any imagery occurring, since those images won’t add any further content to the ingredient beliefs in the ancestors’ inferences. More generally, while the occurrence of mental images might help some subjects carry out various cognitive tasks, it seems contingent and inessential for the resulting judgments. One might well work one’s way through understanding and learning theoretical as well as practical matters without virtually any images occurring. Accordingly, it seems fruitful to think of mental imagery rather in terms of a cognitive tool or heuristic, playing a possibly

Imagination, Inference, and Apriority

177

supportive but generally inessential role for knowledge. Williamson’s own examples involving logic and set theory may ft well this “instrumental” account of imagination, since mental pictures and metaphors seem mostly an aid to understanding and reasoning. However, note that adopting the instrumental view would require some considerable adjustment to the account. Specifcally, it would require introducing some principled criterion for disentangling cases where sensory imagination is supposedly epistemically necessary for knowledge – like in the original examples (CR) and (WW) – from cases where it is merely instrumental and thus neither necessary nor suffcient for knowledge. Importantly, if imagination is a tool or heuristic, it is not a source of knowledge. Are things any different in cases where imagination supposedly operates in a voluntary mode? Perhaps its epistemic role is more straightforward in such cases. Consider Williamson’s example of a hunter who is pondering whether to jump a stream that obstructs his way. The decision is “vital”, since the hunter might die if he tries and fails. Williamson himself acknowledges that the hunter might easily imagine himself failing as well as succeeding in jumping the stream. However, “imagining oneself trying” somehow constitutes a reliable method for forming true beliefs. “Indeed […] under suitable conditions, the method enables one to know what would happen in the hypothetical circumstances” (8). Williamson is confdent that since this is a voluntary use of imagination, that ensures its reliability (7). This optimism is hard to share however, especially since he admits that “hardly anything has been said so far to explain the method’s reliability” (8). Falling short of standards we usually take as basic in epistemology, it seems quite doubtful that the imaginative exercise that the hunter carries out yields knowledge. Indeed, his decision whether or not to jump shouldn’t be based on that simulation. Perhaps, like the hunter, we would all imagine ourselves jumping in the described scenario. Perhaps that might sometimes help our overall assessment of the situation as we make the decision. But that isn’t tantamount to imagination yielding knowledge. For the described imaginative processes might feel familiar, and possibly have heuristic roles, while being again only an accidental byproduct of the actual epistemic process. An alternative account of the hunter’s decision-making process agrees with Williamson that the hunter carries out conditional reasoning via supposition but contends that its epistemic value is independent of the occurrence of any imaginative exercise. Although imagining oneself jumping might perhaps contribute to improve the hunter’s actual performance, all we need to explain his decision-making process is hypothetical inference based on empirical evidence. That includes evidence collected through perception (e.g., measuring by eye the width of the stream, estimating its depth, checking how strong the water stream is), introspection (evaluating how tired one is, whether one feels in shape or not, etc.), and empirical background knowledge (about mountain

178

Antonella Mallozzi

streams, rocks, etc., as well as episodic memories of similar experiences). The hunter decides (and rightly so) whether to jump based on the outcome of inferring from such empirical information, which likely happens largely implicitly and quickly. Since the evidence is perceptual, introspective, and mnemonic, the inference is clearly a posteriori. In conclusion, in order to defend the epistemic powers of imagination qua propositional imagination, one needs to articulate criteria in virtue of which certain processes and states are distinctively imaginative as opposed to broadly inferential. Based on that, one should further make the case that explaining the relevant epistemic processes requires appealing to imagination. Or, short of that, one should make the case that an account that rests on the epistemic roles of imagination offers substantive theoretical advantages over an account that rests on hypothetical inferential reasoning.8

Notes 1 With this criticism Williamson challenges the a priori-a posteriori distinction as introduced by “bottom-up” standards, namely by examples. Williamson further criticizes “top-down” characterizations (namely, via defnition), aiming to show that those are also epistemologically shallow. Since the latter criticism doesn’t involve imagination, I won’t discuss it here. But see Boghossian’s reply to Williamson for an illuminating discussion of the issues involved with top-down characterizations (Boghossian and Williamson 2020: Ch. 9). 2 For discussion, see Casullo (2013) and Jenkins (2008). 3 This problem is especially pressing for Williamson’s counterfactual account of modal knowledge (see Mallozzi 2020). 4 Jenkins and Kasaki (2014) draw a similar parallel with Kant’s imagination. Kind (2020) similarly calls this imaginative operation “scaffolding”. 5 Jenkins and Kasaki (2014) similarly complain that Williamson’s examples are “too easy”. Malmgren (2011) targets the 2007 example mentioned above, ‘If two marks had been nine inches apart, they would have been at least nineteen centimeters apart’. She stresses that this is a peculiar case and it is implausible that it generalizes in the way Williamson wants. 6 Thanks to Michael Devitt for emphasizing this point. 7 Boghossian contrasts epistemic analyticity with metaphysical analyticity, i.e., the idea that certain truths are true only by virtue of meaning. According to him, Quine’s arguments successfully rejected metaphysical analyticity, whereas epistemic analyticity can be salvaged (1996). 8 Thanks to Chris Badura, Paul Boghossian, Michael Devitt, Amy Kind, David Papineau, and an anonymous referee for helpful comments on a previous draft.

References Arcangeli, M. (2019). The Two Faces of Mental Imagery. Philosophy and Phenomenological Research. https://doi.org/10.1111/phpr.12589

Imagination, Inference, and Apriority

179

Balcerak Jackson, M. (2016). On the Epistemic Value of Imagining, Supposing, and Conceiving. In A. Kind and P. Kung (Eds.), Knowledge Through Imagination (pp. 41–60). Oxford: Oxford University Press. Barlassina, L., and Gordon, R. (2017). Folk Psychology as Mental Simulation. The Stanford Encyclopedia of Philosophy. https://plato.stanford.edu/ archives/sum2017/entries/folkpsych-simulation/ Boghossian, P. (1996). Analyticity Reconsidered. Noûs 30: 360–91. ———. (2014). What Is Inference? Philosophical Studies 169: 1–18. Boghossian, P., and Williamson, T. (2020). Debating the A Priori. Oxford: Oxford University Press. Casullo, A. (2013). Articulating the A Priori-A Posteriori Distinction. In A. Casullo and J. Thurow (Eds.), The A Priori in Philosophy (pp. 249–74). Oxford: Oxford University Press. Chalmers, D. (2002). Does Conceivability Entail Possibility? In T. Gendler and J. Hawthorne (Eds.), Conceivability and Possibility (pp. 145–201). Oxford: Oxford University Press. Currie, G. (1995). Visual Imagery as the Simulation of Vision. Mind & Language, 10: 25–44. Currie, G., and Ravenscroft, I. (2002). Recreative Minds. Oxford: Oxford University Press. Gallagher, S. (2007). Simulation Trouble. Social Neuroscience 2: 353–65. Heal, J. (1998). Understanding Other Minds from the Inside. In A. O’Hear (Ed.), Contemporary Issues in the Philosophy of Mind (pp. 83–100). Cambridge: Cambridge University Press. Jenkins, C. (2008). Modal Knowledge, Counterfactual Knowledge and the Role of Experience. The Philosophical Quarterly 58: 693–701. Jenkins, C., and Kasaki, M. (2014). The Traditional Conception of the A Priori. Synthese https://doi.org/10.1007/s11229-013-0394-x Kind, A., and Kung, P. (2016). Knowledge Through Imagination. Oxford: Oxford University Press. Kind, A. (2001). Putting the Image Back in Imagination. Philosophy and Phenomenological Research 62(1): 85–109. ——— (2020). What Imagination Teaches. In John Schwenkler and Enoch Lambert (Eds.), Becoming Someone New: Essays on Transformative Experience, Choice and Change (pp. 133–46). Oxford: Oxford University Press. Kung, P. (2010). Imagining as a Guide to Possibility. Philosophy and Phenomenological Research, 81, 620–663. Langland-Hassan, P. (2016). On Choosing What to Imagine. In A. Kind and P.  Kung (Eds.), Knowledge Through Imagination (pp. 61–84). Oxford: Oxford University Press. Mallozzi, A. (2020). Superexplanations for Counterfactual Knowledge. Philosophical Studies https://doi.org/10.1007/s11098-020-01477-0 Malmgren, A. (2011). Rationalism and the Content of Intuitive Judgements. Mind 120: 263–327. McGinn, C. (2004). Mindsight: Image, Dream, Meaning. Cambridge, MA: Harvard University Press. Nichols, S. (2006). The Architecture of Imagination. Oxford: Oxford University Press.

180

Antonella Mallozzi

Nichols, S., Stich, S., Leslie, A., and Klein, D. (1996). Varieties of OffineSimulation. Theories of Theories of Mind 24: 39–74. Quine, W.V.O. (1951). Two Dogmas of Empiricism. The Philosophical Review 60: 20–43. Spaulding, S. (2016a). Imagination through Knowledge. In A. Kind and P. Kung (Eds.), Knowledge Through Imagination (pp. 207–26). Oxford: Oxford University Press. ———. (2016b). Simulation Theory. In A. Kind (Ed.), The Routledge Handbook of Philosophy of Imagination (pp. 262–73). New York: Routledge. Williamson, T. (2007). The Philosophy of Philosophy. Oxford: Blackwell. ———. (2013). How Deep Is the Distinction between A Priori and A Posteriori Knowledge? In A. Casullo and J. Thurow (Eds.), The A Priori in Philosophy (pp. 291–313). Oxford: Oxford University Press. ———. (2016). Knowing by Imagining. In A. Kind and P. Kung (Eds.), Knowledge Through Imagination (pp. 112–23). Oxford: Oxford University Press. Yablo, S. (1993). Is Conceivability a Guide to Possibility? Philosophy and Phenomenological Research 53, 1–42.

Section III

Thought Experiments

9

Narratives and Thought Experiments Restoring the Role of Imagination Margherita Arcangeli

9.1 Introduction1 According to a common view, fctions are invitations to imagine whereas non-fctions are invitations to believe (loci classici are Walton 1990 and Currie 1990, for a recent defence see Stock 2017). In Fiction and Narrative Derek Matravers rejects this view, and argues that our cognitive interaction with narratives is the same whether they are fctional or not. He provocatively claims that such interaction does not call for imagination, if the latter is defned, as it typically is, as a distinctive attitude. Rather, Matravers gives pride of place to mental models in accounting for our engagement with narratives. There is a nice parallel between this idea and another in the debate on thought experiments. An infuential account of the cognitive processes underlying thought experimentation, the model-based approach, has it that the reasoning employed in thought experiments is closely related to the one used in the consumption of narratives. The most developed version of the model-based approach claims that thought experimenters reason by manipulating a mental model, which simulates the situation described in the thought experimental narrative (Miščević 1992, 2007; Nersessian 1993, 2007). The aim of my contribution is to restore imagination to the forefront of our interactions with (fctional and non-fctional) narratives, including those involved in thought experiments. I take this to be a frst step towards the clarifcation of why the latter can be regarded as a source of knowledge. More generally, the resulting account preserves an important epistemic role for the imagination in grasping narratives in science and the arts. In Section 9.2 I will go into the details of Matravers’ view, specifying the notion of imagination he appeals to, namely propositional imagination (in its belief-like form). I will concede that Matravers might be right in claiming that invoking imagination alone is not suffcient to separate fctional from non-fctional narratives, but I will contest his idea that imagination is not suited to play a key role in our engagement with narratives. I will show that this idea relies on a misinterpretation of the relationship between imagination and belief.

184

Margherita Arcangeli

In Section 9.3 I will turn to thought experiments and to a rather neglected aspect of them, namely their narrative dimension (i.e., the fact that they are presented through narratives). I will show how such a dimension is tightly connected to their cognitive dimension (i.e., what happens in the mind of a thought experimenter). Although imagination has been identifed as a relevant underpinning of thought experimentation, Matravers’s worries can be raised also with respect to thought experiments, thus suggesting that we replace the notion of imagination with that of mental model. Relying on a prominent view of the nature of imagination as a simulative attitude (which is not only belief-like), in Section 9.4 I will distinguish imagination from mental models by arguing that they are different kinds of phenomena, involving different kinds of mental simulation. I will thus show that Matravers appeals to a questionable defnition of imagination and that the model-based approach fails to recognise the important role of imagination in thought experiments.

9.2 Narratives and Imagination Talking about fction very often implies talking about imagination. Imagination seems to be both the means by which we create a fctional world and how we access that world, in the context of all kinds of fction (involving images, novels, plays, operas, ballets, flms). In other words, imagination is the cognitive ability exploited by consumers of fction. The philosophical literature has gone further and claimed that while works of fction are invitations (if not prescriptions) to imagine, works of non-fction instruct their consumers to believe. This is what Matravers calls the “consensus view” (Matravers 2014, 24), which cognitively casts the difference between fction and non-fction in terms of the difference between imagination and belief.2 He rejects this view and argues that, insofar as both fction and non-fction are mediated interactions with situations, they call for the same kind of cognitive interaction. According to him, proponents of the consensus view rely on a notion of imagination that itself suggests this conclusion. The relevant notion seems to be what has been called “propositional imagination” (Kind 2016a). As Walton stresses, propositional imagination “is doing something with a proposition one has in mind” (Walton 1990, 20). In other words, imagination, like belief, desire, or perception, is a psychological attitude, a specifc way of presenting (or representing) a mental content. When such mental content takes a propositional form, imagination is a propositional attitude. Very often this type of imagination has been taken to bear similarities with belief (e.g., as regards the emotions they can trigger), whence the alternative labels “belief-like imagination” (Currie & Ravenscroft 2002), “make-believe” (Walton 1990; Davies 2007a), or “cognitive imagination” (McGinn 2004;

Narratives and Thought Experiments

185

Weinberg & Meskin 2006; Weinberg 2008). 3 A common way to interpret the similarity between belief-like imagination (henceforth imagination) and belief is to say that the former is the mental simulation of the latter.4 Notwithstanding, their affnities with beliefs, imaginings are sui generis mental states, distinct and irreducible to beliefs. 5 What mainly sets imagination and belief apart are their relations to truth and the will (Arcangeli 2018c). Emma can imagine that there are lilacs in the garden, even though she believes, even knows, that this is not true. Moreover, her imagining does not confict with her belief that there are only geraniums in the garden. By contrast, on the basis of the evidence she possesses, Emma cannot deliberately (and without irrationality) form the belief that there are lilacs in the garden. First, in order to believe we need reliable sources of information and reasons. Second, our beliefs are subject to an ideal of coherence. Therefore, imaginings and beliefs typically differ in the inputs they rely on and the outputs they give rise to. For instance, perception is not epistemically connected to imagination as it is with belief: my seeing x justifes my belief about x, but not my imagining about x. Emma’s belief about geraniums is likely to be justifed by her perceptual experience of the fowers in the garden, but the same does not hold for her imagining about lilacs. Similarly, Emma’s belief, in conjunction with appropriate desires, can lead her to undertake certain actions (e.g., buying some fertiliser), whereas her imagining does not typically have the same motivational power. When it comes to our engagement with fctions, imagination can be seen as what allows us to reason in the fctional world, that is, to project ourselves into the fctional situation and to simulate what one (not necessarily the imaginer) would believe from that perspective.6 We do not believe that Remedios, the beautiful woman in One Hundred Years of Solitude, ascended to heaven, body and soul, or that Thérèse Desqueyroux coldly tried to poison her husband. It is as if these events had occurred, we make believe (or imagine) them. This might be an accurate description of what is going on in the readers’ mind, but what Matravers contends is that this is not specifc to our engagement with fctions: imagination seems to be recruited whenever we are not really confronted with a given situation, but we have a mediated access to it through a representation.7 Compare the following cases. Emma is reading Miss Smilla’s Feeling for Snow. She gets fascinated by the idea that there are different types of snow and imagines that, like Smilla, she masters knowledge of these different types. Such reading elicits her desire to learn more about snow and to plan a trip to Greenland. Emma is reading a book based on a personal diary of an actual trip to Greenland. She gets fascinated by the idea that there are different

186

Margherita Arcangeli types of snow and imagines that, like the writer, she masters knowledge of these different types. Such reading elicits her desire to learn more about snow and to plan a trip to Greenland.

Although Emma is engaged with a literary fction in the frst case and a work of non-fction in the second case, there is no difference in the kind of mental state she fnds herself in. In both cases she exploits her imagination and she is led to entertain the same imagining, which eventually indirectly (by generating a desire) causes the same action. According to Matravers this comes as no surprise, since fctions and non-fctions typically do not differ in the perceptual inputs and in the behavioural outputs they cause. He then wonders why we should endorse the consensus view.8 The aforementioned characterisation of imagination is, in fact, neutral with respect to the distinction between fction and non-fction. We may agree with Matravers in claiming that it is not by relying solely on the notion of imagination that we can demarcate fction from non-fction, but we may think that imagination has an important role to play in our interaction with narratives tout court. Matravers rejects this idea as well. Matravers argues that we get a better understanding of the cognitive process at stake in our engagement with narratives by replacing the notion of imagination with that of mental model (a notion he owes to research in cognitive sciences; see Johnson-Laird 1983). The latter would have two explanatory advantages over the former. First, mental models are “compartmentalized but not isolated from our pre-existing structures of belief” (Johnson-Laird 1983, 90), thus explaining the fact that (part of) the content of narratives (or “representations” – see fn 7) can be integrated with our beliefs. Second, in talking about mental models we avoid positing a specifc attitude (i.e., imagination or belief) proper to either fction or non-fction, thus accounting for the intuitive neutrality we have, at least at frst, vis-àvis the contents of the narrative in question (Matravers 2014, 79). With respect to the frst advantage, Matravers starts with the consideration that non-fctions are mainly made up by true propositions, but most fctions are mixtures of true propositions and propositions that are true only in fction. According to the proponent of the consensus view, while the former should be believed, the latter should be imagined. Here lies, Matravers argues, a problem for the proponent of the consensus view: “if the imagination is defned in contrast to belief, how is it that they work seamlessly together?” (Matravers 2014, 38). The conclusion, suggested by Matravers, is that we should get rid of the notion of imagination – at least in our account of narratives – and replace it with a notion that refers to a cognitive underpinning that can easily interact with belief, namely mental models. It is true that imagination is typically defned in contrast with belief, in the sense that it is compared to and distinguished from the latter.

Narratives and Thought Experiments

187

But defning imagination as being different from belief does not imply that these mental states cannot smoothly work together. To see this, consider belief and desire. These are two different propositional attitudes, but in our reasoning they often occur together. The same seems to hold for imagination and belief. It has been advanced that in certain fctional contexts a subject must make her imagination and belief interact seamlessly (Currie & Ravenscroft 2002; Weinberg & Meskin 2016). Matravers (2014, 43) argues that by conceding this point, new problems spring up: if most of the time our engagement with fction implies the involvement of two sorts of mental state (i.e., imagination and belief), but there is no tangible difference (in, e.g., inputs and outputs) between the two, then why should we posit these types of mental state in the frst place? This leads to his second reason to prefer mental models over imagination, which we can only assess by means of a careful analysis of what mental models are and how they relate to imagination. Before tackling these issues, I would like to make Matravers’ case as strong as possible by turning to the debate about thought experiments. In this debate, mental models are used to explain our cognitive interactions with the fctional narratives of thought experiments.

9.3 Thought Experiments and Narratives Thought experiments are a much discussed topic, at the heart of a lively debate among contemporary philosophers. They are commonly used both in philosophical disciplines (e.g., Plato’s Cave – Republic 514b-520a, the brain in the vat – Putnam 1981) and in natural sciences (e.g., Galilei on free-fall – Galilei 1638, Einstein’s lift – Einstein & Infeld 1938). Moreover, although an agreed-upon defnition is lacking, almost all authors involved in the debate agree in considering thought experiments as epistemic tools aimed at somehow testing a theory or specifc hypotheses.9 Thought experiments are extremely important because through them we gain, share, and spread knowledge. It is important to highlight that thought experiments are mostly written and publicly presented through narratives, which may have different levels of detail and rhetoric, and which may be accompanied by pictures.10 This narrative dimension is a rather neglected aspect of thought experiments, but it has led authors to establish a parallelism between them and literary fctions, which are narratives par excellence.11 A thought experiment tells us a story. This story can be about ordinary objects or beings, such as stones, boats, cows, and beetles, but also about extraordinary objects or beings, such as ropes without mass, a substance that looks like water but is not composed of H 2O molecules, demons, and human doppelgängers without consciousness.12 The scenarios depicted by thought experiments are often described as “hypothetical” or

188 Margherita Arcangeli “counterfactual”. Thought experiments tell us stories that could have happened or could happen, if certain conditions were met. In this sense, thought experiments display an affnity with fctional narratives: like works of fction, thought experiments are “typically not, and in any case need not be, representations of anything real” (Elgin 2014, 227). On this view thought experiments can be seen as short novels or fctional tales, and those who put them forward, such as Galilei or Putnam, are novelists or storytellers. The idea that thought experiments are “short fctional narratives” (McAllister 1996, 233) seems to be fairly intuitive and widely accepted.13 Some authors have taken this idea seriously and explored the potential for the debate on thought experiments, as well as for debates in aesthetics, of a comparative analysis bringing together thought experiments and literature (see, e.g., Davenport 1983; Carroll 2002; Davies 2007b, 2010, 2013; Elgin 2007, 2014; Swirski 2007; Gaetens 2009; Ichikawa & Jarvis 2009; McComb 2013; Meynell 2014). Some have even suggested considering thought experiments as a genre, like science fction (see Weinberg 2008; Peterson 2018). Among the authors who study the relationship between thought experimentation and fctional narratives, David Davies has offered the most sustained argument in favour of their similarities. According to him, thought experiments, whether scientifc or philosophical, meet two necessary and suffcient conditions for the fctionality of a narrative. First, fctional narratives are not subject to what he calls the “fdelity constraint” (Davies 2010, 52). A narrative obeying the fdelity constraint requires the author to include in the narrative only those events that she believes to have occurred and to tell them in the order in which she believes they have occurred. A newspaper article exposing a crime or the chronicle of Marie Antoinette’s life both obey the fdelity constraint. Davies maintains that we cannot say the same for fctional narratives. Gabriel García Márquez in his One Hundred Years of Solitude did not intend to describe what really happened in the village of Macondo; the village is imaginary like probably most of the characters who populate it. Although François Mauriac was inspired by the trial of Madame Canaby, the aim of his novel Thérèse Desqueyroux was not to write a faithful account of the latter. Davies argues that in the case of fctional narratives, the purpose of the author is more general, “such as entertaining or perhaps instructing readers in certain specifc ways” (Davies 2010, 52). This does not prevent novelists from intending, and making explicit to the reader, that the situation might have occurred as described. The idea is that, even though Mauriac intended to describe a potentially real story, Thérèse Desqueyroux would not obey the fdelity constraint. Davies also argues that thought experiments do not obey the fdelity constraint. Their authors are not motivated by the goal of describing a real event and how it happened. Frank Jackson (1982), in his thought experiment of Mary the super-scientist, never intended to tell the true

Narratives and Thought Experiments

189

story of a neuroscientist named Mary who grew up in a black-and-white room. With his cow in the feld problem, Martin Cohen (1999) did not aim to provide an accurate account of a farmer’s concern for his favourite cow, although he may have sought to provide a more plausible example than other Gettier cases (see Gettier 1963). Indeed, a farmer might feel reassured in seeing a black-and-white silhouette in the feld and believe that his cow is in the feld, when the cow is actually in the feld, but what he saw was something else. With his thought experiment about running after a ray of light, Albert Einstein (1949) did not intend to give us an account of what he did one afternoon. Galilei too was not motivated by the presentation of a fact, even though his thought experiment on falling bodies described a situation that was possible and that nowadays can be reproduced: we can see (on the internet), and not just imagine, a hammer and a feather falling with the same speed when dropped by an astronaut on the Moon (Strohminger, this volume). By pointing out that thought experiments, like fctions, can represent situations as possibly real, Davies aims to block a potential objection (see also Davies 2007b). It has been argued that at least scientifc thought experiments should not be seen as relying on fctional narratives, because they always describe a situation as potentially real (see Miščević 1992; Nersessian 1993). Even in scientifc thought experiments involving bizarre situations, such as chasing a beam of light or a lift in cosmic space, as Nancy Nersessian puts it: “The assumption is that if the experiment could be performed, the chain of events would unfold according to the way things usually take place in the world” (Nersessian 1993, 295). By contrast fctional narratives do not presuppose any such constraint on the course of the events described in the narrative. As Davies points out, even accepting Nersessian’s hypothesis on scientifc thought experiments (see Cooper 2005 for a critique), it is not clear why it should differentiate them from fctional narratives. What Davies suggests is that Nersessian’s hypothesis is weaker than the fdelity constraint. According to that constraint the author of the non-fctional narrative should not simply present some possible system which unfolds as it would in the real world (as Nersessian demands for scientifc thought experiments), but rather must only present how those described facts, which she believes to have actually occurred, unfolded. Second, according to Davies, the author of a fction intends to induce the reader to pretend to believe the content of the narrative. In other words, he endorses the consensus view and claims that a fctional narrative, as opposed to a non-fctional narrative, involves make-believing (or imagining) that the described situation occurs. Davies argues that thought experiments are invitations to imagine too (see also Stock 2017). When we read the thought experiment of Mary the super-scientist, we do not believe that there is a scientist, Mary, who grew up in a black-and-white room. Likewise, we do not believe that

190

Margherita Arcangeli

there is a cat which is both dead and alive at the same time, when we are confronted with Schrödinger’s cat thought experiment (Schrödinger 1935). The same holds for less fanciful situations: when reading Newton’s thought experiment of the bucket (Newton 1687) or the cow in the feld (Cohen 1999), we do not believe, respectively, that in front of us there is a bucket with spinning water, or that there is a farmer worried about his cow and then reassured by seeing a black-and-white silhouette in the feld. Also in thought experiments we imagine that what is being told really happens or happened. Given Davies’ two points, it would appear reasonable to argue that a thought experiment is presented not only by a narrative, but more precisely by a fctional narrative. Moreover, this analysis, which puts thought experimentation at the same level as literary fction, points at the imagination as a key component of the cognitive process underlying thought experiments.14 The latter is assimilated to the cognitive process underlying the reception of fction, for which imagination seems to be essential too. Although Matravers is not concerned with thought experiments, his worries can be raised against them as well. No matter whether they are conveyed through a fctional or non-fctional narrative, Matravers’ view threatens the idea that imagination is the key cognitive process underlying our engagement with thought experiments. Mental models would instead play this role. This is in line with what has been defended by infuential proponents of the model-based approach to thought experiments. The model-based approach analyses thought experiments in terms of their cognitive functioning, more than in terms of their epistemic outputs. Indeed, in studying thought experiments we can ask, on the one hand, about their results and, on the other hand, about the way they produce these results. Given that philosophers largely agree that thought experiments increase our knowledge, the frst question leads to issues that are mainly epistemological (e.g., How can a thought experiment produce new empirical knowledge without the input of new data? What kind of knowledge does it really produce?), whereas the second question leads us to examine thought experiments from the point of view of the cognitive functioning of the thought experimenter.15 The model-based approach is mainly concerned with the second set of questions. It explores the cognitive dimension of thought experimentation by appealing to research in cognitive science on model-based reasoning. Among the authors who have defended this approach (e.g., Gooding 1993; Bishop 1999; Palmieri 2003; Cooper 2005), two have played a particularly important role in developing it: Nenad Miščević and Nancy Nersessian. These authors start from the idea that the narrative aspect of thought experimentation is crucial to the process of performing a thought experiment. They hold that the cognitive process underlying thought

Narratives and Thought Experiments

191

experimentation is closely related to the one that is exploited in the consumption of narratives, given that thought experiments are presented in a narrative form. Drawing on the literature in cognitive science on narrative comprehension, Miščević and Nersessian have argued that it is through the manipulation of a mental model that a thought experimenter acquires new knowledge (Miščević 1992, 2007; Nersessian 1993, 2007). Should we really replace the notion of imagination with that of mental model to account for our engagement with all sorts of narratives, including those presenting thought experiments? In order to answer this question, we need to know more about the nature of mental models and how they relate to imagination.

9.4 Mental Models and Imagination Let me expand briefy on the notion of mental model. According to Philip Johnson-Laird (1983, 2004), one of the most infuential proponents of mental models, a mental model represents a real-world or imaginary situation. It is a structure stored in short- or long-term memory and is defned as a third type of mental representation, intermediary between propositional and pictorial mental representations. Mental models are structurally analogous to what they represent, but not all of them can be visualised. In the case of narrative comprehension, we can say that mental models simulate what is described by the given narrative. For this reason, Miščević and Nersessian, who explicitly rely on Johnson-Laird’s notion of mental model, see thought experimentation as a species of simulative reasoning.16 We have also seen that imagination has been understood as a form of mental simulation, but interesting differences between mental models and imagination can be highlighted. To do this, we need a useful distinction in philosophy of mind, namely that between psychological attitude and mental content. I have already hinted at it, but something more can be said. This distinction is easily grasped through examples. In a belief that p, p is the content and belief is the attitude; in a desire that q, q is the content and desire is the attitude; in a perception of x, x is the content and perception is the attitude. A psychological attitude contributes to a mental state by dealing with how the latter represents, rather than with what it represents (i.e., the mental content). As there are different types of psychological attitudes, there are also different types of mental content. A usual way of typing mental contents is in terms of their format (or vehicle).17 What a mental state represents can be conveyed in different representational formats. A notable distinction is the one between propositional and non-propositional content: while the former has a propositional format (i.e., it is propositionally organised or carried by propositions), the latter does not (e.g., it has a pictorial or iconic format).

192

Margherita Arcangeli

With such a distinction at hand, we can come back to the relationship between imagination and mental models. Recall that imagination is quite unanimously considered as a psychological attitude. In imagining that there are lilacs in the garden, “there are lilacs in the garden” is the content and imagining is the attitude. What about mental models? The same does not seem to hold for them.18 Talking of mental models as a structure stored in memory, as a type of mental representation (which is neither fully propositional nor fully pictorial), suggests that they have to do with the notion of mental content, more than that of psychological attitude. Mental models can be taken as a type of format in which mental contents can be conveyed. On this view, mental models may be accessed by means of several attitudes (e.g., belief, imagination). If this interpretation is correct, in replacing the notion of imagination with that of mental model we are not dealing with the same dimension, we are moving from the attitudinal to the content dimension. This might be precisely what Matravers has in mind in pleading for mental models. Indeed, he puts forward a two-stage model of our cognitive engagement with narratives according to which at a frst stage we simply build a mental model of the narrative content without having a specifc attitude towards it (Matravers 2014, 90). By contrast, at a second stage (some of) those propositions can be integrated with our pre-existing beliefs and thus believed – this is what happens in response to non-fction usually and to fction less frequently. According to Matravers, the neutrality of the frst stage is lost by analyses of narratives pivoting on the notion of imagination construed as a distinctive attitude. However, Matravers’s frst stage does not seem to capture an explanandum at the same level of interest to proponents of the consensus view. The latter are mainly interested in what goes on at the personal level of a subject engaged with a narrative, whereas the former is a description of what goes on at the sub-personal level. Matravers’ frst stage would capture the phase in which some model-building sub-personal processes generate a mental representation of the narrative’s content. But what happens at the personal level? Arguably at this level a subject cannot have a foating content without an attitude through which she grasps it. Matravers seems to anticipate this kind of worry in claiming that if any attitude is to be invoked, “it is merely one of them [the given propositions] being part of the content of whatever particular representation [narrative] we are reading or remembering” (Matravers 2014, 78–9). This quote suggests that we take entertaining (or a cognate phenomenon, like apprehending, grasping, considering, understanding) to be the relevant attitude involved in the frst stage. The problem for Matravers is that entertaining does not guarantee that he has a theory that does not appeal to imagination. The relationship between entertainment and imagination is controversial. First, imagination has been associated with entertainment. Sartre, for

Narratives and Thought Experiments

193

instance, in his characterisation of imagination as the attitude which “posits its object as a nothingness” (Sartre 1940, 11), puts forward the idea that imagination can be neutral vis-à-vis its content, presenting it under no specifc guise. There is indeed a kind of disinterestedness that imagination exhibits towards its content. Quite often imagination is said to be non-committal, contrary to committal attitudes, such as belief or perception. The same can be said for entertaining, which has been seen as a form of imagination, namely propositional imagination (see, e.g., Gaut 2003). Second, treatments of entertainment have aligned it with imagination. Uriah Kriegel has recently offered a detailed analysis of entertaining which casts it as having a distinctive phenomenology (irreducible to other kinds of phenomenology), and he argues that such a phenomenology and imaginative phenomenology “belong together” (Kriegel 2015, 195). Yet, independently of whether it coincides with entertaining (for sceptical views see, e.g., Doggett & Egan 2007; Kind 2016b; Arcangeli 2018c), imagination fulfls (pace Matravers) his rather undemanding requirement: if any attitude is involved at the frst stage of our engagement with narratives it should be a neutral or non-committal attitude, which is separate from, but can mingle with and lead to beliefs. Therefore, there are no clear reasons to prefer the notion of mental models over that of imagination. Replacing the latter with the former leaves unexplained what happens at the personal level, that is in what attitude a reader (or other consumer of a narrative) fnds herself. Given its non-committal nature, imagination is likely to be this attitude. Moreover, the employment of imagination is compatible with mental models being involved as the format through which the imagined content is conveyed. A plausible alternative description of Matravers’ frst stage thus sees the recruitment of both imagination and mental models. By contrast there are good reasons for granting to the imagination a key role in our engagement with narratives (maybe independently from whether fctional or not). But to see this we need to go beyond the narrow notion of imagination Matravers refers to. He focuses only on belief-like imagination, but arguably this is only one form of the imagination. Matravers might reply that he borrows the notion of belief-like imagination from his opponents (i.e., the proponents of the consensus view). Although most of the debate on fction has focused on belief-like imagination, it has been pointed out that other forms of the imagination are involved in fction (see, e.g., Goldman 2006b; Currie 2014).19 These views draw on a heterogeneous understanding of imagination that pictures it as a family comprising different genera and species. A promising way to account for this complex notion of imagination is in terms of mental simulation (Currie & Ravenscroft 2002; Goldman 2006a, 2006b; Arcangeli 2018b, 2018c). Imagination is thus defned as a complex mental activity that simulates other kinds of mental states

194

Margherita Arcangeli

(e.g., belief, perception); and imaginings are mental states bearing phenomenological and/or functional similarities with their “counterparts” (i.e., the mental states they simulate). These phenomenological and functional similarities are not captured by mental models, which by encoding spatial confgurations (and probably other manifest properties such as shapes and colours) are mental “simulations of […] external or physical states of affairs” (Goldman 2006a, 51, fn 10), rather than of mental states. Imagination and mental models capture two different senses of mental simulation: while the former involves mental mental simulation, the latter involves objectual mental simulation (Arcangeli 2018b, see also Goldman 2006a; Zeimbekis 2011). Objectual mental simulation might encode information concerning the perspective from which the relevant situation is represented, insofar as such perspectival information can be reduced to spatial confgurations or information about manifest properties. Mental mental simulation, however, specifes the type of self “occupying” the given perspective (e.g., the imaginer’s actual self, the self of a friend we imagine being, or even a sort of virtual self – a placeholder for any subject who would have the same experience). In other words, it conveys self-relative information that cannot be explained only in terms of spatial or otherwise perspectival information. This is enough to show that mental mental simulation cannot be reduced to objectual mental simulation. When we engage with a narrative, we are projected into the world described by it and feel as if we “live” in that world, perceiving or believing from that perspective. This aspect can be easily explained using the notion of imagination, conceived as a simulation of mental states (and not restricted to the simulation of belief). An analysis of the interactions with fctions, or narratives in general, based only on the notion of mental models would be unable to account for this mental involvement in narratives. The same holds for narratives presenting thought experiments, which can be seen as “props” inducing the receiver to re-perform, by using different forms of imagination, the thought experiment at issue and thus grasp its conclusion. Imagination is the cognitive ability through which the thought experimenter has access to a scenario that is not directly present to her senses, and enables her to perceive or believe from this new perspective. 20 Thought experimentation is very often described (even by Miščević and Nersessian themselves) in a language that evokes experiential aspects (on this point see Meynell 2014). The notion of imagination is better adapted than the notion of mental model to grasp and simulate these phenomenological aspects. Therefore, an imagination-based approach to thought experimentation is more promising than the model-based approach: like the latter it is a simulative account – it does

Narratives and Thought Experiments

195

justice to the narrative dimension of thought experiments, but it has considerable explanatory advantages (Arcangeli 2010, 2018b). An interesting issue is whether a simulative imagination-based account of thought experiments might help clarify their epistemological valence. The relevant question would then be how we can move from imaginings to knowledge. The simulationist view has the resources to gesture towards some paths to follow. We have seen that via our imagination we can simulate different types of mental state. It has been suggested that some imaginings simulate states of (actual or potential) knowledge. The idea is that imagination can simulate complex mental states which are conducive to knowledge – e.g., “not only the judgement and the experience, but at the same time the epistemic relation between them that converts the former into knowledge” (Dokic 2008, 105). This seems to be what is going on when we engage with thought experiments – and possibly when we engage with other narratives too.

9.5 Conclusion What goes on in the mind of a subject engaged with a fction? A common answer is that fction calls for imagination. In this contribution I have considered a recent critique against this thesis, which holds that mental models, rather than imagination, are the key underpinnings of our engagement with narratives tout court (whether fctional or not). I have pointed out that the analysis of the narrative aspect of thought experimentation can be offered in support of such a critique. Indeed, the fact that thought experiments are presented through narratives has motivated interest on the cognitive processes they require, and mental models have been proposed to ft this role. I have argued that these views are based on a mischaracterisation of imagination. Once we appreciate that imagination is a psychological attitude, which mentally simulates mental states (including states other than beliefs), we see that mental models cannot play a similarly adequate role, not only in thought experiments, but in all kinds of narratives, whether fctional or not. 21

Notes 1 This contribution is partially based on ideas previously published in French in Arcangeli (2018a). 2 Besides the cognitive question of what goes on in the mind of a subject engaged with a fction, one might wonder what makes a proposition (or a work) fctional or not. Such classifcatory question can be treated independently from the cognitive one, although, as Matravers underlines, the consensus view seems to consider them together by suggesting that a proposition is fctional if there is an invitation to imagine it (a fctional work is a compound of fctional propositions). Here I will be concerned with the cognitive question only.

196

Margherita Arcangeli

3 The notion of “propositional imagination” is a slippery one, since it is not clear what is supposed to pick out. Not all propositional imaginings seem to be belief-like, they might be desire-like (Currie & Ravenscroft 2002; Doggett & Egan 2007), hope-like (Goldman 2006a), and even perception-like, if one accepts that the imaginative homologue of epistemic perception (Dretske 1969) exists. 4 Not all authors have endorsed a strong version of simulationism, according to which the very same sub-personal mechanisms of belief are recruited “off-line” by the imagination. Still the gist of the simulationist view, which postulates strong similarities (along with important differences) between imagination and belief, is a shared point in the literature (see, e.g., Schroeder & Matheson 2006; Arcangeli 2018c). 5 For dissenting voices, see Langland-Hassan (2012) and Schellenberg (2013). 6 The projection should be interpreted in a loose way without any commitment to the idea that the imaginer’s self is always part of what is imagined. The experiential perspective involved by the projection can remain “virtual” (Dokic & Arcangeli 2015). In Section 9.4 I will consider the idea that fction can involve forms of the imagination other than belief-like imagination (e.g., perception-like imagination). 7 Matravers holds that the relevant distinction is between “confrontations” and “representations”, rather than between fction and non-fction. Stock (2017, 167) rightly observes that these are not necessarily rival distinctions and the latter can be a meaningful contrast within the “representations” category. 8 See Stock (2017) for a critical discussion of Matravers’ view and a defence of the consensus view. Although I share most of the worries raised by Stock, I can remain neutral here, since my main focus is on imagination as underpinning our interactions with narratives, whether fctional or not. 9 For an introduction presenting examples of thought experiments and the main issues in the debate see Arcangeli (2017). 10 In what follows I will be mainly concerned with literary narratives, but I am not ruling out other types of narratives (e.g., oral ones) through which thought experiments can be conveyed. 11 Quite often the narrative dimension of thought experimentation has been highlighted in order to downplay its value (see, e.g., Norton 1991; Sorensen 1992; Hacking 1993). 12 I am here referring to the following thought experiments: Galilei’s on falling bodies, Theseus’ ship (used since antiquity, see, for example, Hobbes 1655 II, Ch. 11, §7), the cow in the feld (Cohen 1999), Wittgenstein’s beetle (1953, §293), Twin Earth (Putnam 1973 and 1975), Maxwell’s demon (1871), philosophical zombies (Chalmers 1996). 13 Stock, for instance, writes: “In another familiar case, a non-fctional philosophy work can contain a fction in the form of a thought experiment” (Stock 2017, 159). 14 Several analyses of thought experiments dealing with aspects other than the narrative one (e.g., the epistemological aspect) have stressed the prominent role played by imagination in thought experimentation (see, e.g., Mach 1896; Gendler 2004; De Mey 2006; McAllister 2013; Stuart 2017 and 2019; Salis & Frigg 2020). 15 There are two types of thought experimenter: (i) the creator or designer of the thought experiment (e.g., Einstein), (ii) the receiver of it. Both the creator and the receiver of a thought experiment can be said to “perform” it, but it might be that these are two different senses of performance involving

Narratives and Thought Experiments

16

17

18 19

20 21

197

different cognitive processes. It is beyond the scope of the present chapter to go into such details. Given the complexity of the creative dimension of thought experiments and how it relates to the performance dimension (see Arcangeli 2018b; Stuart 2020), here by “thought experimenter” I will simply mean (ii). This idea has been, more or less explicitly, put forward by other authors (see, for example, Mach 1896; Gendler 2004; Chandrasekharan et al. 2013), but not necessarily by invoking the notion of mental model. For a more in-depth analysis of Miščević’s and Nersessian’s views, as well as of other modelbased approaches (e.g., the one proposed by Rachel Cooper, who criticises Miščević and Nersessian), see Häggqvist (1996) and Arcangeli (2010 and 2017). I would use the term “format”, because “vehicle” is ambiguous between the type of representation (i.e., the format) and what concretely realises mental contents (i.e., the medium) – which is likely to include (or to coincide with) brain states or processes (see Crane 1995). The same can be said about mental imagery (Arcangeli 2020). Exploring how the latter relates to the notion of mental models unfortunately goes beyond present purposes. Interestingly, the debate on thought experiments seems to have privileged a perception-like form of imagination (see Arcangeli 2010 and 2017; Salis & Frigg 2020). However, some authors have highlighted the role of nonperceptual forms, such as belief-like imagination, supposition, or conceiving (see, e.g., Weinberg & Meskin 2006; McAllister 2013; Balcerak Jackson 2016). Imagination allows us to clarify the experimental character of thought experimentation: thought experimenters are like observers, like ordinary experimenters (Arcangeli 2018b). I am very grateful to Christopher Badura, Dimitri Coelho Mollo, Jérôme Dokic, Amy Kind, Eric Peterson, and Mike Stuart for critical and helpful comments on earlier versions of this chapter. I would also like to thank organisers and participants of the one-day workshop on thought experiments in Nancy (February 2019) for their feedback. This research has been funded by the SublimAE Project (ANR-18-CE27-0023-01), with the further support of the ANR-17-EURE-0017 FrontCog and the ANR-10IDEX-0001-02 PSL.

References Arcangeli, M. 2010. “Imagination in Thought Experimentation: Sketching a Cognitive Approach to Thought Experiments”. In Model-Based Reasoning in Science and Technology, L. Magnani, W. Carnielli, C. Pizzi (eds.), 571–87 (Springer, Berlin). Arcangeli, M. 2017. “Thought Experiments in Model-Based Reasoning”. In Springer Handbook of Model-Based Science, L. Magnani, T. Bortolotti (eds.), 465–95 (Springer, Berlin). Arcangeli, M. 2018a. “L’expérience de pensée comme récit. Le rôle crucial de l’imagination”. In Perspectives philosophiques sur les fctions, C. Bouriau, G. Schupper (eds.), 111–29 (Kimé, Paris). Arcangeli, M. 2018b. “The Hidden Links between Real, Thought and Numerical Experiments”. Croatian Journal of Philosophy XVIII/52, 3–22.

198

Margherita Arcangeli

Arcangeli, M. 2018c. Supposition and the Imaginative Realm. A Philosophical Inquiry (Routledge, London & New York). Arcangeli, M. 2020. “The Two Faces of Mental Imagery”. Philosophy and Phenomenological Research 101(2), 304–322. doi:10.1111/phpr.12589. Balcerak Jackson, M. 2016. “On the Epistemic Value of Imagining, Supposing, and Conceiving”. In Knowledge Through Imagination, A. Kind, P. Kung (eds.), 41–60 (OUP, Oxford). Bishop, M. 1999. “Why Thought Experiments Are Not Arguments”. Philosophy of Science 66, 534–41. Carroll, N. 2002. “The Wheel of Virtue: Art, Literature, and Moral Knowledge”. Journal of Aesthetics and Art Criticism 60(1), 3–26. Chalmers, D. 1996. The Conscious Mind (OUP, Oxford). Chandrasekharan, S., Nersessian, N.J., Subramanian, V. 2013. “Computational Modeling: Is This the End of Thought Experiments in Science?”. In Thought Experiments in Philosophy, Science, and the Arts, M. Frappier, L. Meynell, J.R. Brown (eds.), 239–60 (Routledge, London & New York). Cohen, M. 1999. 101 Philosophy Problems (Routledge, London & New York). Cooper, R. 2005. “Thought Experiments”. Metaphilosophy 36, 328–47. Crane, T. 1995. The Mechanical Mind (Routledge, London & New York). Currie, G. 1990. The Nature of Fiction (CUP, Cambridge). Currie, G., Ravenscroft, I. 2002. Recreative Minds: Imagination in Philosophy and Psychology (OUP, Oxford). Currie, G. 2014. “Standing in the Last Ditch”. Journal of Aesthetics and Art Criticism 72(4), 351–63. Davenport, E.A. 1983. “Literature as Thought Experiment (On Aiding and Abetting the Muse)”. Philosophy of the Social Sciences 13(3), 279–306. Davies, D. 2007a. Aesthetics and Literature (Continuum, London). Davies, D. 2007b. “Thought Experiments and Fictional Narratives”. Croatian Journal of Philosophy VII, 29–45. Davies, D. 2010. “Learning Through Fictional Narratives in Art and Science”. In Beyond Mimesis and Convention, R. Frigg, M.C. Hunter (eds.), 51–70 (Kluwer Academic Publishers, Dordrecht). Davies, D. 2013. “Can Philosophical Thought Experiment Be ‘Screened’?” In Thought Experiments in Philosophy, Science, and the Arts, M. Frappier, L. Meynell, J.R. Brown (eds.), 223–38 (Routledge, London & New York). De Mey, T. 2006. “Imagination’s Grip on Science”. Metaphilosophy 37, 222–39. Doggett, T., Egan, A. 2007. “Wanting Things You Don’t Want: The Case for an Imaginative Analogue of Desire”. Philosophers’ Imprint 7(9), 1–17. Dokic, J. 2008. “Epistemic Perspectives on Imagination”. Revue Internationale de Philosophie 243(1), 99–118. Dokic, J., Arcangeli, M. 2015. “The Heterogeneity of Experiential Imagination”. In Open MIND: 11(T). Frankfurt am Main: MIND Group, K. Metzinger, J.M. Wind (eds.). doi:10.15502/9783958570085; http://openmind.net/papers/the-heterogeneity-of-experiential-imagination [published in 2016 in Open MIND. Philosophy and the Mind Sciences in the 21st Century, K. Metzinger, J.M. Wind (eds.), 431–50 (MIT Press).] Dretske, F. 1969. Seeing and Knowing (University of Chicago Press, Chicago). Einstein, A., Infeld, L. 1938. The Evolution of Physics. The Growth of Ideas from Early Concepts to Relativity and Quanta (Simon & Schuster, New York).

Narratives and Thought Experiments

199

Einstein, A. 1949. “Autobiographical Notes”. In Albert Einstein: Philosopher-Scientist, P.A. Schilpp (ed.) (Open Court, La Salle). Elgin, C.Z. 2007. “The Laboratory of the Mind”. In A Sense of the World: Essays on Fiction, Narrative, and Knowledge, W. Huerner, J. Gibson, L. Pocci (eds.), 43–54 (Routledge, London). Elgin, C.Z. 2014. “Fiction as Thought Experiment”. Perspectives on Science 22, 221–41. Gaetens, M. 2009. “The art and philosophy of George Eliot”. Philosophy and Literature 33(1), 73–90. Galilei, G. 1638. Discorsi e dimostrazioni matematiche intorno a due nuove scienze (Louis Elsevier, Leida). Gaut, B. 2003. “Imagination and Creativity”. In The Creation of Art, B. Gaut, P. Livingston (eds.), 148–73 (CUP, Cambridge). Gendler, T. 2004. “Thought Experiments Rethought – And Reperceived”. Philosophy of Science 71, 1152–64. Gettier, E. 1963. “Is justifed True Belief Knowledge?”. Analysis 23, 121–3. Goldman, A. 2006a. Simulating Minds: The Philosophy, Psychology, and Neuroscience of Mindreading (OUP, Oxford). Goldman, A. 2006b. “Imagination and Simulation in Audience Responses to Fiction”. In The Architecture of the Imagination: New Essays on Pretence, Possibility, and Fiction, S. Nichols (ed.), 41–56 (OUP, Oxford). Gooding, D. 1993. “What is Experimental About Thought Experiments?”. In PSA 1992, D. Hull, M. Forbes, K. Okruhlik (eds.), 280–90 (Philosophy of Science Association, East Lansing). Hacking, I. 1993. “Do Thought Experiments have a Life of Their Own? Comments on James Brown, Nancy Nersessian and David Gooding”. In PSA 1992, D. Hull, M. Forbes, K. Okruhlik (eds.), 302–8 (Philosophy of Science Association, East Lansing). Häggqvist, S. 1996. Thought Experiments in Philosophy (Almqvist & Wiksel, Stockholm). Hobbes, T. 1655. Elementorum philosophiae sectio prima De corpore (Andrew Crook, London). Ichikawa, J., Jarvis, B. 2009. “Thought-experiment Intuitions and Truth in Fiction”. Philosophical Studies 142, 221–46. Jackson, F. 1982. “Epiphenomenal Qualia”. The Philosophical Quarterly 32, 27–36. Johnson-Laird, P.N. 1983. Mental Models: Toward a Cognitive Science of Language, Inference and Consciousness (HUP, Cambridge). Johnson-Laird, P.N. 2004. “The History of Mental Models”. In Psychology of Reasoning: Theoretical and Historical Perspectives, K. Manktelow, M.C. Chung (eds.), 179–212, (Psychology Press, New York). Kind, A. 2016a. “Introduction: Exploring Imagination”. In The Routledge Handbook of Philosophy of Imagination, A. Kind (ed.), 1–11 (Routledge, London & New York). Kind, A. 2016b. “Imagining Under Constraints”. In Knowledge Through Imagination, A. Kind, P. Kung (eds.), 145–59 (OUP, Oxford). Kriegel, U. 2015. The Varieties of Consciousness (OUP, Oxford). Langland-Hassan, P. 2012. “Pretense, Imagination, and Belief: The Single Attitude Theory”. Philosophical Studies 159(2), 155–79.

200 Margherita Arcangeli Mach, E. 1896. “Über Gedankenexperimente”. Zeitschrift für den physikalischen und chemischen Unterricht 10, 1–5. Trans. W.O. Price, S. Krimsky: “On thought experiments”. Philosophical Forum 4(3), 446–57 (1973). Matravers, D. 2014. Fiction and Narrative (OUP, Oxford). Maxwell, J.C. 1871. Theory of Heat (Longman, London). McAllister, J. 1996. “The Evidential Signifcance of thought Experiment in Science”. Studies in History and Philosophy of Science Part A 27, 233–50. McAllister, J. 2013. “Thought Experiment and the Exercise of Imagination in Science”. In Thought Experiments in Philosophy, Science, and the Arts, M. Frappier, L. Meynell, J.R. Brown (eds.), 11–29 (Routledge, London & New York). McComb, G. 2013. “Thought Experiment, Defnition, and Literary Fiction”. In Thought Experiments in Philosophy, Science, and the Arts, M. Frappier, L. Meynell, J.R. Brown (eds.), 207–22 (Routledge, London & New York). McGinn, C. 2004. Mindsight: Image, Dream, Meaning (HUP, Cambridge). Meynell, L. 2014. “Imagination and Insight: A New Account of the Content of Thought Experiments”. Synthese 191(17), 4149–68. Miščević, N. 1992. “Mental Models and Thought Experiments”. International Studies in the Philosophy of Science 6, 215–26. Miščević, N. 2007. “Modelling Intuitions and Thought Experiments”. Croatian Journal of Philosophy VII, 181–214. Nersessian, N.J. 1993. “In the Theoretician’s Laboratory: Thought Experimenting as Mental Modelling”. In PSA 1992, D. Hull, M. Forbes, K. Okruhlik (eds.), 291–301 (Philosophy of Science Association, East Lansing). Nersessian, N.J. 2007. “Thought Experiments as Mental Modelling: Empiricism without Logic”. Croatian Journal of Philosophy VII, 125–61. Newton, I. 1687. Philosophiae Naturalis Principia Mathematica (Joseph Streater, London). Norton, J. 1991. “Thought Experiments in Einstein’s Work”. In Thought Experiments in Science and Philosophy, T. Horowitz, G. Massey (eds.), 129–48 (Rowman & Littlefeld, Lanham). Palmieri, P. 2003. “Mental Models in Galileo’s Early Mathematization of Nature”. Studies in History and Philosophy of Science 34, 229–64. Peterson, E. 2018. “Thought Experiments as a Kind of Genre”. In The Junkyard. A Scholarly Blog Devoted to the Study of Imagination, May 23, 2018, https://junkyardofthemind.com/blog/2018/5/21/thought-experimentsas-a-kind-of-genre Plato. 2008. The Republic, trans. Benjamin Jowett (Cosimo, New York). Putnam, H. 1973. “Meaning and Reference”. Journal of Philosophy 70, 699–711. Putnam, H. 1975. “Meaning of ‘Meaning’”. Minnesota Studies in the Philosophy of Science 7, 131–93. Putnam, H. 1981. Reason, Truth and History (CUP, Cambridge). Salis, F., Frigg, R. 2020. “Capturing the Scientifc Imagination”. In The Scientifc Imagination, P. Godfrey-Smith, A. Levy (eds.), 17–50 (OUP, Oxford). Sartre, J.-P. 1940. L’imaginaire (Gallimard, Paris – trans. J.M. Webber, The Imaginary, Routledge, London 2004). Schellenberg, S. 2013. “Belief and Desire in Imagination and Immersion”. Journal of Philosophy 110(9), 497–517.

Narratives and Thought Experiments

201

Schrödinger, E. 1935. “Die gegenwärtige Situation in der Quantenmechanik”. Naturwissenschaften 23, 807–12. Schroeder, T., C. Matheson 2006. “Imagination and Emotion”. In The Architecture of the Imagination: New Essays on Pretence, Possibility, and Fiction, S. Nichols (ed.), 19–39 (OUP, Oxford). Sorensen, R. 1992. Thought Experiments (OUP, Oxford). Stock, K. 2017. Only Imagine: Fiction, Interpretation and Imagination (OUP, Oxford). Strohminger, M. 2021. “Two Ways of Imagining Galileo’s Experiment”. In Epistemic uses of imagination, C. Badura, A. Kind (eds.), (Routledge, London & New York). Stuart, M.T. 2017. “Imagination: A Sine Qua Non of Science”. Croatian Journal of Philosophy 17/49, 9–32. Stuart, M.T. 2019. “Towards a Dual Process Epistemology of Imagination”. Synthese. doi:10.1007/s11229-019-02116-w Stuart, M.T. 2020. “The Material Theory of Induction and the Epistemology of thought Experiments”. Studies in History and Philosophy of Science. doi:10.1016/j.shpsa.2020.03.005 Swirski, P. 2007. Of Literature and Knowledge: Explorations in Narrative thought Experiments, Evolution and Game Theory (Routledge, London & New York). Walton, K. 1990. Mimesis as Make-believe: On the Foundations of the Representational Arts (HUP, Cambridge). Weinberg, J., Meskin, A. 2006. “Puzzling over the Imagination: Philosophical Problems, Architectural Solutions”. In The Architecture of the Imagination: New Essays on Pretence, Possibility, and Fiction, S. Nichols (ed.), 175–202 (OUP, Oxford). Weinberg, J. 2008. “Confguring the Cognitive Imagination”. In New Waves in Aesthetics, K. Stock, K. Thomson-Jones (eds.), 203–23 (Palgrave Macmillan, Houndmills and Basingstoke). Wittgenstein, L. 1953. Philosophische Untersuchungen (Blackwell, Oxford). Zeimbekis, J. 2011. “Thought Experiments and Mental Simulations”. In Thought Experiments in Methodological and Historical Contexts, K. Ierodiakonou, S. Roux (eds.), 193–215 (Brill, Leiden and Boston).

10 Two Ways of Imagining Galileo’s Experiment Margot Strohminger

10.1 Introduction Galileo’s Dialogues Concerning Two New Sciences presents a series of considerations against the Aristotelian theory of motion then dominant in his day. One of them is a thought experiment, which one character in the dialogue, ‘Salviati’, presents to his interlocuters ‘Simplicio’ and ‘Sagredo’. It’s worth quoting the relevant part of the exchange in full: Salviati. But without other experiences, by a short and conclusive demonstration, we can prove clearly that it is not true that a heavier moveable is moved more swiftly than another, less heavy, these being of the same material, and in a word, those of which Aristotle speaks. Tell me, Simplicio, whether you assume that for every heavy falling body there is a speed determined by nature such that this cannot be increased or diminished except by using force or opposing some impediment to it. Simplicio. There can be no doubt that a given moveable in a given medium has an established speed determined by nature, which cannot be increased except by conferring on it some new impetus, nor diminished save by some impediment that retards it. Salviati. Then if we had two moveables whose natural speeds were unequal, it is evident that were we to connect the slower to the faster, the latter would be partly retarded by the slower, and this would be partly speeded up by the faster […].But if this is so, and if it is also true that a large stone is moved with eight degrees of speed, for example, and a smaller one with four [degrees], then joining both together, their composite will be moved with a speed less than eight degrees. But the two stones joined together make a larger stone than that frst one which was moved with eight degrees of speed; therefore this greater stone is moved less swiftly than the lesser one. But this is contrary to your assumption. So you see how, from the supposition that the heavier body is moved more swiftly than the less heavy, I conclude that the heavier moves less swiftly. (Galilei 1989, 66–7)1

Two Ways of Imagining Galileo’s Experiment

203

The thought experiment seems designed to challenge the Aristotelian principle that bodies of the same material fall at speeds directly proportional to their weight. Commentators in the philosophy of science have asked whether and how thinking about Galileo’s fctional experiment might accomplish this feat. Commentary has focused primarily on how subjects are typically justifed in believing the thought experiment’s conclusion (insofar as they are justifed at all): do they simply perform an inference from beliefs they held already? Or is more required: for example, do subjects need to experience an ‘intuition’, ‘see’ a mental image, or construct a ‘mental model’?2 An overlooked issue in the debate concerns the conception of body in play in the Aristotelian principle that Galileo is challenging. Contemporary portrayals of the thought experiment standardly assume an extremely liberal conception of bodies on the Aristotelian’s behalf—so liberal that the composite formed as a result of conjoining the two stones in the thought experiment counts as a body, as do the stones themselves (whether they are conjoined or not). The infuential portrayals by James R. Brown (1991) and Tamar Szabó Gendler (1998) are two clear examples. While the issue of how to conceive of bodies in the principle may seem arcane, in fact it ends up placing signifcant constraints on the epistemology of Galileo’s thought experiment. In this chapter I frst argue that it is implausible to interpret the Aristotelian principle targeted by Galileo in the liberal way that is standardly assumed. Instead, the principle under attack conceives of bodies in a less permissive way, so that only the object formed as a result of conjoining the two stones is a body in the thought-experimental scenario. Second I consider how the thought experiment might pose a challenge to the Aristotelian principle, so understood. I will argue that the Aristotelian physicist is expected to imagine the outcome of Galileo’s experiment in two different ways and that in only one of them does she rely on her belief in the Aristotelian principle. In the other she uses her imagination to arrive at a judgment about what would happen, which resembles paradigmatic cases of ‘intuitive’ judgments. In closing I suggest distinguishing between the two ways of imagining Galileo’s experiment by appealing to the distinction between ‘intuitive’ and ‘refective’ thinking, which has become increasingly familiar in recent years with the advent of dual-process theories in psychology. Moreover, while some earlier commentators—in fact Brown (1991) and Gendler (1998) themselves— have suggested a role for (something like) intuitive thinking in explaining how thought experimenters reach Galileo’s intended conclusion, its role is importantly different in the explanation on offer here.

10.2 A Standard Interpretation Galileo uses his thought experiment to challenge the principle that bodies of the same material fall in media (as opposed to a vacuum) at

204  Margot Strohminger speeds directly proportional to their weight.3 Thus, given two bodies of different weights made of the same material dropped from a height, the heavier will fall faster than the lighter. As Salviati puts it, according to the principle of interest, ‘a heavier moveable is moved more swiftly than another, less heavy, these being of the same material, and in a word, those of which Aristotle speaks’ (Galilei 1989, 66).4 Even with this much settled, it is open to debate how to apply the Aristotelian principle to a case such as the one Salviati describes. In the thought-experimental scenario, someone connects two stones and drops them from a height. Is the composite formed as a result of connecting the stones a body (or a ‘moveable’)? What about the stones from which the composite is formed? Are each of them bodies after they are connected? And what about before? It is standardly assumed that Galileo is challenging a version of the Aristotelian principle which conceives of bodies in an extremely liberal way. In particular, it conceives of bodies liberally enough so that the composite made from joining the two stones together is a body, and so too are the stones out of which the composite is formed, both before and after they are joined together. As a result, it is assumed that when we think through what would happen if someone were to conduct Galileo’s experiment given Aristotle’s principle the following reasoning becomes available. First, the composite would be heavier than the heavy stone is on its own. Since they are made of the same material and are both bodies, the composite would fall faster than the heavier stone does on its own. On the other hand, the composite is made up of two bodies of different weights and the same material. So its light stone-component would fall slower than the heavier stone-component. As a result, the former would slow down the latter, and the composite would fall slower than the heavy stone does on its own. Both Brown (1991) and Gendler (1998) make this standard assumption at the outset of their discussions of how Galileo’s thought experiment works. The assumption remains standard.5 Here is how Brown presents the case in the opening paragraphs of The Laboratory of the Mind: It begins by noting Aristotle’s view that heavier bodies fall faster than light ones (H > L). We are then asked to imagine that a heavy cannon ball is attached to a light musket ball. What would happen if they were released together? Reasoning in the Aristotelian manner leads to an absurd conclusion. First, the light ball will slow up the heavy one (acting as a kind of drag), so the speed of the combined system would be slower than the speed of the heavy ball falling alone (H > H+L). On the other hand, the combined system is heavier than the heavy ball alone, so it

Two Ways of Imagining Galileo’s Experiment

205

should fall faster (H+L > H). We now have the absurd consequence that the heavy ball is both faster and slower than the even heavier combined system. Thus, the Aristotelian theory of falling bodies is destroyed. (Brown 1991: 1–2, emphases mine)6 And Gendler presents the case as follows: The famous thought experiment, rephrased slightly, is the following. Imagine that a heavy and a light body are strapped together and dropped from a signifcant height. What would the Aristotelian expect to be the natural speed of their combination? On the one hand, the lighter body should slow down the heavier one while the heavier body speeds up the lighter one, so their combination should fall with a speed that lies between the natural speeds of its components. (That is, if the heavy body falls at a rate of 8, and the light body at a rate of 4, then their combination should fall at a rate between the two (cf. Galilei [1638/1989], p. 107).) On the other hand, since the weight of the two bodies combined is greater than the weight of the heavy body alone, their combination should fall with a natural speed greater than that of the heavy body. (That is, if the heavy body falls at a rate of 8 and the light body with a rate of 4, their combination should fall at a rate greater than 8.) But then the combined body is predicted to fall both more quickly, and more slowly, than the heavy body alone (cf. Galilei [1638/1989, pp. 107–8). The way out of this paradox is to assume that the natural speed with which a body falls is independent of its weight: ‘both great and small bodies  … are moved with like speeds’ (Galilei [1638/1989], p. 109). (Gendler 1998: 403, emphases mine) Both presentations of the case depart in two ways from its original presentation that I will assume are irrelevant. In particular, Brown assumes that metal balls are connected rather than stones (and Gendler doesn’t specify their material at all). Moreover, Brown and Gendler leave unstated the assumption that every body in a given medium has a speed determined by nature, which cannot be changed unless a force acts upon it.7 What is important for my purposes is that Brown and Gendler assume that the Aristotelian of interest to Galileo counts the composite as a body, as well as each of the components making up the composite. This is clear from the fact that the two authors apply the Aristotelian principle to each of the components making up the composite and not just to the composite. In the next section we will present a problem with reading the thought experiment in this way and then propose an interpretation that can avoid the problem.

206

Margot Strohminger

10.3 A Better Interpretation If we attribute to Galileo’s Aristotelian opponent the liberal conception of bodies standardly assumed, then it is remarkably easy to challenge the principle using cases—in fact thinking through any case in which an object falls in a medium will suffce. It then becomes mysterious as to why Galileo relies so single-mindedly on the case involving two stones. Consider on what principled basis one might classify the two stones making up the composite in the thought experiment as well as the composite itself as bodies, as the standard interpretation requires of the Aristotelian. A straightforward option is the following. Any familiar object is a body. So too are its parts and composites made from joining them together. As a result, the composite and the stones making it up all qualify as bodies. Let’s suppose that Galileo’s Aristotelian opponent adopts this principled conception of bodies. Then the following case will be just as effective against the principle that bodies fall at speeds directly proportional to their weight. Suppose Plato were to fall from a tower. What would happen? In particular consider the speed at which Plato’s entire body would fall and how it compares to the speed at which his left foot would fall on its own. On the one hand, Plato’s body would fall faster than his foot since his body is heavier than his foot. On the other hand, Plato’s body would fall slower than his detached foot: his foot will slow down the rest of his body since it is lighter than the rest of his body. This case is silly (and maybe even gruesome), but it illustrates an important point: thinking through what would happen in any case where an object falls in a medium poses a challenge to the Aristotelian given the above conception of bodies. All that we need to do is apply the Aristotelian principle to the fall of a familiar object (made of the same material throughout8) twice over, where we treat the object frst as a single body and then as a pair of (conjoined) bodies. It is implausible that Galileo is criticizing a physicist with this conception of bodies. For if Galileo was he could easily have pointed out that there is nothing special about the thought-experimental scenario he chose. Of course thought experiments generalize to a range of cases beyond the original case described (hence terms like ‘Gettier cases’ and ‘Frankfurt-style cases’9). But even so they do not generalize in the way Galileo’s case does given the standard interpretation. A justifed true belief in the proposition that p can fail to give rise to a Gettier case (namely, when the subject knows that p), and likewise an agent who cannot do otherwise can fail to give rise to a Frankfurt-style case (when the agent is not morally responsible for their action). Given the standard interpretation, the situation here would be like one in which it turns out that whenever one has a justifed true belief, it isn’t a case of knowledge. This is not a knock-down consideration of course. Another

Two Ways of Imagining Galileo’s Experiment

207

consideration is that an alternative interpretation is available, according to which the thought experiment—but not the silly case I sketched—is effective against the principle. Let’s consider that alternative interpretation now. To avoid the problem just raised, we can interpret the conception of bodies accompanying the Aristotelian physics being challenged by Galileo in a less permissive way. On a ‘strict’ interpretation, if something is a body, then the only things it is connected to are its parts. In a slogan: bodies are ‘disconnected’. Thus, when the principle is applied to the thought-experimental scenario, the stones are not bodies so long as they are connected to each other; only the composite is. Similarly Plato’s foot is not a body so long as it is connected to the rest of Plato; only Plato in his entirety is. When we conceive of bodies on the Aristotelian’s behalf in this way, the reasoning that led to a contradiction in both Galileo’s case and the silly case I introduced is no longer possible. Take Galileo’s case. According to the Aristotelian, the composite would fall faster than the heavy stone does on its own: the former is heavier than the latter. It would not fall slower than the heavy stone: the principle doesn’t apply to the heavy stone when it is a part of the composite. Similar remarks apply to a case in which Plato falls from a tower.10 Understood in this way, applying the Aristotelian principle to the case does not, pace Brown and Gendler, lead to the absurd conclusion that the connected stones would fall both faster and slower than the heavy stone on its own. While the strict interpretation addresses the problem facing the standard interpretation, it now faces another. Given the strict interpretation, it now looks as if the thought experiment does not manage to pose a challenge to the Aristotelian principle at all. In the next section I will argue that the Aristotelian is still expected to reach the absurd conclusion that the connected stones would fall both faster and slower than the heavy stone on its own and provide an explanation of how she is supposed to do this without applying her principle to the stones when they are connected. In addition, the explanation does not carry over to the silly case about Plato I introduced. Thus, the thought experiment—but not the silly case I introduced—can present a challenge to the principle. Before moving on, though, it is worth pausing to consider an objection to my argument that the thought experiment does not address an Aristotelian physicist with a liberal conception of bodies. The objector suggests that maybe Galileo really did intend his case to target Aristotelians with a liberal conception of bodies and that he would not have been bothered by its generalizing to every case in which an object falls in a medium (and perhaps he even recognized that it did); the only reason he chose his two-stones case (as opposed to one like my Plato case) is that it is easier to refer to the objects making up the composite.11 I don’t have a convincing refutation of this suggestion; indeed perhaps the suggestion

208

Margot Strohminger

is right (and historical evidence might turn up in its favor). I can only add that I think my interpretation is the more charitable interpretation of Galileo’s case (and it makes Galileo look more charitable in his interpretation of the Aristotelian principle). Even if the suggestion is right though, the alternative interpretation of the thought experiment offered in the remainder of this chapter is not without interest. It shows how Galileo’s thought experiment might also work against an Aristotelian physicist with a stricter (and I think more sophisticated and interesting) conception of bodies.

10.4 Imagining the Experiment without Importing Aristotle’s Principle As we have just seen, when we ask what would happen if someone were to conduct Galileo’s experiment, the strict reading of Aristotle’s principle delivers just one prediction: the composite would fall faster than does the heavy stone on its own. It does not also predict that the composite would fall slower than (or as fast as) the heavy stone on its own. What I will suggest is that the Aristotelian is still able to reach a verdict inconsistent with the verdict delivered by her theory. She reaches the inconsistent verdict by evaluating what would happen in another way, where she temporarily ignores her belief in the principle. Structurally the situation will turn out to be similar to the following mundane case. Consider someone who believes that eating insects is always undesirable. We then ask her to consider a situation in which the only source of sustenance available to her is some crickets; we ask her whether eating insects would be undesirable even then. When she relies on her belief in the principle, she can arrive at the belief that it would still be undesirable in the hypothetical scenario to eat the crickets. But she has another way of evaluating whether it would be desirable to eat the crickets, where she temporarily ignores her belief in the principle. She imagines what she would desire in that situation, and when she does, she can judge that she would want to eat the crickets, her anti-insect principle notwithstanding. More generally, many effective thought experiments produce a similar tension when they try to overturn one of our beliefs in some principle. The situation is just as familiar in the history of philosophy as it is in the history of science. Someone might describe a case in which lying seems right to convince someone who believes that lying is always wrong, or a Gettier case to convince someone that knowledge is different from justifed true belief. Ordinarily thought experimenters seem to somehow exploit their imagination in order to reach competing verdicts about what would happen in a counterfactual scenario such as the one Galileo describes. For one thing, plausibly part of what makes an event a thought experiment

Two Ways of Imagining Galileo’s Experiment

209

is that the agent performing it must exercise her imagination in order to appreciate its intended conclusion.12 Moreover, it is signifcant that the competing verdicts in this case are conditionals: FASTER. If someone were to conduct Galileo’s experiment, then the composite would fall faster than the heavy stone does on its own. SLOWER. If someone were to conduct Galileo’s experiment, then the composite would fall slower than the heavy stone does on its own. It is widely assumed that one of the most typical routes for evaluating conditionals involves the imagination.13 Following Williamson (2007, 2016, 2020), we might characterize the process of evaluating an arbitrary conditional in the imagination as follows.14 One supposes the antecedent of the conditional and then attempts to develop the supposition into the consequent of the conditional in the imagination. When one develops the supposition, one is permitted to ‘import’ or ‘hold fxed’ certain beliefs one has already: the contents of these imported beliefs are taken not only to describe one’s actual circumstances but also the imaginary circumstances. The same goes for information at the subject’s disposal, which is not stored in the form of beliefs. For example, consider how you might evaluate what will happen if a rainbow lorikeet were to fy in front of you right now. To construct the imaginary scenario, you can more or less hold fxed everything as it looks to you. To do this it is natural to think that you ‘import’ some of your visual experiences, and not simply your beliefs, into your imagination. Here’s another way of making the last point within a ‘simulationist’ framework, which conceives of exercising the imagination as the running of various mental capacities ‘offine’.15 The mental capacities running offine in the evaluation of conditionals are not just those used in updating or revising one’s beliefs; they extend also to the capacities underlying sensory perception.16 How one develops the supposition depends in part on the content of the supposition. It also depends on whether the conditional under consideration is indicative or counterfactual.17 When the subject is using her imagination to evaluate a counterfactual—as is the case here—we will call her supposition in the antecedent ‘counterfactual’. A subject is permitted to develop a counterfactual supposition in several ways. Crucial for our purposes is the fact that these ways of imaginatively developing a counterfactual supposition can differ to such an extent that they allow for the same subject to develop the same counterfactual supposition into inconsistent claims. That is, when a subject supposes that p, she may develop the supposition into q by following one route in her imagination, and she may develop the same supposition into not-q by following another. (Similar remarks apply to the imaginative process used to evaluate indicative conditionals.18)

210

Margot Strohminger

Let’s consider how this possibility might be realized in the case of Galileo’s imaginary experiment. When the Aristotelian counterfactually supposes that someone conducts the experiment, she may import her belief in the principle. When she does this, she flls in the details of the counterfactual scenario in such a way that the composite falls faster than the heavy stone does on its own; as a result, she affrms the FASTER conditional. But she is not required to import her belief in the principle. So consider instead the case where she does not import the principle. She flls in the details of the counterfactual scenario by relying on other beliefs she has. For example, suppose she believes that faster objects typically speed up slower objects when they are connected. This belief may be implicit at best. When asked what would happen if someone were to connect two stones, the heavier of which is stipulated to fall faster than the other when they are dropped separately—as in the thought experiment—she may import this implicit belief and then imagine the light stone slowing down the heavier stone and the composite thereby falling slower than the heavy stone does on its own. As a result, she comes to affrm the SLOWER conditional. Perhaps she might even develop the supposition into the claim that the composite would fall slower than the heavy stone does on its own without an implicit belief like the one just suggested. Instead, the subject visualizes what would happen if the experiment were to take place and when she does, she accesses visual information that is innate or accrued from past experiences. This visual information is not stored in her beliefs, but it still enables her to visualize the scenario in such a way that the heavy stone slows down the lighter stone in the thought experiment and she comes to affrm SLOWER.19 The same kind of possibility is arguably realized all the time. When someone who believes that eating insects is always undesirable is asked whether she would want to eat crickets were that the only source of sustenance available for days on end, she arguably can arrive at conficting verdicts in her imagination. She can arrive at the verdict that she would not want to eat the crickets by importing her belief that eating insects is always undesirable in order to imagine that she would not want to eat the crickets in this scenario. But she can also fll in the details of the imagined scenario in such a way that she ends up imagining that she would want to eat the crickets. How she flls in the details here is a diffcult question, but nevertheless it is clear that, to arrive at this second verdict, she must temporarily ignore her belief that eating insects is always undesirable. Now recall the silly case I introduced in the last section. Here the Aristotelian is asked to evaluate in the imagination what would happen if Plato’s body were to fall from a tower. It does not seem as if the Aristotelian will affrm an inconsistent pair of conditionals about it. The same explanation as before applies to how the Aristotelian will come to affrm the conditional analogous to FASTER:

Two Ways of Imagining Galileo’s Experiment

211

FASTER 2: If Plato were to fall from the tower, then he would fall at a speed faster than his body does in the absence of his foot. But what about the conditional corresponding to SLOWER? SLOWER 2: If Plato were to fall from the tower, then he would fall at a speed slower than his body does in the absence of his foot. Even if we suppose that the Aristotelian believes that typically, slower objects slow down faster objects when they are connected and she imports the belief into this counterfactual scenario, it is far less clear that relying on it will help her to imagine that Plato’s foot would slow down the rest of his body. When conjoined, Plato’s foot and his body minus the foot simply aren’t objects to which the implicit folk principle applies. More generally, it is far less natural or tempting for subjects—without the implicit belief—to develop this supposition in such a way that the lighter component (that is, Plato’s foot) would slow down the heavier component to which it is connected (the rest of Plato’s body).

10.5 Two Kinds of Imagination The two ways of using the imagination to evaluate what would happen in Galileo’s experiment might be distinguished by whether they are instances of ‘intuitive’ or ‘refective’ thinking. The distinction between intuitive and refective thinking has become familiar in recent years in psychology and increasingly in philosophy. 20 While these two kinds of thinking typically vary across several dimensions, here we will just highlight one, emphasized by Mercier and Sperber (2009) and Nagel (2012). When a subject makes an intuitive judgment, she is not aware of the process by which she reached it; hence, it feels immediate. The same cannot be said for refective judgments. We can use the intuitive-refective distinction to distinguish two ways of evaluating claims in the imagination. When a subject judges something as a result of her ‘intuitive’ imagination, she is not aware of the steps by which she reached her judgment (even though she may well be aware that she used her imagination somehow). By contrast she is aware of the steps by which she reached her judgment when she deploys her ‘refective’ imagination. (Of course the judgments fall on a spectrum since subjects can be more or less aware of the steps they followed in their imagination.) On the story just given about Galileo’s thought experiment, it is very natural to think that the SLOWER judgment is intuitive: the subject simply ‘sees’ that the light stone would slow down the heavy stone and the composite would fall slower than the heavy stone does on its own, without being able to rehearse her reasons for thinking so. She may, if pressed, perhaps cite her beliefs that slower objects typically slow down

212 Margot Strohminger faster objects when they are connected and that the light stone normally falls slower than the heavy stone when they are dropped separately (assuming she has these beliefs). 21 Part of the reason for thinking the judgment is reached in an intuitive manner is how the situation looks to us from the third-person perspective we are presently occupying. Even from this perspective, it is unclear if subjects draw on specifc beliefs like the two just suggested or indeed what relevant information the Aristotelian has to draw on. But somehow subjects apparently reach the SLOWER judgment. The SLOWER judgment stands in contrast to the Aristotelian’s judgment that FASTER is true: here it is clear that her reason for affrming FASTER is the principle that heavy bodies fall faster than lighter bodies of the same material. When asked why she affrmed FASTER, the Aristotelian can cite this principle as well as other beliefs that she has (in particular, her assumption that the connected stones constitute a body but its parts do not so long as they are connected). 22 That said, I do not mean to suggest anything as strong as the claim that Aristotelians need to, or even typically, arrive at the SLOWER judgment using intuitive, as opposed to refective, imagination. Rather, the suggestion is that we can explain the force of the thought experiment without thinking either that Aristotelians rely on their belief in the principle under challenge to reach SLOWER or even that there are any explicit beliefs that Aristotelians can articulate to justify the SLOWER verdict. The distinction between intuitive and refective imagination features in the explanation I have provided; for all that has been said here, other explanations are possible. If we opt for the standard reading of the thought experiment with which we started, the Aristotelian does not seem to need intuitive imagination to reach the verdicts FASTER and SLOWER at all. She can rely on refective imagination in both cases, consciously importing her belief in the Aristotelian principle to reach both verdicts. Part of what makes this observation striking is the fact that two infuential proponents of the standard interpretation—namely, Brown and Gendler—emphasize that something like what I have called intuitive imagination is needed to understand how subjects reach the thought experiment’s conclusion. In summarizing his view of thought experiments more generally, Brown writes: we perhaps start with some established propositions and rules of inference, but somewhere in the process we make some sort of empirical experience-transcending observation, we have an intuition, we see something with the mind’s eye, and this is essential to reaching the conclusion. Argument alone from established premises is not suffcient. (2004, 40)

Two Ways of Imagining Galileo’s Experiment

213

Much of what Brown is saying here can be captured within the dualprocess framework. 23 Here’s another way of getting at much the same point: thought experimenters must rely partly on intuitive thinking (‘an intuition, […] see[ing] something with the mind’s eye’) to reach the intended conclusion; refective thinking (‘Argument alone from established premises’) is not suffcient. However, Brown and Gendler rely on intuition at a different stage in their psychological explanations than I have. According to Brown (1991, 76–8), thought experimenters use intuition to appreciate the conclusion that natural speed is independent of weight. Gendler (1998, 412) offers a similar explanation of how thought experimenters reach the (weaker) conclusion that natural speed ‘might be’ independent of weight. Neither author uses intuition to explain how thought experimenters come to believe SLOWER earlier on in their reasoning. If the interpretation I have suggested is on the right track, though, then Brown and Gendler are mistaken on this point. Instead, the Aristotelian is expected to deploy her intuitive imagination already in reaching SLOWER. Intuitive imagination therefore comes in already in helping the thought experimenter reject the Aristotelian principle. Whether intuitive imagination yields knowledge of SLOWER is of course another matter. Saying that a judgment is a product of intuitive imagination is to say next to nothing of its epistemology. While a wide variety of mechanisms drive our intuitive imagination, the mechanisms that would seem to be relevant here are those that drive the application of physical concepts such as speed, air resistance, and gravity to cases more generally. I have therefore at most indicated a place to look. 24

10.6 Concluding Remarks Galileo’s famous thought experiment about falling stones poses a challenge to the Aristotelian principle of physics, which says that bodies of the same material fall in media at speeds directly proportional to their weight. This chapter argued that the case tries to do that without assuming that the Aristotelian principle produces inconsistent verdicts about what would happen in the case, pace Brown, Gendler, and others. I defended a new interpretation of how the case might nevertheless still pose a challenge to the principle. Crucial to my interpretation was the idea that thought experimenters are expected to imagine what would happen in Galileo’s experiment in two ways: one of these ways requires the Aristotelian to temporarily ignore or bracket her belief in the principle. Moreover, I suggested, we can understand this way of imagining the experiment as ‘intuitive’ whereas the other is ‘refective’. When an Aristotelian physicist intuitively imagines what would happen were an experimenter to connect two stones of different weights and drop them from a height, she ignores her belief in the principle under challenge and

214 Margot Strohminger automatically judges that the connected stones would fall slower than the heavy stone does on its own. If we wish to understand the epistemology of Galileo’s thought experiment, then we would do well to consider the reliability of intuitive imagination and the folk physics that underlies its deliverances. 25

Notes 1 For an earlier English translation, see Galilei (1954, 62–3). A variant of the case also appears in Galileo’s Dialogue on Motion (Galilei 1969, 371). 2 For examples from the contemporary debate, see Brown (1991, 2004), Nersessian (1991), Miščević (1992), Sorensen (1992), Norton (1996, 2004), Gendler (1998, 2004), Palmieri (2003), Atkinson and Peijnenburg (2004), Brendel (2004), and Meynell (2014). 3 According to Brown (1991, 43), the thought experiment also aims to show that ‘all bodies fall at the same speed’ (in a vacuum). However, for reasons pointed out by Norton (1996, 344–5) and others since (notably Gendler 1998, 419), I disagree. 4 Galileo (1638, 63–4) uses two words in the original Italian that might be translated as ‘body’ (‘corpo’ and ‘mobile’). The Drake translation (Galilei 1989) reserves the word ‘corpo’ for ‘body’ and uses the word ‘moveable’ for ‘mobile’ (unlike the Crew and de Salvio translation [Galilei 1954], which translates both as ‘body’). As far as I can tell, though, Galileo seems to move indifferently between claims about moveables and those about bodies. 5 See also especially Brendel (2004, 93) and McAllister (2004, 1170). That said, Norton (1996, 341–2) is a notable exception. In his reconstruction, the Aristotelian principle is used only to reach the verdict that the composite would fall faster than the heavy stone does on its own (or something similar enough: Norton’s own reconstruction does not use counterfactuals). The principle is not used to reach the verdict that the composite would fall slower than the heavy stone does on its own. I will end up agreeing with Norton on this point. 6 See also Brown (1991, 76). 7 Recall how Salviati gets Simplicio to agree that ‘a given moveable in a given medium has an established speed determined by nature, which cannot be increased except by conferring on it some new impetus, nor diminished save by some impediment that retards it’ (Galilei 1989, 66). 8 We are thus idealizing when we assume Plato’s body parts to be made of the same material throughout. But the same goes for Galileo’s case: here too we idealize and pretend that the composition of stones is uniform throughout. 9 These cases take their name from examples described by Gettier (1963) and Frankfurt (1969). 10 One might try to fnd some conception intermediate between the two conceptions discussed in the main text. According to it, two connected stones count as bodies but Plato’s foot does not when it is attached to the rest of his body. But it’s hard to see on what basis only the former objects get to count as bodies. 11 Thanks to an anonymous reviewer for raising this concern. 12 Initial characterizations of thought experiments sometimes suggest as much. As the frst sentence of the Stanford Encyclopedia of Philosophy entry on the topic announces, ‘Thought experiments are basically devices of the imagination’ (Brown and Fehige 2019).

Two Ways of Imagining Galileo’s Experiment

215

13 For example the assumption appears in discussions of mindreading (see e.g. Nichols and Stich 2003, 28–9, 63–4). For recent discussion, see Langland-Hassan (2020, chaps. 5 and 6). 14 See especially Williamson (2007, 141–55; 2016; 2020, chap. 2). 15 The ‘offine’ metaphor often appears in discussions of mental simulation. For an example that looks at the simulation of vision in particular, see Currie (1995). 16 How so-called ‘mental movies’, or extended imaginings rich in sensory information, develop over time is not well understood (although see Langland-Hassan [2016] and Williams [2021] for relevant discussion). 17 Making this assumption allows us to explain how the indicative conditional ‘If Oswald didn’t kill Kennedy, then someone else did’ can strike us as true while the corresponding counterfactual conditional, ‘If Oswald hadn’t killed Kennedy, then someone else would have’, does not (and in fact strikes us as false). (This example appears in Adams (1970), but other pairs of conditionals that seem to differ in truth-value in spite of only differing in mood are not hard to fnd.) 18 Thus even if one thinks of the pair of the conditionals relevant to the thought experiment as indicative, my points will carry over. Indeed, sometimes we do use indicative conditionals to talk about thought experiments—here, for example, by saying that the composite will fall faster or slower than the heavy stone does on its own. 19 Mach (1960, 36) and others since have suggested that thought experiments derive their power in part from their ability to exploit representations about reality that are not beliefs. For more recent examples, see especially Gendler (1998, 2007) and ‘mental-model’ accounts of thought experiments (see Nersessian 1991 and Miščević 1992 as well as Nersessian 2018 for overview). 20 See especially Sloman (1996), Mercier and Sperber (2009), and Nagel (2012, 497–9; 2014, 226–32). 21 Cf. Sorensen’s (1992, 61) brief discussion of the case, where he suggests that the Aristotelian relies on the claim that ‘when a faster object joins a slower one, the faster one slows down’ to reach the prediction that the composite would fall slower than the heavy stone on its own. The idea is also not that different from Brown’s (1991, 1) suggestion that Aristotelian will rely on the claim that the light object would ‘[act] as a kind of drag’ on the heavier object when they are conjoined. That said, it is important to my interpretation that this claim is not part of the Aristotelian theory of physics that the thought experimenter endorses. 22 An Aristotelian also might reach FASTER in a way that looks less imaginative and more inferential—for example, by reasoning on the basis of various beliefs, arguably some of which are modal (e.g. that it is nomically necessary that heavy bodies fall faster than lighter bodies of the same material). However since such a process looks highly sophisticated, I take it the more typical route for the Aristotelian instantiates the process for evaluating conditionals in the imagination: where one frst supposes the antecedent of the conditional and so on. 23 Although not all: for Brown intuitions in his sense produce a priori beliefs (whereas intuitive thinking in the dual-process theorist’s sense can produce beliefs that are not a priori justifed, let alone justifed). 24 In fact some empirical work on folk physics explores people’s responses to cases not unlike Galileo’s and has even led Oberle et al. (2005) to posit a ‘Galileo bias’, whereby subjects ignore air resistance in judging the speed of falling objects and exhibit related behavior. (Thanks to Joulia Smortchkova for pointing me to this work.)

216 Margot Strohminger 25 Special thanks to John Hawthorne and Juhani Yli-Vakkuri for early discussion of ideas presented in this chapter, as well as to Amy Kind, Chris Badura, and an anonymous reviewer for their comments on a draft of this chapter. I am also grateful to participants in the ‘Fiction, Imagination, and Epistemology’ Workshop at Ruhr-University Bochum for insightful questions and comments. This chapter is part of a project that has received funding from the European Union’s Horizon (2020) research and innovation program under the Marie Skłodowska-Curie grant agreement no. 747658.

References Adams, Ernest W. 1970. “Subjunctive and Indicative Conditionals.” Foundations of Language 6: 89–94. Atkinson, David, and Jeanne Peijnenburg. 2004. “Galileo and Prior Philosophy.” Studies in History and Philosophy of Science Part A 35: 115–36. Brendel, Elke. 2004. “Intuition Pumps and the Proper Use of Thought Experiments.” Dialectica 58: 89–108. Brown, James R. 1991. The Laboratory of the Mind: Thought Experiments in the Natural Sciences. London: Routledge. ———. 2004. “Why Thought Experiments Transcend Empiricism.” In Contemporary Debates in the Philosophy of Science, edited by Christopher Hitchcock, 23–43. Malden, MA: Blackwell. Brown, James R., and Yiftach Fehige. 2019. “Thought Experiments.” In The Stanford Encyclopedia of Philosophy (Winter 2019 edition), edited by Edward N. Zalta. https://plato.stanford.edu/archives/win2019/entries/thought-experiment/. Currie, Gregory. 1995. “Visual Imagery and the Simulation of Vision.” Mind and Language 10: 25–44. Frankfurt, Harry. 1969. “Alternate Possibilities and Moral Responsibility.” Journal of Philosophy 66: 829–39. Galilei, Galileo. 1638. Discorsi e Dimostrazioni Matematiche, Intorno à Due Nuove Scienze. Leiden: Elzevir. ———. 1954. Dialogues Concerning Two New Sciences. Translated by Henry Crew and Alfonso de Salvio. New York: Dover. ———. 1969. “Dialogue on Motion.” In Mechanics in Sixteenth-Century Italy, translated by Stillman Drake and I. E. Drabkin, 331–77. Madison: University of Wisconsin Press. ———. 1989. Two New Sciences, Including Centers of Gravity and Force of Gravity and Force of Percussion. 2nd edition. Translated by Stillman Drake. Madison: University of Wisconsin Press. Gendler, Tamar S. 1998. “Galileo and the Indispensability of Scientifc Thought Experiment.” British Journal for the Philosophy of Science 49: 397–424. ———. 2004. “Thought Experiments Rethought and Reperceived.” Philosophy of Science 71: 1152–63. ———. 2007. “Philosophical Thought Experiments, Intuitions, and Cognitive Equilibrium.” Midwest Studies in Philosophy 31: 68–89. Gettier, Edmund. 1963. “Is Justifed True Belief Knowledge?” Analysis 23: 121–3. Langland-Hassan, Peter. 2016. “On Choosing What to Imagine.” In Knowledge Through Imagination, edited by Amy Kind and Peter Kung, 61–84. Oxford: Oxford University Press.

Two Ways of Imagining Galileo’s Experiment

217

———. 2020. Explaining Imagination. Oxford: Oxford University Press. Mach, Ernst. 1960. The Science of Mechanics. 6th edition. Translated by Thomas J. McCormack. La Salle, IL: Open Court Publishers. McAllister, James W. 2004. “Thought Experiments and the Belief in Phenomena.” Philosophy of Science 71: 1164–75. Mercier, Hugo, and Dan Sperber. 2009. “Intuitive and Refective Inferences.” In In Two Minds: Dual Process and Beyond, edited by Jonathan St. B.T. Evans and Keith Frankish, 149–70. Oxford: Oxford University Press 2009. Meynell, Letitia. 2014. “Imagination and Insight: A New Account of the Content of Thought Experiments.” Synthese 191: 4149–68. Miščević, Nenad. 1992. “Mental Models and Thought Experiments.” International Studies in the Philosophy of Science 6: 215–26. Nagel, Jennifer. 2012. “Intuitions and Experiments: A Defense of the Case Method in Epistemology.” Philosophy and Phenomenological Research 85: 495–527. ———. 2014. “Intuition, Refection, and the Command of Knowledge.” Aristotelian Society Supplementary Volume 88: 219–41. Nersessian, Nancy J. 1991. “Why Do Thought Experiments Work?” Proceedings of the Cognitive Science Society 13: 430–38. Hillsdale, NJ: Lawrence Erlbaum. ———. 2018. “Cognitive Science, Mental Modeling, and Thought Experiments.” In The Routledge Companion to Thought Experiments, edited by Michael T. Stuart, Yiftach Fehige, and James R. Brown, 309–26. London: Routledge. Nichols, Shaun, and Stephen P. Stich. 2003. Mindreading: An Integrated Account of Pretense, Self-Awareness and Understanding Other Minds. Oxford: Oxford University Press. Norton, John D. 1996. “Are Thought Experiments Just What You Thought?” Canadian Journal of Philosophy 26: 333–66. ———. 2004. “Why Thought Experiments Do Not Transcend Empiricism.” In Contemporary Debates in the Philosophy of Science, edited by Christopher Hitchcock, 44–66. Malden, MA: Blackwell. Oberle, Crystal D., McBeath, Michael K., Madigan, Sean C., and Thomas G. Sugar. 2005. “The Galileo Bias: A Naive Conceptual Belief that Infuences People’s Perceptions and Performance in a Ball-Dropping Task.” Journal of Experimental Psychology: Learning, Memory, and Cognition 31: 643–53. Palmieri, Paolo. 2003. “Mental Models in Galileo’s Early Mathematization of Nature.” Studies in History and Philosophy of Science Part A 34: 229–64. Sloman, Steven A. 1996. “The Empirical Case for Two Systems of Reasoning.” Psychological Bulletin 119: 3–22. Sorensen, Roy. 1992. Thought Experiments. Oxford: Oxford University Press. Williams, Daniel. 2021. “Imaginative Constraints and Generative Models.” Australasian Journal of Philosophy 99: 68–82. Williamson, Timothy. 2007. The Philosophy of Philosophy. Oxford: Blackwell. ———. 2016. “Knowing by Imagining.” In Knowledge Through Imagination, edited by Amy Kind and Peter Kung, 113–23. Oxford: Oxford University Press. ———. 2020. Suppose and Tell: The Semantics and Heuristics of Conditionals. Oxford: Oxford University Press.

11 Attention to Details Imagination, Attention, and Epistemic Signifcance Eric Peterson

11.1 Introduction Does imagination carry any epistemic signifcance? The consensus is yes (cf. Kind and Kung 2016; Williamson 2016; Kind 2016; Balcerak Jackson 2016, 2018). On the face of it, the consensus may seem puzzling. After all, imagination is not constitutively constrained by truth in the way that belief is. How is it that imagining some content could provide any epistemic weight to believe that that content is true? It is beyond the scope of this chapter to provide a general account of how imagination can provide us with epistemic signifcance. However, it is important to realize that it is not just any imagining that can provide us with epistemic signifcance, but rather imagination properly constrained (cf. Kind and Kung 2016; Kind 2016; Kung 2010, 2016; Balcerak Jackson 2016, 2018; Ichikawa and Jarvis 2012; Langland-Hassan 2016). While I think that this much is true, it is incomplete in that it misses a crucial role for attention. My goal for this chapter is to highlight a possible role for attention in how some imaginings carry epistemic signifcance. The gist of my argument will be in exploiting similarities between attention in perception and attention in imagination. As it turns out, just as paying attention to the right sort of details matters for epistemic signifcance in perception so it does for imagination as well.

11.2 Preliminaries In this chapter, I will focus only on conscious attention, conscious perception, and conscious imagining. I set aside entirely questions pertaining to unconscious attention, unconscious perception, unconscious beliefs, and the relation between them. While these are interesting in their own right, we need to get clearer on the issues about conscious attention before turning to unconscious attention. Further, as Smithies (2015) claims, there are many dimensions of epistemic evaluation: we evaluate beliefs for whether or not they are justifed, true or false, reliable, amount to knowledge, and so forth. I will use the phrase “epistemic signifcance” to refer to any of the dimensions of epistemic evaluation. By remaining neutral on the exact nature of the

Attention to Details

219

epistemic signifcance that attention provides, I am able to set aside some issues that would take us too far afeld from the goal of this chapter. If I am able here to establish that attention carries epistemic signifcance, then I hope to return in future to the issue of how this epistemic signifcance can best be characterized. By imagination, I will focus only on sensory, experiential, or perception-like imagination. Whether or not the epistemic signifcance of attention extends to other kinds of imagining will need to wait for another project. As an example of a use of experiential imagining, consider this familiar thought experiment: Mary is a brilliant scientist who is, for whatever reason, forced to investigate the world from a black and white room via a black and white television monitor. She specializes in the neurophysiology of vision and acquires, let us suppose, all the physical information there is to obtain about what goes on when we see ripe tomatoes, or the sky, and use terms like ‘red’, ‘blue’, and so on. She discovers, for example, just which wave-length combinations from the sky stimulate the retina, and exactly how this produces via the central nervous system the contraction of the vocal chords and the expulsion of air from the lungs that results in the uttering of the sentence ‘the sky is blue’. What will happen when Mary is released from her black and white room or is given a color television monitor? Will she learn anything or not? (Jackson 1982, 130) When we answer these questions by imagining the case that Jackson has presented us with, this is a paradigmatic case of an epistemically signifcant use of imagination. Even if there is no consensus as to what we learn from this thought experiment (i.e., whether or not physicalism is false, etc.), there seems to be wide agreement that imaginatively engaging with it carries epistemic signifcance—as does engagement with thought experiments more generally. What is often and easily overlooked is the way in which attention factors in our engagements with this and other thought experiments and in particular the way in which attention contributes to the epistemic signifcance of such imaginative activities. I will return to this below, but frst I need to say a little about the nature of attention. What is attention? Like most things in philosophy, providing an intuitive, folk account of “attention” is rather easy; however, providing a robust theory is much more diffcult. Virtually all accounts of attention start with the same quote from James, which introduces the intuitive conception of attention. Far be it from me to buck the trend: Everyone knows what attention is. It is the taking possession by the mind, in clear and vivid form, of one out of what seem several

220 Eric Peterson simultaneously possible objects or trains of thought. Focalization, concentration, of consciousness are of its essence […] (James 1890/1981, 403–4) It seems right to start where James starts, but hardly right for any analysis to end there. To mention just one of the further issues that would need to be resolved in developing a more complete analysis, one needs to determine whether attention should be defned functionally through sub-personal mechanisms (reductionist accounts) or phenomenologically in terms of its person-level properties (anti-reductionist accounts).1 Setting such controversies aside, it is somewhat standard to treat attention as a mental phenomenon, perhaps a mode of consciousness, whose role is to selectively focus on certain content/phenomena. In order to read this sentence, you need to attend to it. This is the voluntary mode of attention (hopefully, no one just forced you to read that sentence). There is also an involuntary aspect to attention as when the coffee maker chirps at you that your coffee is now ready. The sound of the coffee maker draws your attention away (if only temporarily) from reading this chapter. This selective focus on certain content/phenomena plays a critical role in both our perception and our action. And this chapter will add that it plays a critical role in our imaginative activities such as thought experiments. Thus, attention can be voluntary or involuntary, conscious, and have a selective2 function. These are all generally recognized as standard features of attention. Apart from these standard features of attention, I want to highlight two aspects of attention that help us better understand its nature and that play a role in my argument. First, Watzl (2011) discusses what he calls attentional space. To get a grip on this idea, he discusses the way that our perceptual experience seems to come with a spatial structure. We attribute certain relations to our perceptual experience of the world. Some things are closer to us, some farther. Some things are to the left of us, some to the right. Some things are above us and some below us. As Watzl explains, “[t]he ‘spatial structure of experience’ just is the spatial structure of the way the world appears to be” (159). In contrast, he claims that attentional relations are not relations that are attributed to the world, but rather they are relations between the parts of our experience (159). According to Watzl, the most central structure of attentional space has a primitive phenomenal relation of peripherality (160). In conscious attention, what you attend to becomes the center of your conscious experience and all other parts of your experience move to the periphery. One example he discusses is your auditory experience of a jazz concert. Attending to the piano makes it central to your conscious experience. You will still experience the saxophone; however, it will be in the periphery. Watzl uses this peripherality relation to defne what is meant by focus or object of attention:

Attention to Details

221

Some object, event, or property in the world is the focus or object of your attention, just in case you have an experience of it that is at the center of consciousness, where an experience is at the center of consciousness during a certain period of time just in case during that period of time no part of your experience is central to it (or equivalently, just in case during that period of time it is not peripheral to any part of your experience). (160) Some parts of your conscious experience are more central and some are more peripheral. So on Watzl’s account to attend to some object/content x is to make x central in one’s conscious experience. Everything that is not x moves to the periphery. Second, Smithies (2011) claims that the role of attention is “to make information rationally accessible for use in the control of action, reasoning, and verbal report” (248). Call this the Access Role of attention. Importantly, Smithies makes a distinction between rational access and causal access (262). To illustrate the difference consider blindsight patients. Someone with blindsight can access unconscious information at least in the sense that such information can prime their performance in certain tasks. They thus have causal access to this information. However, this information is not rationally accessible to them as a justifying reason for belief and action (262). Following Smithies, I am mostly concerned with rational access whereby the information that is accessed can function as a justifying reason for belief and action. This much characterization of attention will suffce for the purpose of this chapter. I will next demonstrate how attention plays an epistemically signifcant role in our perception, and I will then show how this can be extended to certain imaginative activities such as engaging with Jackson’s Mary case.

11.3 Perception and Attention One intuitive reason for thinking that attention carries epistemic signifcance comes from the widely experienced phenomenon of missing obvious people/events in plain sight as it were. We all have had experiences where our focused attention is strained and we fail to see familiar faces before our very eyes. Suppose you are waiting to pick up a friend from the train station. As the crowds exit the train, you intently scan the crowd of people but completely miss your friend who is waving her arms at you. In such a case, it seems that the demands on our attention preclude us from seeing and therefore presumably from having a certain epistemic success in our perceptual experiences. This plausibly suggests or gestures at the idea that attention is somehow critical or essential for epistemic signifcance in perception. This intuitive assessment has

222

Eric Peterson

empirical support that comes out of inattentional blindness experiments (cf. Simons and Chabris 1999; Mack and Rock 1998). One of the most provocative of these experiments is one in which the subjects were tasked with counting the number of passes of a basketball between members of one team who wore white jerseys and members of the other team who wore black. While attending to the task of counting the passes, a surprising number of subjects do not see a person in a gorilla suit walk into the middle of the scene, pound its chest, and walk out. (For different variations and results of this experiment, see Simons and Chabris 1999.) Such experiments seem to show a plausible connection between attention and the epistemic signifcance of perceptual beliefs. 3 We might say that failure of attention leads to a sort of epistemic failure (e.g., perhaps without attention, one fails to form a perceptual belief or maybe one’s attention is strained such that they miss out on an experience that would otherwise ground a perceptual belief). We can put some more precision to this intuitive assessment by highlighting three roles that attention plays that each carry epistemic signifcance. These are the content role, determinacy role, and the aforementioned access role. I will briefy discuss each of these roles next. The content role of attention refers to the idea that it is attention that fxes the content of perception (and perhaps other cognitive attitudes as well). To explain this requires me to say a bit about perceptual content. I follow Nanay (2010, 2013, 2015 2016) in taking perceptual content to be an attribution of properties to objects within a perceived scene. It is attention that is responsible for this attribution of properties. Watzl’s notion of “attentional space” is helpful here. The whole perceived scene can be structured by attentional space. I can attend to my laptop and shift my attention to my coffee mug, and I can shift my attention to the relation between them (the mug is to the right of the laptop), or I can attend to the colors and shapes of my mug and laptop. Whenever I am attending to one of these distinct properties and relations, this property or relation becomes central to my conscious experience and all of the others move to the periphery. Recall this center/periphery is one relation that characterizes attentional space, and should not be confused with peripheral vision. Although the object of our peripheral vision can become the center of our foveal vision, peripheral vision cannot become foveal vision. The center/periphery relation is a relation between parts of our conscious experience whereas the peripheral/foveal relation is a relation between anatomical structures of our vision. Characterizing perceptual content as constituted by attentional space makes it plausible that perceptual content is fxed by attention. The whole perceived scene of properties and objects is structured by attentional space. Further, perceptual content is essentially representational. However, it is not clear how one can determine what is represented in perception without reference to attention.4 Because attention is essential

Attention to Details

223

in fxing perceptual content and perceptual content is essential in forming perceptual beliefs, it follows that attention is essential in forming perceptual beliefs. The content role of attention provides a weak sense of epistemic signifcance—without perceptual content we do not get perceptual beliefs a fortiori epistemically signifcant perceptual beliefs. The determinacy role of attention is a corollary of the content role. Nanay (2010) argues that the properties that are attributed to the perceived scene are characterized by the determinate-determinable relation, and it is attention that makes an attended property more determinate (266). As Nanay acknowledges, a consequence of this view is that every change in attention brings about a change in one’s perceptual content (268). One consequence of this is that perceptual content is very fnegrained. As he claims, two people could look at the same apple and attend to different properties of the apple, which means they will both have different perceptual content. This difference may make an epistemic difference but it need not. The different perceptual content may just give different reasons for different beliefs each of which are made reasonable by the different content. However, in some cases the difference of perceptual content could carry a greater degree of epistemic signifcance. To see this, consider how our peripheral vision is limited to determinable (rather than determinate) properties. We can form beliefs of this sort: such and such an object having such and such properties is in my peripheral feld. But we cannot form beliefs having more specifc and determinate content than that in our peripheral feld. The point is not that our peripheral perceptual beliefs carry no epistemic signifcance; rather, it is that the epistemic signifcance that they do carry will be limited compared to the epistemic signifcance of foveal vision. Importantly, I am not arguing that the difference between peripheral and foveal vision comes down to attention. Rather, the contrast illustrates the difference between determinable and determinate content and that the latter carries more epistemic signifcance than the former (think about the difference of navigating through a complex terrain with glasses that blocked out your foveal vision but allowed peripheral vision vs. navigating that same terrain without such glasses). And critically, it is attention that makes properties more determinate and so it is attention that contributes to a greater degree of epistemic signifcance within our perceptual experiences. As already mentioned above, the access role of attention makes information rationally accessible for the use of rational thought and action. That this role carries epistemic signifcance is easily shown. If one cannot access information for control of action and reasoning, then this will defnitely preclude knowing certain things. If you cannot access the relevant content, then you cannot form the relevant beliefs. Also, if your attention accesses information that is irrelevant to epistemic signifcance, then attention will be failing in providing epistemic signifcance to one’s

224 Eric Peterson attitude. So one’s attention may be unable to access relevant information or it will only access irrelevant information. Either way epistemic significance can be hindered. These three roles are importantly interconnected. The content role is necessary for the access role. If there is no content, then there is nothing to access for use in rational thought and action. Further, the determinacy role is what accounts for the fact that perceptual content is very fne-grained, which means that two people may have different perceptual content even while attending to the same object or scene, and this may, at times, make an epistemic difference. With these three interrelated roles of attention, we can provide a plausible interpretation of inattentional blindness experiments. In the Gorilla experiment, people are given a specifc task that creates a certain attentional load on participants. This attentional load fxes a determinate content and this determinate content does not include a person in a gorilla suit. And because of this, the content that attention accesses does not include a person in a gorilla suit, which is why it is unsurprising that such information is not used in rational control of thought and action. Between the access role, content role, and determinacy role, it should be obvious that attention carries epistemic signifcance in perception. Admittedly, this is a general and weak thesis as opposed to a stronger thesis that claims that attention is required for one of the many epistemic values such as justifcation. The stronger thesis—that attention is necessary for justifcation—entails the weaker thesis. Not all are convinced of the stronger thesis. And while my argument need not hang on the stronger thesis, it will be fruitful to look at some of the reasons against accepting the stronger thesis. Seeing that these reasons fail will open up the possibility that the stronger thesis is true, which will give us the weaker one for free. Silins and Siegel (2014, 2019) provide reasons to reject the stronger thesis. They argue for what they call the Attention Optional View— attention is not necessary for justifcation. In order to see this, consider the widely endorsed distinction between propositional justifcation and doxastic justifcation that they characterize as follows: Propositional Justifcation: Experience E provides propositional justifcation for P, if it provides some reason for P, whether or not the subject forms the belief that P. Doxastic Justifcation: A belief is doxastically justifed iff it is formed and maintained epistemically well. Against attention being necessary for propositional justifcation, they discuss a non-perceptual case, what they call inferential blindness. For instance, if I have a justifed belief that I have an appointment with X, and a justifed belief that I have an appointment with Y, and a justifed

Attention to Details

225

belief that Y is not X, then I will have reason to believe that I have conficting appointments even though I do not notice the confict. The idea is that one already has the propositional justifcation for the confict, it is just that one does not exploit the evidence at the relevant time. Although this case is not a case about perceptual justifcation, they also point out the precedent for unnoticed justifcation in cases of “change blindness,” which would constitute an alleged case of perceptual, propositional justifcation without attention. For instance, you fail to notice that your friend shaved off their mustache even after they ask you if you notice anything different about them (cf. Dretske 2004, 2006). Both these cases of inferential and change blindness seem to support that idea that one can have a reason to believe that something is the case without exploiting this evidence/reason. Against the idea that attention is necessary for doxastic justifcation, they give the following case: Consider a distracted subject navigating the environment, such as a distracted driver, or a walker lost in thought. Such a subject can still adjust their behavior in response to the environment in a way that is not merely instinctive, operating the brake, the clutch, the defrosting system, the steering wheel, and so on. Further, they arguably can do so while remaining distracted from the environment, without their attention being captured by the obstacles that they are successfully avoiding, or by the equipment that they are manipulating. Despite being superfcially automatic, such behavior is far from being a mere refex, and has a strong claim to being rational. In addition, the subject would satisfy the central diagnostics for having various beliefs about her immediate situation, such as the belief that the car is running and operating as it should be. They are disposed to endorse this proposition if asked, and they are acting in a way that would be advisable, given their desire to continue driving, if the proposition is true. In such a case, their inattentive experience is feeding into well-founded perceptual beliefs. (Silins and Siegel 2014, 159) According to Silins and Siegel, this case demonstrates that attention is not necessary for doxastic justifcation. The distracted driver is obviously forming well-founded beliefs. If this were not the case then we would be left describing the distracted driver’s success of driving from point A to point B as merely getting lucky. But that does not seem right. It is better to describe it as the driver acting on doxastically justifed perceptual beliefs. Suppose we accept this description. Still, that is not enough to establish the claim that attention is not necessary for doxastic justifcation. For need it be the case that attention is absent in such cases? Not all see

226

Eric Peterson

it that way. For instance, Wu (2014) argues that it really depends on what one means by “inattentive experience”. He argues that there can be an “inattentive attention” (253). Wu characterizes inattentive attention as having an increased automaticity and less control. Additionally, he suggests the distinction could be made between conscious and unconscious processes where an increase in inattentiveness leads to a decrease in consciousness—attention comes in degrees. The upshot is that Silins and Siegel’s case really only undermines the notion that conscious attention is necessary for doxastic justifcation; it would not affect the notion of an unconscious attention feeding into doxastic justifcation. But I think that we can push the feature of attention coming in degrees5 even further than Wu does without appealing to unconscious attention (as I mentioned above, I am focusing exclusively on conscious attention). That is, we can question whether it really is the case that the driver’s inattentive experience is outside of their conscious attention. Here I can rely on Watzl’s (2011) “attentional space” as characterized above. Watzl’s account of attentional space gives us a concrete and plausible way to understand the notion of degrees of attention.6 And because of this, it gives us another plausible way to describe the case of the distracted driver without claiming that there is an absence of attention. This is rather simple to show. We can just think of the distracted driver case now with attentional space mapped on to the driver’s experience. Whatever it is that is distracting the driver, it most certainly is central in their attentional space. It could be refecting on how their lecture went or wondering what to make for dinner. But whatever it is all else is peripheral to it. It is not that there is no attention to the road and other cars and the steering wheel, clutch, brake, and so on. It is rather that these perceptual (and sometimes cognitive) experiences are currently less central or more peripheral to the driver’s current thoughts/worries. The point is then that we should not describe the driver as attending to one and only one thing and therefore not attending to things that are important for driving. It is rather that some things are more central in the driver’s (conscious) attentional space and some things are more peripheral. Further, it is important to point out that the distracted driver case is really a mundane case that we all experience all of the time. This sort of “distraction” happens when we are brushing our teeth, taking a shower, working out, shopping for groceries, and so on. And given we can always describe such cases using Watzl’s notion of attentional space, it follows that such cases cannot be used to argue that attention is optional when it comes to doxastic justifcation. This is mainly because we simply cannot say that there is no attention feeding into our perceptual beliefs and the action that follows from them. Note that this argument also can be applied to Silin’s and Siegel’s cases that allegedly show propositional justifcation without attention. As they

Attention to Details

227

put it there are instances of change blindness and inferential blindness where we “fail to notice” the justifcation for a particular belief. The problem here is “failing to notice” could just be taken to mean paying less attention and with the varying degrees of attention within attentional space, we just cannot be confdent that “failing to notice” means a complete lack of attention. Because of this, I am not convinced that their cases support the view that attention is optional for justifcation. Thus, it is plausible to see attention as necessary in perceptual justifcation. And if this is the case, then it follows that attention carries epistemic signifcance in perception. However, even if questions remain about whether attention is not necessary for justifcation, it would not follow that attention is epistemically insignifcant. As I have shown, the content, determinacy, and access roles of attention all to one degree or another carry epistemic signifcance. In the next section, I show how this can extend to our imagination.

11.4 Perception, Attention, and Imagination In general, that there is similarity between perception and imagination (sensory) needs no defense. When we talk of similarity between perception and imagination, we are mostly talking of phenomenal similarity. It is the qualitative experiences that they share. My experience of seeing the famous ivy along Wrigley Field’s outfeld walls and my visualizing the same scene both share the similar qualitative character. And of course, this phenomenal similarity does not only come from anecdotal frst-person experience; it also has some experimental support. This support comes out very clearly in the famous Perky experiment (Perky 1910), where subjects often confused visualizing a banana with perceiving one. This much is widely accepted.7 However, what is in need of defense is whether or not these relevant similarities can do the work that my argument needs them to do. More perspicuously, we can ask given that attention carries epistemic signifcance in perception, why think that it does in imagination? As a start to answering this question, it will be helpful to consider some reasons for thinking that imagination carries epistemic signifcance. It is important to realize that it is not just any imagining that has epistemic signifcance, but rather imagining that is appropriately constrained. This has been argued rather persuasively by Kind (2016). Kind argues that there are two important constraints that provide epistemic significance to our imaginings—what she calls the Reality Constraint and the Change Constraint. Imaginings are constrained by the Reality Constraint when such imaginings are guided by the world as it really is. For instance, when I am considering whether a new table will ft in my dining room, my imagining—if it is to be epistemically useful—will follow the Reality Constraint. That is I will imagine this particular table ftting

228

Eric Peterson

within my imaginative space that includes my dining room. In contrast, when I imagine myself walking through the walls of my dining room, my imagining is no longer being constrained by the Reality Constraint. Our imaginings are constrained by the Change Constraint just in case when we imagine a change; such a change respects logical consequences. For instance, if I replace an old dining room table with a new one that is 10 ft2 larger than the old one, then that necessitates a decrease in the non-furniture square footage of the dining room. Thus, if my imagining this new table in my existing dining room does not take into account such a decrease, then it is not following the Change Constraint. Importantly, Kind points out that these constraints are aspirational in nature. The closer our imaginings come to meeting them, the better chance that we will learn from our imaginings (2016, 153). This, as she points out, is not unlike perception. Our perceptual representations of the world may fall short of ideal perception—representing the world exactly as it is—however, this ought not cause us to distrust what we learn from our perceptions about the world (153). Thus, when our imaginings are suitably constrained, they may carry epistemic signifcance. Now we are ready to answer the question as to why we should think that attention plays any role in this epistemic signifcance. The simple answer is that in many ways our imaginings are constrained by our perceptions and we have already seen how attention in perception carries epistemic signifcance. In order to see how perception can constrain some imaginings (i.e., sensory or perception-like imaginings), consider the imagery mechanism—the capacity of our cognitive architecture responsible for mental imagery. Much of our mental imagery comes from our perceptions, and it is accurate imagery that often helps us imagine in a way that is consistent with reality. We can say more about how the imagery mechanism constrains our imaginings by saying more about the phenomenal similarity between perception and imagining. Nanay (2016) defends a view that grounds the phenomenal similarity between perception and imagining on the sameness of content between perception and imagining. Recall Nanay’s view on perceptual content. In perception, we attribute properties to objects within our perceived space. Importantly, for Nanay, the content of our mental imagery is the same. When we visualize, for instance, we attribute properties to objects within our egocentric space. Whether I am at Wrigley Field perceiving the green ivy on the outfeld walls or whether I am imagining the green ivy on the outfeld walls, the content is the same. The difference, according to Nanay, has to do with the way the properties are attributed. Here is how he puts it: If you are looking at the apple and you are attending to the exact shade of red of the patch on its righthand side, the high determinacy of this attributed color property comes in a bottom-up manner from what you see. But if you close your eyes and visualize an apple, you

Attention to Details

229

can still attribute this very determinate property to the patch on the righthand side of the apple, but in this case, this determinacy comes in a top-down manner: from our memory or belief or expectations. (Nanay 2016, 129) Importantly, our visualizing of an apple depends on our memories of experiencing apples as well as our perceptual beliefs that we have formed from experiencing apples. This dependency relation between the content for perception and the content for imagination is critical in seeing how attention in imagination can carry any epistemic signifcance. Of course, not all imagining depends on perception, but much (all?) of our perception-like imagining depends in some way on our perception. As Balcerak Jackson claims, visual imagining is by its nature a capacity that is dependent on our perceptual capacity (2016, 51). Elsewhere, Balcerak Jackson argues that imaginings can justify beliefs “in virtue of being by their very nature derived from or parasitic on perceptual experience […]” (2018, 221). This dependency reveals how attention plays similar roles in imagination as in perception. In fact, it reveals how attention can play the same three roles in imagination as it does in perception. When we engage in perception-like imagining, it is attention that fxes the content—content role—and it is attention that makes the content more or less determinate—determinacy role. And at the same time, it is attention that allows us to access this content for use in rational thought and action—access role. Like perception, these roles for attention are largely taken for granted. It usually takes attention going wrong such as in cases of attentional blindness in order for us to see its value. The same can be said for imagination. In order to see this, it will be helpful to clarify how imagination can be characterized by attentional space and how this applies to a thought experiment such as Jackson’s Mary. Like perception, imagination can be characterized by attentional space. For complex imaginative activities such as engaging a fctional narrative, there will be many phenomenal relations that can be characterized by the peripherality relation. When engaging a fctional narrative the imagining is guided: imagine x, then imagine y. When guided, even implicitly, to imagine x, it becomes the center of your attention within your imagining. However, our imaginative experience while engaging a fctional narrative is not so fat and one dimensional. Engaging a fctional world can very much mirror the richness of experiencing the real world. There are many parts to the experience. While x is the center of your imagination’s attention, y, z, and so forth are peripheral to it. The same will hold for short fctional narratives like thought experiments. There is a content that the thought experiment is guiding one to imagine. This usually can bring about a rich imaginative experience with many parts. As an example, return to Jackson’s Mary case. As I mentioned above, this is an example of an epistemically signifcant use of perception-like

230

Eric Peterson

imagination. Because it is perception-like imagining, and because of the dependency of imaginative content on perceptual content, we should see that attention can play the same epistemically signifcant roles in this imaginative activity. Or at least, like perception, we can see the importance of attention for epistemic signifcance in imagination, when attention goes awry. When I imaginatively engage the Mary case, I attend to certain content as part of the imaginative experience. I visualize Mary being contained in a black-and-white room. At the same time, I cognitively attend to the alleged fact that she knows all physical facts about color. Then as I am guided by the thought experiment, I visualize her exiting the blackand-white room and experiencing color for the frst time. Attending to this determinate content provides reasons for believing that knowing all of the physical facts of something is not the same as experiencing what that something is like. There are other contents that I could attend to. I could wonder what else is in the room with her, whether she is lonely or sad, and some of these contents will be relevantly peripheral and some will be irrelevantly peripheral to what is the center of my attention as I engage the thought experiment. And if I am not attending to those things that should be central in my experience of the thought experiment, then I will miss out on the justifcatory force/epistemic signifcance of the thought experiment primarily because my attention will not fx the content that is relevant for epistemic signifcance.8 That is attention will fx a determinate content and accessing this content will prove fruitless as it will largely be irrelevant for epistemic purposes. This difference of attending can be a consequence of the determinacy role of attention. Recall that the determinacy role of attention makes perceptual content very fne-grained. There is no reason to think that the same would not also hold for attention in imagination. Just as Nanay suggests that two people could look at the same apple and yet have different perceptual content, so is it the case that two people can engage in the same thought experiment and yet have different imaginary content. And importantly, for my purposes this difference is a difference that can make a difference in epistemic signifcance. Again the reason for this is that if one is attending to the wrong content of the thought experiment— wrong because such content is irrelevant for epistemic purposes—then such a person will miss out on the epistemic signifcance of the thought experiment. The content that they attend to will not give them a reason to form the relevant beliefs. Perhaps it is hard to imagine someone engaging Jackson’s thought experiment in such a way that they attend to the wrong content or to content that is irrelevant for epistemic purposes. This, however, I think has more to do with the fact that we philosophers have been weaned on Jackson’s famous thought experiment. Although this is merely anecdotal, I have found that I often have to coach students who are less

Attention to Details

231

philosophically inclined on how to properly engage with thought experiments. Attending to content that is irrelevant for epistemic purposes may be the rule rather than the exception. Even if it is not, that imaginary content can be fne-grained allows for the fact that there are many and varied contents that one can attend to while engaging thought experiments. Further, even among philosophers who know how to engage thought experiments, there are varying intuitions about whether and what thought experiments teach us. If two people could engage the same thought experiment and yet have different imaginary content, then it seems possible that each could have differing intuitions about that same thought experiment. Attending to different content means that each person accesses different content. I do not pretend that this explanation accounts for all disagreements concerning thought experiments. This is because it would mean that all disagreements over thought experiments would collapse to a sort of talking past each other. While it is implausible to think that all disagreements over thought experiments are of this sort, it is not implausible to think that some disagreements are the result of imagining different content. More importantly for my purposes, however, is that like perception, attention fxes the content of imagining and simultaneously allows for access of such content. Attending to content that is not relevant for epistemic purposes usually distracts me from attending to content that is relevant for epistemic purposes, and this is almost akin to focusing on content that blinds one to seeing a person in a gorilla suit. When attending to the counting task, I am distracted from attending to the rest of the action on the basketball court and so have no reason to form the belief that there is a person in a gorilla suit. Similarly, when attending to the seeming cruelty of Mary being locked in a blackand-white room, I am distracted from attending to the issue about the physical facts and so have no reason to believe that there is more to be known over and above all of the physical facts about something.

11.5 Conclusion Attention plays at least three critical roles in the epistemic signifcance of perception. In particular, attention fxes content (the content role), makes the content more or less determinate (the determinacy role), and allows for access to such content (the access role). Without content, there are no perceptual beliefs. Further, as we have seen, determinacy tends to carry more epistemic usefulness (e.g., navigating through a rough terrain with foveal vision vs. navigating through that same terrain with only peripheral vision). And fnally, the access of such content allows for the control of rational belief and action. When attention goes awry as in attentional blindness, the epistemic signifcance of perception gets hindered. Further, given that perceptual content can be very fne-grained, it follows that not all will have access to the same epistemic signifcance of

232 Eric Peterson perception. Perception-like imagining depends in a crucial way on perception. As I have argued, perception-like imagining inherits these three roles. This gives us a straightforward way of seeing how perception-like imagining carries epistemic signifcance—attention fxes the imaginary content, makes it more determinate, and allows us to access it for the control of rational thought and action. However, like perception, attention in imagination can go awry. Imaginary content is fne-grained. Two people can imagine the same thought experiment and attend to different determinate content. Some of this content will be relevantly epistemically signifcant and some will not. The upshot of all of this is that just as in perception, attention to the right sort of details matters for the epistemic signifcance of imagination. Of course, important questions remain. Perhaps most importantly, I  have not in this chapter attempted to explicate the nature of the epistemic signifcance that attention provides. Some might worry that the role I have identifed for attention will turn out to be merely an enabling one, rather than one that is justifcatory in nature. Attention, that is, plays a role in the generation of beliefs, rather than in the justifcation of them. Though I think that even this enabling role is an important one, my own sense is that the role that attention plays in both perception and imagination goes further than this. In a sense, this chapter can be seen as clearing the ground for that future work. Now that we have brought attention into the epistemic scene of imagination, future work can explore in more detail the precise nature of the epistemic signifcance that it brings. 9

Notes 1 Apart from these debates, attention is called upon to play an explanatory role for a dizzying number of phenomena including consciousness, demonstrative reference, and knowledge of other minds. 2 Fazekas and Nanay (2018) have recently called into question the widely accepted assumption that attention is selection. Their motivation is to offer an account that purports to solve problems for reductionist accounts of attention. Even if they are correct, it is still safe to treat attention as having a selective-like function or feature. 3 Mole (2017) claims that the same thing can be said in an a priori domain. As he claims, what is required in order to recognize an a priori entitlement is the right sort of active attention. There can be a sort of inferential blindness (see above) as well as inattentional blindness. Mole (2017) and Wu (2014) both claim that there is much work that still needs to be done on general relations between attention and knowledge. 4 Recall I am focusing only on conscious perception. If there is unconscious perception, then there may be a sort of unconscious representation. However, there may still be an unconscious attention. 5 Interestingly, Silins and Siegel make much of the fact that attention comes in degrees, but they don’t take that as allowing for a different take on the cases. I  think that my argument above shows that they should see this as being the case.

Attention to Details

233

6 Watzl also acknowledges this, 158. 7 Though see Hopkins (2012) for a dissenting view. 8 One interesting experiment that could lend support to this idea would be test people’s different responses to the Mary Thought Experiment as it was originally intended to be taken in and their responses to the Mary Thought Experiment when it is embedded in the movie Ex Machina. My hypothesis is that in the latter there would be signifcant attentional demands coming from the plot of the movie such that many people would fail to feel the justifcatory force that the Mary Thought Experiment carries. 9 I am grateful for comments from audience members at the Fiction, Imagination, and Epistemology Workshop in Bochum, Germany (June 2019) when I presented some of the seeds of this chapter. I am also grateful for the comments from an anonymous referee. Finally, I am very grateful to the editors of this volume, Amy Kind and Christopher Badura, for their tireless effort and comments to help improve this chapter.

References Dretske, F. (2004). “Change Blindness.” Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition, 120(1/3), 1–18. Dretske, F. (2006). “Perception without Awareness.” In T. S. Gendler & J. Hawthorne (eds.), Perceptual Experience (pp. 147–80). Oxford University Press. Fazekas, P., & Nanay, B. (2018). “Attention is Amplifcation, Not Selection.” The British Journal for the Philosophy of Science, 1–28. https://www.journals. uchicago.edu/action/doSearch?AllField=Fazekas+and+Nanay&SeriesKey=bjps Hopkins, R. (2012). “What Perky Did Not Show.” Analysis, 72(3), 431–9. Ichikawa, J., & Jarvis, B. (2012). “Rational Imagination and Modal Knowledge.” Noûs, 46(1), 127–58. Jackson, F. (1982). “Epiphenomenal Qualia.” The Philosophical Quarterly, 32(127), 127–36. Jackson, M.B. (2016). “Imagining, Supposing, and Conceiving.” In A. Kind & P. Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press. Jackson, M.B. (2018). “Justifcation by Imagination.” In F. Macpherson & F. Dorsch (eds.), Perceptual Imagination and Perceptual Memory (pp. 209– 26). Oxford University Press. James, W. (1890/1981). The Principles of Psychology. Cambridge, MA: Harvard University Press. Kind, A. (2016). “Imagination under Constraints.” In A. Kind & P. Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press. Kind, A., & Kung, P. (eds.) (2016). Knowledge Through Imagination. Oxford: Oxford University Press. Kung, P. (2010). “Imagining as a Guide to Possibility.” Philosophy and Phenomenological Research, 81(3), 620–63. Kung, P. (2016). “Imagination and Modal Epistemology.” In A. Kind (ed.), The Routledge Handbook of Philosophy of Imagination (pp. 437–50). New York: Routledge. Langland-Hassan, P. (2016). “On Choosing What to Imagine.” In A. Kind & P. Kung (eds.), Knowledge Through Imagination. Oxford: Oxford University Press.

234

Eric Peterson

Mack, A., & Rock, I. (1998). Inattentional Blindness. Cambridge, MA: MIT Press. Mole, C., Smithies, D., & Wu, W. (eds.) (2011). Attention: Philosophical and Psychological Essays. Oxford: Oxford University Press. Mole, C. (2017). “Attention.” In E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall edition). https://plato.stanford.edu/archives/fall2017/entries/ attention/ Nanay, B. (2010). “Attention and Perceptual Content.” Analysis, 70(2), 263–70. Nanay, B. (2013). Between Perception and Action. Oxford: Oxford University Press. Nanay, B. (2015). “Perceptual Content and the Content of Mental Imagery.” Philosophical Studies, 172(7), 1723–36. Nanay, B. (2016). “Imagination and Perception.” In A. Kind (ed.), Handbook of the Philosophy of Imagination (pp. 124–34). New York: Routledge. Perky, C. W. (1910). “An Experimental Study of Imagination.” The American Journal of Psychology, 21(3), 422–452. Silins, N., & Siegel, S. (2014). “Consciousness, Attention, and Justifcation.” In E. Zardini & D. Dodd (eds.), Scepticism and Perceptual Justifcation. Oxford: Oxford University Press. Silins, N., & Siegel, S. (2019). 29 “Attention and Perceptual Justifcation.” In A. Pautz & D. Stoljar (eds.), Blockheads!: Essays on Ned Block’s Philosophy of Mind and Consciousness, 487. Cambridge, MA: MIT Press. Simons, D.J., & Chabris, C.F. (1999). “Gorillas in Our Midst: Sustained Inattentional Blindness for Dynamic Events.” Perception, 28(9), 1059–74. Smithies, D. (2011). Attention is Rational-access Consciousness. In C. Mole, D. Smithies, & W. Wu (eds.), Attention: Philosophical and Psychological Essays, 247–73. New York: Oxford University Press. Smithies, D. (2015). “Why Justifcation Matters.” In D. Henderson & J. Greco (eds.), Epistemic Evaluation: Purposeful Epistemology, 224–44. Oxford: Oxford University Press. Watzl, S. (2011). “Attention as Structuring of the Stream of Consciousness.” In C. Mole, D. Smithies, & W. Wu (eds.), Attention: Philosophical and Psychological Essays, 145–73. New York: Oxford University Press. Williamson, T. (2016). “Knowing by Imagining.” In A. Kind & P. Kung (eds.), Knowledge Through Imagination (pp. 113–23). Oxford: Oxford University Press. Wu, W. (2014). Attention (1st ed.). New York: Routledge.

Section IV

Understanding Self and Others

12 Bridging the Divide Imagining Across Experiential Perspectives Amy Kind

Can one have imaginative access to experiential perspectives vastly different from one’s own? Can one successfully imagine what it’s like to live a life very different from one’s own? These questions are particularly pressing in contemporary society as we try to bridge racial, ethnic, and gender divides. Yet philosophers have often expressed considerable pessimism in this regard, a pessimism that is mirrored in non-philosophical contexts throughout popular culture and public discourse. It is often thought that the gulf between vastly different experiential perspectives cannot be bridged. In this chapter, I explore the case for this pessimism. As I will suggest, the case is considerably weaker than it is usually thought.

12.1 Pessimism: The Basic Idea To fesh out the pessimistic position, I want to start with its expression in popular culture and public discourse. Expressions of pessimism about one’s ability to understand experiential perspectives vastly different from one’s own are commonplace in these contexts and occur in a variety of media about a variety of experiential perspectives. Consider, for example, the song “Til It Happens to You” by Lady Gaga. The song’s narrator is a sexual assault survivor who powerfully conveys the sense that you can’t know what it’s like to be such a person unless you have been sexually assaulted yourself: ’Til it happens to you, you don’t know How it feels How it feels ’Til it happens to you, you won’t know It won’t be real (how could you know?) No it won’t be real (how could you know?) Won’t know how I feel ’Til your world burns and crashes ’Til you’re at the end, the end of your rope

238

Amy Kind ’Til you’re standing in my shoes, I don’t wanna hear nothing from you From you, from you, ’cause you don’t know

Among the many arresting images in the video for the song is one of a woman’s arms with “Listen” and “You Will Hear Me” written across them. Another example comes in the musical Hamilton. After Philip Hamilton is killed in a duel, his parents Alexander and Eliza must cope with tremendous grief. In the musical number, “It’s Quiet Uptown,” parental grief is portrayed as both indescribable and unimaginable to those who have not experienced it: There are moments that the words don’t reach There is suffering too terrible to name You hold your child as tight as you can Then push away the unimaginable These frst two examples focus on differences in experiential perspective – what it is like to be a sexual assault survivor, what it is like to be a grieving parent – that arise from a formative event or series of events. But differences in experiential perspective also may arise from social positionality. Here too, the standard treatment of these differences in popular culture and public discourse sees them as unbridgeable. Consider discussions of racial relations in America. In an audio story published by The New York Times, black poet Claudia Rankine (Rankine 2015) notes the limitations that face white people when trying to understand what it is like to be black: Though the white liberal imagination likes to feel temporarily bad about black suffering, there really is no mode of empathy that can replicate the daily strain of knowing that as a black person you can be killed for simply being black: no hands in your pockets, no playing music, no sudden movements, no driving your car, no walking at night, no walking in the day, no turning onto this street, no entering this building, no standing your ground, no standing here, no standing there, no talking back, no playing with toy guns, no living while black. A similar idea is often expressed in philosophical contexts as well. For example, Paul Gilroy notes in Postcolonial Melancholia that “Racial difference obstructs empathy and makes ethnocentrism inescapable. It becomes impossible even to imagine what it is like to be somebody else” (Gilroy 2005, 63). Note the strength of the claims here briefy surveyed. What it’s like to occupy an experiential perspective different from one’s own is

Bridging the Divide

239

something that can’t be known (as in Lady Gaga’s question, “how could you know?”) or even imagined (as in the claim of unimaginability in Hamilton and even more explicitly in Gilroy). The unknowability and unimaginability are meant to be in principle, not merely in practice. I call this the Epistemic Inaccessibility claim. Epistemic Inaccessibility: Any experiential perspective vastly different from the one a person occupies is epistemically inaccessible to that person. While this claim may seem plausible on its face, and while it is certainly widely accepted as plausible, I’m not sure that we should be so easily convinced of its truth. In the remainder of this chapter, I will explore the case for Epistemic Inaccessibility in an effort to determine whether and to what extent we have good reason to believe it. Here it may help to put some of my cards on the table. As a general matter, I am an optimist about imagination. I am optimistic that it can play a role in justifying our beliefs, for example, and I am also optimistic that we can (at least in principle) imagine a wider range of things than is typically recognized (e.g., Kind 2016, Kind 2018, Kind 2020). But this chapter is not meant to make a general case for the optimist position, nor is it even meant to make a specifc case for optimism with respect to bridging experiential divides. I won’t try to establish the conclusion that a person can have epistemic access to experiential perspectives different from their own. Rather, I will attempt something considerably more modest – namely, to make room for the denial of pessimism, or at least, the particularly deep form of pessimism associated with the Epistemic Inaccessibility claim. The frst part of this consists in exploring the case that can be made for Epistemic Inaccessibility. As I will suggest, once this case is feshed out, it cannot stand up to close scrutiny. But if this is right, then a question immediately arises. Why would so many people, philosophers and non-philosophers alike, assume that Epistemic Inaccessibility is obvious, so obvious perhaps that it needn’t even be argued for?1 The answer is straightforward: A commitment to Epistemic Inaccessibility appears to be required by plausible principles about respect and humility. Thus, the second part of making room for the denial of pessimism consists in showing why this is mistaken. As I will suggest, the denial of Epistemic Inaccessibility does not entail that one is either disrespectful or inappropriately arrogant. Though the examples given above were drawn mainly from popular culture and public discourse, in what follows I’ll turn to the philosophical literature. I’ll start by considering arguments that might be adduced in support of Epistemic Inaccessibility. Though these arguments are often not made explicit, I think we can identify two different strands of

240 Amy Kind thought implicit in philosophical discussions of these issues. I call these the Epistemic Arrogance argument and the Too Big a Gulf argument. In the following two sections, I’ll consider these in turn. As I’ll suggest, neither of these arguments is successful in making the case for pessimism. Though this opens the door to a more optimistic approach, I conclude by refecting on the need to proceed with caution in this regard.

12.2 The Epistemic Arrogance Argument To start, it would be helpful to clarify how I’m understanding the notion of experiential perspective. On the one hand, we sometimes talk of individual experiential perspectives – what it’s like to be Christine Blasey Ford, or to be Alexander Hamilton, or to be Serena Williams. On the other hand, we sometimes talk of broader experiential perspectives – what it’s like to be the survivor of sexual assault, or to be a parent who has lost a child, or to be black in America. In talking the latter way, though one need not assume that the perspectives are monolithic, one nonetheless treats the experiential perspective as a broad and shareable type. It’s this latter, broad notion of experiential perspective that I’ll be working with in what follows. To many, the problem with optimism about bridging the divide between two types of vastly different experiential perspectives seems deeper than mere falsity. If we think as a general matter that experiential divides can be bridged, then we are likely to think in at least some particular cases that an experiential divide has been bridged. But this latter thought is taken to embody a certain kind of arrogance. Perhaps it’s even offensive. We saw a hint of this reaction in the lyrics from Lady Gaga, above: “I don’t wanna hear nothing from you.” A more detailed development of this concern can be found in the philosophical literature in connection with what I’m calling the Epistemic Arrogance argument. The Epistemic Arrogance argument has its root in the idea that the attempt to imagine someone else, particularly someone in a vastly different experiential perspective, is fraught with danger. Sometimes this danger is cast as a kind of insensitivity. Consider, for example, this passage from Laurence Thomas: If a woman has been raped, it is clear that the last thing in the world that a […] man should say to her is ‘I can imagine how you feel.’ […] Few actions could be more insensitive to victims of rape than a man’s supposition that via a feat of imagination he can get a grip on the pain that a female victim of rape has experienced. (Thomas 1998, 361) Thomas also casts this danger more specifcally as a kind of moral failing. If someone who is not a Holocaust survivor were to think they could

Bridging the Divide

241

grasp the experiences of a Holocaust survivor by way of “rational imaginative role-taking,” this would, he says, be “moral hubris of the worst sort” (Thomas 1998, 360). For Iris Marion Young, the moral failing is cast as a failing of humility. On her view, when one shows moral humility, one starts with the assumption that one cannot see things from the other’s perspective and waits to learn by listening to the other person to what extent they have had similar experiences. If I assume that there are aspects of where the other person is coming from that I do not understand, I will be more likely to be open to listening to the specifc expression of their experience, interests, and claims. Indeed, one might say that this is what listening to a person means. (Young 1997, 49) In contrast, when one starts with the assumption that one does have imaginative access to a different experiential perspective, one does not show this kind of moral humility. Starting with this kind of assumption, on her view, is more likely to impede communication than to facilitate it, for you will be disinclined to listen to the other person and will likely become defensive if they confront you about any misunderstandings (see Young 1997, 48–9). In what follows, I’ll use “epistemic arrogance” as a relatively broad catch-all that also captures issues of insensitivity, disrespect, and lack of moral humility. We can thus capture the sentiments expressed by Thomas and Young with something like the following argument: 1 2 3

The imaginative project of trying to imagine experiential perspectives vastly different from one’s own displays a certain kind of epistemic arrogance. This kind of epistemic arrogance is morally problematic. Thus, individuals should refrain from engaging in the imaginative project of trying to imagine experiential perspectives vastly different from their own.

Let’s take this schematization as a canonical statement of the Epistemic Arrogance argument. Later, I’ll return to the question of whether this is a good argument. For now, however, what’s important to note is that even if the argument is wholly successful, it doesn’t get to the conclusion that the pessimist is looking for. This argument tells us that attempting to cross experiential divides via imagination is something that we shouldn’t do, not something that we can’t do. As such, it doesn’t present us with an argument for Epistemic Inaccessibility. Perhaps we can tease out a related argument in the vicinity that is lurking behind these sorts of concerns about epistemic arrogance. We can see it expressed, perhaps, in this passage from Elizabeth Spelman:

242  Amy Kind If we reflect a moment on why imagination may seem necessary in this situation – where a member of an oppressor group is trying to learn about a member of the oppressed group – we can see why it is also dangerous […] If we already knew a lot about each other, I wouldn’t have to use my imagination in this way to enter into your world, any more than I would have to, to understand my own. But if I only rely on my imagination to think about you and your world, I’ll never come to know you and it. (Spelman 1988, 179) Like Thomas and Young, Spelman takes this kind of imaginative project, i.e., the project of attempting to cross experiential divides via imagination, to be dangerous. And I take it that she would endorse the Epistemic Arrogance argument outlined above. But unlike the above passages from Thomas and Young, this passage from Spelman moves from a claim about this danger to a claim that sounds closer to an expression of Epistemic Inaccessibility. In claiming that someone engaged in this imaginative project will never come to know the experiential perspective of another, Spelman seems to be endorsing the claim that this kind of imaginative project cannot, in principle, be successful. This itself is not Epistemic Inaccessibility, as it leaves open the possibility that there are other ways to learn about a different experiential perspective. But can we get there from what’s already been said? Let’s leave aside the question of whether Spelman herself would endorse Epistemic Inaccessibility, as sorting this out would require a longer discussion of her work than can be accomplished here. Instead, let’s just explore how we might use the considerations Spelman has offered to mount a defense of Epistemic Inaccessibility. As best as I can determine it, the argument would have to go something like this: 1 Imagination can only enable us to understand an experiential perspective that we already know a lot about. 2 Experiential perspectives vastly different from our own are not ones we already know a lot about. 3 Thus, imagination cannot enable us to understand experiential perspectives vastly different from our own. 4 But since those perspectives are vastly different from our own, there is no other way for us to gain this knowledge. 5 Thus, experiential perspectives vastly different from one’s own are epistemically inaccessible. But note something interesting here: Considerations about epistemic arrogance don’t seem to be doing any work in this argument. Rather, what’s doing the work seems to be the thought that when it comes to some differences between experiential perspectives, there is just too big

Bridging the Divide

243

a divide to cross imaginatively. But that’s a different argument – in fact, it’s the kind of argument that, as I mentioned earlier, I call the Too Big a Gulf argument. I thus propose to proceed as follows. In the next section, I’ll look more closely at the Too Big a Gulf argument. As I’ll suggest, that argument cannot succeed in establishing the kind of deep pessimism that motivates Epistemic Inaccessibility. In the course of that discussion, however, the kinds of concerns that motivate the Epistemic Arrogance argument will return to the fore. Though I do not think these concerns give us reason for pessimism, they do give us good reason to worry that the denial of Epistemic Inaccessibility might confict with other values that are important to us. Thus, in the fnal section of this chapter, I will attempt to address those concerns.

12.3 The Too Big a Gulf Argument One clear expression of the Too Big a Gulf argument comes in Thomas Nagel’s famous paper, “What Is It Like to Be a Bat?” (Nagel 1974). Though both humans and bats are mammals, our species are very different from one another in all sorts of ways. Some of these ways are fairly superfcial, but some are considerably more fundamental. For example, while humans navigate the world primarily by way of our senses of sight and hearing, bats navigate the world by way of their sense of echolocation. As Nagel notes, we can’t really experience what a bat does when it is using its echolocation to navigate the world. 2 Given how central one’s sensory capacities are to one’s experiential perspective of the world, Nagel takes this difference to be of critical importance. The fact that we can’t have this kind of experience, or even anything close to it, leads Nagel to conclude that the bat’s experiential perspective is also closed off to us in imagination. The problem, he says, is that our own experiential resources on this matter are too impoverished. We don’t have what we need to latch on in imagination to the bat’s experiential perspective: Our own experience provides the basic material for our imagination, whose range is therefore limited. It will not help to try to imagine that one has webbing on one’s arms, which enables one to fy around at dusk and dawn catching insects in one’s mouth; that one has very poor vision, and perceives the surrounding world by a system of refected high-frequency sound signals; and that one spends the day hanging upside down by one’s feet in an attic. In so far as I can imagine this (which is not very far), it tells me only what it would be like for me to behave as a bat behaves. But that is not the question. I want to know what it is like for a bat to be a bat. Yet if I try to imagine this, I am restricted to the resources of my own mind,

244

Amy Kind and those resources are inadequate to the task. I cannot perform it either by imagining additions to my present experience, or by imagining segments gradually subtracted from it, or by imagining some combination of additions, subtractions, and modifcations. (Nagel 1974, 439)

Thus, since we can’t experience what it’s like to be a bat, and we can’t imagine what it’s like to be a bat, we have no epistemic access to what it’s like to be a bat. There is just too big a gulf between the bat and the human. Now the gulf between a human and a bat is wider than the gulf between any one human and any other human, no matter how different the experiences the two humans have or how different the experiential perspectives the two humans occupy. But Nagel is also explicit that the same kind of unbridgeable gulf that exists between the human and the bat exists in some cases between two different humans. In particular, he suggests we consider the imaginative gulf between someone who has the capacities for both hearing and seeing and someone who lacks these capacities: The problem is not confned to exotic cases, however, for it exists between one person and another. The subjective character of the experience of a person deaf and blind from birth is not accessible to me, for example, nor presumably is mine to him. (Nagel 1974, 440) In this way, Nagel seems to endorse Epistemic Inaccessibility. But some of his further comments muddy the waters. For Nagel explicitly cautions us against overgeneralizing this point: “It is often possible to take up a point of view other than one’s own, so the comprehension of such facts is not limited to one’s own case” (Nagel 1974, 441–2). To sort this out, it will be helpful to be clear about what the proponent of Epistemic Inaccessibility is (and is not) committed to. As stated, the principle that we are working with restricts the epistemic inaccessibility to perspectives that are vastly different from one’s own. One could, however, make an even stronger claim than this: Unrestricted Epistemic Inaccessibility: All experiential perspectives different from the one a person occupies are epistemically inaccessible to that person. It’s this stronger principle that Nagel’s remarks suggest he would reject. To my mind, the fact that he limits his conclusion this way is often underappreciated. Yes, in discussing this chapter with students in philosophy of mind classes we might note how important it was for Nagel to

Bridging the Divide

245

fnd just the right sort of example from the animal kingdom to make his point. If he had chosen an example from too far down the phylogenetic chain, his claims might have been met with skepticism that the animal in question even had an experiential perspective. If he had chosen an example from too far up the phylogenetic chain, his claims might have been met with skepticism that the animal in question had a suffciently different experiential perspective such that access to it was ruled out. But in my own teaching experience, even after this point is made and seemingly appreciated, students are still very quick to leap from Nagel’s claim that what it’s like to be a bat is inaccessible to anyone (i.e., to anyone who is not a bat) to the claim that what it’s like to be anyone other than oneself is inaccessible. More generally, there often seems to be a kind of amnesia about the restricted nature of Nagel’s claim when the issue of epistemic accessibility is directly discussed. Why might someone like Nagel choose to endorse the more restrictive Epistemic Inaccessibility principle over Unrestricted Epistemic Inaccessibility? Here it helps to return to a point we saw earlier. For Nagel, imagination works by operating on material provided by experience. Imagination performs various kinds of transformations on this material – combining it with other material, adding or subtracting elements, and modifying it in various other ways. This means that if we want to understand experiential perspectives different from our own, we have to fnd some way to leverage the experiences that we’ve had to get to experiences that we haven’t. In my own previous work, I’ve referred to this process as imaginative scaffolding (see Kind 2020, 137). We scaffold out from experiences we’ve had to experiences that we haven’t. On Nagel’s view, when the experiences we haven’t had are similar to the ones we have had, this imaginative work is tractable. But, as he notes, “The more different from oneself the other experiencer is, the less success one can expect with this enterprise” (Nagel 1974, 442) – and, as we’ve seen, he thinks there are some cases in which success is impossible. Presumably, this is also what Spelman has in mind when she claims that imagination can only enable us to understand an experiential perspective that we already know a lot about, i.e., it’s only in the cases of perspectives that we know a lot about that we have the materials needed for successful imaginative scaffolding. On this line of reasoning, then, what matters for imaginative access is how different we are from each other, how big a gulf there is. This line of reasoning can be roughly schematized as follows, a schematization that I’ll take as the canonical statement of the Too Big a Gulf argument: 1 2

One cannot have direct experiential access to an experiential perspective other than one’s own. Thus, to have epistemic access to an experiential perspective other than one’s own, one must do so by way of imaginative access.

246 Amy Kind 3 4 5 6

To achieve imaginative access to an experiential perspective other than one’s own, one must perform various imaginative operations on material provided by past experiences. When an experiential perspective is vastly different from one’s own, one’s past experiences do not provide suffcient material for this imaginative work. Thus, one cannot imaginatively access an experiential perspective vastly different from one’s own. Thus, experiential perspectives vastly different from one’s own are epistemically inaccessible.

The key premise here seems to be premise 4. The basic thought seems to be that whatever kinds of past experiences one has had, they aren’t derived from the relevant experiential perspective, and since that experiential perspective is vastly different from one’s own, those past experiences can’t adequately support the imaginative work that needs to be done. Considerable weight is being put on the notion of “vastly different,” a notion that remains largely undeveloped. We know that for Nagel the experiential perspective of a bat is meant to be taken as an example of one that is vastly different from the experiential perspective of a human, and that the experiential perspective of a person born without the capacity for sight or hearing is meant to be taken as an example of one that is vastly different from the experiential perspective of a person with these capacities. But, to return to the examples from popular culture and public discourse offered at the start of the chapter, what about the experiential perspective of someone who has been sexually assaulted, or someone who is grieving the loss of a child, compared to the experiential perspective of someone who has not undergone the relevant experience? Or what about the experiential perspective of a black person in America compared to the experiential perspective of a white person in America? Though it’s not clear how Nagel would answer these questions, these do seem to be the sorts of examples that philosophers like Thomas and Young have in mind when they offer the Epistemic Arrogance argument and discuss matters relating to Epistemic Accessibility. More generally, it seems commonplace in discussion of these issues in both philosophical and public discourse to treat as unbridgeable the experiential divides owing to factors such as race, gender, ethnicity, socioeconomic status, and disability, alone or in combination, along with experiential divides owing to certain kinds of experiences such as trauma (sexual assault, grief, war).3 Though I won’t here attempt a principled delineation of what counts as “vastly different,” I’ll assume going forward that it is meant to capture at least these kinds of experiential divides. We’ll thus be working with what might be thought of as a more expansive interpretation of the notion of “vastly different.”

Bridging the Divide

247

Interestingly, despite the importance of premise 4 in the argument, very little is typically said on its behalf. Perhaps the main line of defense that is offered concerns considerations of embodiment. For example, in an insightful discussion of moral imagination, Catriona Mackenzie and Jackie Leach Scully argue that “imagination is fundamentally an embodied capacity of mind” and thus that our specifc forms of embodiment thereby place signifcant constraints on our abilities to “imaginatively put ourselves in the place of others” (Mackenzie and Scully 2007, 342; see also Clavel-Vázquez and Clavel Vázquez, ms). While their discussion is focused on considerations about disability, and about the limitations facing non-disabled people when they attempt to imagine what it’s like to be disabled, these considerations presumably extend to other factors relating to embodiment as well. Whether they would extend more broadly to all kinds of “vastly different” experiential perspectives would depend on the extent to which these differences depend on differences in embodiment. But while their focus is on embodiment, Mackenzie and Scully also suggest that there are further limitations on our capacities for imaginative projection, namely those arising from “our social and cultural context, specifc histories, relationships with others, and patterns of emotional response” (Mackenzie and Scully 2007, 344). It’s my sense that it’s considerations of roughly this sort – i.e., the idea that our imaginative capabilities are limited by our specifc circumstances, both bodily and cultural – that are generally taken to undergird premise 4. If what one can imagine is signifcantly limited by who one is, then we would have reason to think that experiential perspectives vastly different from one’s own are imaginatively out of reach.

12.4 Pushing Back Against the Pessimism The Too Big a Gulf argument strikes many as intuitively plausible, as do the considerations adduced in its favor. But despite this intuitive plausibility, I want to try to push back against it, in an effort to show how pessimism might reasonably be questioned. First, consider a widely shared attitude about literature and literary non-fction, namely, that one of its values is precisely to acquaint readers with a variety of experiential perspectives that they might not themselves occupy.4 One philosopher who has consistently argued for this view is Martha Nussbaum. As she’s suggested: “Narrative art has the power to make us see the lives of the different with more than a casual tourist’s interest—with involvement and sympathetic understanding” (Nussbaum 1997, 88). But it’s not just philosophers who take this kind of attitude toward literature and literary non-fction. Former President Barack Obama, refecting on his friendship with the novelist Marilynne Robinson, notes that in reading her books he was able to draw connections between the

248 Amy Kind people he was seeing everyday – the people whom he was shaking hands with and making speeches to – and his grandparents, who were from Kansas and ended up journeying all the way to Hawaii, but whose foundation had been set in a very similar setting. It was Robinson’s descriptions of interior life that enabled this. As he put it: I think that I found myself better able to imagine what’s going on in the lives of people throughout my presidency because of not just a specifc novel but the act of reading fction. It exercises those muscles, and I think that has been helpful. In the wake of the brutal murder of George Floyd in May 2020, just one in a long line of horrifc incidents in which unarmed black men and women have been killed by the police, articles containing lists of relevant book recommendations fooded news sites and social media feeds. While many of these books are non-fction works that are being put forward as a way to help white people better understand the workings of structural racism, many are memoirs or works of fction that are suggested as a way of helping whites better understand what it is like to be a black person in the United States. From recent books like The Hate U Give, Between the World and Me, and The Underground Railroad to classics like Native Son and The Invisible Man, the suggestion seems to be that it is worthwhile for whites to engage with these books that present experiential perspectives different from their own, and moreover, that it is worthwhile at least partly because it may enable a better understanding of these perspectives. 5 Philosopher Susan Brison provides an especially clear statement of this kind of sentiment in connection with her discussion of the autobiographical essay The Alchemy of Race and Rights by Patricia J. Williams. Describing this work as “groundbreaking,” Brison notes that Williams’ descriptions of the experiences of her great-great-grandmother, as well as her descriptions of the racism that she herself has experienced in ordinary life, provide us “imaginative access to what it’s like to be the victim of racial discrimination” (Brison 2003, 6).6 Elaine Scarry, who is generally skeptical about our ability to imagine other people, mentions Uncle Tom’s Cabin and Passage to India as playing a similar function of enabling imaginative access, though she thinks these are exceptional cases (Scarry 2002, 104–5). If literature can do this, then we should reject the Too Big a Gulf argument. This quotation from Brison – and more generally, the quotations we have just seen in our discussions of the value of literature and literary non-fction – typically describes this value in terms of imaginative access to an experiential perspective. I take it that imaginative access is meant to be a success term, i.e., the relevant imaginings are meant to succeed in providing epistemic access to the experiential perspective in question.

Bridging the Divide

249

But it may be important that these claims are not put in terms of knowing what it’s like to occupy that experiential perspective. This brings me to a second point. In resisting the case for pessimism and leaving the door open for people to have epistemic access to experiential perspectives vastly different from their own, one might try to differentiate epistemic access from knowledge. Epistemic access may well consist in a different kind of epistemic state. Perhaps it’s when the claim is put in terms of knowledge, e.g., when it’s claimed that someone who hasn’t been sexually assaulted can know what that experience is like, that the claim seems particularly callous or arrogant, and when it seems most open to the charge that it’s been overstated. It also opens the door to more general worries about skepticism that it might be better to avoid. But someone who wants to resist pessimism need not make the claim this way. One possibility, for example, would be to describe the relevant epistemic state in terms of understanding, where understanding is meant to be importantly different from knowledge. For example, while knowledge is typically taken to be all-or-nothing, understanding is thought to come in degrees (see, e.g., Kvanvig 2003, Elgin 2009). This allows for to the following line of resistance to pessimism: For someone to count as having epistemic access to an experiential perspective vastly different from their own, what matters is that they have deep or signifcant understanding of it, even if that understanding is not complete.7 Third, it’s important to note that Epistemic Inaccessibility, even in the restricted form that we’ve been considering, makes a very strong claim. What’s claimed is not just that it is hard to imaginatively cross certain kinds of experiential divides, or that we should be less confdent in our ability to imaginatively cross certain kinds of experiential divides. Rather, what’s claimed is that it’s in principle impossible to cross such experiential divides. Thus, one can resist Epistemic Inaccessibility without committing oneself to the claim that crossing such experiential divides is easy or commonplace. One might insist, in fact, that in many cases, the epistemic work is likely to be exceptionally diffcult.8 Once we recognize that distinction, I think we can come to see that many of the expressions of pessimism that we fnd in the literature are better understood as expressions of this weaker form of pessimism, namely, that actually bridging these epistemic divides is extraordinarily diffcult and likely quite a rare achievement. Consider, for example, Cliff Sosis’s interview with George Yancy as part of the What Is It Like to Be a Philosopher series. Throughout the interview, Yancy points to various failures of understanding across the racial divide, in particular, failures by white people to understand the experiences of black people. In response, Sosis worries that Yancy may “underestimate our ability to empathize with each other a bit […]. We don’t all have the same struggles, but we can understand the struggles of

250

Amy Kind

others. I mean, I’m not a bat.” Yancy’s long response is well worth reading in its entirety, as is the whole interview, but I’ll just quote the section of it that is especially important for our purposes: you don’t need to be a bat to fail to understand what it is like to be Black or a person of color. Being white in America will do the trick. […] Your question is a good one, but I don’t think that I’m underestimating the extent to which white people can’t or don’t empathize with Black people or people of color. Again, this might also be linked to the ways in which so much of our culture (visual or not) requires Black people and people of color to empathize with white people. This is because it is necessary for Black people, for example, to have a kind of dual cognitive skill where we are forced to understand what goes on within the white world and what goes on within our own worlds. White people, can, for the most part, avoid our world, avoid Black children’s literature (the very few books out there dealing with Black children and their lives), avoid serious Black characters playing serious roles in movies. I don’t think the imagination and intelligence of white people under white supremacy help them to empathize with Black people or people of color. White history has proven that; it isn’t just my pessimism. (Yancy 2016, my emphasis) In describing this empathetic understanding as something white people “can’t or don’t” do, Yancy seems to remain agnostic between the two forms of pessimism just distinguished. Moreover, in pointing to white supremacy as at least partly responsible for this failure, Yancy seems to leave open the possibility that things might in principle be different – if somehow we were to eradicate white supremacy, for example. In fact, the references to black literature and black flm suggest one thing that could be done in an effort to make things different (a point that underscores our above discussion of the value of literature). Yancy’s suggestion that black people are capable of achieving an understanding of white people suggests that this kind of gulf between experiential perspectives – that is, a racial gulf – is not the kind of gulf that Epistemic Inaccessibility should rule out. Granted, Yancy does offer reasons to think that the gulf is not a symmetrical one, reasons that show why white-to-black understanding should be treated differently from black-to-white understanding. This point is also familiar from discussions of W.E.B. Du Bois’s notion of double consciousness. (See also the discussion of the insider-outsider in Wylie 2003.) But even in this forceful expression of pessimism, the case made seems to favor it’s-extraordinarily-diffcult pessimism rather than it’s-impossible-pessimism. Another forceful expression of pessimism comes in a paper by Janine Jones addressing similar issues about white-to-black understanding.

Bridging the Divide

251

Jones focuses on the class of white people she calls “goodwill whites.” Whereas some white people who harm blacks know that they’re doing so and mean to be doing so, goodwill whites do not realize that this is what they’re doing. Jones argues that goodwill whites are often unable to articulate what they are, since they are unable to articulate their identity in terms of whiteness. They see race as “something that others possess. Whites are just ‘normal’.” As she goes on to argue: Whites’ inability to form the belief that they are white skews the nature of the relationships that exist between whites and blacks. It affects their ability to empathize because they are unable to import an ingredient essential to empathy: an appreciation of their own situation. (Jones 2004, 70)9 Jones speaks of inability here, but I don’t think she means this to be an inability in principle. Later in the paper this becomes clear when she takes up the question of whether it would be possible for a goodwill white person to “see into [the] heart” of a black person, even though they often don’t. In a passage addressed to a young black child, she answers this question in the affrmative: I would say that in some instances they may not be able to do so because they do not possess a retrievable source analogue to match your experience. Moreover, unable to appreciate either your situation or their own, they are unable to map a constructed source analogue of their experience onto yours. And yet in some cases I think they can see in your heart. They can empathize with you because as human beings living in a society with at least some important shared lived experiences and shared stories, they can either retrieve or construct experiences that map onto some of your experiences. (2004, 78) Jones’ description here of the process that a goodwill white person might undertake looks to be a form of imaginative scaffolding mentioned earlier. Though Jones’ essay undoubtedly puts her in the pessimistic camp, her pessimism too looks to be a form of it’s-extraordinarily-diffcult pessimism rather than of it’s-impossible-pessimism.10 To my mind, however, this former kind of pessimism is really a kind of optimism – or at least it gestures in the direction of optimism. If one comes to recognize that something thought to be impossible is in fact possible, even though it’s really, really hard, that might give one some reason for hope. I won’t here quibble over how exactly this position should be classifed. But insofar as I’m pushing back in this chapter against pessimism, my target is it’s-impossible-pessimism.

252

Amy Kind

So where does all this leave the Too Big a Gulf argument? In my view, the argument as it stands should be rejected. The problem, as our discussion has made clear, is with the fourth premise: 4

When an experiential perspective is vastly different from one’s own, one’s past experiences do not provide suffcient material for this imaginative work.

Given the intended sense of “vastly different,” one that takes the class of experiential perspectives thereby captured to be an expansive one, the premise is false. One option to salvage the argument, then, would be to adjust this premise. Perhaps one might fnd a different way to categorize the relevant class of experiential perspectives that are imaginatively inaccessible, or perhaps one could develop a less expansive understanding of what counts as “vastly different.” Either way, though some version of the Epistemic Inaccessibility principle might then be supported, it will not be a version that excludes, in principle, the possibility of one person having epistemic access to an experiential perspective different from her own with respect to race and gender, for example. Though it may turn out to be true that some experiential perspectives are indeed epistemically inaccessible to someone who does not occupy that perspective, these will be much fewer than has been thought, and it will not include many of the kinds of experiential divides that people have often claimed to be unbridgeable. Absent another kind of argument, one that does not simply rely on the kinds of considerations that allowed Nagel to draw his conclusions about the in-principle unbridgeable divide between humans and bats, we are not entitled to draw the same conclusion about inprinciple unbridgeable divides between humans occupying these kinds of experiential perspectives.

12.5 Epistemic Arrogance Reconsidered At this point, however, the concerns we looked at earlier in the context of discussing the Epistemic Arrogance argument return with even more force. The upshot of those considerations, recall, was that individuals should refrain from engaging in the imaginative project of trying to imagine experiential perspectives vastly different from their own; engaging in this project is taken to be morally problematic. If this is right, then successfully resisting pessimism is a kind of Pyrrhic victory for the optimist about imagination. Thus, in this section I want to push back against the Epistemic Arrogance argument. Though the attempt to imagine across experiential perspectives may indeed be fraught with danger, there are ways to do so while avoiding moral vice.

Bridging the Divide

253

Behind the Epistemic Arrogance argument are concerns about respect, or, perhaps better put, disrespect. Recall the points we saw above. When we engage imaginatively with someone’s experiential perspective rather than listening to them, we fail to treat them with respect. When we tell someone that we know how they feel rather than listening to them, we fail to treat them with respect. Moreover, to extend the points we saw above, when we take the deliverances of our own imaginings to outweigh what others are telling us, we also fail to treat them with respect. Thomas puts this last point especially clearly, noting the wrongness inherent in “discount[ing] the feelings and experiences of persons in diminished social category groups simply because their articulation of matters does not resonate with one’s imaginative-take on their experiences” (Thomas 1998, 375). To avoid this wrongness, Thomas advises that we adopt an attitude of moral deference, an attitude that requires that we really listen to the stories that others tell – where really listening requires not only that we pay close attention to the nuances of what’s being said and to the emotions being expressed, but that we also understand the vulnerabilities in play and are appropriately moved on account of all of these things. All of these points, however, seem to presuppose a certain incompatibility between imagining and listening. The assumption seems to be something like the following: When we engage with someone else, we can either listen to them or we can engage imaginatively with them – where the “or” here is meant to be exclusive. But this strikes me as a false dichotomy. Why can’t we do both? And moreover, why shouldn’t we do both? In particular, suppose we were to do both as part of an ongoing practice where what we learn from listening to someone else – from really listening to them, in Thomas’ sense – helps inform our imagination. Perhaps it’s precisely what we learn from listening that enables us to make sure that our process of imaginative scaffolding is an effective one. In fact, I’m inclined to think that part of what makes someone a good imaginer – what makes them able to have broader and better imaginative access – is that they are a good listener. As a general matter, imagining that operates in a vacuum is unlikely to be helpful to us epistemically. We know this to be true when we’re putting imagination to epistemic use in a wide variety of other contexts – whether we’re trying to plan a vacation or problem-solve or make an important decision. So it’s no surprise that it’s true in this context of epistemic use as well. Successful uses of imagination will be informed by the world, informed by what we’re hearing, informed by what we’re seeing, and informed by what others are telling us they are hearing and seeing. Sometimes, as Thomas notes, there will be conficts between what others are telling us and what we imagine. And, as he says, it would be problematic were we to discount what they are telling us simply because

254

Amy Kind

their articulation of matters does not resonate with our imaginative take on their experiences. But I take there to be some importance to the fact that this rough principle is put in terms of simply because. Thomas thus leaves open the possibility that there may be times when, given other factors that may be in play, our imaginings do allow us to discount what others are telling us. Determining whether, and if so when, such discounting might be appropriate is an extremely complicated matter. But we might recognize these complications, and recognize that there are many hard questions here unanswered, without taking these facts to entail that we have no imaginative access to the experiential perspectives of other people.11 Clearly there are important insights behind the Epistemic Arrogance argument. Moral deference and humility are important. Imagining cannot be a substitute for listening. We need to listen to other people and to take seriously what they are saying – if we don’t, our imaginings are likely to be inaccurate, uninformative, and of little value. We need to be humble about what our imaginings are telling us. But none of this suggests any in-principle limits on imagination. To my mind, the main upshot of the Epistemic Arrogance argument is that we must proceed with caution – in fact, with extreme caution. This general upshot can be made more specifc in various ways. Let me give two examples. First, even if there do turn out to be cases in which we can imagine an experiential perspective different from our own, cases in which it turns out that we do know what it’s like to have an experience that we haven’t had, it doesn’t mean that it will be acceptable in those cases to say to someone who has had the relevant experience: “I know how you feel.” There are all sorts of reasons that, in many circumstances, this is an inappropriate response.12 Indeed, this is often a problematic response even when you do share someone’s experiential perspective or when you have been through the same type of experience that they have been through. To say something like this may seem to rob the person of their dignity and of their sense of the uniqueness of their own experience. Moreover, saying something like “I know how you feel” typically functions to shift the conversation. Now it’s about you and your feelings rather than about them and their feelings. Thus, this kind of claim might be seen as a kind of conversational narcissism. But here again, we can recognize the importance of not saying this kind of thing, of listening, without denying the capabilities and reach of imagination. Second, when one achieves understanding of a certain experiential perspective via imaginative projection, one might not have the same license to act as someone who has this understanding in virtue of occupying the experiential perspective. Consider the following analogy. Suppose a friend has an important insight about the current political situation and tells you about it. Later, when the two of you are with some

Bridging the Divide

255

other friends, the relevant topic comes up, and you jump in to share your friend’s insightful point instead of letting her share it herself. Even if you give her credit for the insight in the retelling, it seems perfectly reasonable for your friend to be annoyed. It was her point, and she wanted to be the one to share it. The fact that you have just as good an understanding of the point doesn’t change things. A similar dynamic will be in play with respect to understanding of an experiential perspective that you’ve gained by imaginative projection. Given that it’s not your experiential perspective, there will be various limits on your capacity to act on that understanding – representing that perspective in conversation, using certain pieces of language, and so on. Finally, it’s worth drawing out one last moral from our discussion of the Epistemic Arrogance argument. As we have seen, defenders of this argument point to problems inherent in the imaginative project of trying to imagine experiential perspectives vastly different from one’s own. As we have also seen, they suggest that we would be better off listening to what people with vastly different experiential perspectives have to say. I have already pointed out that these projects need not be seen as oppositional to one another. But there’s a further important point here to be made. The suggestion that we can learn about experiential perspectives different from our own by way of listening to what people tell us is itself reason to reject Epistemic Inaccessibility. To return to a passage quoted earlier, when Spelman says that “if I only rely on my imagination to think about you and your world, I’ll never come to know you and it,” this seems to imply that there is some way that I can come to know about you and your world – even if that way is not by imagination (or not only by imagination). These gulfs aren’t too big to cross. Ironically, perhaps, consideration of the Epistemic Arrogance argument may thus give us a different sort of way to push back against pessimism.

12.6 Concluding Remarks My goal in this chapter has been to make room for the denial of pessimism, or, at least, the particularly deep form of pessimism that sees an impossibility in bridging experiential divides. As I have suggested, the two arguments considered – the Epistemic Arrogance argument and the Too Big a Gulf argument – do not provide adequate reason for us to think Epistemic Inaccessibility is true. Even if the sorts of epistemic divides that we’ve been considering might be extraordinarily diffcult to cross, we have not yet been given any reason to think that they are in principle unbridgeable. Perhaps there are additional arguments that might be given, but in the absence of any such arguments, there is no special reason to take pessimism as the default assumption, and the way seems to be cleared for a defense of optimism.

256

Amy Kind

Of course, even with the way cleared for this defense of optimism, considerable work would need to be done to mount it. Much more needs to be said about how these imaginative projects might work and about how, precisely, the imaginative scaffolding might be built. Moreover, there are all sorts of good reasons to think that the scaffolding process will be diffcult. For example, as noted above, our own situatedness – bodily, societally, and culturally – affects the experiences that we have to draw on. And even in cases where the situatedness of one experiential perspective shares much in common with the situatedness of another, there will still be all sorts of important differences, many of which are diffcult to tease out. (For discussion, see Thomas 1998, 365.) Several other hard questions still remain unaddressed. One was noted in passing earlier: What, exactly, is the epistemic state yielded via imaginative access? Is it knowledge, understanding, or something else entirely? Another especially hard question that I have not done justice to in this chapter concerns what, exactly, we are trying to access in imagination. As I noted at the start of Section II, the notion of experiential perspective in play throughout my discussion is one regarding broad types of experiential perspectives: What it’s like to be the survivor of sexual assault, or to be a parent who has lost a child, or to be black in America. As I also noted, these perspectives are not monolithic ones. Not everyone who has been sexually assaulted, or every grieving parent, or every black American shares exactly the same perspective. Nor does every bat, for that matter – or at least, so I assume. I’ve treated the question before us as if it’s a question about broad types largely because that is how it’s often treated in the literature.13 But whether this is the right way to pursue these imaginative projects needs further scrutiny, as do the related questions of what exactly this broad type is and what it would mean to access it. Any defense of optimism would need to grapple with all of these issues.14

Notes 1 See, e.g., Clavel-Vázquez and Clavel Vázquez (ms): “That it is not possible to fully know what it is like being someone else is an uncontroversial philosophical claim.” 2 For the purposes of this chapter, I’m granting this assumption. But one might question this by pointing to blind people who have developed echolocatory skills, for example. The possibility of virtual reality simulations of bat-hood would also be relevant. One might also question how different echolocation is from audition. See Allen-Hermanson (2019). 3 See, for example, L.A. Paul’s claim: If you are a man who has grown up and always lived in a rich Western country, you cannot know what it is like to be an impoverished woman living in Ethiopia, and if she has never left her village, she cannot know what it is like to be a man like you. If you are a white businessman living

Bridging the Divide

257

in San Francisco in 2013 you cannot know what it was like to be a black man involved in the Jamaican rebellion in 1760, hiding out in the forest in the dead of night while British troops comb the island trying to hunt you down, or know what it was like to be a slave in the American south. (Paul 2014, 7) 4 I am focusing only on literature and literary non-fction, but these considerations might be extended to other kinds of works ranging from music and art to virtual reality simulations and video games. 5 For example, in “A Year of Anti-Racism Work,” a week-by-week list compiled by Michelle Panchuk, people are tasked not just to check out a book written by a black woman, or a book by and about Asian-Americans, but also to “make sure you are empathetically engaged with the book you are reading.” 6 Interestingly, Brison’s own book Aftermath has also been described as one that provides imaginative access to an experiential perspective that many readers may not share, namely, that of the experience of a rape survivor. See the blurb on the publisher’s website at https://press.princeton.edu/books/ paperback/9780691115702/aftermath. 7 Moreover, while knowledge is generally viewed as factive, some epistemologists have argued that understanding is not factive. Scientists might increase their understanding when they move from one false theory to another, better theory – even when that second theory is still false. See Elgin 2009 for an argument to this effect. That understanding need not be factive seems directly relevant to the debate about epistemic arrogance. 8 Perhaps one way to raise a worry for the “impossible in principle” claim would be to say something like this: Someone who isn’t a survivor of sexual assault could come to know what it’s like to be such a survivor if they were to be sexually assaulted; thus, it’s not impossible in principle for them to come have this knowledge. This is not the kind of worry I mean to be raising. My worry about the “impossible in principle” claim of Epistemic Inaccessibility arises even if we keep fxed the fact that the individual in question does not occupy the relevant experiential perspective. 9 Though the point is put in terms of “empathy” in this quotation, Jones elsewhere talks of “empathetic understanding,” and I think her overall discussion suggests that she has something very similar in mind to the kind of epistemic access we’ve been talking about. 10 Likewise for Mackenzie and Scully, discussed earlier (see Section III). In discussing whether someone with “normal” hands could imaginatively project themselves into the perspective of someone with ectrodactyly (a genetic condition that results in missing fngers and toes), they note that “it is likely to be extremely diffcult” (Mackenzie and Scully 2007, 344). As should be clear, this is less pessimistic than Epistemic Inaccessibility. 11 Analogous issues arise, I think, in discussions within feminist standpoint epistemology about the epistemic privilege thesis. As many feminist theorists have pointed out, the claim of epistemic privilege here does not mean that such knowledge is automatic. Alison Wylie, for example, explicitly argues that feminist standpoint epistemology “must not be aligned with a thesis of automatic epistemic privilege; standpoint theorists cannot claim that those who occupy particular standpoints […] automatically know more, or know better, by virtue of their social, political location” (Wylie 2003, 28; see also McKinnon 2015). 12 That’s not to say it will always be a problematic response. There might well be times when an acknowledgment that someone else understands how you

258

Amy Kind

feel may be both helpful and reassuring. Along these lines, Adam Smith remarks that “nothing pleases us more than to observe in other men a fellow-feeling with all the emotions of our own breast” (Smith 1759/2002, 17). Thanks to Adrienne Martin for helping me to see this point. 13 See, e.g., Nagel’s claim that “The point of view in question is not one accessible only to a single individual. Rather it is a type” (Nagel 1974, 441). Or see Thomas’s discussion of moral deference where he talks of perspectives in terms of “social category groups” (see, e.g., 1998, 364). 14 I am grateful to my students in the fall 2016 and fall 2017 iterations of my “Experience” course for sparking many of the ideas underlying this chapter. These ideas were frst publicly aired in a talk called “Imagining Others” at the Claremont McKenna College Athenaeum in 2018. I subsequently presented this work in a version of the paper much closer to the present one at the Southern Society for Philosophy and Psychology in San Antonio, the University of Fribourg, California State University Long Beach, Institut Jean Nicod, and the London Aesthetics Forum. I am grateful to those audiences for their questions and helpful feedback. In preparing the fnal version of this chapter, I benefted greatly from comments by Christopher Badura, Peter Kung, Frank Menetrez, and Nick Wiltsher.

References Allen-Hermanson, Sean. 2019. “So that’s What It’s Like?” In Kristin Andrews and Jacob Beck, eds., The Routledge Handbook of Philosophy of Animal Minds. New York: Routledge, 157–68. Brison, Susan. 2003. Aftermath: Violence and the Remaking of a Self. Princeton, NJ: Princeton University Press. Clavel-Vázquez, Adriana, and Clavel Vázquez, María Jimena. ms. “Embodied Imagination.” Unpublished. Elgin, Catherine. 2009. “In Understanding Factive?” In Duncan Pritchard, Allan Miller, and Adrian Hadock, eds., Epistemic Value. Oxford: Oxford University Press, 322–30. Gilroy, Paul. 2005. Postcolonial Melancholia. New York: Columbia University Press. Jones, Janine. 2004. “The Impairment of Empathy in Goodwill Whites for African Americans.” In George Yancy, ed., What White Looks Like. New York: Routledge, 65–86. Kind, Amy. 2016. “Imagining Under Constraints.” In Amy Kind and Peter Kung, eds., Knowledge Through Imagination. Oxford: Oxford University Press, 145–59. Kind, Amy. 2018. “How Imagination Gives Rise to Knowledge.” In Fabian Dorsch and Fiona Macpherson, eds., Perceptual Memory and Perceptual Imagination. Oxford: Oxford University Press, 227–46. Kind, Amy. 2020. “What Imagination Teaches.” In Enoch Lambert and John Schwenkler, eds., Becoming Someone New: Essays on Transformative Experience, Choice, and Change. Oxford: Oxford University Press, 133–46. Kvanvig, John. 2003. The Value of Knowledge and the Pursuit of Understanding. Cambridge: Cambridge University Press. Mackenzie, Catriona and Jackie Leach Scully. 2007. “Moral Imagination, Disability and Embodiment.” Journal of Applied Philosophy 24 (4): 335–51.

Bridging the Divide

259

McKinnon, Rachel. 2015. “Trans*formative Experiences.” Res Philosophica 92 (2): 419–40. Nagel, Thomas. 1974. “What Is It Like to Be a Bat?” The Philosophical Review 83 (4): 435–50. Nussbaum, Martha. 1997. Cultivating Humanity: A Classical Defense of Reform in Liberal Education. Cambridge, MA: Harvard University Press. Paul, L.A. 2014. Transformative Experience. New York: Oxford University Press. Rankine, Claudia. 2015. “The Condition of Black Life is One of Mourning.” The New York Times. Available at https://www.nytimes.com/2015/06/22/ magazine/the-condition-of-black-life-is-one-of-mourning.html. Scarry, Elaine. 2002. “The Diffculty of Imagining Other People.” In Martha Nussbaum and Joshua Cohen, eds., For Love of Country? Boston, MA: Beacon Press, 98–110. Smith, Adam. 1759/2002. Theory of Moral Sentiments. Edited by David D. Raphael and Knud Haakonssen. Cambridge: Cambridge University Press. Spelman, Elizabeth. 1988. Inessential Woman. Boston, MA: Beacon Press. Thomas, Laurence. 1998. “Moral Deference.” In Cynthia Willet, ed., Theorizing Multiculturalism. Oxford: Blackwell Publishers, 359–81. Wylie, Allison. 2003. “Why Standpoint Matters.” In Robert Figueroa and Sandra Harding, eds., Science and Other Cultures. New York: Routledge, 26–48. Yancy, George. 2016. Interview by Cliff Sosis: What it is Like to Be a Philosopher. Available at http://www.whatisitliketobeaphilosopher.com/george-yancy Young, Iris Marion. 1997. Intersecting Voices. Princeton, NJ: Princeton University Press.

13 On Imagining Being Someone Else Julia Langkau

13.1 Perspective-Taking and Its Relatives Perspective-taking can and often does fail, and we sometimes learn about its failure when our reactions are inappropriate or our predictions are wrong. For instance, suppose I thought I understood how my friend is feeling after her favorite hockey team lost an important game, but she tells me I don’t. Or suppose I tried hard to understand what it is like for another friend to undergo certain relationship struggles, yet I end up being surprised by his decision to stay with his partner. The reason for my failure could be psychological in nature: it could be a matter of my ability to take over the other person’s perspective. Maybe I am not very good at it, or maybe we are generally not very good at it because it is hard. But the reason for my failure could also be epistemic in nature: it could be a matter of the explicit information I get, of how I interpret certain behavior, or of the standards that apply to a particular situation. For instance, my second friend could have concealed some crucial facts about his relationship or may have exaggerated out of temporary anger. My frst friend could have very high standards concerning what counts as understanding her, and what I said to her may have been in the ballpark but not exactly right. Finally, the reason for my failure could be that I’m after something impossible. Throughout his work, Peter Goldie expresses doubts concerning the role perspective-taking or empathy should play in understanding the other and predicting their actions (Goldie 2002, 2003, 2005, 2011). In his paper ‘Anti-Empathy’ (2011), Goldie claims to fnally be able to identify the problem with perspective-taking: it is not ‘a contingent one, concerned only with the limits to our powers of imagination’. Rather, Goldie argues that the problem is a conceptual one: we can never fully grasp what another person is feeling, thinking, or deciding to do from their perspective. Only a subject themselves can take the relevant stance toward their own mental states (Goldie 2011, 302–3). In order to understand Goldie’s claim and argument, we have to distinguish perspective-taking from similar activities or states. The relevant kind of perspective-taking is understood as involving a conscious, often intentional imaginative act. This distinguishes it from lower-level

On Imagining Being Someone Else

261

empathy such as emotional contagion as well as from personal distress. In emotional contagion, the subject’s emotional state is caused by another subject’s emotional state, but the subject may not be aware of this causal relation (see, e.g., Goldman 2006). Personal distress is caused by another person and their distress, but refers to a self-centered state of mind (see, e.g., Stueber 2016, 369), in which the subject focuses on their own experience and as a result is less likely to engage in helping behavior. While empathy is sometimes understood to include pro-social motivation and helping behavior (e.g., Batson 2009), for our purposes, the relevant aspects are the mental phenomena. Importantly, perspective-taking can involve all kinds of mental states such as thoughts, beliefs, intentions, and emotions (see, e.g., Goldman 2006).1 Goldie argues for the advantages of taking an external, yet still experiential, sympathetic perspective (e.g., in Goldie 2005). Sympathetic imagining explicitly allows us to take into account the differences between ourselves and the other and to evaluate the other’s perspective from our own perspective. For instance, when trying to understand my frst friend, I can take into account that her sadness and frustration about the lost game usually only last for a couple of days – something she’s not aware of while being in that state. In philosophy as well as psychology, it is often assumed that perspective-taking can take two different forms. First, we can take over the perspective of another by imaginatively putting ourselves in their situation. I can try to imagine what I would feel like if my favorite hockey team lost an important game or what it would be like for me to undergo certain relationship struggles. Second, we can take over the perspective of another by trying to imagine what it would be like for them, not for us, to be in a certain situation. I can try to imagine what it would be like for my friend after her favorite hockey team lost an important game, or what it would be like for my friend to undergo certain relationship struggles. Goldie (2011) calls the frst kind ‘in-his-shoes perspective-shifting’ (‘in-their-shoes perspective-shifting’ in what follows) and the second kind ‘empathetic perspective-shifting’. In-their-shoes perspectiveshifting is defned as follows: […] consciously and intentionally shifting your perspective in order to imagine what thoughts, feelings, decisions, and so on you would arrive at if you were in the other’s circumstances. (Goldie 2011, 302) Goldie defnes empathetic perspective-shifting as follows: […] consciously and intentionally shifting your perspective in order to imagine being the other person, and thereby sharing […] his or her thoughts, feelings, decisions, and other aspects of their psychology. (Goldie 2011, 302)

262

Julia Langkau

Goldie thinks the difference between these kinds of perspective-taking lies in the content of the imaginative project, more precisely: in who is supposed to do the thinking in the imaginative project. In the frst case, I am supposed to imagine myself in the other person’s situation, i.e., to imagine the thinking to be done by myself. In the second case, I am supposed to imagine being another person in a certain situation, i.e., to imagine the thinking to be done by the other person, whom I imagine myself to be. An intuitive way of understanding the distinction, according to Goldie, is the following: while I can imagine myself meeting the person I’m empathizing with in the case of in-their-shoes perspective-shifting, I cannot imagine so in the case of empathetic perspective-shifting, because I am imagining being the other person (Goldie 2011, 306; originally Wollheim 1984, 75). But Goldie has suggested yet another way of thinking about the two kinds of perspectivetaking, namely in terms of two aspects – frst, a narrative: the ‘sequence of events which is the content of the imaginative process’ (Goldie 2002, 98); and second, the target’s characterization, which includes intellectual traits and abilities as well as dispositions to experience emotions. In the case of in-their-shoes perspective-shifting, all the subject has to do is imagine experiencing the narrative, i.e., having the relevant experiences, responding to them emotionally, deliberating, deciding what to do, etc. (Goldie 2002, 98). Empathetic perspective-shifting requires that the subject also takes into account the target’s characterization (i.e., their psychological make-up) in order to imagine having the relevant experiences given their intellectual traits, abilities, and emotional dispositions. It is commonly held that if we have suffcient information about their situation and characterization, we can get to know a target’s thoughts, beliefs, intentions, emotions, etc., through empathetic perspective-shifting. If we have information about their situation but lack suffcient information about their characterization, in-their-shoes perspective-shifting can be helpful. However, we can’t always rely on it: if the person is much different from us, in-their-shoes perspective-shifting is misleading (see Batson 2009, 276; Stueber 2016, 373–4). Goldie (2011) argues that empathetic perspective-shifting is impossible for conceptual reasons. In the following section, I will present his argument. In Section 13.3, I will argue that skepticism toward empathetic perspective-shifting leads to skepticism toward in-their-shoes perspectiveshifting, but that optimism toward the second justifes optimism with respect to the frst. I will conclude that we can distinguish two different aims in perspective-taking, but that both kinds of perspective-taking face the same epistemic challenges, and that the underlying process can be the same. In Section 13.4, I will show how an epistemically ideal context can pose psychological challenges, and how an epistemically less ideal context can make it easier to imagine being someone else.

On Imagining Being Someone Else

263

13.2 Goldie’s Argument against Empathetic Perspective-Shifting Goldie’s argument against empathetic perspective-shifting consists in two main steps. First, he claims that the conceptual impossibility of empathetic perspective-shifting can easily be overlooked if we focus on simple cases. In simple cases, dissimilarity between the characterization of subject and target isn’t relevant. All we usually need in simple cases is a thin notion of rational agency, one that is shared between the subject and the target, assuming that they are both minimally rational. In such cases, it doesn’t matter ‘what sort of person’ is doing the thinking, deliberating, etc. For instance (see Goldie 2011, 307), assume a subject B wants to have a sandwich for lunch and believes that a certain restaurant is the only nearby place to serve sandwiches. In taking over B’s perspective, A imagines being hungry and imagines believing that there is only one place nearby that serves sandwiches. No matter what kind of person B is, if they are minimally rational, A will be able to predict their decision (of course, B also has to have the intention to go to a restaurant, and has to be able to afford eating at a restaurant, and A has to have this information about B’s situation). However, we cannot generalize from simple cases to cases in which the psychological make-up of a person matters. The problem becomes relevant in cases where at least one of the following conditions concerning a subject A and a target B occurs (Goldie 2011, 308): 1 2 3 4

There are relevant differences in the psychological dispositions between A and B. There are relevant non-rational infuences on B’s psychological make-up or decision-making process. There is signifcant confusion in B’s psychological make-up. B is faced with a psychological confict, such as having to make a choice between two or more alternatives where it is not clear to B which alternative is to be preferred.

Cases involving one or more of these points reveal the conceptual impossibility of empathetic perspective-shifting, according to Goldie. Here, a more full-blooded notion of agency is relevant for understanding a person’s situation and predicting their thoughts, actions, etc. This fullblooded notion of agency crucially involves a person’s characterization. To understand this frst step of the argument better, let us have a closer look at these four conditions. Condition (1) is uncontroversial: there sometimes are differences in the psychological dispositions of two people, and empathetic perspective-shifting would have to take these differences into account. In fact, such differences are likely to be very common and, if empathetic perspective-shifting is possible, will pose

264

Julia Langkau

one of the core challenges to it. Condition (2) seems uncontroversial as well: sometimes, we don’t think, decide, or act rationally. In his paper ‘Imagination and the Distorting Power of Emotion’ (Goldie, 2005), Goldie addresses the issue of how our emotions, e.g., fear or anger, often distort our practical reasoning: they function as reason-giving. While we are in a particular situation, we don’t realize that our practical reasoning is being distorted. For instance, my irrational fear of dogs makes me avoid or run away from dogs, even though I am usually aware of the fact that most dogs are harmless. Goldie wonders whether we can take the distorting power of emotion into account when imagining, even when imagining ourselves in a particular situation: Can I imagine myself from the inside, reasoning about what I ought to do in this situation, and then imagine doing what I imagine deciding to do, whilst knowing, outside the scope of what I imagine, that this is not what I ought to do […]? (Goldie 2005, 133) Goldie emphasizes the normative aspect of what we have to imagine in such a situation: the relevant mental state is not about what we will do, but what we should do (Goldie 2005, 134). What we should do, reasoning from the outside, differs from what we should do, subjectively, in the situation. Note that when we are looking at a way of gaining knowledge or at any epistemic process, we usually don’t take into account ways in which people’s irrationality could distort the process. The reason is that any epistemic process will get distorted if there are relevant non-rational infuences on the psychological make-up or decision-making process of any of the people involved. However, if non-rational infuences are typical for our everyday decision-making and acting, they should indeed be taken into account when we consider general challenges to an epistemic process – in this case to everyday perspective-taking. In the second step of his argument against empathetic perspectiveshifting, Goldie aims to establish that it is not possible for a subject to take a target’s characterization into account in the relevant way, and that this is why empathetic perspective-shifting must fail. His argument builds on a distinction introduced by Richard Moran (2001), according to which there are two different stances one can take toward oneself: a ‘theoretical or empirical’ and a ‘frst-personal deliberative or practical’ stance. The theoretical or empirical stance toward oneself involves explicit access to facts about oneself. However, when deciding what one ought to do (or think, feel, etc.), our characterization is not treated as an empirical fact about ourselves. We rather take a frst-personal deliberative stance in which our characterization implicitly shapes our thinking, feeling, etc.:

On Imagining Being Someone Else

265

The kind person has kind thoughts and feelings, but these thoughts and feelings are not typically about the disposition of which they are an expression. The person who loves his spouse seldom thinks that he is doing loving things because he loves her […] (Goldie 2011, 309) Having introduced this distinction, Goldie argues as follows: since, in a conscious attempt to engage in empathetic perspective-shifting, A cannot keep B’s characterization in the non-conscious background, A is forced to treat B’s characterization from the empirical or theoretical stance. Since this is not what B does, it ‘produces a fundamentally distorted model of B’s thinking’ (Goldie 2011, 309) and thus must lead to a failure of perspective-taking. For the purpose of this chapter, I will leave (3) and (4) aside, since they both present special circumstances under which empathetic perspective-shifting may be particularly desirable, but under which it is also particularly diffcult. If we succeed in (1) showing that empathetic perspective-shifting is possible and can be successful in cases with relevant differences in the subject’s and the target’s psychological dispositions and (2) considering the distorting power of emotion, this shows more generally that empathetic perspective-shifting is possible, at least, in certain typical circumstances. Let us use our two examples from the beginning of the chapter, alter them slightly, and fll in some details. Suppose A aims to take the perspective of B, who loves hockey and watches their favorite hockey team lose an important game. B is frustrated and sad. A, in an attempt to take over B’s perspective, uses their knowledge of B’s love of hockey and of a particular team. B’s emotional reaction is based on (and somehow causally connected to) their love of hockey and that particular team, but B is not aware of their own dispositions; they take a frst-personal deliberative stance. A, however, uses their knowledge as an explicit premise when trying to take over B’s perspective; they take an empirical stance. Further, suppose A aims to take the perspective of C, who struggles in their relationship. Given everything C tells A about their relationship, it seems that C is convinced the best thing to do would be to leave their partner. However, A also knows that C is easily manipulated by others. So, after confronting their partner, they end up not leaving them after all. While C is not aware of their disposition at the moment they confront their partner (frst-personal deliberative stance), A knows about it and takes it into account (empirical stance). The frst example refects signifcant differences in the psychological make-up of subject A and target B (1), and the second example refects non-rational infuences in C’s decision-making process (2). According to Goldie, A cannot be successful in taking over the perspective of B because A has to take an empirical stance on B’s love of hockey. Neither

266  Julia Langkau can A be successful in taking over the perspective of C because A is aware that C’s reasoning is distorted, while C themselves are not. As may be obvious, the focus in both cases is on understanding the other from their perspective, and predicting their actions or decisions from their perspective. By distinguishing the two kinds of perspective-taking on the basis of Moran’s two stances toward oneself, Goldie suggests that they are categorically different, and that one of them is impossible. But Goldie’s notion of empathetic perspective-shifting is unnecessarily ambitious. As Ylwa Wirling (2014) observes, it seems to be driven by the idea that being someone else is metaphysically impossible. And indeed, that we cannot imagine being somebody else is relatively uncontroversial. But imagining being identical with the target has never been thought to be the point of empathetic perspective-shifting (see Walton 2015, 4), and it could not have been Goldie’s point. 2 His argument is directed toward the distinction of two perspectives as used and applied by philosophers and psychologists in their theoretical and empirical work. For instance, Goldie claims that Batson et al. (1997), in their study on psychological differences between the two kinds of perspective-taking, do not properly distinguish empathetic from sympathetic perspective-taking. In particular, he argues that they give ambiguous instructions to participants in their experiments (Goldie 2011, 306). If this observation is correct, then Batson et al.’s study does not confirm the distinction between two kinds of perspective-taking3, but it obviously does not mean that there is no such distinction. Hence, while it can’t be the case that Goldie thinks the aim is to imagine being identical with the target, he still sets the bar too high for empathetic perspective-shifting. The next section argues that even though we can distinguish between two different aims in perspective-taking, the processes involved in both cases are similar, and we face similar epistemic challenges. Whether the subject imagines themselves in a certain situation as they are now (intheir-shoes perspective-shifting), as they are sometime in the future, or as the target is (empathetic perspective-shifting) all require the subject to imagine experiences that are different from their current ones.4

13.3  Epistemic Challenges to Imagining Experiences Goldie’s main point is that if A takes account of B’s love of hockey by using explicit knowledge of it, this will necessarily distort the simulation. Similarly, if A takes into account that C is easily manipulated, A will not succeed in simulating C’s mental state. In what follows, I will argue that two plausible assumptions about simulation allow us to question these conclusions: first, that what matters is the outcome of a process, and second, that the outcome does not have to be qualitatively identical to the target of the simulation.

On Imagining Being Someone Else

267

In explaining what leads to the distortion in empathetic perspectiveshifting, Goldie claims that from a frst-personal deliberative stance, any thoughts and feelings ‘are not typically about the disposition of which they are an expression’ (Goldie 2011, 309), implying that from an empirical stance, they are. But note that the fact that A aims to take B’s characterization and C’s irrationality into account does not mean that A’s feeling or thinking is necessarily about B’s love of hockey or about C’s being easily manipulated, as suggested by Goldie. To see this, we have to specify whether we are concerned with the process of empathetic perspective-shifting or with its outcome (see Goldie 2011, 303–4). If we understand empathy as a process (as, for instance, Feagin 2011), then trivially A, who explicitly refers to B’s characterization, does not simulate exactly B’s mental processes, and A’s thinking is indeed partly about B’s characterization. But if we are interested in empathy as the outcome of this process, i.e., if we are interested in whether a subject can simulate the relevant mental state or states of a target in a particular situation, the fact that the process of getting to this mental state or to these states is different for the subject does not necessarily mean that empathy fails. Empathy considered as the outcome of the process is not about B’s love of hockey or about C’s being easily manipulated, it rather consists in a certain attitude toward the lost game and toward the relationship and the thinking or reasoning in response to or as part of these attitudes, respectively. Hence, whether or not A is successful in taking over B’s or C’s perspective depends on the simulated attitude toward the lost game or the relationship. Here is how I wish to think about it. Say S and T are going on vacation to Mexico, but while S is traveling from the USA, T is traveling from Europe. In some sense, S and T are not going on the same vacation, since they are coming from different places. But this sense is trivial, because the vacation is crucially about the destination, not about how they get there. Once they are in Mexico, they will be able to have all the same experiences. Similarly, A’s aim is to obtain certain mental states which simulate B’s and C’s original mental states, and in order to achieve this aim, A consciously takes B’s characterization and C’s irrationality into account. We should also be clear that successful perspective-taking does not imply that the subject’s mental state is qualitatively identical to the target’s original mental state. First of all, imagining X is never exactly like experiencing or perceiving X. When looking at an apple in front of me and later imagining that very apple, both the mental attitude and the content of that mental attitude arguably are different in some respect. Second, as many have observed and discussed (see, e.g., Wirling 2014), perspective-taking doesn’t imply that we are no longer aware that the mental states we are imagining are not ours. In contrast, it is part of perspective-taking that we are aware of that fact. Third, note that complex simulations (of complex processes) usually come with a purpose.

268

Julia Langkau

For instance, a horseback riding simulator is designed to help people improve their horseback riding skills on real horses. While the horseback riding simulator largely ‘behaves’ like a horse, it does not have realistic legs; it is made of plastic and metal and might not move its neck. Its purpose is not to look and behave like a horse in every respect, but to react to the practicing rider such that they can learn something which will later be relevant on a real horse. So, clearly riding on a horseback simulator does not give one the same experience as riding on a horse, but some or all of the relevant aspects will be similar enough to justify the rider’s practicing on a machine. Similarly, the purpose of taking over  the perspective of another is to put us into the position to understand the other person’s mental state without having to be in that very situation ourselves and despite psychological differences. But not all aspects, and in particular not all psychological differences are relevant to the respective situation. Suppose that B also loves to eat asparagus, which A doesn’t. In that case, A arguably doesn’t have to imagine what it is like to be hockey-and-asparagus-loving B; A only has to imagine what it is like to be hockey-loving B. Hence, in order to empathize with B, A should not have to imagine being B where this goes beyond simulating the relevant aspects of B’s mental state in the relevant situation.5 If we agree that empathizing is mainly about the outcome of a simulation process, and that successful perspective-taking does not imply that the mental state the subject reaches is qualitatively identical to the target’s mental state, then a subject does not necessarily have to take a merely theoretical stance on the target’s characterization. The subject can put their own, similar experience to use in order to imagine the target’s experience. Amy Kind has argued that we can imagine many experiences we haven’t had by assembling them, in the imagination, from experiences we have had. Referring to Thomas Nagel, she has named this process ‘imaginative scaffolding’: If you haven’t hiked the Lost Palms Oasis train but you’ve hiked other trains in Joshua Tree, and you’ve seen an oasis before, or even just seen pictures of oases before, then by using your imagination – as Nagel says, by ‘imagining some combination of additions, subtractions, and modifcations’ to other experiences you’ve had – it’s plausible to say that you could know what hiking the Lost Palms Oasis train is like even without having done it. (Kind 2020, 137) Imagine A loves the Summer Olympics and cheers for the athletes of their native country. When A is taking over B’s perspective in the empathetic sense, A can use their experience of loving the Summer Olympics and apply it, in the imagination, to B’s situation. In the process of taking over B’s perspective, A is aware of the fact that they have to take B’s

On Imagining Being Someone Else

269

psychological disposition into account, but once A has taken over B’s perspective, they will not have a merely theoretical understanding. There are two potential challenges a subject faces when engaging in ‘imaginative scaffolding’. First, A might not love any sport or anything similar enough to sport and hence having an emotional reaction to a certain team or person losing in sports would be a new kind of experience they fail to imagine. I will come back to this challenge below. Second, suppose that in C’s situation, A would have left their partner, because A is not easily manipulated. If A uses their own experience or an experience similar to C’s experience, they will fail to take C’s perspective. But suppose, as given in the example, A knows that C will act irrationally when confronting their partner with the initial decision, because they know that C is easily manipulated by others. 6 While C is not aware of this, A takes it into account when they choose the kind of experience most similar to C’s experience. In Section 13.2, we said it is relevant to take emotional distortions into account because such distortions are very common. And if they are common, a subject aiming to take the perspective of another will likely have had some such experience and hence will be able to use some such experience. Say, A is afraid of dogs, and while they know that most dogs are harmless and that they shouldn’t run away from a tiny, harmless dog, this is exactly what they feel like doing when they encounter one. In such a situation, A irrationally thinks they should run. A can now use this experience and apply it to C’s situation and the experience they aim to simulate.7 Goldie moreover thinks that the subject can’t understand the longlasting consequences a decision may have for the target. In perspectivetaking, we cannot account for each other’s ‘[…] conception of oneself as having a past and a future in the light of which decisions and choices are made – decisions and choices which one has to live with’ (Goldie 2011, 317). But assuming that we all share crucial psychological features, we all (at least most of us) have an understanding of ourselves as having a past and a future. We all know what it means to treat a certain decision as one that matters to our future, even if we are not necessarily aware of it. Hence, imagining a situation as one that matters for the future of the target is not a problem; we can account for it through ‘imaginative scaffolding’. If ‘having a past’ just means having any kind of past, there is again no problem, since we all have a past. However, if it is supposed to mean having a particular past, i.e., the past of the target, it often won’t be relevant for taking over the perspective of a target in a particular situation. Take again the example of A irrationally running away from any kind of dog. How exactly A acquired this disposition is not important. Equally, the reason why C has the disposition of being easily manipulated is not important for the fact that they are. It might broaden A’s understanding of C to know how C acquired this disposition, but it

270 Julia Langkau is not necessary for A’s understanding of C’s current mental state and situation, which is such that C is easily manipulated. We have seen that if we are concerned with the result of perspectivetaking, it doesn’t matter how the subject reaches that result, and since the aim cannot be to have exactly the same mental state, a subject can use their own repertoire of experiences. Sometimes, however, we might not have had an experience similar enough to the one we’re aiming to simulate. Maybe we have been in a situation in which what we decide matters for our future, but we haven’t been in a situation where our decision would have a signifcant effect on our future. This leads me back to the frst challenge stated above: what if we haven’t had an experience similar enough for it to be useful when taking over the perspective of another? Recall that Goldie distinguishes between two different aspects of perspective-taking: the narrative (or the situation; to be taken into account in both kinds of perspective-shifting) and the characterization (only to be taken into account in empathetic perspective-shifting). In an earlier paper (Goldie 2002), he says the following: Putting yourself in the other person’s shoes is, in a sense, less demanding on the imagination as it does not require you to take on the other’s characterization as part of the imaginative process; all that is required is, as a starting-point, centrally imagining yourself experiencing the narrative, or imagining yourself ‘from the inside’ having those experiences, responding emotionally, deliberating, deciding what to do, and so on. (Goldie 2002, 98) However, imagining being myself only works if I have had experiences similar enough to the ones I ought to imagine, and imagining myself in a completely new situation might simply fail. The question of whether we can imagine experiences we haven’t had before has recently been much discussed. L.A. Paul (2014) thinks we cannot imagine new kinds of experiences, especially when they are ‘transformative experiences’. Transformative experiences, according to Paul, are new kinds of experiences that change our epistemic situation (epistemically transformative experiences), and in some cases change who we are (personally transformative experiences). Kind (2020), in contrast, argues that nothing is in principle in the way of our imagining new experiences, or that the class of unimaginable experiences is at least very small. If there is a small or even a large class of experiences a subject cannot imagine because they lack the relevant experiences, the problem concerns the narrative just as much as the characterization, and hence is not only a problem for empathetic perspective-shifting. If we are skeptical about our ability to take into consideration the characterization of a

On Imagining Being Someone Else

271

target, there is no reason not to be skeptical about the narrative as well. But if we are confdent concerning the narrative and we assume that we can, at least in many cases, imagine having a certain experience we haven’t had, then there is no reason to assume that we cannot also imagine having different characterizations. Of course, we may fail sometimes and instead take a theoretical stance and use our empirical knowledge to predict another’s decisions and actions. How should we then distinguish the two kinds of perspective-taking? Wirling considers the plausibility of their being the end points on a continuum. While she fnds it implausible to think that one extreme is imagining oneself in the other’s situation and the other extreme is really taking over the other’s perspective, she thinks that there is a continuum concerning the complexity of the situation and the differences between subject and target (Wirling 2014, 218–9). I agree that we could speak of a continuum with respect to the imaginative work we have to invest and the possible challenges we face. The beginning point is my current or a recent experience, and the end point is an experience much different, possibly different from any experience I have ever had. The more psychological differences there are between the subject and the target, the more the subject has to engage in ‘scaffolding’ to imagine the target’s experience, but the same can be said about myself today and in the future: the more I change, and the further away in the future the experience is that I aim to imagine, the more work I have to do in the imagination. Suppose I am aiming to take the perspective of my grandmother on the one hand and also trying to imagine myself in, say, 40 years from now. While my grandmother is okay with moving into a care facility because she’ll be able to spend her time reading autobiographies, I imagine myself being okay with moving into a care facility because I will be able to spend my time reading novels. What I am doing in both cases of perspectivetaking is imagining an experience different from my current experience, and the experience I imagine is similar in both cases. The same is true for the situation I am aiming to imagine: the further away it is from a situation I have ever been in, the more I have to engage in imaginative ‘scaffolding’, and this is true whether I imagine myself in this situation or somebody else. I thus conclude that the underlying process of both kinds of perspective-taking can be the same. However, the aim, what we are trying to achieve, and for whom this is relevant, can be much different. In the case of in-their-shoes perspective-shifting, my aim is to imagine being myself, as the person I am now, in a certain situation different from my current situation. In the case of empathetic perspective-shifting, my aim is to imagine being someone else in a certain situation. But ‘being someone else’ only means imagining having a different psychological make-up than I have now, and consequently imagining a different kind of experience.

272

Julia Langkau

Empirical studies have suggested that we naturally engage in empathetic perspective-shifting, provided we have suffcient information about the target’s situation and character. Hence, the standard perspective we take if enough information is given is the one that includes the target’s characterization, and in-their-shoes perspective-shifting requires an extra step we only take if we realize that we don’t have enough information (see, e.g., Batson 2009; Hygge, 1976; Jarymowicz, 1992). This sits well with my claim that we don’t separate narrative and characterization in the process of perspective-taking and instead aim directly at imagining the relevant experience. To conclude this part of the discussion: what distinguishes in-theirshoes perspective-shifting from empathetic perspective-shifting is not that in the frst case the subject only has to take into account the narrative, while in the second case they also have to take into account the characterization of the target. Rather, whether the subject imagines themselves in a particular situation as they are now, as they are, for instance, in 40 years, or as a certain target is in that particular situation – all require the subject to imagine an experience that is different from their own current experience.8 Whether the subject is successful or not depends on various factors. First and foremost, the subject has to have suffcient information. But they also have to have the relevant repertoire of experiences. If the situation they are aiming to imagine is too far away from what they have ever experienced, perspective-taking will become harder, if not impossible. I have argued that empathetic perspective-shifting is not conceptually impossible, but rather poses epistemic challenges that are not in principle different from the epistemic challenges in-their-shoes perspectiveshifting poses. Besides these two kinds of epistemic challenges (suffcient information and the relevant repertoire of experiences), a subject faces psychological challenges. In the next section, I will discuss psychological challenges in the case of our engagement with fction.

13.4 Psychological Challenges in an Epistemically Ideal Context Empathetic perspective-shifting is more likely to be successful when the subject knows the target well, especially when the situation is one where the target’s emotions are likely to distort their practical reasoning. Maybe the intuition that imagining being someone else in a certain situation is impossible comes partly from the fact that we rarely know enough about a person’s psychological make-up to empathize successfully in complex situations. But there is nothing in principle in the way of knowing a person well enough to take account of such distortions. One context in which we can be in an epistemically privileged situation is fction (see, e.g., Currie 2016). Fiction often gives us direct access to the

On Imagining Being Someone Else

273

thoughts and feelings of a character, either through internal monologue or through detailed third-personal description. A famous example of the frst is Molly Bloom’s fnal monologue, in the fnal chapter of James Joyce’s Ulysses. Here are the very last words of it: […] and Gibraltar as a girl where I was a Flower of the mountain yes when I put the rose in my hair like the Andalusian girls used or shall I wear a red yes and how he kissed me under the Moorish wall and I thought well as well him as another and then I asked him with my eyes to ask again yes and then he asked me would I yes to say yes my mountain fower and frst I put my arms around him yes and drew him down to me so he could feel my breasts all perfume yes and his heart was going like mad and yes I said yes I will Yes. (Joyce 2010, 682) The particular style (no punctuation, incomplete sentences, repetition, etc.) of this passage suggests that the words we read are exactly the ones going through Molly’s mind. However, the text doesn’t explicitly inform us about any emotions or feelings Molly may have. We have to infer such experiential states from the content of her thoughts and the literary means of expression used to express those thoughts. One might say that Molly’s mental states are presented to us from a frst-personal deliberative stance, and the reason why we don’t explicitly learn anything about her emotional state is that she’s not thinking about her emotional state, but she’s rather in a certain emotional state that underlies and maybe guides her thinking. A detailed description of the emotional life of a character in the third person can provide us with more direct information about their experiential mental states, as in Virginia Woolf’s Mrs. Dalloway: So he was deserted. The whole world was clamouring: Kill yourself, kill yourself, for our sakes. But why should he kill himself for their sakes? Food was pleasant; the sun hot; and this killing oneself, how does one set about it, with a table knife, uglily, with foods of blood, – by sucking a gaspipe? He was too weak; he could scarcely raise his hand. Besides, now that he was quite alone, condemned, deserted, as those who are about to die are alone, there was a luxury in it, an isolation full of sublimity; a freedom which the attached can never know. (Woolf 2003, 104) This passage includes a description of thoughts as well as feelings. The character, Septimus, is feeling deserted, condemned, and lonely on the one hand; recognizes societal pressure to kill himself; considers positive aspects of life that speak against killing himself; and thinks about how he should go about killing himself. However, besides his overall negative

274

Julia Langkau

state of mind, he also experiences a feeling of luxury, sublimity, and freedom in his decision to commit suicide. This passage thus gives us a very detailed description of the character’s complex emotional state. From an epistemic point of view, it provides an ideal context for empathetic perspective-shifting: we have all the relevant information about Septimus’ mental state. The information is presented at least partly from an empirical stance. Epley and Caruso (2009) have identifed three barriers a subject has to overcome in order to successfully take over a target’s perspective. Besides using accurate information and overcoming their own perspective,9 the subject has to activate the mental process of perspectivetaking, either automatically or as the result of an active attempt (Epley and Caruso 2009, 299–307). Empirical research (e.g., Kotovych et  al. 2011) has shown that we have more problems empathizing when we are given explicit information about the target’s mental state than when we have to make inferences as to what the target’s thoughts and feelings might be. In an experiment, Kotovych et  al. asked some subjects to read Alice Munro’s short story, The Offce, which only hints at the character’s feeling. Other subjects were given an altered version of the text which included a summary of the beginning, directly reporting on the character’s mental states (e.g., ‘I feel embarrassed’) (Kotovych et al. 2011, 270). Subjects who read the original short story and had to make inferences to access the character’s mental states seem to have found it easier to ‘appreciate’ and ‘understand’ the character’s feelings than subjects who were directly told how the character felt. The authors interpret these fndings as giving the reader the opportunity to use their own experience to attribute it to the character (Kotovych et al. 2011, 274). These fndings suggest that we fnd it easier to activate experiential states when trying to empathize with Molly than when trying to empathize with Septimus, even though our epistemic situation is much better, if not ideal, in the second case. The problem we face thus when trying to empathize with Molly is that while we experience something, we might not get her mental state right (as intended by the author). Literary fction often employs ‘stylistic means to invite emotional refection and reappraisal’ (Oatley 2016, 622). Such means are, for instance, metaphors, pictures associated with the character’s inner life, the particular structure of the text, etc.10 In Joyce’s quote, no emotions or feelings are mentioned, but the way in which the thoughts are presented is clearly suggestive of certain affective states. The information, however, is presented in a much different way than in real life, where body language, facial expression, etc. serve as clues. Accordingly, inferring from literary means of expressions requires a different skill than reading the evidence in real-life confrontations (for more details, see Langkau 2020). In this respect, movies come much closer to real-life confrontations. In the movie Mrs. Dalloway (Gorris 1997),

On Imagining Being Someone Else

275

the scene in which Septimus’ mental states are described is presented through a close-up of the actor’s face and body. Septimus is visibly in a very dark place, but there is a moment in which his head suddenly rises up, his facial expression changes, and he looks determined and full of dignity. This moment may indeed enable us better than the explicit description given by Woolf to actually experience, in the imagination, some of Septimus’ new-found freedom. However, we might not exactly recognize the feeling as one of luxury, which maybe only the explicit text can inform us about. Hence, it seems that when it comes to fction, we are either in an epistemically privileged context but face psychological challenges, or it is psychologically easier to take a frst-personal deliberative stance, but we are in an epistemically challenging context: we have to use literary means of expression as evidence for the character’s mental states.

13.5 Conclusion I have argued that we can distinguish two different kinds of perspectivetaking, but that they are more similar to each other than Goldie thought. Their similarity suggests that skepticism toward our ability to take into account a target’s characterization leads to skepticism toward imagining any future experience we ourselves might make, and that optimism concerning the second justifes optimism concerning the frst. The distinction between characterization and narrative in perspective-shifting is thus misleading, or at least does not characterize the relevant distinction. Rather, the relevant distinction is between our current experience and the experience we are aiming to imagine. In an epistemically ideal situation, a subject has access to two different kinds of knowledge: suffcient knowledge about the target’s situation as well as their psychological make-up and some relevant experiential knowledge. Given the direct access to a character’s mental states we usually get in the context of fction, fction seems to provide an ideal context for empathetic perspective-shifting. However, this is only correct from an epistemic point of view, since explicit information about a character’s mental states seems to make it harder to activate the relevant experiences, in the imagination.11

Notes 1 Recently, psychologists have also distinguished ‘experience-taking’ from ‘perspective-taking’. Experience-taking is supposed to be a spontaneous, non-intentional, experientially driven process of simulating the mind-set and persona of a fctional character while reading. The idea is that we let go of our own beliefs, memories, and personality traits in order to take on a character’s mental states and thus imagine ourselves becoming the character (Kaufman and Libby 2012, 15).

276

Julia Langkau

2 Karsten Stueber, however, seems to think that this was Goldie’s main point: “Importantly, [empathetic perspective-shifting] should not be conceived of as requiring that we imagine being the other person […]. As Peter Goldie (2011) has argued, such complete identifcation with another person even in the imagination is a conceptual impossibility” (Stueber 2016, 374). 3 One problem with empirical studies is that it is not easy to distinguish both kinds of perspective-taking verbally such that the subjects are guaranteed to understand the difference, especially in an empirical study with untrained subjects, and given that there are various different ways of naming both perspectives. 4 See also Chapter 5 (Roelofs, this volume) which agrees with me that Goldie is wrong about the impossibility of empathetic perspective-shifting, and that the underlying process is the same in both kinds of perspective-taking. He distinguishes two aspects of an imaginative act – what it recreates and what it targets – and argues that what we recreate is often the same in both kinds of perspective-shifting, even if the target differs. 5 Which aspects are important is of course a diffcult question. As Coeckelbergh observes, “[…] anything approaching identifcation far exceeds the minimum requirement” (Coeckelbergh 2007, 65). 6 This may be suffcient for A to predict C’s actions, but we are interested in understanding the other from their perspective. 7 Goldie sees that this problem doesn’t only occur with perspective-shifting, but also when I try to imagine myself in a situation in which irrationality would affect my decision-making and acting. 8 Note that even though I use the term ‘experience’, ‘experience-taking’ is different from ‘perspective-taking’ (see endnote 2). Perspective-taking is the result of a conscious process, which involves the knowledge of the subject that what they are imagining are not their own current experiences. 9 One advantage of sympathetic imagining Goldie (2005) emphasizes is that it avoids the dangers of egocentric bias (see, e.g., Epley and Caruso 2009), because it does not require that we overcome egocentric bias. 10 See also Oatley (2016, 621–4). 11 I would like to thank Christopher Badura, Patrik Engisch, Amy Kind, Luke Roelofs, and an anonymous reviewer for discussion and helpful comments. Work on this chapter was funded by the Swiss National Science Foundation.

References Batson, C.D. 2009. “Two forms of perspective taking: Imagining how another feels and imagining how you would feel.” In Handbook of Imagination and Mental Simulation, edited by K.D. Markman, W.M.P. Klein, and J.A. Suhr, 267–79. New York: Psychology Press. Batson, C.D., Early, S., and Salvarani, G. 1997. “Perspective taking: Imagining how another feels versus imagining how you would feel.” Personality and Social Psychology Bulletin, 23: 751–8. Coeckelbergh, Mark. 2007. “Who needs empathy? A response to Goldie’s arguments against empathy and suggestions for an account of mutual perspectiveshifting in contexts of helping and care.” Ethics and Education, 2(1): 61–72. Currie, Gregory. 2016. “Does fction make us less empathic?” Teorema, XXXV(3): 47–68.

On Imagining Being Someone Else

277

Epley, Nicholas and Caruso, Eugene M. 2009. “Perspective taking: Misstepping into others’ shoes.” In Handbook of Imagination and Mental Simulation, edited by K.D. Markman, W.M.P. Klein, and J.A. Suhr, 295–309. New York: Psychology Press. Feagin, Susan L. 2011. Empathizing as simulating. In Empathy. Philosophical and Psychological Perspectives, edited by A. Coplan and P. Goldie, 149–61. Oxford: Oxford University Press. Goldie, Peter. 2002. “Emotion, personality and simulation.” In Understanding Emotions: Mind and Morals, edited by Peter Goldie, 97–109. Aldershot: Ashgate Publishing. Goldie, Peter. 2003. “Narration, emotion, and perspective.” In Imagination, Philosophy, and the Arts, edited by Matthew Kieran and Dominic McIver Lopes, 54–68. London: Routledge. Goldie, Peter. 2005. “Imagination and the distorting power of emotion.” Journal of Consciousness Studies, 12(8–10): 127–39. Goldie, Peter. 2011. “Anti-empathy.” In Empathy. Philosophical and Psychological Perspectives, edited by A. Coplan and P. Goldie, 302–17. Oxford: Oxford University Press. Goldman, Alvin. 2006. Simulating Minds: The Philosophy, Psychology, and Neuroscience of Mindreading. Oxford: Oxford University Press. Gorris, Marleen. 1997. Mrs Dalloway. Co-production United Kingdom-United States-Netherlands; First Look Studios, Bergen Film & TV, Newmarket Capital Group, BBC Films, NPS Televisie, Netherlands Film Fund. Hygge, Staffan. 1976. “Information about the model’s unconditioned stimulus and response in vicarious classical conditioning”. Journal of Personality and Social Psychology, 33: 764–71. Jarymowicz, Maria. 1992. “Self, we, and other(s): Schemata, distinctiveness, and altruism”. In Embracing the other: Philosophical, psychological, and historical perspectives on altruism, edited by P. Oliner, et al., 194–212. New York: New York University Press. Joyce, James. 2010. Ulysses. London: Wordsworth Classics. Kaufman, G.F., and Libby, L.K. 2012. “Changing beliefs and behavior through experience-taking.” Journal of Personality and Social Psychology, 103(1): 1–19. Kind, Amy. 2020. “What imagination teaches.” In Becoming Someone New: Essays on Transformative Experience, Choice, and Change, edited by J. Schwenkler and E. Lambert. Oxford: Oxford University Press. Kotovych, M., Dixon, P., Bortolussi, M., and Holden, M. 2011. “Textual determinants of a component of literary identifcation.” Scientifc Study of Literature, 1(2): 260–91. Langkau, Julia. 2020. “The empathic skill fction can’t teach us.” Philosophical Psychology, 33: 313–31. Moran, Richard. 2001. Authority and Estrangement: An Essay on Selfknowledge. Princeton: Princeton University Press. Oatley, Keith. 2016. “Fiction: simulation of social worlds.” Trends in Cognitive Sciences, 20(8): 618–28. Paul, L.A. 2014. Transformative Experience. Oxford: Oxford University Press.

278

Julia Langkau

Stueber, Karsten R. 2016. “Empathy and the imagination.” In The Routledge Handbook of Philosophy of Imagination, edited by Amy Kind, 368–79. New York: Routledge. Walton, Kendall L. 2015. In Other Shoes. Music, Metaphor, Empathy, Existence. Oxford: Oxford University Press. Wirling, Ylwa. 2014. “Imagining oneself being someone else. The role of the self in the shoes of another.” Journal of Consciousness Studies, 21(9–10): 205–25. Wollheim, Richard. 1984. The Thread of Life. Cambridge: Harvard University Press. Woolf, Virginia. 2003. Mrs Dalloway. London: CRW Publishing Limited.

14 “Imagine If They Did That to You!” The Complexity of Empathy Luke Roelofs

We often remonstrate with each other by appealing to imagined role reversals. If I’ve done something that seems to you obviously wrong but to me seems fne or trivial, you might try and make me see the problem by having me imagine the same action done to me by someone else. Indeed, the idea of switching places with those our actions affect is central to the so-called “golden rule”, to treat others as one would wish them to treat oneself, which is sometimes held up as a near-universally endorsed moral principle. But what is going on, cognitively, when we imagine such a role reversal? Why does imagining something that did not happen change my view of what did happen? What is the relationship between imagining this counterfactual scenario, and imagining the actual scenario from the other person’s point of view? And how does either relate to “imagining being” the other person? This question connects to a much bigger question about the place of empathy in a world increasingly threatened by climate change, a pandemic, and a rising global tide of fascism. It’s common and popular to argue that we need more empathy. Shortly before Barack Obama’s election, he wrote that the USA suffers from an “an empathy defcit”, and that “a stronger sense of empathy would tilt the balance of our politics in favour of those who are struggling” (Obama 2006, pp. 67–8); halfway through his term, Ta-Nehisi Coates suggested that diffculty with the “basic extension of empathy is one of the great barriers in understanding race in this country” (Coates 2011). A recent book about why white Evangelicals support Trump is titled The End of Empathy (Compton 2020); a call to arms after Trump’s inauguration laid out the hope for an “Empathy Politics” (Serano 2017), and it’s always tempting, when trying to get middle-class White people to care about racism, to show them a video of police brutalising a Black person and ask “how would you feel?” But this “empathy economy” (Stinson 2020) invites pushback; critics claim that “Antiracism is far less about empathy, about appealing to similarities, than it is about love, about honoring and protecting differences” (Salesses 2020). Even those who call for more empathy often feel a need to add caveats: Coates (2011) says “I do not mean a soft, fattering, hand-holding empathy. I mean a muscular empathy rooted in

280

Luke Roelofs

curiosity”, and Serano (2017) says “I am not talking about the shallow empathy of those who only seem capable of identifying with people who they perceive to be their own kind”. I think this sort of qualifed position is right: it’s not simply that empathy is good, or that empathy is bad, but that empathy is diffcult (cf. Epley and Caruso 2008; Langkau this volume; Kind this volume; and Jamison 2014, who writes “empathy is an edifce we build like a home or offce – with architecture and design, scaffolding and electricity”, p. 15). A key part of this diffculty is balancing the recognition of otherness with the search for commonality, and my aim in this chapter is to contribute to untangling some of the conceptual issues involved in fnding this balance. I will keep my focus on “empathic role reversal” without trying to defne or delimit “empathy” as a general category. I argue that we need to separate two distinct questions about any given act of empathetic imagining, namely, “what mental states does it create?” and “what thing or possibility does it target?”. When we do so, we can see how the various linguistic forms available (“imagine if”, “imagine being”, “imagine from their point of view”, etc.) do not name distinct cognitive acts so much as they emphasise different steps of a complex cognitive process, which is primarily directed at the actual other, but often instrumentally directed at anything counterfactual. In doing so I hope to show why certain lines of criticism of empathy fail, and why empathic role reversal has value despite the complex diffculties of doing it well.

14.1 Empathic Role Reversal To focus the discussion, consider case 1: Case 1: A has wronged B, but doesn’t really care. C remonstrates with A, saying “how could you do that to B? Imagine if B did that to you! How would you feel?” A imagines that: they imagine B wronging A in the same way. They fnd themselves feeling outraged and hurt about B’s imagined act. As a result, they revise their attitude to their own actions, and seek out B to make amends. In this case, imagining a counterfactual scenario with roles reversed changes A’s attitude and feelings about the actual world. Of course there is a large question here about how the fnal step works, where A changes their moral judgements and is moved to new actions. But my interest in the previous step, the imagining, and particularly in the apparent contrast with case 2: Case 2: A has wronged B, but doesn’t really care. C remonstrates with A, saying “How could you do that to B? Look at this situation from B’s perspective! Imagine how they must feel!” A imagines that:

The Complexity of Empathy

281

they imagine the situation from B’s perspective, and fnd themselves feeling outraged and hurt about the actual action which they themselves have performed. As a result, they revise their attitude to their actions, and seek out B to make amends. In this case, it might seem, no counterfactual scenario has been imagined, only the actual situation itself. Yet in some sense, cases 1 and 2 seem very similar; they both result in the same change of heart, and pre-theoretically it might seem strange to insist on a profound cognitive difference in what A does in the two cases. But now fnally consider a third case: Case 3: A has wronged B, but doesn’t really care. C remonstrates with A, saying “How could you do that to B? Imagine if you were B in this situation.” They follow up with one or the other of the following instructions: (a) “Imagine how they would feel! (b) Imagine how you would feel! (c) Imagine how that would feel!” A complies with the instructions given: they imagine being B, suffering the wrong they have themselves just inficted. Finding themselves feeling outraged and hurt, they revise their attitude to their actions, and seek out B to make amends. What is going on here? It seems very similar to cases 1 and 2: A is imaginatively “putting themselves into the shoes of” their victim, and fnds that this exercise changes their feelings about their actions. But are they imagining the actual situation (where B is the victim, and A the agent), or a counterfactual situation (where A is the victim, and someone else the agent), or some metaphysically impossible situation (where A is B, and A is the victim, and B is the victim)? It would be very natural to say that in case 3, A is “imaginatively identifying themselves with B”. But then who is A identifying themselves with in case 2, if not B? Are they identifying themselves with themselves in case 1? Is case 3 actually just a more perspicuously described case 2? And in case 2, where A “imagines how B feels”, are they necessarily thereby “imagining being B”? I will argue that these three cases involve substantively the same mental process (what I have been calling “empathic role reversal”). They may differ in certain subsidiary respects, but these differences are only loosely tied to the different forms of language employed. In each case, C’s aim is to for A to understand and be moved by B’s experience, and towards this end they may direct A’s attention either to actual or counterfactual scenarios, depending on their priorities. In other words, there is here no very sharp or deep distinction between “self-imagining” and “other-imagining”. Many writers have claimed that some version of this distinction is of major importance: Gordon, for example, claims that

282

Luke Roelofs

Adam Smith’s Theory of Moral Sentiments is fawed because it runs the two together (1995a, cf. 1995b, p. 55, Nanay 2010); Batson claims to fnd empirical differences between the two, which he fears too much study has neglected (Batson et al. 1997; Batson 2009; cf. Stotland 1969); competing accounts of engagement with fctional settings and characters turn on the distinction (e.g. Currie 1995; Smith 1995, 1997; Gaut 1998; Giovannelli 2008); and many philosophers have wondered if there is something inherently impossible or misguided about one or the other (Williams 1973; cf. Wollheim 1984; Walton 1990; Velleman 1996; Wirling 2014; Langton 2019). In particular, Peter Goldie has argued that while self-imagining is a valid and useful cognitive tool, other-imagining is both conceptually impossible and ethically suspect (2011). I do not want to claim that there is no distinction at all between “imagining being myself (in some situation)” and “imagining being the other”. But I do deny that there is a single, clear, stable distinction to draw, that is signifcant across contexts. On the contrary, I argue, there are multiple distinctions here, which do not always line up, and which often show intermediate, indeterminate, or mixed forms. Moreover, I claim that in empathic role reversal, imaginers often employ multiple forms of imagination, at once or sequentially, criss-crossing these distinctions, in service of a single goal. As a result, theories like Smith’s account of sympathy, which do not carefully distinguish between self-imagining and other-imagining, may be right not to do so, since for certain purposes that distinction obscures as much as it clarifes. To be precise, I think there are at least three orthogonal distinctions in this vicinity. A full treatment of them will require the distinction between model and target that I will draw in Section 14.3, but they may be sketched briefy here. First, there is what we may call the identity distinction, between imaginings that represent the imaginer, in some actual or possible circumstances, and those that represent another person (this is the standard way to distinguish “self-imagining” from “other-imagining”). Second, there is what we may call the adjustment distinction, between imaginings whose content matches the imaginer’s own personality, character, and background, and those that make adjustments to simulate the personality, character, and background of another person. Third, there is what we may call the internal-external distinction, between imaginings that take an “external” view on something and imaginings that are “from the inside”, recreating the experience of a participant. It is the distinction between, for instance, simply imagining that some events occur, or imagining seeing them from some impersonal viewpoint, and imagining performing or suffering some active role in those events. This distinction is itself highly disputed, drawn in multiple ways and given different labels (such as “central” and “acentral”,

The Complexity of Empathy

283

“subjective” and “objective”, “experiential” and “non-experiential”), so I will try not to lean too heavily on any particular account of it; for discussion see Wollheim (1984), Vendler (1984), Recanati (2007), Giovannelli (2008), and Dokic and Arcangeli (2015). I will suggest, in Section 14.4, that keeping all three of these rough distinctions in mind will give us a better sense of how empathic role reversal functions. But frst I want to say a bit more about Goldie’s critique of empathy, because he articulates most sharply the idea that there is a large difference between self-imagining and other-imagining, and that it matters greatly which one we are involved in on a particular occasion.

14.2 Goldie’s Critique of Empathy Here is how Goldie puts his position: …what I am against is what I will call empathetic perspectiveshifting: consciously and intentionally shifting your perspective in order to imagine being the other person, and thereby sharing in his or her thoughts, feelings, decisions, and other aspects of their psychology. I am not against what I will call in-his-shoes perspectiveshifting: consciously and intentionally shifting your perspective in order to imagine what thoughts, feelings, decisions, and so on you would arrive at if you were in the other’s circumstances. These two […] are often either confused or not suffciently distinguished. What I want to do here is to show just how different they are, and how deep the problems are for empathetic perspective-shifting. (Goldie 2011, p. 302) I think the distinction Goldie draws here (empathetic vs. in-their-shoes) is confused, running together two distinct questions. He draws his distinction both in terms of what sort of mental states you imagine having (the sort you would have vs. the sort the other actually has), and in terms of the identity of the subject imagined (whether it is you or them). These two dimensions, I will argue, come apart (they are what I above called the identity distinction and the adjustment distinction). Later on he re-asserts the latter, identity-based, version of the distinction: […] the difference between in-his-shoes perspective-shifting and empathetic perspective-shifting lies in the content of the imaginative project: who, in the imaginative project, is doing the thinking […]. [in] in-his-shoes perspective-taking […] A imagines himself in that situation […] [while in] empathetic perspective-shifting […] A imagines being B in that situation […] (Goldie 2011, p. 305)

284

Luke Roelofs

Despite committing to this identity-based way of drawing the distinction, the objections Goldie raises revolve around the adjusted or unadjusted character of the mental states one imagines having. Indeed, he admits that in cases where A and B are psychologically similar enough, in relevant respects, and not relevantly irrational, the two forms of perspective-shifting “will each produce the same result, the same ‘output’” (Goldie 2011, p. 307). Consequently, he thinks, there will be no detectable problem with empathetic perspective-shifting in such cases. The inherent problems with it become manifest only in more “ambitious” cases, in particular in cases where A and B have importantly different characters, so that what A would think and feel in a given situation diverges from what B would think and feel. (This also applies in cases where irrational or unconscious factors strongly infuence B’s thought process, so that what “any rational agent” would think and feel in a given situation diverges from what B would think and feel.) In both cases, he insists, A cannot successfully and accurately imagine being B, and even if A recognises these differences, and tries to adjust their imagining to incorporate them, the very fact that they must consciously and explicitly make those adjustments will “falsify” the character of B’s perspective, for whom they are not explicit. Of course, A might try to imagine what A themselves would think and feel if they were (i) in B’s situation, and (ii) had B’s psychological character – this would seemingly be in-their-shoes perspective-shifting, but with all the same problems as empathetic perspective-shifting. Goldie admits this, but says that: […]at this point, this more ambitious variety of in-his-shoes perspective-shifting begins to suffer from exactly the same problems as empathetic perspective-shifting[…] the lesson from this for inhis-shoes perspective-shifting, I think, is that it should not be too ambitious. (Goldie 2011, p. 312) I think some good direct answers to Goldie’s critique are available; Langkau (this volume) presents a number (see also Coeckelbergh 2007). My aim here is not to directly defend empathetic perspective-taking, but to challenge the framework within which Goldie makes the critique: that there are two distinct cognitive acts here, that we have to choose which to engage in, and that it makes sense to be “against” one but not the other. On the contrary, the identity of the person imagined is a separate question from the adjusted or unadjusted character of the mental states imagined – and Goldie’s own discussions show the cracks on this score, with the need to restrict his objections to “ambitious cases” rather than base cases, and to admit that even his favoured form of perspective-shifting, the “in-their-shoes” form, is problematic in

The Complexity of Empathy

285

ambitious cases. The force of his critique concerns not the identity of the person imagined but the amount of psychological adjustment attempted, and so whatever force there is to the critique (which, as noted, I think Langkau addresses quite effectively) it affects “in-their-shoes” and “empathetic” perspective-shifting equally. Moreover, I hope to show, neither the identity distinction nor the character distinction is sharp; many cases fall somewhere in between. And this intermediate character is often not just a result of sloppiness or inattention, but a positive part of the skill of empathising well.

14.3 Two Readings of “What Do I Imagine?” I claim there is a systematic ambiguity in the way we describe imagination’s objects: there are two very different ways of asking and answering the question “what is imagined?” Roughly, this question can either ask what states (images, feelings, propositional thoughts, etc.) are created in the imaginer’s mind or ask what external thing the imaginer thereby represents. I will distinguish these by talking about what an imaginer mentally creates, and what their imagining targets. Note that this is not a matter of there being two senses of imagination, two different mental processes: it is two aspects of a single process, which a single linguistic form (using “imagine” as a verb with a direct object) ambiguously expresses. To see what this distinction comes to, consider some simple examples. I might form in my mind a visual image as of a soft-bodied creature with two eyes, an elongated conical body, and ten limbs. This might be as part of a few different imaginative projects: I might be trying (badly) to imagine what octopuses look like, or to imagine what the octopus I will meet tomorrow (when I visit the zoo) will look like, or to imagine what I would look like if I were turned into an octopus. These are three very different targets: one is (roughly) a general category, one is a specifc individual, and one is a hypothetical version of myself. In that sense, “what I imagine” is very different. Yet in another sense, “what I imagine” is in each case the same: I imagine this image. If someone asked me “what are you imagining?”, they might be asking about either. I would say: the image is what I create, while the general category, specifc individual, or hypothetical version of myself is what I target. One notable thing is that the image may be evaluated for accuracy, but the standard for accuracy is set by the target: if the octopus I will meet tomorrow looks very unusual, then an image that would be acceptably accurate as “what octopuses look like” might be importantly inaccurate as “what the octopus I will meet tomorrow will look like”. And “what I would look like if I were turned into an octopus” is arguably not defnite enough to give much in the way of accuracy conditions. As a matter of fact, of course, this image would be seriously inaccurate in any of these three cases: its shape is distinctly un-octopuslike, because

286 Luke Roelofs the description I gave above is a description of a squid, not an octopus. Nevertheless, if my intention is to imagine an octopus, the image I form is an inaccurate image of an octopus – not an accurate image of a squid. But if I had instead been aiming to imagine what squid look like, I might have formed the same image, and then it would have been an accurate image of a squid. It is, I think, easier to present this distinction with regard to visual images, since it is fairly clear that the image by itself under-determines the target. But I think the same point extends to other forms of imagining: including, in particular, the imagining of thoughts, feelings, decisions, and so on that we are interested in here. When A imagines being wronged by B, they will recreate in their mind a certain complex of connected thoughts, images, emotions, wishes, assumptions, dispositions, etc. It would be very odd to call this complex (the thing that they mentally create) an “image”, but I am unsure what exactly to call it. The best term I have found is “model”: on my account, imagining involves creating in my mind a model, and using it to represent (to target) something outside my mind (the target). This account is meant to be neutral on how to analyse the states involved in imagining. I am most attracted to what is sometimes called a “recreativist” or “simulationist” account of the imagination (for discussion see Currie and Ravenscroft 2002; Goldman 2006; Tagliafco 2011; Kind 2013; Balcerak Jackson 2018), on which it is essentially the capacity to create and manipulate “offine versions” of other mental states. A mental image is an offine copy of perception, and thus just one species of the general class of imaginings. Then we would say that whenever we create an imaginative model, it is formed of offine states which are recreations of other, online, mental state types. Recreativism is congenial to my analysis because it offers a unifed account of the sort of mental states that form the model (they are “offine copies”), but other accounts, including those with a more heterogeneous account of the states involved (e.g. Dorsch 2012; Kind 2013; Langland-Hassan 2020), are also compatible with the distinction drawn here. What we mentally create is, by defnition, some set of mental states. By contrast, what we target is, I think, more or less unlimited (though there may be targets which we cannot imagine accurately or usefully). We can target defnite scenarios like “what will happen”, or “what it would have looked like”, or “how she must feel now”, or “what it would sound like if…”, but we can also target generic possibilities (“what would a human-sized frog look like?”), and our targets can be governed by the rules of a game (“pretend that I’m a human-sized frog”) or a public fction (“once upon a time there was a human-sized frog”). My distinction closely resembles the distinction common in discussions of pictorial representation (e.g. Abell 2005, 2007; Greenberg 2018), between a picture’s content (what it says) and its object or target

The Complexity of Empathy

287

(what it is a picture of). The former is what can be judged accurate or inaccurate, but the latter is what provides the standard of accuracy. My aim is to extend this distinction to apply not just to physical pictures and mental images, but to the full range of imaginings (an extension which Greenberg suggests but does not develop, 2018, p. 895). Models express contents, relative to the context (especially the intentions) surrounding their creation, and those contexts also specify a target relative to which the model is accurate if its content holds. Following Greenberg, I deny that targets can be identifed with any element of content (such as “singular” content as opposed to “attributive content”). Likewise, my model-target distinction does not directly map onto Currie and Ravenscroft’s (2002, p. 12 ff) distinction between an imagining’s “content” and its “character”. They draw this distinction primarily to insist that while visual imagining, for instance, always has a visual “character”, it does not therefore have “seeing” as part of its content. I very much agree with this point, and in particular with the thought that the phenomenal character of an imagining does not dictate its target, but the content-character distinction they draw does not directly match my target-model distinction. More broadly, I do not think the model-target distinction can be straightforwardly identifed with familiar distinctions like attitude-content, sense-reference, or predicatesubject. This is for the simple reason that imaginative models can contain recreations of the states within which we draw all of these distinctions. For example, hoping that Hesperus is Phosphorus and imagining hoping that involves the same attitude (hoping), content (that Hesperus is Phosphorus), subject (Hesperus), predicate (being Phosphorus), senses (“Hesperus”, “Phosphorus”), and referent (the planet Venus). There may be many interesting connections between model-target and these other distinctions, but they are not simply the same. Although recreation and targeting work together, their connection is quite loose: as my above examples illustrated, the same model can stand in for many targets, and the same object can be targeted using many different models. Moreover, not all aspects of what we recreate need to be representing anything about the target. For example, I might tell you about the shape of a country by asking you to “imagine a hexagon”, and telling you that the country is roughly that shape. What you create is an image of a straight-edged regular hexagon (maybe a blue one!), but the straightness of the edges (and the blueness) is irrelevant to what you are targeting (the approximate shape of the country), and is not to be evaluated for accuracy. We might wonder at this point about “idle” conscious intentions. Since there is no obvious way for such imaginings to be inaccurate, we might think they have no target: this would be perfectly compatible with the account I am offering. Not all imaginings need to have targets. Alternatively, we might think of them as targeting something like “objects

288

Luke Roelofs

that look like this”, so that they are always trivially accurate. This would have the advantage of allowing idle imaginings to constrain what counts as accuracy for any further thoughts that aim to “build on” them. I think that recreating and targeting can be varied independently of each other more or less however we want, at least when we form explicit intentions. However, in practice we often do not, and then the two intertwine because our minds (fortunately) tend to work in ways that tend towards accuracy. If I set out to imagine an octopus, I will usually form an image that would be fairly accurate, and if I form an image that is inaccurate (like the squid-image above), it will usually not be wildly inaccurate (not an image of the empire state building, for instance). Conversely, if an octopus-like image pops into my head unbidden, I am likely to take it as standing in for octopuses, and not for American architecture. By “I am likely to take it as” standing in for (i.e. targetting) octopuses, I mean two things: frst, that if I do have any explicit thoughts about “what it is” (i.e. about what I represent by forming this image), those explicit thoughts are likely to say “it’s an octopus”, and second, that even if I never think explicitly about “what it is”, I am likely to “proceed from it” (i.e. draw inferences, form further images, have thoughts, dredge up memories) in ways that pertain to octopuses, not to American architecture. Plausibly, in the second sort of case (where there is no explicit thought about what I am targetting), these functional facts about how I proceed (what role the image plays in my ongoing mental life) are what constitute it being about, i.e. targetting, one thing rather than another. Because of the psychological and constitutive facts just canvassed, it will often be sensible to assume that an image (or other sort of imaginative model) targets whatever it would be most accurate in targeting, and thus that what I imaginatively target is fxed by what I imaginatively recreate. But, I claim, this is just a rough-and-ready heuristic, and the two imaginative relations have no fxed correspondence.

14.4 The Complexity of Empathy Using this distinction between model and target, let us return to the three cases in Section 14.1. I said there that I think all three involve substantively the same, complex, cognitive process: the differences in language used refect differences in the presence or prominence of different parts of this process, but not differences in what is, in the frst instance, going on. Now I can explain this claim of substantive sameness more fully. 14.4.1 Building Models and Shifting Targets My basic claim is that cases 1 and 2 can involve different targets (a counterfactual situation, and the other’s perspective on the actual situation) but the very same model. They involve the same model insofar as they

The Complexity of Empathy

289

all recreate the same suite of connected mental states, namely those that are typically involved in suffering the kind of wrong done by A. B actually underwent that experience, in the counterfactual scenario where A is the victim, A has that experience. The model matches both, but just like the same image can represent different sights, this model could be used to represent B’s actual experience or A’s possible experience with equal accuracy. Of course in practice there may be lots of variation in the model created: A might include or emphasise different details in case 1 that they do not in case 2 (see below for fuller discussion), but there is no necessary difference in model between the two cases. In that sense, A “imagines the same thing” in both cases. I said that this shared model was being used to represent two different targets, but that is not the whole story. It is true that in case 1, it seems that A sets out to comply with C’s instruction to imagine a counterfactual (“if they did that to you”), and so we should take their intention as fxing their target: the counterfactual scenario. No such counterfactual is mentioned in case 2, where A seems to intentionally target the actual situation. But despite this difference in initial targets, I think that in both cases A ends up targeting B’s perspective on the actual situation. In case 1 they just come at that target more indirectly, frst recreating the suite of mental states typical of being a victim of that kind of wrong (organised by the intention to imagine the counterfactual scenario), but then keeping that model while shifting its target to be the actual situation. This shift, I think, is easy precisely because there is no change in the model, but only in what the subject takes it to stand in for. To motivate the thought that there is this fnal target-shift, contrast case 1 with case 4: Case 4: A has wronged B, but doesn’t really care. C remonstrates with A, saying “how could you do that to B? Imagine if B did that to you! How would you feel?” A imagines that: they imagine B wronging A in the same way. They fnd themselves feeling outraged and hurt about “what B has done”. They turn to C and say “you’re right! Thank goodness B didn’t do that to me, that would be awful. I’m so relieved that this is just a counterfactual, it really helps me appreciate my actual situation more by contrast”. I think A in this case has complied with the letter of C’s instruction perfectly well: they have imagined if B did that to them, just as accurately and insightfully as in case 1. But they have clearly not done everything that C hoped for, because they have not changed their attitude towards B’s actual situation. C’s aim (and A’s aim insofar as A is trying to go along with C) is to use the counterfactual to sharpen A’s grasp of the actual situation, and when that does not happen, C’s remonstrance has failed, even if A successfully imagines the counterfactual.

290

Luke Roelofs

All I will insist on is that the overall aim and target of the imagining is the same in cases 1 and 2, with the difference in immediate targets being subsidiary to this. But a stronger claim could also be made: that at least sometimes, the imagining process involved cannot be cleanly divided between self-imagining and other-imagining. I suggested above that what a given imagining targets might be fxed partly by functional facts about what further role it plays in the subject’s mind. If A, as soon as they form this model (i.e. connected offine versions of the states typical of being wronged), is disposed both to infer things from it about a fanciful counterfactual (e.g. to worry they would not respond gracefully, or to resolve to take precautions against it) and to change their attitude to the actual B (e.g. to feel regret and shame, and resolve to apologise), then these dispositions might support thinking that they target two different targets at once with the same model, or that what they target with that model is indeterminate, or (if the dispositions develop at different rates) that they targeted frst one thing and then another, but without any determinate moment when the shift happened. And arguably this is implicit in C’s instruction in case 1: case 4 illustrates that simply imagining the counterfactual scenario is not really complying with what C clearly meant. So if A’s target is fxed by their intention, and their intention is to follow C’s instruction, then their target again might be both self and other. So the frst of the three distinctions I noted in Section 14.1, regarding the identity of the person imagined, is not always in practice sharp. It is, I think, easy to show that the other two are likewise prone to intermediate forms. With the issue of psychological adjustment, it is clear that more or less adjustment may be made: there are a hundred ways in which any two people differ, and even if I make some adjustments to my imagining to refect some of these differences, there are more that I could make. And although the distinction between imagining “from the outside” and “from the inside” is formally binary, the same imagining may be from inside some objects but from outside others. In Vendler’s famous example, when I imagine swimming in the sea “from the outside”, i.e. imagine seeing myself swimming, I may still be imagining “looking down at the sea from a cliff” from the inside. So without fxing which particular feature we are talking about, we will not be able to exhaustively distinguish imaginings “from the inside” and “from the outside”. So the very same, dual-targetted, imagining could be appropriate for both case 1 and case 2. This point is reinforced when we turn to consider case 3. I don’t think this case really involves a different mental process than cases 1 and 2; rather, I think the linguistic form used here (“imagine if you were B”) is just a tool for effciently gesturing towards the same overall mental process as in cases 1 and 2: recreating the experience of victimhood and targeting it at B’s actual experience, perhaps by frst setting out to target a self-involving counterfactual. The seemingly impossible counterfactual of “if A were B” is a red herring: the point is to

The Complexity of Empathy

291

convey concisely that it is B whose perspective one is concerned with, but that what one should imagine is their frst-person perspective. 14.4.2 Imagining From the Inside and From the Outside The variety of idioms here, and the complex connections of ultimate and immediate targets, might seem messy and confused. But I think they are actually functional: different forms of language can make salient different challenges and risks involved in empathic role reversal. Consider frst the question: why bother with the self-involving counterfactual? Why not simply tell A to imagine how B feels? I think we can see the value of the detour through self-imagining by considering another case where the remonstrance fails: Case 5: A has wronged B, but doesn’t really care. C remonstrates with A, saying “How could you do that to B? Imagine doing that to B, how do you think they feel?” A, a little confused, dutifully imagines doing what they did to B, and imagines that B feels hurt and outraged. But they are not moved; instead they reply: “I don’t have to imagine it, I know it already: I did that, and yeah B is mad about it, but whatever. Imagining the actual situation isn’t going to change how I feel about it”. Like in case 2, here the only situation targeted is the actual situation; yet there is seemingly a big difference in how it is imagined. Case 2 is more like case 1 than like case 5, because in cases 1 and 2, the model constructed is “from the victim’s perspective”, i.e. it recreates the sorts of mental states a victim would typically have (beliefs, perceptions, memories, and then frustration, hurt, a sense of betrayal, etc.). But in case 5, A is clearly using a very different model – one that imagines B’s experience “from the outside”, to use the third distinction I noted in Section 14.1. They might be using a model that is “from the perpetrator’s perspective”, i.e. that recreates the same kinds of mental states that A already went through, when they committed the wrong in the frst place. Alternatively, they might be using a model that is “from a third-person perspective”, recreating little more than beliefs about what in fact happened. In either case, since they remember performing the action, and already know it happened, there is nothing new for them to learn from this imagining. This sort of imaginative model is clearly ill-suited to accomplish what B is trying to accomplish. And the formation of models like this is arguably a major problem when trying to cultivate empathy. If A simply thinks about the actual case, about which they already have many beliefs, feelings, memories, etc., they might struggle to avoid forming a model that is contaminated by their own perspective – that privileges

292

Luke Roelofs

how they see things and thereby obscures how B sees things. Their natural, self-involved, view on their own life forms a barrier to whatever in B’s perspective might be uncomfortable or unfamiliar. The value of the detour through self-imagining in case 1 is to help bypass this barrier. By turning aside from the actual world, A is better able to form a model that really matches the experience of suffering, as opposed to committing, the wrong in question. Their natural selfinvolvement is recruited to subserve empathy, rather than standing in its way. By starting with that self-referential intention, they cannot help but imagine “from the victim’s point of view”, i.e. what they recreate will be the vivid, outraged, experiences of a victim, and not the bored experiences of a perpetrator, or the bloodless belief that such and such has taken place. And when that model is in place, the remonstrator’s hope might run, the shift in target is not so hard. 14.4.3 The Risk of Egocentric Bias But there is an opposite risk, which a detour through self-imagining exacerbates rather than mitigates, namely the risk of “egocentric bias”, of forming an imaginative model that refects the peculiarities of one’s own character, or fails to refect the peculiarities of the other’s character. This risk is at the centre of Goldie’s critique of empathetic perspective-shifting: he ties the risk of imagining the other too much like ourselves, or at least not enough like the other themselves, specifcally to other-imagining, which takes the other as its target. But the degree of egocentrism is a matter of the particular psychological states that form the imaginative model, which is dissociable from the imaginative target. To see this, consider fnal two cases: Case 6: A has wronged B, but doesn’t really care. C remonstrates with A, saying “how could you do that to B? Imagine if B did that to you! How would you feel?” At frst, A imagines that: they imagine B wronging A in the same way. But they fnd themselves feeling no outrage or hurt, but merely a mild annoyance. When they report this, C objects: “that’s because you and B are such different people: here are some ways that B’s background and character make what you did especially bad…” After hearing these differences pointed out, A tries again, imagining themselves not only suffering the same actions as B did but also having the same relevant background and character as B. The hurt and outrage are now vivid, and A realises how badly they’ve acted. Over the course of the conversation, A changes their model in such a way as to make it a more accurate model of B’s actual perspective. Both before and after, the explicit target of their imagining is a self-involving

The Complexity of Empathy

293

counterfactual scenario, just a more remote counterfactual in the latter case. But this fact is not the important one, as we see if we compare case 6 with the very similar case 7: Case 7: A has wronged B, but doesn’t really care. C remonstrates with A, saying “how could you do that to B? Imagine if you were B! How would that feel?” (a) A imagines that: they imagine being B, suffering this action performed by A. But they fnd themselves feeling no outrage or hurt, but merely a mild annoyance. (b) When they report this, C objects: “that’s because you and B are such different people: here are some ways that B’s background and character make what you did especially bad…” After hearing these differences pointed out, A tries again, imagining being B, but now with a more accurate model that incorporates relevant background and character. The hurt and outrage are now vivid, and A realises how badly they’ve acted. Here the change in model, to reduce its egocentrism, has nothing to do with a change in target; it is just about getting a more accurate model of the same target. But it seems to me that what A is doing here is still the same complex activity as in case 6 (just like cases 1, 2, and 3 all involve the same complex activity), and that the ultimate point in both cases is to end with the right kind of model for understanding, and being moved by, B’s actual experience. The diverse explicit targets, like the counterfactual where A has different character traits, are only instrumental tools for helping A to construct that model. Telling A specifcally to imagine B’s perspective may be a way to make salient the risk of egocentric bias. Conversely, telling A to imagine themselves in B’s situation may be a way to make salient the opposite risk, of imagining what happened only “from the outside”, and fail to escape from one’s existing view of the events. To pair with “egocentric bias”, we might term this “allocentric indifference”. The point is that empathic role reversal can go wrong in at least two opposite ways: the imaginer might fail to construct a model that captures the relevant type of experience “from the inside” (as in case 5), succumbing to allocentric indifference, or they might fail to construct an accurate model, by incorporating their own character and not the relevantly different character of the other (as in cases 6 and 7), succumbing to egocentric bias. Different linguistic forms serve to highlight and guard against different such failings: “imagine if it happened to you” emphasises that the model constructed should involve the kinds of states associated with suffering the wrong, not with committing it or simply knowing about it; “imagine if you were them” emphasises that the model constructed should refect the specifc characteristics of the other person; “imagine how they must (actually) feel” emphasises that the model is ultimately meant to target

294

Luke Roelofs

something actual, not just a counterfactual. But it would be a mistake, I think, to fx on these different linguistic forms, and take them to be calling for distinct cognitive acts, which could then be compared and critiqued relative to one another. The act called for is grasping the other’s situation from their perspective, which may require frst targetting other scenarios in order to get the right model: this act is complex, and the same in each case. But all of these challenges are what Langkau calls “psychological” and “epistemic” challenges (this volume, cf. Epley and Caruso 2009, compares the more general challenges of imagining skilfully discussed in Kind 2018, 2020): they involve things which are hard for us to do, or which we may lack the necessary knowledge to do. They are not, as Langkau puts it, “conceptual challenges”, inherent problems with the very possibility of empathic role reversal, or with any particular form of it.

14.5 Conclusions Of course my “build the model, then shift the target” account is not the only way to theorise empathic role reversal. In particular, one might naturally think of C’s aim as being for A to learn a general truth from considering the counterfactual. For example, Nagel writes: If something I do will cause another creature to suffer, that counts against doing it. I can come to see that this is true by generalizing from the evident disvalue of my own suffering. (Nagel 2012, p. 77) If A learns a general principle about the disvalue of some experience from self-imagining, they could then apply that general principle to their actual situation: the principle is what mediates between imagining the counterfactual and grasping the actual. I have not here refuted or objected to that account, and I think we often do draw general lessons from our imaginings, and apply them to new cases. But I think sometimes we also perform the more direct target-shifting process I have described, and having that account in view is useful for, I believe, a few reasons. It underlines how exceedingly close the processes in cases 1, 2, and 3 may be, and thereby explains why we often feel that it makes little difference which of the three linguistic forms we use. The target-shift account also, I think, better captures the concrete, frst-personal character of the cognition in these cases – A does not need to try and formulate a general principle that explicitly captures all the relevant similarities between the counterfactual and actual scenarios: they can simply recognise the two scenarios as similar. And if we thought that imagining things from others’ perspectives was of foundational importance in morality (as on the

The Complexity of Empathy

295

views of Smith 1976; Nichols 2004; and others), then we might think that the general-principle account puts the cart before the horse: we arrive at general moral principles by abstracting from empathy with others, rather than relying on them to enable it. I would add that the complexity of empathic role reversal, and the diversity of seemingly conficting terminology for describing it, refects the diffculty of empathising well. Empathy is a skill we have to keep learning, an ongoing balancing act between drawing the other too close, substituting an image of them modelled after ourselves, and letting them drift so far away that we lose sight of their humanity. Trying to understand others through imagination is, I believe, vital and noble, but it needs to be done in the keen awareness of how limited that understanding will remain, and with a resolve not to let our compassion be restricted by differences that challenge it.

References Abell, Catharine. 2005. “Against Depictive Conventionalism.” American Philosophical Quarterly 42(3), 185–97. Abell, Catharine. 2007. “Pictorial Realism.” Australasian Journal of Philosophy 85(1), 1–17. Balcerak Jackson, Magdalena. 2018. “Justifcation by Imagination.” In Perceptual Memory and Perceptual Imagination, edited by F. Dorsch and F. Macpherson, 209–26. Oxford: OUP. Batson, C.D. 2009. “Two Forms of Perspective Taking: Imagining How Another Feels and Imagining How You Would Feel.” In Handbook of Imagination and Mental Simulation, edited by K.D. Markman, W.M.P. Klein, and J.A. Suhr, 267–79. New York: Psychology Press. Batson, C.D., Early, S., and Salvarani, G. 1997. “Perspective Taking: Imagining How Another Feels versus Imagining How You Would Feel.” Personality and Social Psychology Bulletin 23, 751–8. Coates, Ta-Nehisi. 2011. “A Muscular Empathy.” The Atlantic. https://www. theatlantic.com/national/archive/2011/12/a-muscular-empathy/249984/ Coeckelbergh, Mark. 2007. “Who Needs Empathy? A Response to Goldie’s Arguments against Empathy and Suggestions for an Account of Mutual Perspective-shifting in Contexts of Helping and Care.” Ethics and Education 2(1), 61–72. Compton, John. (2020). The End of Empathy: Why White Protestants Stopped Loving Their Neighbors. Oxford: Oxford University Press. Currie, Gregory. 1995. Image and Mind: Film, Philosophy and Cognitive Science. Cambridge: Cambridge University Press. Currie, Gregory, and Ravenscroft, Iain. 2002. Recreative Minds: Imagination in Philosophy and Psychology. Oxford: Clarendon Press. Dokic, Jérôme, and Arcangeli, Margherita. 2015. “The Heterogeneity of Experiential Imagination.” In Open MIND. Philosophy and the Mind Sciences in the 21st Century, edited by T.K. Metzinger and J.M. Windt, 431–50. Cambridge, MA: MIT Press.

296

Luke Roelofs

Dorsch, Fabian. 2012. The Unity of Imagining. Germany: De Gruyter. Epley, Nicholas and Caruso, Eugene M. 2009. “Perspective Taking: Misstepping into Others’ Shoes.” In Handbook of Imagination and Mental Simulation, edited by K.D. Markman, W.M.P. Klein, and J.A. Suhr, 295–309. New York: Psychology Press. Giovannelli, Alessandro. 2008. “In and Out: The Dynamics of Imagination in the Engagement with Narratives.” Journal of Aesthetics and Art Criticism 66(1), 11–24. Goldie, Peter. 2011. “Anti-Empathy.” In Empathy: Philosophical and Psychological Perspectives, edited by P. Goldie and A. Coplan, 302–18. Oxford: Oxford University Press. Goldman, Alvin. 2006. Simulating Minds: The Philosophy, Psychology, and Neuroscience of Mindreading. Oxford: Oxford University Press. Gordon, R.M. 1995a. “Sympathy, Simulation, and the Impartial Spectator.” Ethics 105, 727–42. Gordon, Robert. 1995b. “Simulation without Introspection or Inference from Me to You.” In Mental Simulation: Evaluations and Applications - Reading in Mind and Language, edited by Martin Davies and Tony Stone, 352–66. United Kingdom: Wiley. Greenberg, Gabriel. 2018. “Content and Target in Pictorial Representation.” Ergo 5, 865–98. Gaut, Berys. 1998. “Imagination, Interpretation, and Film.” Philosophical Studies 89, 331–41. Jamison, Leslie. 2014. The Empathy Exams: Essays. United States: Graywolf Press. Kind, Amy. 2013. “The Heterogeneity of the Imagination.” Erkenntnis 78(1), 141–59. Kind, Amy. 2018. “How Imagination Gives Rise to Knowledge.” In Perceptual Memory and Perceptual Imagination, edited by Fabian Dorsch and Fiona Macpherson, 227–46. Oxford: Oxford University Press. Kind, Amy. 2020. “What Imagination Teaches.” In Becoming Someone New: Essays on Transformative Experience, Choice, and Change, edited by J. Schwenkler and E. Lambert, 133–46. Oxford: Oxford University Press. Langland-Hassan, Peter. 2020. Explaining Imagination. Oxford: Oxford University Press. Langton, Rae. 2019. “Empathy and First-Personal Imagining.” Proceedings of the Aristotelian Society 69(1), 77–104. Nagel, Thomas. 2012. Mind and Cosmos: Why the Materialist Neo-Darwinian Conception of Nature is Almost Certainly False. Oxford: Oxford University Press. Nanay, Bence. 2010. “Adam Smith’s Concept of Sympathy and its Contemporary Interpretations.” Adam Smith Review 5, 85–105. Nichols, Shaun. 2004. “Review of Currie and Ravenscroft, Recreative Minds.” Mind 113, 329–34. Obama, Barack. 2006. The Audacity of Hope: Thoughts on Reclaiming the American Dream. New York: Crown Press. Recanati, François. 2007. “Imagining de se.” https://jeannicod.ccsd.cnrs.fr/ ijn_00160757

The Complexity of Empathy

297

Salesses, Matthew. 2020. “The Empathy Economy Is a Sham. The Protest Movement Is Real.” The Daily Beast. https://www.thedailybeast.com/ the-empathy-economy-is-a-sham-the-protest-movement-is-real Serano, Julia. 2017. Empathy Politics. Medium.com. https://medium.com/ @juliaserano/empathy-politics-d7f62aa90e75 Smith, A. 1976. The Theory of Moral Sentiments, D.D. Raphael and A.L Macfe (eds), Oxford: Clarendon Press. Smith, M. 1995. Engaging Characters. Oxford: Oxford University Press. Smith, M. 1997. “Imagining from the Inside.” In Film Theory and Philosophy, edited by R. Allen and M. Smith, 412–30. Oxford: Oxford University Press. Stinson, Liz. 2020. “The Empathy Economy Is Booming, but What Happens When Our Emotional Connections to Others Are Designed, Packaged, and Sold?” AIGA Eye of Design. https://eyeondesign.aiga.org/the-empathyeconomy-is-booming-but-what-happens-when-our-emotional-connectionsto-others-are-designed-packaged-and-sold/ Stotland, E. 1969. “Exploratory Investigations of Empathy.” In Advances in Experimental Social Psychology, edited by L. Berkowitz, vol. 4, 271–313. New York: Academic Press. Tagliafco, Daniela. 2011. “Can we really speak of ‘pretend desire’?” Logic and Philosophy of Science 9(1), 461–67. Vendler, Zeno. 1984. The Matter of Minds. Oxford: Clarendon Press. Velleman, David. 1996. “Self to Self.” The Philosophical Review 105(1), 39–76. Walton, Kendall. 1990. Mimesis and Make-Believe: On the Foundations of the Representational Arts. Cambridge, MA: Harvard University Press. Williams, Bernard. 1973. “Imagination and the Self.” In Problems of the Self, edited by B. Williams, 26–45. Cambridge: Cambridge University Press. Wirling, Ylwa. 2014. “Imagining Oneself Being Someone Else – The Role of the Self in the Shoes of Another.” Journal of Consciousness Studies 21(9–10), 205–25. Wollheim, Richard. 1984. The Thread of Life. Cambridge, MA: Harvard University Press.

15 Imagination, Selves, and Knowledge of Self Pessoa’s Dreams in The Book of Disquiet Nick Wiltsher and Bence Nanay 15.1 Introduction to the Text(s) The Portuguese writer Fernando Pessoa was extraordinary, not least because he was multiple. Pessoa cultivated numerous “heteronyms” and “semi-heteronyms”: alter egos with their own manners, styles, and biographies. Critics differ over their number, but there were certainly many; Paulo de Medeiros says that the tally “could easily approach, if not exceed, a hundred” (Medeiros 2013, 10). Most of the works attributed to Pessoa’s heteronyms were among the vast cache of his writings discovered and published following his premature death in 1935. As one might expect, given this multiplicity, Pessoa is especially insightful regarding the relations and importance of imagined selves to the actual self. One would not so proliferate alter egos unless one were convinced of the artistic and existential signifcance of such shadowing of the self. We wish here to explore some aspects of what Pessoa has to say on this subject in The Book of Disquiet. For Pessoa, we suggest, imagined selves facilitate the adoption of perspectives on the actual self that disclose valuable self-knowledge. This suggestion bears upon philosophical questions about imagination’s relation to the self, including those of the extent to which one can adopt perspectives dissociated from one’s own, and of the role of such imagined perspectives in the constitution of the self. Pessoa’s authorial practice piques interest even before one considers the contents of The Book of Disquiet. This posthumously published book consists mostly of writings Pessoa composed under the name of Bernardo Soares. Soares is employed as a clerk in Lisbon in the late 1920s and early 1930s. He defrays his dreary existence by recording it in a series of reveries, refections, and reports. These are presented in the book as numbered fragments, with almost no narrative links among them. Pessoa published very few of these fragments during his lifetime, despite working on the project for years. He left behind the title, and some writings clearly intended to fall under it; but he left many more among his papers, notebooks, and scraps which may or may not belong to the project, and only the sketchiest indication of how they ought to be ordered and arranged.1 Selection and presentation are thus editorial matters, so there are numerous editions, none of which claims absolute

Imagination, Selves, and Knowledge of Self

299

authority. We will refer to the 2017 New Directions edition, which is a translation into English by Margaret Jull Costa, mostly following the selection and ordering of Jerónimo Pizarro’s 2010 critical edition (Livro do Desassossego, published by Imprensa Nacional-Casa da Moeda). 2 So far, so relatively conventional. But Soares is neither a pseudonym nor a fctional character. A pseudonym is usually a false name adopted for practical reasons. Mary Ann Evans, David Cornwell, and Erika Leonard were all motivated to invent pen names, but none were trying to create a substantial alter ego. A fctional character, on the other hand, has substance and personality to go with their name, but their relation to their author is quite clear: one creates the other, and adopts their perspective only for the ends of the fction. 3 Some of Pessoa’s heteronyms approach the condition of either pseudonym or character, but the most important ones are neither. By his own account, heteronyms such as Alberto Caeiro were created in extended periods of inspiration, in which they became mysteriously manifest within him and worked through him. Medeiros quotes from Pessoa’s letters thus: I wrote thirty-some poems at once, in a kind of ecstasy […] I began with a title […] this was followed by the appearance in me of someone I instantly named Albert Caeiro. Excuse the absurdity of this statement: my master had appeared in me. (Pessoa 2001, 256, quoted in Medeiros 2013, 11) There is romantic excess in this telling; Pessoa deliberately created his heteronyms, crafting their biographies and cultivating their outputs. Nonetheless, he thought of the main heteronyms as substantial selves, distinct and partially autonomous from their creator, and standing in a complicated relation with him. As Medeiros puts it, Pessoa’s practice reveals “a recognition of the importance of relating to the Other in order to be himself, but also a defection of the Self’s own authority and authorship” (Medeiros 2013, 11). Unlike Caeiro, Soares is what Pessoa describes as a “semi-heteronym”, less distinct from the author. As Pessoa himself put it: even though [Soares] is not my own personality, he is not so much different from myself as he is a mere distortion of that personality. He is me without my rational and emotional aspects. The prose, except for what in mine seems reasoned, is the same as mine, and the Portuguese is completely the same.4 So the personal, confessional, diaristic thoughts Soares expresses bear close relation to Pessoa’s own. Several scholars nonetheless attribute the ideas in the book to Soares (see 217 for textual support). However, Jerome Boyd Maunsell argues that this attribution is contestable,

300

Nick Wiltsher and Bence Nanay

principally because Pessoa wrote some of the fragments under the aspect of Vicente Guedes; he says that Portuguese editions often attribute authorship to Guedes/Soares, while English editions generally do so to Soares alone (Boyd Maunsell 2012, 122). Medeiros, meanwhile, argues that it is best to treat the work as Pessoa’s, rather than the work of any alter ego (Medeiros 2013, 10–4). So it is reasonable, and simplest for our purposes, to follow the practice of critics who talk of the ideas in the book as Pessoa’s. What of those ideas, then? The mood of The Book of Disquiet is uneven: passages are ironic and literal, bleak and redemptive, funny and frightening, oppressive and enlightening. A somewhat surprising kind of happiness might be the dominant chord throughout the book. But, for the most part, Pessoa’s meditations tend to the titular disquiet. For all his occasional ability to see light and fnd joy in his situation, Pessoa is troubled by it, troubled in some deep and obscure way. His trouble is related, though not identical, to the fact that he is bored, consumed by tedium and surrounded by monotony. The line of thought we wish to pursue concerns how imagined selves, and imagined situations, contribute both to his happiness and to his disquiet. The line is as follows. We frst set out a tempting account of the importance of imagination in The Book of Disquiet. On this reading, Pessoa is immersed in miasmatic boredom, but able to temporarily rise above it through the restorative powers of escapist imagination. We next introduce a puzzle for this reading. If imagination is important because it allows escape, those who imagine often should be content. But Pessoa often dreams, and yet is often disquieted. To resolve this puzzle, we argue that his disquiet owes to failures of imagination; the book’s great irony is the contrast between Pessoa’s keen awareness of the importance of a successful, active imaginative life, and his failure to maintain such a thing. We argue further that these failures point us toward a better account of imagination’s importance to Pessoa, an account that both prefgures and goes beyond Sartre’s ideas about the relations between nothingness, imagination, and self-consciousness. We return briefy in conclusion to how we might see Pessoa’s own practice in the light of this account. Two caveats should be entered, one terminological, one interpretative. Terminology: Pessoa often refers to “dreaming” and cognates where we might more usually say “imagining”, and he uses “imagination” and cognates sometimes to refer to something like the ability to dream, and sometimes to refer to something more like a grand Kantian faculty. 5 We are concerned with the ability to conjure alternative worlds and lives, and so we ignore passages where Pessoa seems to be invoking the second kind of “imagination”, and we assimilate “dreams” to “imaginings”. Interpretation: The Book of Disquiet is vast. Complete English translations run to nearly 900 pages. It is also dense, rich, and twisting.

Imagination, Selves, and Knowledge of Self

301

Further, as Boyd Maunsell points out, the book was written in distinct phases, each with its own characteristics and themes (Boyd Maunsell 2012, 120). Given all this, we do not think our reading of Pessoa is exhaustive or defnitive. There are many relevant passages to which we make no reference, and many that would contradict or complicate our suggestions.6 The fact that other ideas about imagination can be found in his work shouldn’t vitiate the interest of those we discuss.

15.2 Imagination as Escape? Pessoa is certainly convinced that imagination is important to him: “periods of daydreaming”, he says, “though devoid of either purpose or dignity, still constitute the greater part of the spiritual substance of my life” (188). Moreover, these periods occupy a great deal of his life and his attention. But why is imagination important to him? A tempting, plausible answer to this question appeals to the escapist possibilities of imagination. According to this reading, Pessoa is utterly bored with his job and his life, and cannot foresee either progressing in a satisfactory direction. To alleviate this tedium, he indulges in fights of fantasy, fnding happiness in imagined alternatives to his existence. Pursued at length, such imaginings allow him not just to accommodate himself to his life, but even to relish it, precisely because its strictures allow for endless escapism. Thus, Abigail Schoneboom writes, Pessoa’s life becomes a “carefully managed monotony that renders small events thrilling and frees the mind to dream”, in which he recognizes that “actual achievement of ambitions destroys the infnite pleasure that derives from unfulflled dreams”, and so remains “opposed to advancement”, instead “[r]elishing his clerkdom as a sublime realm of the possible” (Schoneboom 2015, 839). With different emphasis but similar effect, Thomas J. Cousineau observes that individual fragments of The Book of Disquiet often interweave experience (or feeling), thought, and imagining, and suggests that, often, unpleasant experience and “disillusioned” thought “leads [Pessoa] to fnd relief from self-awareness in a compensatory dream” (Cousineau 2013, 18; emphasis in original). This reading is tempting because many fragments describing (day) dreams do indeed have a favor of escapist fantasy. For example, Pessoa writes: In the midst of my day-to-day work, dull, repetitive, and pointless, visions of escape surface in me, vestiges of dreams of far-off islands, parties held in the avenues of gardens in some other ages, different landscapes, different feelings, a different me. (344) Similarly, he describes an episode where

302

Nick Wiltsher and Bence Nanay I write carefully, bent over the book […] and at the same time and with equal attention, my thoughts follow the route of an imaginary ship through oriental landscapes that have never existed […]. I watch intently the rows of deckchairs and the stretched-out legs of people relaxing on the voyage. (293)

Here and elsewhere, Pessoa describes the interpolation of visions and memories and daydreams with his awareness of his surroundings and activities. Though usually described in sensory terms, some of these dreams are less obviously based on imagery, as when, [o]n the same page that contains the names of unfamiliar textiles, the doors to the Indus and to Samarkand swing open, and the poetry of Persia […] lends distant support to my disquiet in quatrains whose every third line is left unrhymed. (187) The escapist interpretation of imagination’s importance is thus textually supported, but it is also textually frustrated. Both the displayed quotations above are more complicated when read in full. The frst continues thus: I realize that if I had all that, none of it would be mine […]. I can enjoy my inner visions of non-existent landscapes. But if the [dreams] were mine, what would I have to dream about? (344) The second, meanwhile, is prefaced by the assertion that I’m always thinking and listening to two things at once […]. [I]n my case the two realities I attend to have equal weight […]. [I]n that, perhaps, lies both my tragedy and the comedy of my tragedy. (293) A good escapist, a Walter Mitty or a Don Quixote, enjoys their escapes (see also Currie and Ravenscroft 2002). Moreover, as Schoenboom suggests, escapist fantasies are supposed to make everyday life more tolerable. Pessoa’s periods of daydreaming are prolonged and almost refexive; he says that “[i]n me the habit of dreaming and the ability to dream are primordial […]. I’m not merely a dreamer, I am exclusively a dreamer” (65). But he does not straightforwardly enjoy his imaginings. If he did, it would be hard to make sense of his claim that “my happiest hours are those in which I think nothing, I want nothing, when I do not even dream” (160). And his escapes do nothing much to alleviate the

Imagination, Selves, and Knowledge of Self

303

problems of his quotidian existence. He imagines often, yet he remains disquieted. So here is the puzzle: if Pessoa is an accomplished escapist, he ought to enjoy his imaginings, and he ought to fnd his daily life more tolerable for that indulgence. He imagines prolifcally, yet he fnds little enjoyment in his dreams and little to like about his everyday existence. We wish now to suggest that, despite appearances, he fails as an escapist because he is not good at imagining, and that the ways in which he fails suggest a better account of the importance of imagination.

15.3 Why does Imagination Fail? Our claim that Pessoa is not very good at imagining is at odds with the escapist reading, and seemingly contradicted by passages such as this: Because I am nothing, I can imagine myself to be anything. If I were somebody, I wouldn’t be able to. An assistant book-keeper can imagine himself to be a Roman emperor; the King of England can’t do that, because the King of England has lost the ability in his dreams to be any other king than the one he is.7 (344) In fact, this quotation contains the key to understanding Pessoa’s failures of imagination. The general idea is that our reality limits what we can imagine. The more one is certain about one’s own reality, the less well one is able to imagine beyond it. This is a major insight, and an interesting converse of the thought that a lack of knowledge about other’s situations inhibits attempts to imagine them (e.g., Jones 2004; Paul 2014, 7). Much of the remainder of this chapter is about the insight’s consequences. So we need to elucidate exactly what limits one’s “reality” places on imagining. In the quotation, Pessoa might be suggesting that the territory available to imagined selves is that unoccupied by the actual self. The contours of the territory are provided by more or less objective markers of personal success: money, status, fame. If this is the right reading, then Pessoa, manifest as Soares, is well-off: Soares is an assistant bookkeeper, lacking such trappings of success, and so can imagine himself to be pretty much anybody. The uninhabited territory of imaginative success is vast. The poor old King of England, on the other hand, is too squarely situated in kingly realms; the only imaginative hinterland available to him is impoverished. Other passages support this reading, such as the following (although it is not unambiguously about imagination): I often wonder what kind of person I would be if I had been protected from the cold wind of fate by the screen of wealth […] I know that had that non-existent past existed, I would not now be capable

304 Nick Wiltsher and Bence Nanay of writing these pages, which, though they are few, are at least better than all the pages I would undoubtedly have only day-dreamed about given more comfortable circumstances. (242) Here, again, Pessoa suggests that he is a better writer, creator, dreamer for being relatively impoverished. But does lacking a trust fund really provide Pessoa (or anyone else) with such freedom of imagination? In other passages, he claims just as clearly that it does not: Anyone reading the earlier part of this book will doubtless have formed the opinion that I’m a dreamer. If so, they’re wrong. I don’t have enough money to be a dreamer […]. The great dream demands certain social circumstances. One day, captivated by a certain musical plaintiveness in what I had written, I imagined myself to be another Chateaubriand but brought myself up sharply with the realization that I was neither a viscount nor a Breton. On another occasion, when I seemed to notice in my own words a similarity to Rousseau, again it did not take me long to see that I did not have the advantage of being a nobleman or a castellan and, moreover, was neither Swiss nor a vagabond. (436) Besides his failures to inhabit Chateaubriand or Rousseau, he can’t get Caesar right either: How many Caesars have I been, yet I was never like the real Caesars. I was truly imperial in my dreams and for that reason came to nothing. (253) In these quotations, Pessoa’s efforts are frustrated by the distance between his actual life and the one he is trying to imagine—precisely the distance that, in the earlier quotations, is supposed to facilitate imagining. So we get something like a paradox. If Pessoa (or Soares) is free, as a result of lacking luxury, to imagine being anyone, why can’t he imagine being Chateaubriand, Rousseau, or Caesar? Why does he fail repeatedly and systematically at these imaginative projects? Either he can imagine himself royal, imperial, noble, or he can’t. Either he is a dreamer, or he isn’t. The following two quotations hint at a solution to this contradiction: I imagined myself free of Rua dos Douradores, of my boss […] but suddenly, even as I imagined this […] a feeling of displeasure erupted into the dream: I would be sad […]. they have all become part of my life […] I could never leave all that behind […] losing them would be akin to death […]. Moreover, if I left them all tomorrow and

Imagination, Selves, and Knowledge of Self

305

discarded this […] suit, what else would I do? Because I would have to do something. And what suit would I wear? Because I would have to wear another suit. (188) Today is one of those days when the monotony of everything closes about me as if I had just entered a prison […] I’d like to run away, to fee from what I know […] for anywhere, be it village or desert, that has the virtue of not being here […]. but would I really dare go off to this hut or cave, knowing and understanding that, since the monotony exists in me alone, I would never be free of it? (312) Pessoa’s imagination, it seems, fails when the self is salient. The successful imaginings, such as the wanderings of the ship, are impersonal: visions or “vestiges of dreams”, without any personal involvement.8 The imaginings that fail are those in which Pessoa tries to adopt the perspective of a relatively substantial self—to be Caesar or Chateaubriand, or just to be someone near to but distinct from his own self. And this is Pessoa’s tragedy, because successful imagining of other perspectives is what matters for the self (as we will argue later).9 The fact that these imaginings fail suggests a general story about the limits on our imaginative capacities of personal projection. The limitations do not owe to objective aspects of our concrete situation—how rich or successful we are—but rather to our self-image: how much concrete reality we attribute to ourselves. Even an assistant book-keeper can have a frm, immovable self-image, whose fxity makes it diffcult for him to imagine being someone else. If the assistant book-keeper can only see himself in the fxed self-image of an assistant book-keeper, he can’t imagine not wearing a suit. He could imagine wearing a different suit, but not wearing no suit at all. What frustrates imagination is not wealth and fame, but rather taking yourself to be unchangeable: an essentialized self, or sense of self. We will now elaborate two aspects of this general story. The frst concerns the importance of imagination. Somehow, imagining other selves is important for one’s own self: why so? The second concerns the limits of imagination: why does a concrete self-conception inhibit this vital kind of imagining? We start with the second, since answering it allows us to rule out a tempting answer to the frst, and so sets us on course for a satisfactory account of the importance of imagination: briefy, that it opens up salutary perspectives on one’s self.

15.4 Imagination’s Limits If it is true that failures of imagination are due to excessive essentializing of the self, successful imagination seems to require de-essentializing the

306 Nick Wiltsher and Bence Nanay self. After all, if you consider yourself to have certain essential properties, you will not be able to imagine being any self that lacks those properties. The more one loses oneself, the more selves one can imagine freely. And many passages in The Book of Disquiet are exactly about the link between imagination and losing oneself: living so much on one’s imagination actually erodes one’s ability to imagine, especially one’s ability to imagine the real. Living mentally on what is not and cannot be, we are, in the end, unable to ponder even what might really be. (427) I stand on the verandah of life, but it’s not quite of this life […]. If I shut my eyes I continue to see it just because I can’t see it. If I open my eyes I see it no more, because I never really saw anything. Every part of me is a vague nostalgia neither for the past nor for the future. (390) Everything about me is fading away. My whole life, my memories, my imagination and its contents, my personality, it’s all fading away. I continually feel that I was someone else, that I felt and thought as another. (366) I’m a perpetual unfolding of connected and disconnected images— always disguised as something external—that stand between men and the light if I’m awake, or between ghosts and the visible dark if I’m asleep. I really do not know how to distinguish one from the other, nor would I venture to affrm that I’m not sleeping when I’m awake, or that I’m not on the point of waking when I’m asleep. (382) In these quotations, Pessoa describes a kind of de-essentialization, with uncanny, disconcerting aspects. But this loss of self is never, for him, unambiguously negative. Its chief positive aspect is that it allows him to imagine being someone else. The same themes of de-essentializing the self and the benefts of doing so loom large in the works of another 20th-century author known for his unfnished oeuvre, Robert Musil. The title of Chapter 4 of The Man Without Qualities, the chapter in which he explains what the mysterioussounding title of the book is supposed to mean, is “If there is a sense of reality, there must be a sense of possibility”. The sense of reality is about how things are and the sense of possibility is about how things could be. Much of the novel could be interpreted as an interplay between these two attitudes. Crucially, being a man without qualities amounts to

Imagination, Selves, and Knowledge of Self

307

applying the sense of possibility to oneself. Applying the sense of reality to oneself would amount to taking our qualities for granted—taking them to be essential. But applying the sense of possibility to oneself (that is, being able to imagine ourselves to be different from the way we are) is a key anti-essentialist move. To put it simply, according to Musil, one can have a frm, essentialist view of oneself or one can have a less essentialist view. The extremity of this latter would amount to being a man without qualities (Nanay 2014). Here is an everyday example. You go to the local cafe and order an espresso. Why? Just a momentary whim? Trying something new? Maybe you know that the Italian owner would disapprove if you ordered a cappuccino after 11 am? Or are you just an espresso kind of person? It is very likely that the last of these options best refects the reasons for your choice. You do much of what you do because you think it accords with the kind of person you think you are. You order eggs Benedict because you’re an eggs Benedict kind of person. It’s part of who you are. And this goes for many of our daily choices. You go to the philosophy section of the bookshop and the fair-trade section at the grocer’s because you are a philosopher who cares about global justice, and that’s what philosophers who care about global justice do. We all have fairly stable ideas about what kind of people we are. And this is possible because we tend to put ourselves into preexisting categories—espresso person, foodie, progressive—and keep ourselves in them. But this is not a necessity. We can also take our self-image less seriously, or at least recognize that it does not derive from impartial observation of our essential qualities. This seems to be Musil’s preferred option. Pessoa almost certainly read at least the frst volume of Musil’s book, given how closely he was engaged with literary happenings in Europe and given Musil’s popularity among the 1920s French literati (Pessoa’s French was impeccable). And it is diffcult not to see Musil’s idea of the sense of possibility behind some of Pessoa’s thoughts regarding dreaming and being a dreamer. Pessoa’s distaste for labeling or essentializing oneself is explicit in his writings, for example: From birth to death man lives life enslaved by the same external concept of self […] He follows norms he neither knows exist nor knows himself to be guided by. (176) And this Pessoa passage sums up the general idea of being a man without qualities succinctly: I prefer not to label myself but to be only obscurely who I am and to enjoy the piquancy of being unpredictable even to myself. (323)

308

Nick Wiltsher and Bence Nanay

But Musil and Pessoa pursued their shared concerns in different directions. While Musil explored various sides (not just the positive, but also the negative sides) of being a person without qualities, Pessoa is more preoccupied with the idea that successful imagination requires being a person without qualities. This line of interpretation needs to be defended against the view that imagination’s importance to Pessoa is to do with “knowing yourself”. For if imagining well requires that one has no essential qualities, it seems there is no stable, metaphysically substantial self that one can know. So our reading and the self-knowledge view cannot both be correct. Like the escapist interpretation, the self-knowledge view has scholarly and textual support. Rhian Atkin, for example, identifes Pessoa as an “intra-fâneur”, arguing that his “apparent observations of the world almost always instigate a closer examination of the self”, and that he uses dreaming and imagining to “reach a heightened understanding of himself and of the world in relation to him” (Atkin 2010, 161, 163). Atkin’s interpretation is supported by passages in which Pessoa does indeed seem to suggest that the point of imagination is to achieve a kind of self-insight: To know nothing about oneself is to live. To know a little about oneself is to think. To know oneself precipitately, as I did in that moment of pure enlightenment, is suddenly to grasp Leibniz’s notion of the dominant monad, the magic password to the soul […]. it was only a moment, but I saw myself. Now I cannot even say what I was. And, after it all, I just felt sleepy because, though I don’t really know why, I suspect the meaning of it all is simply to sleep. (221) But note that here the grand idea that knowing yourself is somehow important is immediately and ironically defated by the conclusion that the “meaning” of it all is just to sleep—to forget about yourself, as much as you can (Pessoa contrasts sleep with dreaming: sleep is a state of non-consciousness for which he often longs when beset by dreams). Elsewhere, he says that [i]f there is one thing life gives us, apart from life itself, and for which we must thank the gods, it is the gift of not knowing ourselves: of not knowing ourselves and of not knowing one another. (333) In this rejection of the demand that one should know oneself, Pessoa allies himself with early 20th-century skepticism about the very idea, contrasted with the Romantic notion of imagination as a facility for spelunking the caverns of one’s soul. André Gide, for example, wrote in Autumn Leaves

Imagination, Selves, and Knowledge of Self

309

(1950/2011, 8) that “a caterpillar who seeks to know himself would never become a butterfy”—a very Pessoa-esque thought indeed. So we suggest that Pessoa did not think that one ought to know oneself, whatever that means, and nor did he think that the importance of imagination is that it helps one to do so. Furthermore, his rejection of these ideas is bound up with the important line of thought we highlighted above: knowing oneself requires an essential self that one can know; thinking that one can know oneself leads one to think that one has an essential self; thinking that one has an essentialized self stands in the way of imagination.10 Instead of knowledge, Pessoa suggests, we ought to pursue ignorance about ourselves: The possession of defnite, frm opinions, instincts, passions and a fxed, recognizable character, all this contributes to the horror of making of our mind a fact, of making it material and external. Living is a sweet, fuid state of ignorance about all things and about oneself. (56) Here, Pessoa gives a very explicit description of the risks, the “horrors”, of a known, essentialized self, then advocates ignorance instead. But what is ignorance about oneself, as opposed to knowing oneself? This is a vital question for Pessoa, as the ability to imagine depends on it. Given how much he dwells on the self, ignorance cannot be merely ignoring, refusing to engage with, one’s self. It must be some mode of engagement that acknowledges the self but denies it any essential qualities. One possible mode of such engagement would be constant questioning, which Pessoa does indeed invoke on occasion: “I question myself but I do not know myself” (174). But his most consistent answer is that the best way to engage with the self without essentializing it is to take ourselves less seriously. The opposite of knowing oneself is irony.11 To consciously unknow oneself, that is the right path to follow. And to consciously unknow oneself is the active task of irony. (334) We began this discussion by asking why Pessoa thinks imagination is important. We rejected an answer appealing to escapism, as it fails to explain why Pessoa remains disquieted despite his frequent fantasies. We suggested that this is explained by the fact that imagining of other selves is important to Pessoa, and that he consistently fails to do this well. We suggested that this failure owes to Pessoa’s excessively essentialized view of himself, which prevents him imagining other selves. We elaborated on, justifed, and generalized the suggestion, and we used this elaboration to suggest that the importance of imagination to the self, for Pessoa, cannot be that it allows one to know oneself better.

310 Nick Wiltsher and Bence Nanay Rather, it allows one to take an ironic, defationary perspective on one’s inessential being. This is all to the good, but important questions remain unresolved: why exactly is such ironic imaginative perspective important to the self, if not for self-knowledge? Why is Pessoa’s failure to imagine other selves the root of his disquiet? We now address these questions.

15.5 Imagination and the Self We began this chapter by noting that Pessoa had many heteronyms. While most of us do not have such well-developed alter egos, he did not think that he was unique in this fundamental multiplicity; he asserts that “each of us is more than one person, many people, a proliferation of our one self” (399). And those proliferating selves are produced by imagination, almost automatically: I created various personalities within myself. I create them constantly. Every dream, as soon as it is dreamed, is immediately embodied by another person who dreams it for me. (153) So here is a hypothesis: the proliferation of selves is what is important, and imagination is important because it allows such proliferation. The question, then, is why proliferating selves are so important. Once again we will begin by abjuring a tempting answer. Given Pessoa’s obsession with multiple alter egos, one might think that it is just the act of imagining yourself to be your alter ego(s) that is important; somehow, such imagining contributes to a healthy inner life. But this suggestion is no real explanation. Why would it be a good thing to have alter egos or alternative selves? Isn’t a form of this supposed to be a sign of multiple personality disorder (which is generally thought to be less than healthy)? Furthermore, most people do not have dozens of alter egos, like Pessoa did. Are we doing something wrong? Pessoa was clearly idiosyncratic in this respect. But then why should we take advice from him about the importance of imagination? We suggest that, for Pessoa, the point of multiplying selves is to allow refection upon the central self, in the ironic mode mentioned above. Imagination is important because it facilitates the change of perspective that allows one to see oneself from the outside. Such change of perspective requires, minimally, ownership of the perspective; the owner is produced by the imagination. Pessoa makes suggestions along these lines in several places, for example: those occasional moments of detachment in which we become aware of ourselves as individuals whom other people perceive as other. (211)

Imagination, Selves, and Knowledge of Self

311

I am present at a play with different scenery and the drama I watch is me […]. Dear God, who is this person I attend on? How many people am I? Who is me? What is this gap that exists between me and myself? (366) Again, Pessoa is here aligned with a trend of his near contemporaries (see Nanay 2015), stretching back to Oscar Wilde’s insistence that we should become the spectators of our own life, and forward to Julio Cortazar’s concept of paravision (the altered perceptual state that characterizes aesthetic experiences), which he describes as An instantaneous aptitude for going out, so that suddenly I can grasp myself from outside […] as if I were somebody who was looking at me. (Cortazar 1966, 414) Pessoa’s contribution to this tradition of literary thought is his distinctive answer to the question of why seeing yourself from the outside is a good thing, which he summarizes very simply: “Seeing myself frees me from myself” (414). “Seeing” here is fgurative. To see is to achieve comprehension or insight from a perspective, the point of view of a person. It can, but need not, be comprehension gained through vision. Seeing is different from mere looking (cf. 238); one can look (fguratively) without achieving insight. The moments in which one sees properly are uncommon and arresting: as Pessoa puts it, “to see clearly is to stop” (240).12 An indication that one is seeing is that practical considerations are deprecated in one’s experience; an opposition between seeing and action is a recurring motif of The Book of Disquiet. This line of thought is arguably an extension of the Kantian conception of aesthetic experience as divorced from practical considerations regarding its object. In Pessoa’s extension, properly seeing is likewise divorced from action. So when you adopt an imaginative perspective on your self, and when, from that perspective, you truly see yourself from the outside, any practical, pragmatic, action-oriented considerations fade away, and one has a kind of distanced, peculiarly aesthetic experience of oneself. That Pessoa intended this extension of the Kantian conception is strongly suggested by one of the few chapter titles he left indicating how the book should be organized: “An aesthetics of indifference” (28). And the suggestion is substantiated by passages such as this, when he describes seeing in the true sense of the word: I simply see unintentionally, unwittingly, the attentive spectator of a nonexistent spectacle. (425)

312 Nick Wiltsher and Bence Nanay Pessoa is the attentive spectator of a nonexistent spectacle of himself. And if imaginative episodes of seeing oneself from the outside are important, this explains why multiple imaginative selves with such perspective are important; this, in turn, explains why multiplicity of self is such an important theme both in The Book of Disquiet and in Pessoa’s authorial life. But, again, why is such seeing from the outside important? Perhaps it is because one needs to see oneself as others do. Pessoa at times draws attention to an asymmetry between our attitude toward ourselves and our attitude toward others: We are all accustomed to think of ourselves as essentially mental realities and of others as merely physical realities. (211) The idea is familiar from philosophy. Our own subjectivity is inescapable, and we exist frst in our own consciousness; we see others as and through their bodies, and have to make inferential leaps to acknowledge their existence as minds (on the problem of other minds, see, inter many alia, Dretske 1973). So one might think that, when seeing ourselves from the outside, our attitude to ourselves is akin to our normal attitude toward others: we would see ourselves as merely physical realities. But this cannot be quite right, since it skirts close to the kind of self-essentializing gaze that we have already argued cannot be what Pessoa wants. So the self-refective perspective must somehow be both detached and indifferent, without being essentializing: not wholly involved with the self on which it refects, but not entirely separate from it either. This is not just an imagined other’s perspective on oneself. It is a radically different kind of perspective, whose role is to restore the unity of one’s perspective on oneself as both mental and physical. And this requires a very specifc kind of imaginative episode of seeing yourself from the outside. To explore fully the nature and signifcance of this kind of imaginative perspective, we will frst return to the thought that its signifcance does not inhere in pursuit of self-knowledge. Pursuit of self-knowledge requires (belief in) a self with essential qualities: properties of personality that are, if not eternal and unchanging, at least stable and secure enough to support confdent ascription. One could pursue such self-knowledge by adopting outside perspectives on oneself with an essentializing gaze: a viewpoint that reifes the accidents and contingencies of situations and assigns them to an essential self. But Pessoa’s point is that the gaze one trains on oneself need not be of that sort. Throughout the book he sharply contrasts self-knowledge and self-love (and, to a lesser extent, knowledge and love):

Imagination, Selves, and Knowledge of Self

313

If there is one thing life gives us, apart from life itself, and for which we must thank the gods, it is the gift of not knowing ourselves […] No one would love himself if he really knew himself. (333) To understand is to forget to love. I know nothing at once so false and so meaningful as that saying of Leonardo da Vinci that one can only love or hate something once one has understood it. (212) To love oneself is not to know oneself: it is to recognize that there is nothing essential to know. Seeing ourselves from the outside allows us to see, precisely, our fundamental inessence. It allows us to love ourselves, not for the categories into which we place ourselves (the drinker of espresso, the eater of eggs Benedict), but for who or what we happen to be at the moment of refection. Pessoa’s thinking prefgures that of Sartre, perhaps in several respects. Gary J. Shipley says that, like Sartre, Pessoa “does not posit the reality of selves, but instead sees selves as imaginary devices, through which we can transcend Reality [sic], the reality in which the self is a nothing” (Shipley 2011, 118). The comparison is striking, though we wish to tease out its signifcance in a different manner. In Being and Nothingness (1943), Sartre describes the nature of the self, or rather of being-for-itself. “Being-for-itself” designates the kind of being that is capable of consciousness, as opposed to being-in-itself (Sartre 1943, 33). The consciousness of being-for-itself is a sort of “nothingness”, an opposing of consciousness to the world. All such consciousness involves a certain sort of self-consciousness (Sartre 1943, 19), and all such consciousness is the ground of its own existence via that selfconsciousness (Sartre 1943, 22). This presence of both consciousness and self-consciousness in every conscious act means that being-for-itself is never quite unifed: it always contains two slightly different perspectives, one distant from and observant of the other (Sartre 1943, 77). Scholars debate whether Sartre’s account is coherent, never mind plausible, but we don’t want to engage with that discussion here. What we note is a confuence of thought with Pessoa: both he and Sartre insist on the importance for consciousness of a self-conscious, self-constituted perspective. For Sartre, this is the constitutive ground of consciousness; for Pessoa, operating in a more literary mode, it is when such perspective is lacking that a grip on oneself as a real consciousness begins to slip. The comparison can be pushed further, in two ways. First, as we have argued, Pessoa sees imagination as the way in which an outside perspective on the self is achieved. Sartre does not explicitly draw this link in Being and Nothingness. This is peculiar, because he argued in The Imaginary

314 Nick Wiltsher and Bence Nanay (1940) that imagination is essentially a mode of consciousness that posits its objects as nothingness. So, if the consciousness of being-for-itself is a “nothingness”, one would think that self-consciousness, consciousness of consciousness, would precisely be an act of imagination. Sartre gestures toward such ideas in the conclusion of The Imaginary, but they are not developed as one might expect in Being and Nothingness. We suggest (and we can only suggest here) that adequate development of them might be pursued via Pessoa. Second, Pessoa might just provide unexpected relief from some of the gloomier ideas that Sartre does, famously, develop adequately in Part Three of Being and Nothingness. There, Sartre positions the being-for-itself, the individual, among other such beings, and explores the nature of being-for-others. Its nature is almost unrelentingly antagonistic. Being-for-itself, says Sartre, enters into recurrent confict with the essentializing gaze of others: the self sees that it is nothing, deduces from this its freedom, and so aims for transcendence of its concrete situation; the others ascribe to it essential properties that reduce and confne it to the concrete. The antagonism is so recurrent that the gaze of others becomes internalized, and so one begins to see oneself as they do: essentializing self-consciousness is the dear price of social existence. But it could be that Pessoa’s illustration of existential disquiet shows us a way to evade this fate, and instead turn the internal gaze into one of self-love. It is true, Pessoa thinks, that one must adopt perspectives of (imagined) others on oneself; indeed, imaginative proliferation of other’s perspectives is almost automatic. But one can use one’s imagination to adopt, own, those perspectives. If one succeeds in so inhabiting them, one can force them to adopt the same view of one’s essential nothingness that one has oneself; one need not allow the imaginative gazes to be essentializing in pursuit of spurious self-knowledge. Rather, the imaginative gazes can be accepting, loving, kind; above all, ironic. The comedy and tragedy of The Book of Disquiet owe to the fact that, even though Pessoa is successful in generating dreams and perspectives, he is unsuccessful in owning them. As he says: I’ve always been an ironic dreamer, unfaithful to promises I make to myself. I’ve always savoured the shipwreck of my daydreams as if I were someone else, a stranger. (323) Here the threads of discussion are pulled together: the desirability of an ironic stance on oneself, the necessity of imaginative selves for the adoption of such a stance, the importance of imagination for imagining those selves, and the disquieting consequences if one fails to do so. We earlier quoted Medeiros on Pessoa’s practice of multiplying heteronyms. Medeiros says that it reveals “a recognition of the importance

Imagination, Selves, and Knowledge of Self

315

of relating to the Other in order to be himself, but also a defection of the Self’s own authority and authorship” (Medeiros 2013, 11). We can now see how Pessoa’s ideas about imagination in The Book of Disquiet support this description of his practice. The others that Pessoa created were, for him, internalized gazes that allowed him to fully see himself in opposition to them. They thus played a role in making him who he was. But, at the same time, the very act of multiplying those selves and according to them a certain autonomy undermined Pessoa’s claim to be his own authoritative author; some degree of control was ceded to the heteronyms. In our terms, the creation of heteronyms allowed him to resist essentialization. And this is the principal difference between Pessoa himself and the Pessoa fltered through the semi-heteronym of Soares: where Soares fails to imagine, Pessoa undoubtedly succeeds. All this give us a fairly consistent picture of the human mind and imagination—much more consistent than the fragmentary nature of The Book of Disquiet would suggest. We can have an essentialist view of ourselves. This would make it diffcult to imagine ourselves to be someone else. Or we can have a less essentialist view of ourselves. This amounts to approaching oneself with a fair amount of irony, to taking oneself less seriously (see also Nanay forthcoming). If we have a less essentialist view of ourselves, this opens up the possibility of not just imagining ourselves to be someone else, but also imagining ourselves from the outside, which allows important kinds of experiences. They are important because, handled properly, they allow us to turn the internalized gazes of others from acts of antagonism to ones of self-love. This is a complicated, sophisticated, almost philosophically elaborated account of the connections between self, self-knowledge, and imagination. Pessoa contrasts self-knowledge of a spuriously essential self, and the disquiet that this brings, with the possibility of detached imagined perspectives on the self that resist essentialization, embrace irony, and allow a form of self-love. It is pretty clear which of these we should choose.

Notes 1 “The organization of the book should be based on as rigorous a selection as possible of the various existing texts, adapting any older ones that are untrue to the psychology of Bernardo Soares…. Apart from that, there needs to be a general revision of style, without it losing the personal tone or the drifting, disconnected logic that characterizes it” (“two notes”, appendices to 2017 New Directions edition). 2 All unattributed references are to this edition (Pessoa 2017); numbering refers to the fragment cited. Most of those we cite also appear in the concise 1991 Serpent’s Tail edition, again translated by Costa. This follows a thematic selection made by Maria José de Lancastre for an Italian edition. The numbering in that edition follows that of the magisterial Portuguese edition, edited by Maria Aliete Galhoz, Teresa Sobral Cunha, and Jacinto do

316

3 4 5

6

7 8 9 10 11 12

Nick Wiltsher and Bence Nanay

Prado Coelho and published in 1982. This is the source of most published editions (the other major English translation is by Richard Zenith: Penguin Classics, 2001). However, the exemplary online presentation of the material in the Arquivo LdoD, built and maintained principally by the Centre for Portuguese Literature at the University of Coimbra, is increasingly popular (Portela and Silva 2017). This resource allows for comparisons of notable editions and the compilation of one’s own selection. While we appreciate the scholarly and ludic possibilities of the online edition, the fact that it is in Portuguese precludes using it here. This is schematic; there are borderline and complicated cases. From a letter to Adolfo Casais Monteiro on January 13, 1915; translation from Susan Margaret Brown’s Selected Letters of Fernando Pessoa (Sheep Meadow Press, 2016). The terminological points hold both in the original Portuguese and in English, though the correspondence is complicated by the slightly different connotations of “dream”, “imagination”, and cognates compared with “sonho”, “imaginação”, and cognates. Pessoa’s own writings on philosophy, collected and edited by Nuno Ribero as Philosophical Essays (Pessoa 2012), are an eclectic, fragmented mix of musings on and inspired by the works of historical philosophical fgures, strongly infuenced by his wide-ranging interests, which encompassed such esoterica as astrology, theosophy, numerology, occultism, and pantheism. A thorough analysis of them may well yield much of value, but we are doubtful that they would illuminate the questions we are pursuing here. One reason to be dubious is that the philosophical essays and fragments were mostly composed in the early 1900s, well before any of the material collated in The Book of Disquiet. Given Pessoa’s restless intellect, there is no reason to presume consistency between ideas in works composed decades apart. Here we revert to Costa’s slightly different translation in the 1991 Serpent’s Tail edition, though for continuity the numbered citation refers to the 2017 New Directions edition. We use “(im)personal imagination” in a non-technical sense. On the technical senses, see Williams (1973); Currie (1995). On the connection between imagination and perspective, see Williams (1973); Wollheim (1974); Peacocke (1985); Martin (2002); Noordhof (2002); Smith (1997); Lopes (1998). See also the rich literature on imaginings about the self, e.g. Williams (1973); Wollheim (1973); Reynolds (1989); Velleman (1996); Wirling (2014). On irony and imagination, see Currie (2006). The connection between irony and the self also prefgures many of Albert Camus’s thoughts on the absurd: see Camus (1942). A similar idea is expressed in fragment 202 (467) in the Serpent’s Tail edition: “to see clearly is to be unable to act”. Since it is not included in the main text of Pizarro’s selection (it appears numbered 503 in an appendix), this fragment is not included in the New Directions edition.

References Atkin, Rhian. 2010. “Bernardo Soares, Flânerie, and the Philosophy of Inaction.” Hispanic Research Journal 11, no. 2: 157–71. Boyd Maunsell, Jerome. 2012. “The Hauntings of Fernando Pessoa.” Modernism/modernity 19, no. 1: 115–37.

Imagination, Selves, and Knowledge of Self

317

Camus, Albert. 1942. Le Mythe de Sisyphe. Paris: Galimard. Cortazar, Julio. 1966. Hopscotch. New York: Avon Books. Cousineau, Thomas J. 2013. An Unwritten Novel: Fernando Pessoa’s The Book of Disquiet. Champaign/London/Dublin: Dalkey Archive Press. Currie, Gregory. 1995. Image and Mind. Cambridge: Cambridge University Press. Currie, Gregory and Ian Ravenscroft. 2002. Recreative Imagination. Oxford: Oxford University Press. Currie, Gregory. 2006. “Why Irony is Pretence.” In The Architecture of Imagination, edited by Shaun Nichols, 111–3. Oxford: Oxford University Press. Dretske, Fred. 1973. “Perception and Other Minds.” Nous 7, no. 1: 34–44. Gide, Andre. 1950/2011. Autumn Leaves. New York: Philosophical Library. Jones, Janine. 2004. “The Impairment of Empathy in Goodwill Whites for African Americans.” In What White Looks Like: African American Philosophers on the Whiteness Question, edited by George Yancy, 65–80. New York and London: Routledge. Lopes, Dominic McIver. 1998. “Imagination, Illusion, and Experience in Film.” Philosophical Studies 89, no. 2–3: 343–53. Martin, Michael G.F. 2002. “The Transparency of Experience.” Mind and Language 17, no. 4: 376–425. Medeiros, Paulo de. 2013. Pessoa’s Geometry of the Abyss: Modernity and the Book of Disquiet. London: Modern Humanities Research Association and Maney Publishing. Nanay, Bence. 2014. “The Dethroning of Ideocracy: Robert Musil as a Philosopher.” The Monist 97, no. 1: 3–11. Nanay, Bence. 2015. “Aesthetic Attention.” Journal of Consciousness Studies 22, no. 5–6: 96–118. Nanay, Bence. Forthcoming. The Fragmented Mind. New York: W. W. Norton. Noordhof, Paul. 2002. “Imagining Objects and Imagining Experiences.” Mind and Language 17, no. 4: 426–55. Paul, L.A. 2014. Transformative Experience. Oxford: Oxford University Press. Peacocke, Christopher. 1985. “Imagination, Experience, and Possibility: A Berkeleyan View Defended.” In Essays on Berkeley, edited by John Foster and Howard Robinson, 19–35. Oxford: Clarendon. Pessoa, Fernando. 2001. The Selected Prose of Fernando Pessoa. Translated by Richard Zenith (2001). New York: Grove Press. Pesso, Fernando. 2012. Philosophical Essays: A Critical Edition. Edited by Nuno Ribero (2010). New York: Contra Mundum Press. Pessoa, Fernando. 2017. The Book of Disquiet. Edited by Jerónimo Pizzaro (2010). Translated by Margaret Jull Costa (2017). New York: New Directions. Portela, Manuel, and António Rito Silva, eds. 2017. LdoD Archive: Collaborative Digital Archive of the Book of Disquiet. Coimbra: Centre for Portuguese Literature at the University of Coimbra. https://ldod.uc.pt/ Reynolds, Steven L. 1989. “Imagining Oneself to Be Another.” Nous 23, no. 5: 615–33. Sartre, Jean-Paul. 1940. The Imaginary. Translated by Jonathan Webber (2004). Abingdon: Routledge. Sartre, John-Paul. 1943. Being and Nothingness. Translated by Hazel E. Barnes (1958). Methuen: London.

318

Nick Wiltsher and Bence Nanay

Schoneboom, Abigail. 2015. “The Romance of the Lowly Clerk: Recognizing the Tradition of Offce Intellectualism.” Organization 22, no. 6: 832–46. Shipley, Gary J. 2011. “Dreaming Death: The Onanistic and Self-A nnihilative Principles of Love in Fernando Pessoa’s Book of Disquiet.” Glossator 5: 107–38. Smith, Murray. 1997. “Imagining From the Inside.” In Film Theory and Philosophy, edited by Richard Allen and Murray Smith, 412–30. Oxford: Oxford University Press. Velleman, J. David. 1996. “Self to Self.” The Philosophical Review 105, no. 1: 39–76. Williams, Bernard. 1973. “Imagination and the Self.” In Problems of the Self, edited by Bernard Williams, 26–45. Cambridge: Cambridge University Press. Wirling, Ylwa. 2014. “Imagining Oneself Being Someone Else: The Role of the Self in the Shoes of Another.” Journal of Consciousness Studies 21, no. 9–10: 205–25. Wollheim, Richard. 1973. “Imagination and Identifcation.” In On Art and the Mind, edited by Richard Wollheim, 54–83. Cambridge, MA: Harvard University Press. Wollheim, Richard. 1974. “Identifcation and Imagination.” In Freud: A Collection of Critical Essays, edited by Richard Wollheim, 172–95. New York: Anchor Press.

Notes on Contributors

Margherita Arcangeli is an associate professor at the Ecole des Hautes Etudes en Sciences Sociales (EHESS). She held postdoctoral positions at Jean Nicod Institute, University of Geneva and Swiss Center for Affective Sciences, Humboldt-Universität zu Berlin, and EHESS. She is the author of several articles and a book on imagination, and other topics in philosophy of mind, philosophy of science, and aesthetics, among which are thought experiments, memory, and the sublime. Christopher Badura  is a PhD student in philosophy at the Ruhr University Bochum, Germany, working on logics of imagination. His research interest is philosophical logic and its application to philosophical issues concerning imagination. Francesco Berto works at the Department of Philosophy and the Arché Research Centre at the University of St Andrews, and at the Institute for Logic, Language and Computation at the University of Amsterdam. He has published on logic, ontology, epistemology, and the philosophy of computation. Rebecca Hanrahan is an associate professor of philosophy at Whitman College. She specializes in the epistemology and metaphysics of modality. She has published in such journals as Philosophia, Philosophical Forum, and The Journal of Philosophical Research. Amy Kind  is Russell K. Pitzer Professor of Philosophy at Claremont McKenna College, where she also serves as Director of the Gould Center for Humanistic Studies. In addition to authoring the introductory textbooks Persons and Personal Identity and Philosophy of Mind: The Basics, she has edited Philosophy of Mind in the 20th and 21th Centuries, The Routledge Handbook of Philosophy of Imagination, and (with Peter Kung) Knowledge Through Imagination. Peter Kung is Associate Professor of Philosophy at Arizona State University. His principal area of research is imagination and philosophical methodology. He has co-edited, with Amy Kind, Knowledge Through Imagination (Oxford University Press, 2016). He also works on epistemology, particularly foundational questions about justification.

320

Notes on Contributors

Derek Lam  is an assistant professor in the Department of Philosophy and Religious Studies at the University of Wisconsin-Whitewater, where he has been since he completed his PhD at the University of Virginia in 2018. His current research focuses on the metaphysics of time, the nature of intentional action, and modal epistemology. Julia Langkau  is a lecturer at the University of Flensburg. She has been a research fellow of the Swiss National Science Foundation at the University of Fribourg, a visiting research fellow at the University of Miami, a ZIF Marie Curie fellow at the University of Konstanz, and a postdoctoral researcher at the University of Zürich. Her areas of research are epistemology, philosophy of mind, and aesthetics. She is co-editor of the forthcoming Routledge volume The Philosophy of Fiction: Imagination and Cognition. Antonella Mallozzi  is Assistant Professor of Philosophy at Providence College and an associate editor of Analysis Reviews. She works primarily in metaphysics and epistemology. She is the editor of the forthcoming special issue of Synthese “New Directions in the Epistemology of Modality,” a co-author of the Stanford Encyclopedia of Philosophy entry on “The Epistemology of Modality,” and the author of several recent articles on essence, grounding, explanation, and the a priori. Joshua Myers is a PhD candidate in philosophy at New York University. His research interests are in philosophy of mind, epistemology, and cognitive science, with a focus on the representational and epistemic structure of the imagination. Bence Nanay is professor of philosophy and BOF research professor at the University of Antwerp and Senior Research Associate at Peterhouse, Cambridge University. He published three monographs with Oxford University Press (Between Perception and Action, 2013, Aesthetics as Philosophy of Perception, 2016, Aesthetics: A Very Short Introduction, 2019) with one more (Seeing Things You Don’t See) under contract. He has won the prestigious Bessel Award of the Humboldt Foundation, Germany. He is the principal investigator of a two million Euro ERC grant and the director of the European Network for Sensory Research. Michael Omoge is a postdoctoral fellow at the University of the Western Cape. He completed his PhD in 2020 at the University of KwaZuluNatal. His research interests intersect philosophy of mind, cognitive science, and modal epistemology. Eric Peterson  is Director of Scott Scholars Program and Instructor of Business Ethics at Creighton University. His research interests include imagination, philosophy of mind, ethics, and much more. He is cofounder (with Amy Kind) and Managing Editor of The Junkyard, a scholarly blog devoted to the study of imagination.

Notes on Contributors

321

Luke Roelofs  is a postdoctoral associate at NYU’s Centre for Mind, Brain, and Consciousness. They grew up in England and have worked at universities in Canada, Australia, and Germany. They work on the metaphysics of consciousness and the epistemology of empathy. Margot Strohminger is Lecturer in Philosophy at the Australian Catholic University, Melbourne. She moved to ACU from the University of Oxford, where she was Marie Curie Fellow at the Faculty of Philosophy and Junior Research Fellow at St Cross College. She obtained her PhD in Philosophy from the University of St Andrews in 2014. She specializes in epistemology and the philosophy of mind, with a focus on modal knowledge and the imagination. Nick Wiltsher  is an associate senior lecturer in philosophy at Uppsala University. He has previously worked in Antwerp, Auburn (AL), Porto Alegre, and Leeds. He works on imagination and aesthetics, jointly and severally.

Index

Note: Page numbers followed by “n” denote endnotes. Abelson, Robert 8 absolute possibility 92 access role in attention 221, 223–24, 229 acknowledged propositions 147–49 act of imagination 2, 9, 16, 310, 314 actual possibilities 67 actual world 26–27, 33–37, 74, 91, 104–5, 119n8, 280, 292 Adjunction in imagining 131–32, 135, 137 adjustment distinction 282, 283 admissibility vs. inadmissibility 49–50 aesthetic experience 311 affordances of environment 72 Alchemy of Race and Rights, The (Williams) 248 alignment in perception 40 alleged imaginings 54 all-or-nothing knowledge 249 alter egos 298, 299 antecedent belief 144, 152, 156n3 ‘Anti-Empathy’ (Goldie) 260–61 anti-essentialist concerns 52–54, 307 anti-exceptionalism 160, 169, 175 antiracism 279–80 anti-realism 81 appeal to evidence 58 a priori/a posteriori distinction: cognitive architecture of simulation 168–70; in conceptual possibility 25–31; Generality Claim 164, 170–73; Imagination Claim 164, 165–70; imagination vs. inference 173–78; introduction to 4, 11–12, 160–61; sensory imagination and 160–62, 164–73; Williamson’s challenge with 161–64, 178n1

a priori entailment (APE) thesis 27–31 Aristotle 2–4, 37–38 Atkin, Rhian 308 attention: access role in 221, 223–24, 229; Access Role of 221, 223–24; auditory experiences and 220–21; concentration and 220; conscious attention 218, 220, 222, 226; content role in 222–24, 229, 231; defned 219–20; determinatedeterminable role in 223–24, 229, 230; epistemic signifcance and 221–32; focalization and 220, 221; imagination and 227–31; inattentional blindness 14, 222, 224, 231, 232n3; inattentive attention 225–26; introduction to 218; involuntary attention 220; justifcation and 224–27; perception and 222–23, 227–31; perception and imagination 227–32; as selection 232n2; summary of 231–32; unconscious attention 218, 226, 232n4; voluntary attention 220 attentional blindness 229, 231 attentional space 220–21, 222, 226, 229 Attention Optional View 224–25 attitude-content distinction 287 auditory experiences and attention 220–21 automatic reasoning 68 Autumn Leaves (Gide) 308–9 Balcerak Jackson, Magdalena 155 basic facts 27–28 Bayes’ theorem 7, 50 behavior scripts 84–85, 88–89, 96n8

324 Index Being and Nothingness (Sartre) 313–14 belief-like imagination 184–85 beliefs: antecedent belief 144, 152, 156n3; belief revision theory 156n2; conceiving and 23–24; conditional beliefs 123–24, 126–27, 142–49; conditional beliefs vs. beliefs in conditionals 152–54; consequent belief 144, 156n3; consistently true beliefs 73; epistemic logics for revision of 130; experiential perspectives and 152–53; of goodwill whites 251; imagination defned in contrast with 186–87; imaginative episodes in 141–42; immediate belief 63, 65, 70–71; justifcation of 35–37, 75, 110–11, 229; non-basic beliefs 146; perception/belief/desire architecture 82, 222; pre-existing beliefs 83–85, 192; process of formation 116; psychologically immediate beliefs 68–71; rational belief 231; rational imagination and 32; reasoning with 113–17; revision theory of 127, 130, 146–48, 156n2–156n3, 157n16; simulated vs. actual 14; unconscious beliefs 218; unjustifed beliefs 108, 110–11; see also modal beliefs black experience 238, 240, 248, 250, 256 black-to-white understanding 250–51 blindsight 221 body in play principle (Aristotle) 203, 204–13, 215n21–215n22 body translated by Galileo 214n4 Book of Disquiet, The (Pessoa) 17; failure of imagination 303–5; imagination and the Self 310–15; imagination as escape 301–3, 309; introduction to 298–301; limits of imagination 305–10 Boyd Maunsell, Jerome 299–301 Brison, Susan 248 Chalmers, David 6, 26–31 change blindness 225, 227 Change Constraint (CC) 142, 227–28 characterization of perspective-taking 270–71 classical logic 126, 150–51

Coates, Ta-Nehisi 279–80 cognitive imagination 81, 82–83, 122, 184 cognitive process of thought experiments 190–91 cognitive skills and perception 165–68, 228 cognitive synonymy 127–28, 136 Cohen, Martin 188–89 coherence, justifcation for 35–37, 146–47 compositional scripts 89 concentration and attention 220 conceptual absurdity 25 conceptual entailment 32–34, 36 conceptual possibility: Chalmers and Jackson on 6, 26–31; conceptual entailment 32–33; detecting incoherence 34–37; Ichikawa and Jarvis on 6, 32–37; imagery in 37–40; imaginability-possibility link 7, 23–24; introduction to 6–7, 23–24; metaphysical possibility link 23–27, 33; overview of 25–26; philosophical theorizing in 24–25; a priori entailment (APE) thesis 27–31 conditionalization principle 148 conditional reasoning 118n1, 119n5, 177 conditionals: beliefs 123–24, 126–27, 142–49, 152–54; indicative conditionals 209, 215n17–215n18; Jeffrey’s principle of conditionalization 148; probabilities in 123, 148, 153–54; Ramsey test for 123–24; subjective conditional probabilities 153; truthand-falsity conditions 134–35 conscious attention 218, 220, 222, 226 conscious imagining 218 consciousness: of consciousness 314; “idle” conscious intentions 287–88; nature of 6; phenomenal consciousness 86–87, 122; selfconsciousness 300, 313–14 conscious perception 218, 232n4 consensus view 184, 186, 189, 192–93, 195n2, 196n8 consequent belief 144, 156n3 consistently true beliefs 73 constrained imagining 2–3, 105, 115, 122, 188

Index 325 content-character distinction 287 content role in attention 222–24, 229, 231 Copy Principle 38–39 Cornwall, David 299 Cosmoscope 30–31 Costa, Margaret Jull 299 counterfactuals: counterfactual conditionals 126, 132, 144, 156n2, 160, 215n17; in empathy 289–91, 293–94; in imagination 279–81; in imaginative reasoning 79, 91, 165, 173; in modal knowledge 178n3; simulation devices and 31; in thought experiments 188, 208–11, 215n17 Cousineau, Thomas J. 301 Davies, David 188–89 daydreams/daydreaming 1, 105, 122, 302 decision-making 1, 141, 177, 264–65 deduction vs. supposing 175 deductive reasoning 108, 117, 119n7 de-essentializing the self 17, 305–7 defeasibility types 7, 52, 58, 126 defeaters 46–47, 52–53, 54 demonstrated success 94 detecting incoherence 34–37 determinate-determinable role in attention 223–24, 229, 230 Dialogues Concerning Two New Sciences (Galileo) 202 disposition: to believe 11, 71; conditional analysis of 156n2; conditional belief and 143–46; to experience emotions 262; to form modal beliefs 63, 70–76, 77n11– 77n12; manifestation of 144; offine manifestation of 145; psychological dispositions 263, 265, 269. see what is possible 69, 71 distorting power of emotion 264, 265 distracted driver case 226 domain-specifc knowledge 86, 88 doxastic-epistemic logic 142 doxastic justifcation 224–26 dreaming as imagining 300, 308 dualists/dualism 53–54, 87 dual-process theories 203, 213, 215n23 dual-targetted imagining 290 Du Bois, W.E.B. 250 Dutch book arguments 147

egocentric bias 276n9, 292–94 Einstein, Albert 189 embodied capacity of mind 247 emotional contagion 261 emotion/emotional state: belief and 184; distorting power of 264, 265; fellow-feeling and 257n12; imaginative projection and 247; imagining of 286; listening and 253; perspective-taking and 261–65, 269–74; reason-giving emotions 264; semi-heteronym and 299; simulation of 12; thought experiments and 12–13 empathetic understanding 257n9 empathic perspective-shifting 15, 260–61, 261–62, 263–66, 270–72, 283–85, 284–85, 292–94 empathy: complexity of 288–94; egocentric bias 276n9, 292–94; Goldie’s critique of 283–85; from the inside/from the outside imagining 290, 291–92; models and targets of 288–91; role of 279–80; as role-reversal 16, 280–83, 294–95; in “what is imagined?” question 285–88 empathy defcit 279 empathy economy 279–80 empirical generalizations 36, 170–71 empirical stance toward oneself 264–65 End of Empathy, The (Compton) 279 episodic memory 163, 172–73, 176, 178 epistemic analytic truth 172, 178n7 Epistemic Arrogance argument 15, 240–43, 246, 252–55 epistemic asymmetry 59–60 epistemic challenges to perspectivetaking 266–72 epistemic failure to understand 260 Epistemic Inaccessibility argument 239, 241, 243, 244–45, 249, 252 epistemic privilege thesis 257n11 epistemic signifcance: attention and perception 5, 221–27; defned 218–19; of perception-like imagination 219, 229–32; thought experimentation and 219, 230–31 epistemic uses of imagination: cognitive theory of imagination 81; imaginative reasoning and 9–12,

326 Index 103–5, 110, 114–15, 117–18; introduction to 1–18; key themes 2–5; overview of 5–17; see also justifcation of imaginative episodes epistemology of modality see modal epistemology equivalence in imagination 126–28, 136–37 Error Theory 54–55, 60n11 escapist interpretation of imagination 17, 301–3, 309 essentialist view of ourselves 315 ethics 6–7, 37, 282 etiological defeat 46–48 etiological immunity: defned 48–49; explanation of 50–52, 54, 56; modal justifcation and 57–59 etiological vulnerability: defned 48–49; explanation of 55–57; modal justifcation and 57–59 Euclidean geometry 93 Evans, Mary Ann 299 evolutionary advantage 66–67, 69, 70–72, 77n11, 94–95 evolutionary usefulness 94 experiences: aesthetic experience 311; auditory experiences and attention 220–21; black experience 238, 240, 248, 250, 256; Holocaust survivor experience 241; inability to experience 251; inattentive experience 225–26; racism experience 248, 279; sexual assault experience 237–38, 240, 256, 257n8; spatial structure of experience 220–21; transformative experiences 149, 270; traumatic experiences 15 experience-taking 275n1, 276n8 experiential imagination 14, 18, 149, 219 experiential perspectives: beliefs and 152–53; epistemic access to 3–4; Epistemic Arrogance argument 15, 240–43, 246, 252–55; Epistemic Inaccessibility argument 239, 241, 243, 244–45, 249, 252; optimism and 239, 251; pessimism and 237–40, 247–52; summary of 255–56; Too Big a Gulf argument 243–47; vastly different perspectives 246–47, 252 explicit reasoning 85

face value 48, 58 failing to notice 227 failure of imagination 303–5 failure to understand 260 fallibility expressions in imagination 35–36, 46, 52 fantasies/fantastical imagination 1, 3, 47, 83, 119n8, 301–2, 309 feminist theory of epistemic privilege thesis 257n11 fctional narratives 105, 183, 185–90, 195n2, 229, 272–73 Fiction and Narrative (Matravers) 183 fdelity constraint 188–89 First Degree Entailment (FDE) logic 134 frst-personal deliberative stance toward oneself 264–65, 275 Floyd, George 248 focalization and attention 220, 221 folk physics 214, 215n24 foundationalist internalism 146–47 from the inside/from the outside imagining 290, 291–92 functional-action scripts 89, 90 Galileo bias 215n24 Galileo’s thought experiment: Aristotelian principle of body in play 203, 204–13; better interpretation 206–8; introduction to 3, 13, 202–3; mundane case of 208–11; standard interpretation 203–5; summary of 213–14; two kinds of imagination 211–13 gaze-related directional prepositions 64, 76n4 Generality Claim (GC) 164, 170–73 geometric reasoning 93–94 Gettier cases 189 Gide, André 308–9 Gilroy, Paul 238–39 God/gods 6, 308, 311, 313 Goldie, Peter 260–70, 282–85 goodwill whites 251 Gorilla experiment 224 Grandin, Temple 38 grief, coping with 238, 246 guided chosen imaginings 142 hallucination/hallucinating 46, 109, 122 Hamilton (musical) 238

Index 327 hearer testimony 61n15 Hesperus 53 heteronyms 298, 299, 315 hidden inconsistencies/incoherences 37, 38 Hintikkan approach to intentionality 124–25 holistic enterprise 55 Holocaust survivor experience 241 human-sized knowledge 77n10 Hume, David 38–39, 46, 51, 169 hyperintensional imagination 133–34 hypothetical reasoning 155 hypothetical thought experiments 188 Ichikawa, Jonathan 6, 32–37, 39 idealization 6–7, 23–27, 29–30, 39 ideal knowers 125 ideal perception 228 ideal reasoning 31, 34 Identifcation Defeater 52–53, 54, 56, 58–59 identity distinction 282, 283–84 “idle” conscious intentions 287–88 ignorance position 25, 29, 309 illusions 109 imagery: in conceptual possibility 37–40; mental imagery 3, 38, 132–33, 167, 170, 176–77, 228–29; model construction 40; vivid imagery 38 imaginability-possibility link 7, 23–24 imagination: act of imagination 2, 9, 16, 310, 314; Adjunction in imagining 131–32, 135, 137; attention and 227–31; cognitive imagination 81, 82–83, 122, 184; cognitive theory of 81, 82–83; conscious imagining 218; constrained imagining 2–3, 105, 115, 122, 188; counterfactual scenario in 289–91, 294; defned in contrast with belief 186–87; as embodied capacity of mind 247; epistemic asymmetry and 59–60; equivalence in 126–28, 136–37; as escape 17, 301–3, 309; experiential imagination 14, 18, 149; explicit reasoning in 85; failure of 303–5; fallibility expressions in 35–36, 46, 52; guided chosen imaginings 142; as guide to current environment 72–75; hyperintensional

imagination 133–34; imaginationbased modal justifcation 46–48, 51–52, 56, 59–60; inference in 32, 116; inference vs. 173–78; from the inside/from the outside 290, 291–92; interferences in 90; intuitive imagination 13, 54, 92, 95, 116, 211–14; invitations to 90, 183; knowledge by 176; limits of 305–10; mereological structure of imagined scenarios 131–32; metaphysical modality and 44–45; metaphysical possibility and 38–39, 45; narratives and 184–87; neutral imagination 193; non-imaginistic elements 6, 23–24, 31, 40, 122, 173; non-sensory imagination 48; propositional imagination 18, 82, 122, 125, 143, 145, 150, 173–74, 184, 196n3; rational imagination 32–37, 241; reality-oriented imagination 124; recreativist account of 286; as ROMS 122–24; Self and 310–15; sensory imagination 18, 73; skepticism and 79–81, 92–95; “what is imagined?” question 285–88; see also epistemic uses of imagination; psychology of imagination; sensory imagination ‘Imagination and the Distorting Power of Emotion’ (Goldie) 264 imagination-based modal justifcation 46–48, 51–52, 56, 59–60 imagination box 82–83 Imagination Claim (IC) 164, 165–70 imaginative episodes see justifcation of imaginative episodes imaginative identifcation 281 imaginative justifcation 4–5, 9, 104, 107, 108, 110, 119n7; see also justifcation of imaginative episodes imaginative reasoning 106–9; automatic reasoning 68; conditional reasoning 118n1, 119n5, 177; deductive reasoning 108, 117, 119n7; epistemic structure of reasoning 106–9; epistemic uses of imagination and 9–12, 103–5, 110, 114–15, 117–18; explicit reasoning 85; geometric reasoning 93–94; hypothetical reasoning 155; ideal reasoning 31, 34; input states of reasoning 106–11, 116–17,

328 Index 119n4–119n5; introduction to 9–12, 103; justifcation and 104, 107, 108, 110, 119n7; logical reasoning 4, 12, 95, 133, 152–53, 175; mathematical reasoning 116– 17; mediating states of reasoning 106, 110, 116, 119n5, 119n7; modal-conclusion-frst reasoning 54; non-ideal reasoning 34; output states of reasoning 106–8, 110, 117, 119n4; overview of 110–13; particular theory of reasoning 106; platitude about reasoning 107, 108, 111–13; poor reasoning in 107, 112; reason-giving emotions 264; reasoning with beliefs 113–17; sensory imagination and 104, 105, 110; sui generis reasoning 103, 114, 118; theoretical reasoning 103, 106, 118; unconscious reasoning 68 imaginative scaffolding 143, 149–52, 154–55, 245, 269, 271 imagistic imagination 31, 38–40 immediate belief 63, 65, 70–71 immediate perceptual modal epistemology 67–70 impossibility: of empathetic perspective-shifting 263, 276n2, 276n4; evidence for 6, 48, 55–56, 60, 60n6; imaginability and 92; imagination and 92; it’s-impossible-pessimism 251; Kripkean a posteriori impossibility 25; necessary vs. impossible 55; pessimism and 255; in principle 257n8; in scripts 88; semantics approaches to possibility/ impossibility 126; unimaginabilityimpossibility link 7, 45 inability to experience 251 inadmissible evidence in objective chance 49–50 inattentional blindness 14, 222, 224, 231, 232n3 inattentive attention 225–26 inattentive experience 225–26 incoherence, detecting in conceptual possibility 34–37 inconceivability-based model justifcation 56 indeterminacy 28, 41n13 indicative conditionals 209, 215n17–215n18

inference 32, 108, 116, 173–78 inferential blindness 224, 227, 232n3 input states of reasoning 106–11, 116–17, 119n4–119n5 intensionally equivalent sentences 127, 129 intentional control 3 intentional states 126 interferences 8–9, 90–92 internal-external distinction 282–83 in-their-shoes perspective-shifting 15, 261–62, 271–72, 283–85 intuitions: justifcation of belief 203; linguistic intuitions 76n4; minimal alteration principle and 130; modal intuitions 59, 61n14, 83; perspective-shifting and 272; a priori beliefs from 215n23; scripts and 85; as subskills 14; thought experiments and 212–13, 231 intuitive imagination 13, 54, 92, 95, 116, 211–14 intuitive neutrality 186 intuitive thinking 203, 211–13, 215n23 invitations to imagine 90, 183 involuntary attention 220 involuntary mode of imagination 176 it’s-extraordinarily-diffcult pessimism 250, 251 it’s-impossible-pessimism 251 Jackson, Frank 6, 26–31, 188–89 Jarvis, Benjamin 6, 32–37 Jeffrey’s principle of conditionalization 148 Johnson-Laird, Philip 191 Jones, Janine 250–51 Joyce, James 273 justifcation: attention and 224–27; of belief 35–37, 75, 110–11, 229; for coherence 35–37, 146–47; by deduction 108; doxastic justifcation 224–26; inconceivability-based modal justifcation 56; perceptual justifcation 47, 60n5, 108–9, 227; prima facie justifcation 46–47; process-reliabilist account of 146; propositional justifcation 224–25, 226; testimonial justifcation 49, 57–59; unjustifed beliefs 108, 110–11; see also imaginative justifcation; modal justifcation

Index 329 justifcation of imaginative episodes: conditional beliefs and 142–49; conditional beliefs vs. beliefs in conditionals 152–54; imaginative scaffolding 143, 149–52, 154–55; introduction to 141–43; summary of 154–56; topics/subject matters 129–30, 157n11 justifcatorily immediate 67–69 justifcatory force in reasoning 107, 108 Kant, Immanuel 311 knowing yourself 308–12 knowledge: all-or-nothing knowledge 249; domain-specifc knowledge 86, 88; experiential perspectives on 249, 256, 257n7; human-sized knowledge 77n10; ideal knowers 125; by imagination 176; nonperceptual knowledge 67–68, 197n19; self-knowledge 2, 14–17, 298, 308–12; see also modal knowledge; perceptual knowledge Knowledge Through Imagination (Kind, Kung) 122 knows, in modal logic 124 Kolmogorov axioms of probability 147 Kriegel, Uriah 193 Kripkean defense of dualism 53 Kripke-Putnam a posteriori 33 Laboratory of the Mind, The (Brown) 204–5 Langland-Hassan, Peter 39, 142–43, 148 Leitgeb, Hannes 144 Leonard, Erika 299 less-than-ideal epistemic position 25 Lewis, David 153 Lewisian chance-credence norm 48–50 limits of imagination 305–10 linguistic intuitions 76n4 listening to others 253 literary fction 186–87, 190, 274; see also fctional narratives logical consequences 125, 131, 135, 144, 228 logical de-idealization 127 logical reasoning 4, 12, 95, 133, 152–53, 175

logic of suppositional thinking 128 long-term memory 191 make-believe 3, 184, 189 Márquez, Gabriel García 185, 188 Mary Thought Experiment 14, 188–90, 219, 229–31, 233n8 mathematical reasoning 116–17 Matravers, Derek 12, 183, 184, 186–87 Mauriac, François 188 Medeiros, Paulo de 298, 299, 314–15 mediate perceptual modal epistemology 67–70 mediating states of reasoning 106, 110, 116, 119n5, 119n7 memory/memories: cognitive limitations 29; episodic memory 163, 172–73, 176, 178; fading of 306; idealizations in 6, 30; of images 288; imaginative model in 105, 111–12, 114–15, 291; limits to 33; long-term memory 191; of mathematical proofs 161; metal states and 275n1; mnemonics and 161, 178; short-term memory 191; storage of mental models 192; visualization and 229, 302 mental attitude 267 mental imagery 3, 38, 132–33, 167, 170, 176–77, 228–29 mental mental simulation 194 mental models 12, 183–87, 190–95, 197n16 mental movies 215n16 mental scripts 89 mental simulation see RealityOriented Mental Simulation mereological structure 129, 131–32 metaphysical modality: disagreements about scripts 86–92; introduction to 4, 8–9; metaphysics of 81; modal reality and 44–45; psychology of imagination and 82–96; skepticism in 79; true logic for 56–57; as trustworthy 96n3 metaphysical possibility: conceptual entailment and 32–33; conceptual possibility and 23–27, 33; imagination and 38–39, 45; role of 24–25 miasmatic boredom 17, 300 microphysical facts 28 Millar, Allan 65

330

Index

minimal alteration 124, 130, 138 mnemonics 161, 178 modal auxiliaries 64–65, 76n4 modal beliefs: formation of 8, 63–66, 69–70, 73–76, 77n10; justifcation of 67, 70, 97n14; psychologically immediate 70; reliable processes and 75; testimony as source of 61n15 modal-conclusion-frst reasoning 54 modal epistemology: imagining the impossible 54; introduction to 3–4, 6–8, 23–24, 32; as naturalistic 81, 96n4; perception and 63, 69, 74; as psychologically unrealistic 77n10; as rationalistic 80; real possibilities and 63, 67, 75; theory of perceptual modal epistemology 64–67, 73 modal intuitions 59, 61n14, 83 modal justifcation: etiological defeat 46–48; etiological immunity and 57–59; etiological vulnerability and 57–59; imagination-based 46–48, 51–52, 56; objective chance and 44–45 modal knowledge: epistemic powers of conceivability 174; introduction to 1, 5, 6–9, 14; as perceptual knowledge 8, 63–76; self-understanding and 14–17; Williamson’s counterfactual account of 178n3 modal logic 7, 51, 55–58, 60n11, 124 modal monism 27 modal psychology: defned 80; introduction to 8, 79–81; psychology of imagination 82–85 modal reality 7, 44–45, 48, 56, 60 modal scenarios 84–86 model-target distinction 287 modus ponens 55 modus tollens 55 moral failing 240–41, 252 moral humility 241, 254 moral imagination 247 Moran, Richard 264 movement of thought 174 Mrs. Dalloway (flm) 274–75 Mrs. Dalloway (Woolf) 273 Munro, Alice 274 Musil, Robert 306–8 Nagel, Thomas 243, 268 narratives: fctional narratives 105, 183, 185–90, 195n2, 229, 272–73;

imagination and 184–87; nonfctional narratives 183, 189–90, 247–48; in perspective-taking 270; short fctional narratives 188; thought experimentation and 184–91 narratives tout court 195 naturalized modal epistemology 80 nature of consciousness 6 necessary vs. impossible 55 Nersessian, Nancy 189 neutral imagination 193 Newtonian mechanics 114 New York Times 238 Nichols, Shaun 8, 80 Nolan, Daniel 80–81 non-actual possibilities 7, 63, 64, 104 non-basic beliefs 146 non-basic facts 27–28, 41n10 non-classical logics 126 non-doxastic states 114–15 non-epistemic uses of imagination 119n8 non-fctional narratives 183, 189–90, 247–48 non-idealized cognizer/thinker 31 non-ideal reasoning 34 non-imaginistic elements of imagination 6, 23–24, 31, 40, 122, 173 non-inferable details 83, 85 non-modal beliefs 23, 35 non-monotonic logic of ROMS 127, 136 non-perceptual knowledge 67–68, 197n19 nonphysical phenomena 91 non-rational infuences 16 non-sensory imagination 48 non-trivial empirical information 28 non-zero chance 50, 59, 148 normative epistemology 168 Nussbaum, Martha 247 Obama, Barack 247–48, 279 objective chance: etiological immunity and 49–52; inadmissible evidence in 49–50; modal justifcation and 44–45, 48; non-zero chance 50 objective modality 45, 48, 56–57, 60n11 objective possibilities 45, 48, 67, 71–72

Index 331 objectual mental simulation 194 occurrent perceptions 114 Offce, The (Munro) 274 “offine” 144–46, 149, 152, 156n3, 160, 163; see also simulation offine manifestation of a disposition 144–46, 149, 152, 156n3 offine simulation theory of behavior prediction 166 One Hundred Years of Solitude (Márquez) 185, 188 online manifestation of a disposition 146 optimism 4, 81, 92, 113, 177, 239–40, 251–56, 262, 275 order-insensitive conjunction 127 order-of-presentation effects 127 other-imagining 281–83, 290, 292 output states of reasoning 106–8, 110, 117, 119n4 overstipulation 34 particular theory of reasoning 106 Paul, L.A. 270 pen names 298, 299 perception: analogy of 13–14; attention and 222–23, 227–31; causal origin of 47; cognitive skills and 165–68; conscious perception 218, 232n4; empirical evidence through 177; epistemic signifcance of 5; imagination constraints and 39; modal epistemology and 63, 69, 74; occurrent perceptions 114; simulated vs. actual 14; veridical perception 5, 73–74; worldsensitivity of 2 perception/belief/desire architecture 82, 222 perception-like imagination 110, 219, 229–32; see also sensory imagination perceptual justifcation 47, 60n5, 108–9, 227 perceptual knowledge: beliefs and 77n5, 77n8; gaze-directional prepositions and 76n4; immediate vs. mediate 67–70; modal knowledge as 8, 63–76; of nonactual possibilities 64–67 perceptually manifest 65 perceptual modal epistemology: actual possibilities and 67; evolutionary

advantage 66–67, 69, 70–72; evolutionary advantage and 66–67, 69, 70–72, 77n11; imagination as guide to current environment 72–75; immediate vs. mediate 67–70; non-perceptual knowledge 67–68; objective possibilities 71–72; perceptual knowledge 64–67; real possibilities 63–64, 67, 71–72; real possibilities and 63–64, 67; summary of 75–76 Perky experiment 227 personal distress 261 perspective-taking: empathic perspective-shifting 15, 260–61, 261–62, 263–66, 270–72, 283–85, 284–85, 292–94; epistemic challenges to 266–72; experiencetaking vs. 275n1, 276n8; in-theirshoes perspective-shifting 261–62, 271–72, 283–85; overview of 15–17, 260–62; summary of 275 pessimism 15, 237–40, 247–52 Pessoa, Fernando 17, 298–315; see also Book of Disquiet, The (Pessoa) phenomenal-action scripts 89, 90 phenomenal consciousness 86–87, 122 phenomenal similarity 227–28 phenomenology of imagination 166–67 philosophical theorizing 24–25 Phosphorus 53 physicalists 53–54 physical scripts 89 pictorial representation 286–87 Pizzaro, Jeronimo 299 platitude about reasoning 107, 108, 111–13 plausibility requirement 126 poor reasoning 107, 112 possibility approaches 126 Postcolonial Melancholia (Gilroy) 238–39 practical modality 79, 93, 94–95 predicate-subject distinction 287 pre-existing beliefs 83–85, 192 preface paradox 35–37 pretense 1, 79, 82, 165 prima facie justifcation 46–47 Principal Principle (PP) 44, 49–51, 54–55, 55–58, 60, 153 Principle of the Necessity of Distinctness 67, 68, 70

332 Index process-reliabilist account of justifcation 146 proof, defned 52 propositional imagination 18, 82, 122, 125, 143, 145, 150, 173–74, 184, 196n3 propositional justifcation 224–25, 226 pseudonyms 299 psychological adjustment 285, 290 psychological failure to understand 260 psychologically immediate beliefs 68–71 psychologically realistic 68–70 psychology of imagination: conditional beliefs and 153; imagination box 82–83; metaphysical modality and 82–96; script elaborator 84–85, 90; skepticism and imagination 79–81, 92–95; UpDater model 83–84 racism experience 248, 279 Ramsey test for conditionals 123–24 Rankine, Claudia 238 rational belief 231 rational imagination 32–37, 241 rationalism about modalizing 79–80 reaction vs. imagination 52 Reality Constraint (RC) 142, 227–28 Reality-Oriented Mental Simulation (ROMS): belief-like imagination 184–85; equivalence in imagination 126–28, 136–37; imagination as 122–24; logic of 124–26; semantics of 128–36, 138; summary of 138 real possibilities 63–64, 67, 71–72 reason-giving emotions 264 reasoning and imagination see imaginative reasoning recreativist account of imagination 286 refective thinking 203, 211–13 refexivity of identity 166 relevance principle 124, 125, 150–51 reliability 75, 95, 146, 167–68 restricted substitutivity principle 127 Restricted Transitivity (RT) 137, 138 restrictive Epistemic Inaccessibility 245 Robinson, Marilynne 247–48 role-reversal as empathy 16, 280–83, 294–95

Sartre, Jean-Paul 313–14 Scarry, Elaine 248 Schank, Roger 8 Schoneboom, Abigail 301, 302 Schrödinger’s cat thought experiment 190 scientifc thought experiments 189 script elaborator 82–85, 90 script/scripts: behavior scripts 84–85, 88–89, 96n8; compositional scripts 89; disagreements about 86–92; disagreements in metaphysical modality 86–92; functional-action scripts 89, 90; impossibility in 8, 88; intuitions and 85; mental scripts 89; phenomenal-action scripts 89, 90; physical scripts 89 Self and imagination 310–15 self-ascriptions of imaginings 54–55 self-awareness 301–2 self-centered state of mind 261 self-conception 17, 305 self-consciousness 300, 313–14 self-essentializing gaze 312 self-image 305, 307 self-imagining 281–82, 290, 292, 294 self-knowledge 2, 14–17, 298, 308–12 self-love 17, 312–15 semantics: approaches to possibility/impossibility 126; conditional probabilities and 153; disagreements in 86–87; of ROMS 128–36, 138 semi-heteronyms 298, 299 sense-reference distinction 287 sensory imagination: introduction to 5, 13–14, 18, 73; as perceptionlike 110, 219, 229–32; a priori-a posteriori distinction and 160–62, 164–73; reasoning and 104, 105, 110 sexual assault experience 237–38, 240, 256, 257n8 short-term memory 191 simplifcation in imagining 132, 135, 137 simulation: cognitive architecture of 168–70; in imagination 5, 162, 166–67, 286; in mental models 191, 194, 209; propositional imagination and 174 skepticism and imagination 79–81, 92–95, 245

Index 333 Smith, Adam 257n12, 282 social phenomenon of testimony 58 Sosis, Cliff 249–50 spatial structure of experience 220–21 Spaulding, Shannon 115 speaker testimony 61n15 Special Transitivity 137, 138 Spelman, Elizabeth 241–42, 245, 255 Stitch, Stephen 8 Strohminger, Margot: introduction to 8, 13; modal auxiliaries 76n4; non-perceptual knowledge 67–68; perceptual modal epistemology 63–67, 75; Principle of the Necessity of Distinctness 67, 68, 70; psychologically realistic 68–70 subjective conditional probabilities 153 Substitutivity in ROMS 137 sui generis reasoning 103, 114, 118 supposing vs. imagining 174 suppositional input 123–28, 130–33, 136–38 suppositional state 155 sympathetic perspective-taking 261, 266 Tesla, Nikola 38 testimonial justifcation 49, 57–59 testimony, social phenomenon of 58 theoretical reasoning 103, 106, 118 theoretical stance toward oneself 264–65 Theory of Moral Sentiments (Smith) 282 Thérèse Desqueyroux (Mauriac) 188 “thinking of” in offine manifestation of disposition 145 Thomas, Laurence 240, 241–42, 253–54 thought experimentation: epistemic signifcance and 219, 230–31; ethics in 37; experiential imagining in 219; fctional narratives in imagination 105, 183, 185–90, 195n2; introduction to 3, 5, 12–14, 183–84; Mary Thought Experiment 14, 188–90, 219, 229–31, 233n8; mental models 186, 191–95, 197n16; narratives and 184–91; power of 215n19; propositional imagination and 184, 196n3; summary of 195; two types of

196n15; see also Galileo’s thought experiment “Til It Happens to You” (Lady Gaga) 237–38, 240 Too Big a Gulf argument 15, 243–47 topics/subject matters 129–30, 157n11 transformative experiences 149, 270 traumatic experiences 15 Trump, Donald 279 truth-and-falsity conditions 134–35 truth-value of mathematical statements 52 two-dimensional (2D) semantics 27 Ulysses (Joyce) 273 unconscious attention 218, 226, 232n4 unconscious beliefs 218 unconscious perception 218 unconscious reasoning 68 undermining/undercutting defeater 46–47 understanding: empathetic understanding 257n9; experiential perspectives on 249, 256, 257n7; failure to understand 260; perspective-taking challenges with 269–70 unfolding process 2–3, 154 unimaginability 60n6, 238–39 unimaginability-impossibility link 7, 45 unimaginability-possibility link 7 unjustifed beliefs 108, 110–11 unrestrictive Epistemic Inaccessibility 245 UpDater model 83–84 vagueness 28, 30, 41n13, 306 Van Cleve, James 36 Van Leeuwen, Neil 37–38 vastly different perspectives 246–47, 252 veridical perception 5, 73–74 vestiges of dreams 305 Vetter, Barbara 67 victim’s point of view 291–92 visions 301–2, 305 visual hallucination 109 visualization and memory 229, 302 vivid imagery 38 voluntary attention 220

334

Index

voluntary mode of imagination 176–77 “what is imagined?” question 285–88 “What Is It Like to Be a Bat?” (Nagel) 243–44 white liberal imagination 238 white-to-black understanding 250–51 Wilde, Oscar 311 Williams, Patricia J. 248

Williamson, Timothy 11–12, 39, 160–78; see also a priori/a posteriori distinction Wirling, Ylwa 266 wishful thinking 113 Woolf, Virginia 273 world-sensitivity of perception 2 Yancy, George 249–50 Young, Iris Marion 241–42