New Directions in the Philosophy of Memory 9781138065604, 9781315159591

Although philosophers have explored memory since antiquity, recent years have seen the birth of philosophy of memory as

709 54 2MB

English Pages [362] Year 2018

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

New Directions in the Philosophy of Memory
 9781138065604, 9781315159591

  • Commentary
  • downloaded from Memory of the World Library

Table of contents :
Cover
Title
Copyright
Contents
The Philosophy of Memory Today and Tomorrow: Editors’ Introduction
Part I Challenges and Alternatives to the Causal Theory of Memory
1 Beyond the Causal Theory? Fifty Years After Martin and Deutscher
2 A Case for Procedural Causality in Episodic Recollection
3 The Functional Character of Memory
Part II Activity and Passivity in Remembering
4 Remembering as a Mental Action
5 The Roots of Remembering: Radically Enactive Recollecting
6 Handle With Care: Activity, Passivity, and the Epistemological Role of Recollective Memories
Part III The Affective Dimension of Memory
7 Affective Memory: A Little Help From Our Imagination
8 Painful Memories
Part IV Memory in Groups
9 Shared Remembering and Distributed Affect: Varieties of Psychological Interdependence
10 Memory, Attention, and Joint Reminiscing
Part V Memory Failures: Concepts and Ethical Implications
11 Forgetting
12 On the Blameworthiness of Forgetting
13 Consent Without Memory
Part VI The Content and Phenomenology of Episodic and Semantic Memory
14 The Remembered: Understanding the Content of Episodic Memory
15 The Past Made Present: Mental Time Travel in Episodic Recollection
16 Remembering Past Experiences: Episodic Memory, Semantic Memory, and the Epistemic Asymmetry
17 On Seeming to Remember
Contributors
Index

Citation preview

New Directions in the Philosophy of Memory

Although philosophers have explored memory since antiquity, recent years have seen the birth of philosophy of memory as a distinct field. This book— the first of its kind—charts emerging directions of research in the field. The book’s seventeen newly commissioned chapters develop novel theories of remembering and forgetting, analyze the phenomenology and content of memory, debate issues in the ethics and epistemology of remembering, and explore the relationship between memory and affectivity. Written by leading researchers in the philosophy of memory, the chapters collectively present an exciting vision of the future of this dynamic area of research. Kourken Michaelian is a senior lecturer at the University of Otago. He is the author of Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past (MIT, 2016) and coeditor of Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel (2016, Oxford University Press), and The Routledge Handbook of Philosophy of Memory (2017). Dorothea Debus teaches philosophy at the University of York. She has written on philosophical questions relating to the phenomena of memory, the imagination, attention, and emotions; her current research project (“Shaping Our Mental Lives”) investigates our active involvement with our own mental lives. Denis Perrin is a professor at the Grenoble Alpes University. He is the author of Qu’est-ce que se souvenir? (2012), the editor of a special issue “Episodic memory” of the Review of Philosophy and Psychology (2014), and the author of several papers on episodic memory and mental time travel.

Routledge Studies in Contemporary Philosophy

Subjectivity and the Political Contemporary Perspectives Edited by Gavin Rae and Emma Ingala Aspect Perception after Wittgenstein Seeing-As and Novelty Edited by Michael Beaney, Brendan Harrington and Dominic Shaw Nature and Normativity Biology, Teleology, and Meaning Mark Okrent Formal Epistemology and Cartesian Skepticism In Defense of Belief in the Natural World Tomoji Shogenji Epistemic Rationality and Epistemic Normativity Patrick Bondy From Rules to Meanings New Essays on Inferentialism Edited by Ondřej Beran, Vojtěch Kolman, and Ladislav Koreň Toleration and Freedom from Harm Liberalism Reconceived Andrew Jason Cohen Voicing Dissent The Ethics and Epistemology of Making Disagreement Public Edited by Casey Rebecca Johnson New Directions in the Philosophy of Memory Edited by Kourken Michaelian, Dorothea Debus, and Denis Perrin For a full list of titles in this series, please visit www.routledge.com

New Directions in the Philosophy of Memory Edited by Kourken Michaelian, Dorothea Debus, and Denis Perrin

First published 2018 by Routledge 711 Third 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 © 2018 Taylor & Francis The right of the editors to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with Sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Names: Michaelian, Kourken, editor. Title: New directions in the philosophy of memory / edited by Kourken Michaelian, Dorothea Debus, and Denis Perrin. Description: 1 [edition]. | New York : Routledge, 2018. | Series: Routledge studies in contemporary philosophy ; 106 | Includes bibliographical references and index. Identifiers: LCCN 2017061298 | ISBN 9781138065604 (hardback : alk. paper) Subjects: LCSH: Memory (Philosophy) Classification: LCC BD181.7 .N49 2018 | DDC 128/.3—dc23 LC record available at https://lccn.loc.gov/2017061298 ISBN: 978-1-138-06560-4 (hbk) ISBN: 978-1-315-15959-1 (ebk) Typeset in Sabon by Apex CoVantage, LLC

Contents

ContentsContents

The Philosophy of Memory Today and Tomorrow: Editors’ Introduction

1

KOURKEN MICHAELIAN, DOROTHEA DEBUS, AND DENIS PERRIN

PART I

Challenges and Alternatives to the Causal Theory of Memory

11

  1 Beyond the Causal Theory? Fifty Years After Martin and Deutscher

13

KOURKEN MICHAELIAN AND SARAH K. ROBINS

  2 A Case for Procedural Causality in Episodic Recollection

33

DENIS PERRIN

  3 The Functional Character of Memory

52

JORDI FERNÁNDEZ

PART II

Activity and Passivity in Remembering

73

  4 Remembering as a Mental Action

75

SANTIAGO ARANGO-MUÑOZ AND JUAN PABLO BERMÚDEZ

  5 The Roots of Remembering: Radically Enactive Recollecting

97

DANIEL D. HUTTO AND ANCO PEETERS

  6 Handle With Care: Activity, Passivity, and the Epistemological Role of Recollective Memories DOROTHEA DEBUS

119

vi Contents PART III

The Affective Dimension of Memory

137

  7 Affective Memory: A Little Help From Our Imagination

139

MARGHERITA ARCANGELI AND JÉRÔME DOKIC

  8 Painful Memories

158

PHILIP GERRANS

PART IV

Memory in Groups

179

  9 Shared Remembering and Distributed Affect: Varieties of Psychological Interdependence

181

JOHN SUTTON

10 Memory, Attention, and Joint Reminiscing

200

FELIPE DE BRIGARD

PART V

Memory Failures: Concepts and Ethical Implications 

221

11 Forgetting

223

MATTHEW FRISE

12 On the Blameworthiness of Forgetting

241

SVEN BERNECKER

13 Consent Without Memory

259

CARL F. CRAVER AND R. SHAYNA ROSENBAUM

PART VI

The Content and Phenomenology of Episodic and Semantic Memory

277

14 The Remembered: Understanding the Content of Episodic Memory279 MARK ROWLANDS

15 The Past Made Present: Mental Time Travel in Episodic Recollection294 MATTHEW SOTERIOU

Contents  vii 16 Remembering Past Experiences: Episodic Memory, Semantic Memory, and the Epistemic Asymmetry

313

CHRISTOPH HOERL

17 On Seeming to Remember

329

FABRICE TERONI

Contributors Index

347 350

The Philosophy of Memory Today and Tomorrow: Editors’ Introduction Kourken Michaelian et al. Editors’ Introduction

Kourken Michaelian, Dorothea Debus, and Denis Perrin

1  The Philosophy of Memory Today As Bernecker and Michaelian point out in their introduction to the recent Routledge Handbook of Philosophy of Memory (2017), it is hardly surprising, given that memory is one of our most fundamental cognitive capacities and the source of an enormous fraction of our knowledge, that philosophical interest in memory dates back to the dawn of philosophy and has remained strong throughout the history of the discipline. What is surprising is the fact that the philosophy of memory has only recently emerged as a recognized field of research. Whatever the explanation for the delayed emergence of the field, it now most definitely has emerged, with the publication of the Handbook being only the most obvious sign of this development. Other signs include the rapid proliferation of workshops, conferences, and special issues on the topic of memory and the establishment, at the Issues in Philosophy of Memory conference held in Cologne in 2017, of the PHilosophy Of Memory Organization (phomo.org). There is, in short, a lot going on in the philosophy of memory today, and, whereas the entries for the Handbook were compiled so as to provide a systematic overview of historical and contemporary philosophical research on memory, the present volume, New Directions in the Philosophy of Memory, offers a snapshot of some of the most active and dynamic areas of current research.

2  Overview of the Book The book has seventeen chapters (not including this introduction). The chapters are grouped into six parts, of which we here provide brief overviews. 2.1  Part I: Challenges and Alternatives to the Causal Theory of Memory Though the causal theory, which was given its classical formulation by Martin and Deutscher (1966), has for long been the dominant philosophical theory of remembering (see, e.g., Bernecker, 2010; Cheng & Werning, 2016), it has in recent years come under increasing pressure, and the

2  Kourken Michaelian et al. three contributions to Part I consider challenges and alternatives to the theory. According to the classical causal theory, the occurrence of genuine remembering presupposes the existence of a certain sort of causal connection between the subject’s apparent memory and his earlier experience of the remembered event, namely, a causal connection sustained by a memory trace originating in that experience. As Michaelian and Robins demonstrate in their chapter, increasing recognition among philosophers of the thoroughly reconstructive character of remembering (e.g., Robins, forthcoming; Salvaggio, forthcoming) has led to the formulation not only of versions of the causal theory that attempt to do without memory traces but also of postcausal theories that reject the necessity of causal connection for genuine remembering outright. Inspired both by Sutton’s (1998) philosophical work on distributed memory storage and attributionalist approaches to memory in psychology (Whittlesea, 1997), Perrin develops a theory of the former sort. According to his procedural causal theory, the causal connection characteristic of remembering obtains not between the retrieved memory and the earlier experience but rather between the reconstructive process that produces the memory and the constructive process that produced the experience. Though quite far from the theory defended by Martin and Deutscher, Perrin’s theory is recognizably a causal theory. The functionalist theory developed by Fernández, in contrast, joins postcausal theories such as the simulation theory (Michaelian, 2016; cf. De Brigard, 2014) in advocating the outright rejection of the necessity of causal connection for remembering. Applying general functionalist approaches developed in philosophy of mind, Fernández argues that a memory state should be understood as a state of a sort that tends to be caused by the corresponding earlier experience but that it need not in fact be caused by that experience. 2.2  Part II: Activity and Passivity in Remembering Both Perrin’s argument for the procedural causal theory and Fernández’s argument for the functionalist theory are informed by a conception of remembering as an active, reconstructive process. The three chapters that make up Part II share this conception, but rather than proposing general theories of reconstructive remembering, they focus specifically on its active dimension. Building on recent work on feeling-based metacognition (see, e.g., Proust, 2013; Dokic, 2014), Arango-Muñoz and Bermúdez argue that, far from being a matter of the passive storage and retrieval of information, remembering is a matter of reconstructing the past and that this reconstruction qualifies, in virtue of the role played in it by metacognitive feelings, as a form of full-fledged mental action. Like Arango-Muñoz and Bermúdez, Hutto and Peeters abandon the passive, storage-and-retrieval picture of remembering for a picture of remembering as the active reconstruction of the past. Building on a range of empirical research, including research on episodic memory as a form of mental time

Editors’ Introduction  3 travel (Perrin & Michaelian, 2017), they argue that radical enactivism (e.g., Hutto & Myin, 2013, 2017) grounds a view of remembering, including experientially rich forms of episodic remembering, that dispenses entirely with the notion of stored content. There are potential links here both with the procedural causal theory defended by Perrin in his chapter, which likewise rejects the claim that remembering involves stored content, and with the relationalist account of memory defended elsewhere by Debus (2008), which rejects the view that retrieved memories have representational content. Debus’ contribution to this volume may nevertheless be at odds with those of Hutto and Peeters and of Arango-Muñoz and Bermúdez, as it stresses that episodic or recollective memories have characteristic features of both activity and passivity. Adopting an epistemological perspective, Debus argues that memory is able to serve as a source and ground of knowledge because, when the subject remembers, the past impinges on him as a passive recipient. There may be less incompatibility with more reconstructive views here than one might initially think, however, as Debus also argues that the relevant type of passivity is compatible with at least some active intervention by the subject, such as that involved in switching from field to observer perspective (see McCarroll, forthcoming). 2.3  Part III: The Affective Dimension of Memory Like a majority of the contributors to Part II, Gerrans understands memory as a form of mental time travel into the past, performed by the same neural systems responsible for mental time travel into the future. Whereas the contributions to Part II are concerned with the active dimension of remembering, understood as past-oriented mental time travel, Gerrans’ contribution to Part III is concerned with another specific dimension of pastand future-oriented mental time travel, namely, its affective dimension. Mental time travel, he argues, in line with Goldie’s (2012) narrative account and with accounts stressing the sense of ownership involved in memory (Fernández, forthcoming), can be understood as involving empathy for one’s past or future selves. Focusing on the case of painful memories, he argues that, when one remembers, one can in a very literal sense share feelings with one’s past self. One does this not, he claims, by simulating one’s past experience in full, including one’s past feeling of pain, but rather by simulating the feeling of being the person undergoing the pain. Arcangeli and Dokic are similarly concerned with the affective dimension of remembering, understood as past-oriented mental time travel. On one view of the relationship between memory and emotion, a memory can be about a past emotion. On another view, a memory can cause a present emotion. But on neither view can a memory itself have an emotional component. Arcangeli and Dokic, drawing, like Gerrans, on Goldie’s account, argue for a novel view of the relationship between memory and emotion, suggesting that we must acknowledge, in addition to the perspective of the remembered

4  Kourken Michaelian et al. subject and the perspective of the remembering subject, the perspective of the narrator of the remembered event and that doing so enables us to acknowledge the possibility of memories with a genuinely emotional component, the idea being that genuinely affective memories occur as the narrator’s emotional perspective directly influences the remembering subject’s emotional perspective. 2.4  Part IV: Memory in Groups The affective dimension of remembering is also at issue in Sutton’s chapter, but whereas the contributions to Part III largely bracket the social dimension of remembering, Sutton seeks to bring these two dimensions together in a treatment of memory at the group level. Picking up on recent approaches to remembering as a collective process (Michaelian & Sutton, 2017) and recent arguments for the possibility of distributed affectivity (Krueger & Szanto, 2016), he argues that both remembering and feeling are sometimes socially shared, rather than internal or strictly individual, activities. In line with his previous argument for a complementarity-based approach to extended cognition (Sutton, 2010), however, he maintains that complementary relations between subjects in different affective states are often more important for the occurrence of collective remembering than is convergence among individuals in the same affective state. De Brigard’s chapter likewise concerns the social or collective dimension of remembering, but whereas Sutton’s focus is on affect, De Brigard’s focus is on the abilities necessary for participation in joint reminiscing. He draws on earlier suggestions by Hoerl and McCormack (2004) and Campbell (2002) to explore the abilities on which this form of collective remembering depends, identifying three such abilities. Mental ostension consists in inward attention to a specific component of a memorial content. Deferred mental ostension is a matter of indirect attention through deference to a present mental content. Finally, concerted deferred mental ostension is the ability to guide someone else’s attention inwardly toward the relevant aspect of the mental content. 2.5  Part V: Memory Failure: Concepts and Ethical Implications Many of the chapters in the preceding parts of the book seek to understand how memory works when it does work; the chapters in Part V, in contrast, seek to understand how memory works when it fails. Frise’s chapter takes up a topic that has so far been neglected in the philosophy of memory despite its obvious importance to the field, that of the nature of forgetting. Reviewing a range of discussions of forgetting in the philosophical and psychological literatures on memory, Frise considers and rejects the theories of the nature of forgetting implicit in these discussions on the ground that they fail two tests: they do not accommodate the existence of the feeling of forgetting (see Arango-Muñoz, 2013), and they do not accommodate

Editors’ Introduction  5 the fact that some failures of prospective memory (see Szpunar, Spreng, & Schacter, 2016) qualify as instances of forgetting. He therefore introduces a new theory, the LEarning, Access failure, Dispositional (LEAD) theory, and argues that it passes these tests. Bernecker’s chapter is likewise concerned with forgetting, focusing not on its nature but rather on its ethics. Many have claimed that one cannot be held responsible for forgetting because it is not under one’s control. Some (e.g., Matheson, 2017) have argued that forgetting is sometimes under one’s control and that, when it is, one can be held responsible for it. Bernecker concurs but goes further, arguing that one can also sometimes be held responsible for forgetting even when it is not under one’s control. Craver and Rosenbaum’s chapter likewise has an ethical focus, but they are concerned with the ethical implications of a much more dramatic sort of memory failure, that manifested in cases of episodic amnesia. Writing from the point of view of researchers working with amnesic subjects, they interrogate the ability of such subjects to give meaningful consent to participate in experiments. While it might be thought that episodic amnesiacs are trapped in a “permanent present tense” (Corkin, 2013), they discuss evidence that amnesiacs are likely to have the capacities necessary for meaningful consent (e.g., Craver et al., 2014) and argue that a subject need not be able to recall the moment of consent in order for his consent to endure. 2.6 Part VI: The Content and Phenomenology of Episodic and Semantic Memory The chapters making up the final part of the book are all concerned, in one way or another, with the content and phenomenology of episodic and semantic memory. Rowlands, in his chapter, sets out the idea, developed in his own recent work (Rowlands, 2017) and in that of others (Fernández, 2008), that the content of a retrieved episodic memory must be understood as referring not only to the remembered episode but also to the location of the remembered episode in the subject’s personal past, suggesting that the latter aspect of episodic memory content can be described using the Fregean notion of a mode of presentation. He argues further that this implies that the content of an episodic memory is not independent of the act of remembering and that this entails that remembering is reconstructive in character. Soteriou similarly suggests that episodic remembering involves not only a representation of a past episode but also a representation of a temporal perspective on that episode. In opposition to recent arguments that have sought to undermine the distinction between remembering the past and imagining it (Michaelian, 2016), he argues that the distinction is underwritten by the different ways in which the temporal location of the represented event is determined: by the actual temporal location of the past event, in the case of remembering, and by the subject’s intentions, in the case of imagining (cf. Hoerl, 2014). Whereas Soteriou is concerned with

6  Kourken Michaelian et al. the distinction between episodic memory and episodic imagination, Hoerl, in his chapter, is concerned with the distinction between episodic memory and semantic memory. The question of the distinguishing mark of episodic memory—of the nature of episodicity (Perrin & Rousset, 2014)—is familiar, but Hoerl defends a new answer to the question, arguing that, while both episodic memory and semantic memory display an epistemic asymmetry between past and future, they do so in different ways, with episodic memory providing the subject with knowledge not only of past events but also of what it was like to experience them. Teroni, finally, is concerned with semantic rather than episodic memory, suggesting that, while the attitude of semantic remembering can be explained in terms of the feeling of familiarity, the feeling of familiarity does not itself justify semantic memory belief. Instead, he argues, it is the subject’s past reasons for the belief, rather than the associated feeling of familiarity, that justifies a current semantic memory belief.

3  The Philosophy of Memory Tomorrow Our intention, in putting this book together, was for it to provide a snapshot of current research directions in the philosophy of memory. What, then, does the snapshot reveal? Our overview of the book’s contents suggests four key observations. 3.1  Methodological Consensus First, there is now a stable consensus on the relevance of empirical research in psychology and other disciplines to the resolution of philosophical debates about memory. There remain, to be sure, differences of degree between those philosophers of memory who proceed in a more a priori manner and those who rely more heavily on empirical research. But virtually every chapter in the book is informed to some extent by empirical research, suggesting that the difference between relevant underlying methodological assumptions is much less stark than it was only a few years ago. 3.2  Substantive Consensus Second, there is now a stable consensus on the active, reconstructive character of remembering. Whereas philosophers of memory until recently treated this as something to be defended (or attacked) by means of argument, the claim that memory has a reconstructive character now most often serves as a starting point for further argument. The psychology of memory is univocal in its endorsement of a reconstructive view, and this substantive consensus is no doubt in part a product of the methodological consensus noted previously. The consensus is on display throughout the book, but see especially the chapters in Part II.

Editors’ Introduction  7 3.3  New Questions About Memory Third, as consensus emerges on certain basic features of memory, philosophers of memory are asking new questions. These include more finegrained questions about previously neglected aspects of individual memory, such as the role of affect discussed in Part III. They include questions about previously neglected mnemic phenomena such as the sorts of memory failure discussed in Part V. And they include questions about remembering beyond the individual level, such as the forms of group memory discussed in Part IV. 3.4  Back to Basics Finally, even as they ask new questions, philosophers of memory are displaying a new willingness to tackle afresh what is perhaps the most basic question of the philosophy of memory, that of the nature of remembering as such, by proposing radically new theories of remembering. As the contributions to Part I demonstrate, increasing methodological reliance on empirical results has suggested new ways of understanding the nature of remembering; these challenge the long-dominant causal theory without reverting to older epistemic or empiricist theories of memory. Along the same lines, the contributions to Part VI illustrate continuing attempts to solve the difficult puzzles posed by the relationship between episodic memory and episodic imagination and that between episodic memory and semantic memory. Overall, this combination of methodological consensus and consensus on the broad contours of the phenomenon under investigation, on the one hand, with, on the other hand, active debate over specific features of the phenomenon and rival general theoretical characterizations of it suggests a field in good health. The state of the philosophy of memory today thus suggests a bright tomorrow for the field. Acknowledgments: The initial idea for this book emerged through conversation among the editors at a conference in Grenoble in 2015. Many of the chapters were presented in preliminary form at two subsequent meetings, Memory and Subjectivity, held in Grenoble in mid-2016 and funded by a grant to Denis Perrin from the Institut Universitaire de France, and New Directions in the Philosophy of Memory, held in Dunedin in late 2016 and funded by a grant to Kourken Michaelian from the University of Otago. In addition to the funders, the editors wish to thank the contributors, all of whom delivered excellent chapters on time despite having to respect a tight deadline, as well as those colleagues (some themselves contributors and some not) who served as reviewers for the chapters.

References Arango-Muñoz, S. (2013). Scaffolded memory and metacognitive feelings. Review of Philosophy and Psychology, 4(1), 135–152.

8  Kourken Michaelian et al. Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press. Bernecker, S., & Michaelian, K. (Eds.). (2017). The Routledge handbook of philosophy of memory. London: Routledge. Campbell, J. (2002). Consciousness and reference. Oxford: Oxford University Press. Cheng, S., & Werning, M. (2016). What is episodic memory if it is a natural kind? Synthese, 193(5), 1345–1385. Corkin, S. (2013). Permanent present tense: The unforgettable life of the amnesic patient, HM. New York: Basic Books. Craver, C. F., Kwan, D., Steindam, D., & Rosenbaum, R. S. (2014). Individuals with episodic amnesia are not stuck in time. Neuropsychologia, 57, 191–195. De Brigard, F. (2014). Is memory for remembering? Recollection as a form of episodic hypothetical thinking. Synthese, 191(2), 155–185. Debus, D. (2008). Experiencing the past: A relational account of recollective memory. dialectica, 62(4), 405–432. Dokic, J. (2014). Feeling the past: A two-tiered account of episodic memory. Review of Philosophy and Psychology, 5(3), 413–426. Fernández, J. (2008). Memory and time. Philosophical Studies, 141(3), 333–356. Fernández, J. (Forthcoming). The ownership of memories. In M. García-Carpintero & M. Guillot (Eds.), The sense of mineness. Oxford: Oxford University Press. Goldie, P. (2012). The mess inside: Narrative, emotion, and the mind. Oxford: Oxford University Press. Hoerl, C. (2014). Remembering events and remembering looks. Review of Philosophy and Psychology, 5(3), 351–372. Hoerl, C., & McCormack, T. (2004). Joint reminiscing as joint attention to the past. In N. Eilan, C. Hoerl, T. McCormack, & J. Roessler (Eds.), Joint attention: Communication and other minds (pp. 260–286). Oxford: Oxford University Press. Hutto, D. D., & Myin, E. (2013). Radicalizing enactivism: Basic minds without content. Cambridge, MA: MIT Press. Hutto, D. D., & Myin, E. (2017). Evolving enactivism: Basic minds meet content. Cambridge, MA: MIT Press. Krueger, J., & Szanto, T. (2016). Extended emotions. Philosophy Compass, 11(12), 863–878. Martin, C. B., & Deutscher, M. (1966). Remembering. The Philosophical Review, 75(2), 161–196. Matheson, D. (2017). An obligation to forget. In S. Bernecker & K. Michaelian (Eds.), The Routledge handbook of philosophy of memory (pp. 364–372). London: Routledge. McCarroll, C. (Forthcoming). Remembering from the outside: Personal memory and the perspectival mind. Oxford: Oxford University Press. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Michaelian, K., & Sutton, J. (2017). Collective memory. In M. Jankovic & K. Ludwig (Eds.), The Routledge handbook of collective intentionality (pp. 140– 151). London: Routledge. Perrin, D., & Michaelian, K. (2017). Memory as mental time travel. In S. Bernecker & K. Michaelian (Eds.), The Routledge handbook of philosophy of memory (pp. 228–239). London: Routledge.

Editors’ Introduction  9 Perrin, D., & Rousset, S. (2014). The episodicity of memory. Review of Philosophy and Psychology, 5(3), 291–312. Proust, J. (2013). The philosophy of metacognition: Mental agency and selfawareness. Oxford: Oxford University Press. Robins, S. K. (Forthcoming). Confabulation and constructive memory. Synthese. Rowlands, M. (2017). Memory and the self: Phenomenology, science, and autobiography. Oxford: Oxford University Press. Salvaggio, M. (Forthcoming). The justification of reconstructive and reproductive memory beliefs. Philosophical Studies. Sutton, J. (1998). Philosophy and memory traces: Descartes to connectionism. Cambridge: Cambridge University Press. Sutton, J. (2010). Exograms and interdisciplinarity: History, the extended mind, and the civilizing process. In R. Menary (Ed.), The extended mind (pp. 189–225). Cambridge, MA: MIT Press. Szpunar, K. K., Spreng, R. N., & Schacter, D. L. (2016). Toward a taxonomy of future thinking. In K. Michaelian, S. B. Klein, & K. K. Szpunar (Eds.), Seeing the future: Theoretical perspectives on future-oriented mental time travel (pp. 21–35). Oxford: Oxford University Press. Whittlesea, B. (1997). Production, evaluation, and preservation of experiences: Constructive processing in remembering and performance tasks. In D. L. Medin (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol. 37, pp. 211–264). Cambridge, MA: Academic Press.

Part I

Challenges and Alternatives to the Causal Theory of Memory

1 Beyond the Causal Theory?

Kourken Michaelian and Sarah K. RobinsBeyond the Causal Theory?

Fifty Years After Martin and Deutscher1 Kourken Michaelian and Sarah K. Robins

1 Introduction It is natural to think of remembering in terms of causation: I can recall a recent dinner with a friend because I experienced that dinner. Some fifty years ago, Martin and Deutscher (1966) turned this basic thought into a full-fledged theory of memory, a theory that—due both to its intuitive plausibility and its apparent success in distinguishing remembering from related processes, including imagining—came over the following decades to dominate the landscape in the philosophy of memory. Previous approaches, such as the empiricist theory,2 had attempted to capture the nature of remembering from a first-person perspective, in terms of its characteristic phenomenology. The causal theory, in contrast, offered a third-personal account of the nature of remembering. Remembering, Martin and Deutscher argue, boils down to the existence of a specific sort of causal connection between the rememberer’s original experience of an event and his later representation of that event: a causal connection sustained by a memory trace. Though it initially seemed to boost the theory’s fit with the empirical science of memory, it has become apparent in recent years that this reference to memory traces in fact threatens to undermine the causal theory. As older conceptions of memory in terms of storage and retrieval have given way to new conceptions of remembering as a constructive or simulational process, contemporary memory science has appeared to overturn not only the particular view of traces advocated by Martin and Deutscher but also the more general claim that traces—of one sort or another—are essential to remembering. Contemporary proponents of the causal theory have thus been confronted with the question: is it possible to develop an empirically adequate version of the theory, or is it time to move beyond the causal theory? The purpose of this chapter is to trace the recent history of the causal theory, showing how increased awareness of problems for the classical causal theory has led to the development of a variety of updated versions of the theory and ultimately to the emergence of postcausal theories.

14  Kourken Michaelian and Sarah K. Robins

2  The Classical Causal Theory Like most subsequent causal theorists, Martin and Deutscher focus on episodic memory, memory for past events. Omitting certain technical details, they argue that a subject remembers a past event if and only if (1) he now represents the event, (2) he experienced the event when it took place, and (3) there is a causal connection between his current representation of the event and his experience of it. This account treats memory as a diachronic capacity, in the sense that it claims that for remembering to occur is for there to be a particular relationship between representations located at two different points in time: the subject’s original experiential representation of the event and his current retrieved representation3 of the event. Conditions (1) and (2) require the existence of these representations and are widely accepted constraints on remembering. It is in virtue of the third condition, stipulating a causal connection between the two representations, that Martin and Deutscher’s account qualifies as a causal theory. Anticausal approaches to the mind in general (e.g., Wittgenstein, 1953; Holland, 1954) and to memory in particular (e.g., Malcolm, 1963; Squires, 1969) were popular when Martin and Deutscher wrote, and the causal condition was therefore objectionable to many of their contemporaries. Nevertheless, though acausal accounts of memory are still occasionally defended (e.g., Martin, 2001; Hamilton, 2003), the causal theory, arguably due to the attention devoted by Martin and Deutscher to refining condition (3), gradually won out over the alternatives. In formulating the causal condition, Martin and Deutscher’s primary concern was to differentiate remembering from imagining. Even if a subject somehow manages to produce a representation that is accurate with respect to a past experience that she has had, her representation will intuitively fail to qualify as a memory if it lacks a causal connection to that experience. Suppose that Roger attends a magic show. Later, he suffers an accident, the result of which is complete retrograde amnesia: he no longer remembers events from his past, including the magic show. Also as a result of the accident, he is prone to producing confabulatory accounts of past events. Suppose that he produces a story that happens to correspond in perfect detail to his experience of the magic show. Conditions (1) and (2) are satisfied, but Roger is clearly not remembering. This sort of coincidental correspondence between an experiential representation and a retrieved representation may be unlikely, but its very possibility suggests the need for a causal connection between the representations—absent such a connection, the subject would seem to be merely imagining. Hence condition (3). Martin and Deutscher argue further that not just any causal connection between an experiential representation and a retrieved representation suffices for remembering: remembering requires a causal connection sustained by a memory trace. The inclusion of a reference to memory traces in the theory is necessary in part in order to differentiate remembering from relearning,

Beyond the Causal Theory?  15 which occurs when one acquires information through experience, forgets it, and then reacquires it from another source. Suppose, again, that Roger attends a magic show; later, he suffers an accident, the result of which is complete retrograde amnesia. If, at some point between the show and the accident, Roger told his friend Lane about the show, then he might later relearn of it from him. Suppose that Lane comforts Roger after his trauma by repeating stories of his past, including that of the magic show. As a result, Roger is again able to represent the event. In this case, the experiential representation and the retrieved representation are causally connected: Roger’s experience of the magic show is the cause of his conversation with Lane, which is in turn causally implicated in Lane’s relaying the information to him. But intuitively this is a case of relearning, not remembering. In differentiating remembering from relearning, Martin and Deutscher were sensitive to the fact that the occurrence of remembering is compatible with the use of external prompts. Drawing a distinction between remembering and relearning requires saying when external information serves as a mere supplement to memory and when it serves as a replacement for it; that is, we need a way of excluding relearning while permitting prompting. Martin and Deutscher do not draw the distinction in terms of the quantity of external information involved in the process of (apparent) remembering but rather in terms of the role it plays. Remembering, for them, is compatible with extensive prompting from external sources. What matters is whether there is also an internal state of the (apparent) rememberer that is active—a state acquired as a result of the experience that he is trying to remember, that is, a memory trace. Condition (3) thus becomes: there is a causal connection, sustained by a memory trace, between the subject’s retrieved representation of the event and his experiential representation of the event. Just as the bare requirement of a causal connection was objectionable to many of Martin and Deutscher’s contemporaries, so was the more specific requirement of a causal connection sustained by a memory trace. Some worried that to include a reference to memory traces in a philosophical theory of remembering was to allow philosophy to “dictate to science what to discover in the human brain” (Zemach, 1983: 32). Others were concerned about influence in the opposite direction, worrying that Martin and Deutscher’s reference to memory traces was an attempt to import a scientific notion into the everyday concept of memory that philosophy was meant to analyze (Malcolm, 1977). The relationship of the causal theory of memory to the science of memory remains an open question, and we return to this question in subsequent sections. Martin and Deutscher were also sensitive to the possibility that a cognitive capacity other than memory, also acquired during the subject’s experience of an event, might result in a later representation of the event. The desire to preclude this possibility led them to add further details to the memory trace requirement. Suppose that Roger, while attending the magic show, is hypnotized and as a result can be placed in a highly suggestible

16  Kourken Michaelian and Sarah K. Robins state. Suppose that Lane tells Roger about the magic show while he is in this suggestible state and that Roger endorses Lane’s account. Intuitively, though there is a causal connection between his experience of the magic show and his representation of it, he does not remember the magic show. This case fails to be a case of remembering because, while Roger might have a suitable memory trace, his memory trace is not doing the relevant causal work—it is some other, nonmemorial capacity that is responsible for the representation. To exclude such cases of nonmemorial retention,4 Martin and Deutscher argue that remembering requires the preservation of a trace that represents the past and provides the content of the retrieved representation. In particular, they see traces as “structural analogues” of past experiences: a memory trace is an entity that contains a quantity of information that matches or exceeds what the subject recalls about the relevant event. In other words, remembering, for them, necessarily involves the transmission of content from experience to retrieval and is incompatible with the generation of new content between experience and retrieval. Martin and Deutscher’s appeal to memory traces is simultaneously a nod to convention and a bold innovation. On the one hand, the claim that memory traces are structural analogues of past experience is a longstanding and widespread assumption of both philosophical and everyday thinking about memory (see Draaisma, 2000; De Brigard, 2014b): just as Martin and Deutscher compare memory to the grooves of a record, Plato, for example, compared it to impressions in a wax tablet. On the other hand, Martin and Deutscher offer a new reason for this old view of memory traces, treating traces not as the objects of remembering but rather as the bearers of the right sort of causal connection between the experiential representation and the retrieved representation. Despite the fact that the characterization of memory traces as structural analogues of past experiences is traditional, however, there is reason to prune it from the causal theory. To say that memory traces are structural analogues of past experiences is to say that a memory trace represents an experience in virtue of its standing in a relationship of structural isomorphism with that event. As an account of mental representation, structural isomorphism provides a way of ensuring that the inferential interactions between the contents of thought are reflected in the causal interactions between the vehicles by which they are represented. Although this view of mental representations was popular at the time at which Martin and Deutscher were writing, it is controversial and is not now widely endorsed (e.g., Shepard & Chipman, 1970). Moreover, the characterization of memory traces as structural analogues of past experience makes a claim about how mental representation works, and this specific claim goes beyond the general claim, required by the causal theory, that memory traces must be mental representations.5 In what follows, we therefore do not interpret the classical causal theory as incorporating a characterization of memory traces as structural analogues of past experiences.

Beyond the Causal Theory?  17 According to the classical causal theory, then, it is memory traces that make the difference between a mere causal connection between an experiential representation and a retrieved representation and what we can refer to as an appropriate causal connection, a causal connection of the sort required to underwrite remembering. Pruned of the structural analogy requirement, the causal theory makes an empirical bet regarding the existence of traces but stops short of betting on any particular account of the physical nature of memory traces. The physical details do not matter; only certain very general features do. In line with their treatment of imagining, relearning, and nonmemorial retention, Martin and Deutscher are committed to viewing memory traces, first, as being distinct states and second, as having distinct contents. Regarding the first commitment, a memory trace must be a distinct, internal state of the rememberer. In order for this condition to be met, the causal chain leading back to the experience must be distinguishable from other causal chains. After all, people have multiple memories and therefore multiple memory traces. Roger, from our previous example, has a memory of attending a magic show but presumably many other memories as well. To determine whether he remembers the magic show requires establishing that that this particular causal chain has been sustained. To determine whether he remembers another experience—his fifth birthday, for instance—requires establishing the existence of a different causal chain. This is only possible if the chain supported by each internal state is distinct. This distinctness serves as a marker of the unique causal history of each memory trace, which becomes especially important in establishing the difference between remembering and relearning. Remembering and relearning might produce exactly similar representations; the only way of differentiating between them is by when and how they were acquired. Regarding the second commitment, the memory trace must not only provide a distinct causal link via an internal state that serves as a representation of that experience. As in the earlier example of hypnosis, it is possible that other aspects of a preserved internal state could result in a representation of a past experience. If we are to establish the difference between remembering and nonmemorial forms of retention, then there must be some way in which the memory trace is distinct from these other processes. The memory trace must be a distinct component of the internal state in which it features, distinguishable from all other components this state may have. The memory trace is distinctive, Martin and Deutscher argue, because it alone represents that past experience. By preserving information about that event or experience across time, the memory trace is distinguishable from other retained states that could in one or another way result in representations of the experience. Moreover, by preserving information over time, the memory trace provides an explanation of how accurate retention of information from the past is possible.

18  Kourken Michaelian and Sarah K. Robins

3  Neoclassical Causal Theories Fifty years after Martin and Deutscher wrote, the classical causal theory continues to be influential, and a number of causal theories that may be characterized as neoclassical have recently been developed. Neoclassical causal theories retain the core claim of Martin and Deutscher’s theory— that an appropriate causal connection (where appropriate causation is understood as causation going via a memory trace) is both necessary and, along with other suitable conditions, sufficient for memory—while modifying certain less central elements of the theory. The theories proposed by Bernecker (2008, 2010) and Cheng and Werning (2016) are illustrative of the neoclassical approach. Offering a systematic argument for the superiority of the causal theory over noncausal theories, Bernecker offers a detailed development of a causal theory in the spirit of Martin and Deutscher’s. In particular, he understands appropriate causation in terms of contiguity, maintaining that it is the presence of an uninterrupted chain of memory traces between learning and remembering that distinguishes remembering from imagining and relearning. Bernecker’s analysis also updates Martin and Deutscher’s in certain respects. First, he denies that the content of the experiential representation and the content of the retrieved representation must be identical. Instead, they must be “sufficiently similar” (2010: 217): content can change over time (e.g., one might initially remember receiving a new bicycle and later only remember receiving a gift), but no new content can be generated. Second, he endorses a distributed view of traces. Unlike the distributed conceptions of traces that we discuss in the next section, however, Bernecker’s view is that traces are distributed at the implementational level only, allowing content transmission to occur at the psychological level. Cheng and Werning’s approach differs from Bernecker’s in terms of both scope and method. In terms of scope, Bernecker discusses a range of forms of memory, including memory for persons and things, memory for properties, memory for events, and memory for facts and propositions, focusing on the latter. Cheng and Werning focus specifically on memory for events—more precisely, on episodic memory, their understanding of which, in line with the psychological literature on mental time travel (Suddendorf & Corballis, 1997), includes a role for autonoesis, or consciousness of the self in subjective time (Tulving, 1985), a topic to which we return in Section 3. In terms of method, whereas Bernecker relies primarily on the tools of conceptual analysis. Cheng and Werning’s approach is naturalistic in spirit, appealing to data on the role of specific brain structures, primarily the hippocampus, in remembering; like Michaelian (2011b), they seek to understand memory as a natural kind. While this naturalistic approach lends a degree of methodological novelty to their approach, the main substantive novelty of their version of the causal theory consists in its characterization of memory representations as being sequential in nature, a characterization they derive from

Beyond the Causal Theory?  19 their understanding of the role of hippocampal processes in remembering (cf. Cheng, Werning, & Suddendorf, 2016). Ultimately, however, the gist of their theory—which requires that the retrieved representation be causally grounded in the corresponding earlier experience via a memory trace—is similar to that of Bernecker’s, which, as we have seen, is in turn similar to that of Martin and Deutscher’s. Both Bernecker (2008, 2010) and Cheng and Werning (2016) can thus be classified as neoclassical causal theorists,6 and the differences between their respective versions of the causal theory, as well as those between their versions of the causal theory and Martin and Deutscher’s version of the theory, can be set aside for present purposes. Both classical and neoclassical causal theorists assume, first, that remembering involves the transmission of content from experience to retrieval and second, that remembering is incompatible with the generation of new content between experience and retrieval. Each of these assumptions has, however, been denied by other recent versions of the causal theory. We consider theories that deny the former assumption in Section 5 and theories that deny the latter in Section 6.

4  Hybrid Theories Setting the issues of transmission and generation aside for the moment, we emphasize that Martin and Deutscher’s core claim—that an appropriate causal connection is both necessary and, along with other suitable conditions, sufficient for memory—is accepted in one form or another by many contemporary philosophers of memory (see Debus, 2017).7 In particular, we note that the literature contains few challenges to the claim that appropriate causal connection is necessary for memory. Some invoke this claim in passing while focusing on other issues (e.g., Debus, 2008, 2014; Hopkins, 2014). Others do not invoke it but nevertheless say nothing to challenge it. In contrast, the literature does contain a number of challenges to the claim that appropriate causation is sufficient for memory. If one of these challenges were to succeed, it would be necessary to supplement the appropriate causation condition—along with the other basic conditions required by the causal theory—with a further condition, thus producing a hybrid theory of remembering. Debus (2010; cf. James, forthcoming), for example, argues that genuine memories are, in addition to being causally connected to the subject’s past experiences, necessarily epistemically relevant to the subject, in the sense that he is disposed to take them into account when forming judgments about the past, typically (but not always) by forming a belief that the remembered event occurred. Because the classical causal theory does not treat epistemic relevance as necessary for remembering, Debus maintains, it is bound to classify certain cases as instances of genuine memory when in fact they are instances of merely apparent memory. (Consider Martin and Deutscher’s oft-discussed case of a painter who paints a scene from his past without

20  Kourken Michaelian and Sarah K. Robins realizing that it is a scene from his past.) This argument, which, if it works, applies equally to neoclassical causal theories, in effect suggests that the causal theory should be replaced with a hybrid causal-epistemic theory. Similarly, Klein (2014, 2015; cf. Dokic, 2014) argues that genuine memories necessarily involve, in addition to causal connection, a specific phenomenology: autonoetic consciousness, or a sense of the self in subjective time. Klein and Nichols (2012; cf. Fernández, forthcoming), for example, discuss the case of patient RB, whom they characterize as having retained the capacity to retrieve information deriving from his past experiences but as lacking a “sense of mineness” for the memories thus produced. Though the causal theory would classify the case of RB as one in which the subject is capable of remembering, on Klein’s view RB is, because he lacks the capacity for autonoesis, incapable of genuine memory. This argument, which, if it works, applies, like Debus’s argument, equally to the other versions of the causal theory considered so far, in effect suggests that the causal theory should be replaced with a hybrid causal-autonoetic theory. The causal-autonoetic theory and the causal-epistemic theory are close cousins: as Mahr and Csibra (forthcoming) have emphasized, the involvement of autonoesis in remembering explains the subject’s tendency to believe that remembered events occurred. And they are thus vulnerable to similar challenges. In particular, both the causal-autonoetic and the causal-epistemic theory imply that semantic memory (memory for facts) differs radically in kind from episodic memory (memory for experienced events) (Michaelian, 2015), suggesting that they conflate a requirement for episodicity (Perrin & Rousset, 2014) with a requirement for mnemicity (Michaelian & Sutton, 2017). Even if autonoesis or epistemic relevance turns out to be a requirement for mnemicity, however, there would appear to be nothing that would prevent an advocate of a given version of the causal theory (or a given postcausal theory; see Section 6) from adding an appropriate condition to his theory. Hybrid theories will therefore be set aside in what follows.

5  Distributed and Procedural Causal Theories Though it accepts the sufficiency of appropriate causation, there is a sense in which the distributed causal theory departs more radically from the classical causal theory than do hybrid theories. Hybrid theorists posit conditions on remembering in addition to the appropriate causation condition. Distributed causal theorists take a different tack, modifying the concept of a memory trace in such a way that the appropriate causation condition can arguably no longer be understood as requiring transmission of content from experience to retrieval. We say “arguably,” for distributed causal theorists have not always been clear about whether they deny that appropriate causation involves transmission of content. Indeed, the literature contains no detailed articulation of the distributed causal theory. Sutton (1998, 2010) has

Beyond the Causal Theory?  21 provided a detailed account of the distributed conception of traces but has said little about how this conception of traces might be combined with the causal theory. Bernecker (2010) and Michaelian (2011), meanwhile, have developed detailed versions of the causal theory that endorse distributed traces in principle but have said little about their own distributed conceptions of traces. Notwithstanding this gap in the literature, it is uncontroversial that the inspiration for the distributed causal theory comes primarily from debates and developments regarding the nature of mental representation more generally. Much as a general view of mental representation in terms of structural analogy influenced Martin and Deutscher’s account of traces as structural analogues of experience, proponents of distributed conceptions of traces have been influenced by connectionist, dynamicist, and distributed views of mental representation. The traditional conception of traces involves fixed, explicit contents carried by distinct local vehicles. The vehicles in question might be distributed in the sense that they are complex entities the parts of which are stored in different locations, but they are local in the sense that each memory content is carried by a distinct vehicle. Proponents of distributed conceptions challenge this matrix of ideas, arguing that we should give up at least some of the features of the traditional conception. Sutton’s account of distributed traces comes closest to a full-blown rejection of the traditional conception: memories, he argues, “are blended, not laid down independently once and for all, and are reconstructed rather than reproduced” (1998: 2). On this account, a subject’s memory is a network in which various items of information are connected as a function of the frequency with which they cooccur in his experience. Each experience activates a certain pattern in the network, but the patterns overlap in a way that precludes distinct contents or vehicles. If this view is right, we may be able to refer to memory traces in a loose sense, since a specific experience will result in a specific modification of connections in the network, but these are traces of a sort that require us to reject the two key commitments of (neo)classical causal theories (identified in Section 2): there are no traces in the sense of distinct vehicles carrying distinct contents. Due to the gap in the literature noted earlier, it remains unclear how, in view of the fact that they reject these commitments, distributed causal theorists would have us understand the nature of the causal connection that they take to hold between retrieved memories and experiences, and there is a pressing need for further work on this question. Some distributed causal theorists have been less specific about the nature of memory traces but have tried to reconcile a distributed conception of traces with the appropriate causation condition. These authors reject the (neo) classical assumption that remembering involves the transmission of content from experience to retrieval, instead maintaining that content is reconstructed at the time of retrieval. To say that remembering is reconstructive, rather than reproductive, is to say that the content of a retrieved representation is,

22  Kourken Michaelian and Sarah K. Robins at least in part, produced at the time of retrieval, rather than transmitted from the corresponding experience. There is a long-standing consensus in the empirical literature that remembering is reconstructive in this sense (see, e.g., Schacter & Addis, 2007; Schacter et al., 2012). One possible reaction to the reconstructive character of remembering would be to continue to understand traces as distinct entities but to hold that their content is implicit in the sense that it needs to be activated or made explicit at the time of retrieval (see Vosgerau, 2010). Another possible reaction is provided by the procedural causal theory developed by Perrin (this volume). The procedural causal theory explicitly denies that remembering involves the transmission of content. Perrin retains a generic version of the core claim of the causal theory—that an appropriate causal connection is both necessary and along with other suitable conditions, sufficient for memory—but understands it in a radically different manner than do (neo)classical causal theorists. Whereas (neo)classical causal theorists understand causal connection in terms of the transmission of content via memory traces, procedural causal theorists take the reconstructive character of remembering to undermine this understanding of causation in memory. Inspired by older attributionalist approaches in psychology (e.g., Kolers & Roediger, 1984; Jacoby &Whitehouse, 1989; Whittlesea, 1997), Perrin proposes an alternative understanding of the nature of causation in memory. The key idea is that, rather than the content of the retrieved representation being causally related to the content of the corresponding experience, it is the process that produces the retrieved representation that is causally related to the process that produced the corresponding experience. Adopting a view of perception as itself being a constructive process, Perrin’s suggestion is that the constructive process of perceiving may bear certain similarities to the reconstructive process of remembering and thus give rise to a degree of fluency in the latter—it is in general easier to reconstruct a scene that one has previously constructed—despite the fact that no content is transmitted. The procedural causal theory may succeed in providing a description of a kind of causal connection that can obtain between experience and retrieval despite the fact that no content is transmitted from the former to the latter. But it does not yet provide a description of what it is for such a causal connection to be appropriate. While this is a potential problem, perhaps a more pressing question for both procedural and distributed causal theorists is whether they mean to retain the (neo)classical assumption that remembering is incompatible with the generation of new content between experience and retrieval. In one sense, of course, distributed and procedural causal theories necessarily acknowledge that remembering involves the generation of content, since they claim that content from previous experience is retained at best only implicitly, which implies that content must be “regenerated” at the time of retrieval. But this is just to say that they deny what might be called “transmissionism,” the view that (explicit) content is stored between

Beyond the Causal Theory?  23 experience and retrieval. In another sense—and this is the sense that matters here—they may deny that remembering involves the generation of content, since it is open to them to deny that “regenerated” retrieved content may include information going beyond that of the experience. That is, it is open to them to accept preservationism, the view that a retrieved representation may not include content not included in the original experience. More conservative versions of the theories will accept preservationism, but the basic distributed and procedural causal theories can be conjoined with a range of views on the generation of content. The more extreme the views become, the more likely they are to reject the core commitments of the causal theory. The most conservative view available is that the content of the retrieved representation is identical to the content of the experiential representation. This extreme form of preservationism is incompatible even with the occurrence of forgetting and is not to be taken seriously. An intermediate view is that the content of the retrieved representation must be contained in or in some sense implied by the content of the experiential representation. This more moderate form of preservationism is compatible with the occurrence of forgetting but not with the generation of new content between experience and retrieval and is explicitly endorsed by some (e.g., Bernecker, 2008, 2010; Cheng & Werning, 2016) and implicitly assumed by many others. While it is always possible in principle to hold on to preservationism by enriching the content of experience (McCarroll, 2017), we will see later that there is a real tension between even the moderate form of preservationism and the reconstructive character of remembering, which suggests a form of generationism according to which the content of the retrieved representation may indeed include information not included in the content of the experiential representation.8

6  Postcausal Theories As we saw in Section 4, the sufficiency of appropriate causation is challenged by hybrid theories on phenomenological or epistemic grounds. A different sort of challenge to the sufficiency of appropriate causation arises due to the reconstructive character of remembering, i.e., due to the fact that the content of retrieved representations is, at least in part, produced at the time of retrieval, rather than derived from the content of the corresponding experience. Reconstruction, in fact, challenges not only the sufficiency of appropriate causation but also its necessity and has therefore led to the emergence of theories that may be characterized as postcausal, in the sense that they claim that a causal connection—“appropriate” or otherwise—is not necessary for memory, even while recognizably descending from the causal theory. Postcausal theories in effect treat memory as a synchronic rather than a diachronic capacity, in the sense that they see the occurrence of remembering as depending on what happens when the subject (apparently) remembers, rather than on whether there is a suitable relationship between

24  Kourken Michaelian and Sarah K. Robins the subject’s retrieved representation and his experiential representation; thus, unlike hybrid theories, they move decisively beyond the causal theory. One intriguing postcausal theory is the functionalist theory, which Fernández (this volume) offers as an alternative to both the causal theory and the narrative theory of memory (e.g., Schechtman, 1994; Goldie, 2012; Brockmeier, 2015). Fernández argues that the causal theory is both too strict, in that it is incompatible with the generation of new content during reconstructive remembering, and too weak, in that it ignores the tendency (emphasized by hybrid theories) for memory to give rise to belief. He likewise argues that the narrative theory—which, emphasizing reconstruction, views remembering as an imaginative process in which the subject draws on stored information deriving from his experiences, along with information deriving from other sources, to create narratives about his past—is both too strict, in that it does not acknowledge the possibility of memories that are not embedded in narratives, and too weak, in that it does not acknowledge any role at all for the causal history of memories. The alternative that Fernández offers is a theory on which a mental state qualifies as a memory just in case it plays the functional role that memories typically play, where this role is a matter, first, of tending to cause belief and second, of tending to be caused by past experience. What is most important about the functionalist theory, in the present context, is the second of these claims: while the functionalist theory requires, in order for a mental state to qualify as a memory, that it tend to be caused by the subject’s past experience of the remembered event, it does not require that the mental state actually be caused by the experience. The functionalist theory thus rejects the core claim of the causal theory. In line with our previous discussion of the causal-epistemic theory, the second of the functionalist’s claims, regarding the link between memory and belief, may be understood as concerning episodicity, rather than mnemicity. If we therefore disregard this claim, Fernández’ functionalist theory and the simulation theories recently developed by a number of authors (Shanton & Goldman, 2010; De Brigard, 2014a; Michaelian, 2016) come to broadly similar conclusions about the nature of remembering. The path taken by the simulation theorist is, however, somewhat less direct, involving a close consideration of the role of traces in remembering. It might be thought, given the association between reconstruction and distributed/procedural theories, that local trace theories can avoid the challenge posed by reconstruction, but the causal theory cannot in fact be protected by retreating to the local conception. Even if, as noted previously, the distributed conception has in many cases been adopted only in a nominal sense, most philosophers of memory have been in principle convinced by the arguments in favor of the distributed conception. And even if some have not yet been convinced by the arguments and so deliberately continue to work with the local conception, they are nevertheless bound, given the weight of the evidence in its favor, to acknowledge the reconstructive character of remembering within the

Beyond the Causal Theory?  25 parameters of the local conception. The challenge must thus be faced by all causal theorists. Is the existence of an appropriate causal connection between the retrieved representation and the experiential representation sufficient for remembering, given the local conception of traces? Given reconstruction, the local trace theorist must acknowledge what we might refer to as “the fact of multiple experiences”: multiple experiences may contribute to the content of a single stored trace. He must also acknowledge what we might refer to as “the fact of multiple traces”: multiple traces may contribute to the content of a single retrieved representation. These facts together imply that, if a given retrieved representation is appropriately causally connected to a given experience, it may also be appropriately causally connected to other experiences. The existence of an appropriate causal connection thus does not suffice, given the local conception, to determine whether the subject is remembering a given event. Is the existence of an appropriate causal connection between the retrieved representation and the experiential representation sufficient for remembering, given the distributed conception of traces? Given the distributed conception, retrieval is a matter of activating certain ideas—nodes in a larger network of ideas—together. The tendency for certain ideas to be activated together is, however, not attributable to a unique event, since the relevant connection weights have inevitably been affected by multiple experiences (Robins, 2016b). Nor is there any guarantee that a given retrieved representation matches a unique experiential representation. It is, as noted in Section 5, not entirely clear how the notion of appropriate causation is to be understood by the distributed trace theorist. But however it is understood, it would appear that the distributed conception implies that, if a given retrieved representation is appropriately causally connected to a given experience, it may also be appropriately causally connected to other experiences. The existence of an appropriate causal connection thus does not suffice, given the distributed conception, to determine whether the subject is remembering a given event. If appropriate causation were merely to fail to be sufficient for memory, it would be possible to save the causal theory by means of the incorporation of an additional condition, in the manner of the hybrid theories discussed in Section 3. But reconstruction appears to undermine not only the sufficiency of appropriate causation but also its necessity. Beginning with the local conception of traces, the fact of multiple experiences and the fact of multiple traces together imply that the content of a retrieved representation will typically not derive entirely from that of the relevant earlier experience. In some cases, a majority of the content may so derive. In other cases, however, only a minority of the content so derives. And in some cases, none of the content so derives. As long as some of the content derives from the experience, of course, a causal connection obtains, and it is intuitively plausible that there is a difference in kind between such cases and cases in

26  Kourken Michaelian and Sarah K. Robins which none of the content derives from the experience. On the basis of this intuition, Michaelian (2011a) has argued for a constructive causal theory, a theory that is like the causal theory in that it requires the transmission of content from experience to retrieval (accepting transmissionism) but unlike it in that it permits the generation of new content between experience and retrieval (rejecting preservationism and accepting generationism). Robins (2016a), who likewise seeks to retain the causal theory while acknowledging the reconstructive character of remembering, has defended a broadly similar approach. While constructive causal approaches provide an appealing means of reconciling the causal theory with reconstruction, however, the empirical research on reconstruction suggests, as Michaelian has pointed out in subsequent work (2016c) that the very same cognitive process may be at work both in cases in which some content is transmitted from the experience and in cases in which no content is transmitted. This in turn implies that, given the local conception of traces, causal connection does not mark the difference between genuine and merely apparent memory. Turning to the distributed conception of traces, we find a similar implication. Due to the blended nature of distributed storage, not all of the ideas that compose a given retrieved memory are activated because of the relevant earlier experience. In some cases, a majority of the ideas may be activated due to the earlier experience. In some cases, however, only a minority are. And in some cases, none are. There is, however, no reason to suppose that there is a difference in kind between cases in which none of the ideas are activated because of the relevant earlier experience and cases in which at least some are—in cases of both sorts, the same process may be at work. This implies that, given the distributed conception of traces, causal connection does not mark the difference between genuine and merely apparent memory. A distributed or procedural causal theorist might object that this argument presupposes transmissionism, which distributed and procedural theories reject. The idea would be that a distributed/procedural theory can reject transmission but accept either preservationism or a moderate form of generationism according to which there must be some degree of overlap between the content of the retrieved representation and the content of the earlier representation in order for genuine remembering to occur. The distributed/procedural theorist can then maintain that genuine remembering occurs only if, first, there is such overlap, and second, this overlap is due to the presence of an appropriate causal connection, understood in nontransmissionist terms. While this is an interesting objection, it assumes that a convincing nontransmissionist account of appropriate causation can be formulated, and this remains to be done. The argument given previously does not presuppose transmissionism but does bet that there will turn out to be no interesting difference between cases in which the activation of at least some of the relevant ideas is due to the earlier experience and cases in which the activation of none of them is due to the earlier experience.

Beyond the Causal Theory?  27 Reacting to these difficulties for the constructive causal theory, Michaelian has proposed a simulation theory of remembering, the key idea of which is that, contrary to the basic assumption of the causal theorist, there is no difference between remembering the past and imagining it, in which case memory does not presuppose a causal connection—to remember just is to imagine the past. De Brigard (2014a), though he is less explicit about his stance on the necessity of causal connection, has developed a similar view, treating episodic memory as a form of episodic hypothetical thought, or thought about possible events. And Shanton and Goldman (2010) have likewise argued that remembering is to be understood in simulational terms, linking remembering to theory of mind. Evidence for the simulation theory comes from research on episodic memory as a form of mental time travel analogous to episodic future thought (Suddendorf & Corballis, 1997). A large body of research now supports the view that the process of remembering the past is executed by the same cognitive system as the process of imagining the future and indeed, that imagining the future is the primary function of the system in question (see Michaelian, Klein, & Szpunar, 2016). Both imagining the future and remembering the past draw on stored content originating in experience of past events. Imagining a future event does not, of course, draw on content originating in experience of the particular event imagined. By the same token, the mental time travel framework suggests that remembering a past event does not necessarily draw on content originating in experience of the particular event remembered. From a broadly naturalistic point of view, this, in turn, suggests that remembering does not presuppose a causal connection. If remembering does not presuppose a causal connection, a fortiori it does not presuppose an appropriate causal connection. But this does not mean that the process of imagining the past cannot itself be appropriate or inappropriate: if the subject imagines the past in the wrong way, the representation he produces may fail to qualify as a memory, even if it should happen to be accurate. Not only simulation theorists but also constructive causal theorists, who acknowledge that memories may be in part the product of imagination, even if they deny that they can be wholly the product of imagination, thus must provide an account of the appropriateness of the process of imagining the past. Michaelian’s version of the constructive causal theory therefore incorporates a reliability condition—a condition requiring that the system function in such a way that it tends to produce mostly accurate representations—and this condition is inherited by his version of the simulation theory, which, strictly speaking, says that to remember a past event is to imagine it in a reliable manner. The reliability condition enables the simulation theory to distinguish remembering, understood as imagining the past, from confabulation and other ways of merely imagining the past. It remains to be seen whether further conditions must be added to the simulation theory to enable it to distinguish between

Figure 1.1  Relationships among causal and postcausal theories. (Neo)classical causal theories (Martin & Deutscher, 1966; Bernecker, 2008, 2010; Cheng & Werning, 2016) maintain that appropriate causation is both necessary and sufficient for remembering and endorse both transmissionism and generationism. Distributed and procedural causal theories (Sutton, 1998; Perrin, this volume) agree with (neo)classical causal theories that appropriate causation is both necessary and sufficient for remembering, but their distributed conception of traces leads them to reject transmissionism. Constructive causal theories (Michaelian, 2011; Robins, 2016b) likewise agree that appropriate causation is both necessary and sufficient for remembering, but their constructive view of remembering leads them to reject preservationism; the constructive view is compatible with both local and distributed conceptions of traces. Hybrid causal theories, including epistemic-causal theories (Debus, 2010) and autonoetic-causal theories (Dokic, 2014; Klein, 2015), depart to some extent from the causal tradition by maintaining that appropriate causation is necessary but not sufficient for remembering; they do not take an explicit stand with respect to transmissionism or preservationism, and the feasibility of the various views in this space remains to be explored. Postcausal theories, including the functionalist theory (Fernández, this volume) and the simulation theory (Michaelian, 2016c; cf. De Brigard, 2014a and Shanton & Goldman, 2010), make a decisive break with the causal tradition by maintaining that appropriate causation is neither necessary nor sufficient for remembering. The functionalist theory does not take an explicit stand for or against transmissionism or preservationism. The simulation theory explicitly rejects preservationism but like the constructive causal theory, might in principle be combined with either a local or a distributed conception of traces and hence might or might not reject transmissionism. Other theories: In principle, theories that maintain that appropriate causation is sufficient but not necessary for remembering might be described, but the motivation for such theories is unclear, and none have so far been proposed.

Beyond the Causal Theory?  29 remembering and relearning and between remembering and nonmemorial retention.

7 Conclusions Fifty years after Martin and Deutscher, causal theories of various sorts— neoclassical, hybrid, and distributed/procedural—continue to dominate the landscape in the philosophy of memory (see Figure 1.1). Clearly, the field as a whole has yet to move decisively beyond the causal theory. The emergence of postcausal theories, however, hints at increased awareness of the tension between the causal theory and the reconstructive character of remembering. Of course, while postcausal theories may be better suited than causal theories to accommodating the reconstructive character of remembering, they will themselves inevitably face objections. The functionalist theory is too new for objections to it to have emerged. But objections to the simulation theory— focusing on the “continuist” view of past- and future-oriented mental time travel that it presupposes (Perrin, 2016; Michaelian, 2016a; Perrin & Michaelian, 2017) and on its ability to distinguish between remembering and misremembering or confabulating (Robins, 2016b; Michaelian, 2016b; Robins, forthcoming)—have already begun to be voiced. Time will tell whether postcausal theorists are able to address these and other objections and convince significant numbers of philosophers of memory to move beyond the causal theory.

Notes 1 Thanks for feedback to audiences at the Université Grenoble Alpes, the University of Otago, and Issues in Philosophy of Memory (Cologne, 2017), and thanks for written comments to Steven James and Denis Perrin. 2 For background on the empiricist theory, see Bernecker, 2008. Bernecker also discusses the epistemic theory, which has likewise been eclipsed in popularity by the causal theory; the epistemic theory is not to be confused with the hybrid causal-epistemic theory reviewed in Section 5. Martin and Deutscher were not the first to state the causal theory, but they offer the canonical statement of the theory, and so we do not discuss earlier formulations here. 3 Throughout, “retrieved representation” refers to the representation entertained by the subject at the time of (apparent) remembering, regardless of whether the process responsible for the production of the representation in question in fact involved retrieval of information and regardless of whether remembering in general is understood as involving retrieval. 4 For an extended discussion of nonmemorial retention, see Robins, 2016b. 5 For a detailed argument to this effect, see Robins, 2016a. 6 Cf. Deutscher’s (2017) comparison of Bernecker’s to Martin and Deutscher’s theory, which provides a more detailed discussion of the points of similarity between the two. 7 Some have argued for a return to epistemic (e.g., Adams, 2011) or even empiricist theories (Byrne, 2010) of remembering, but such arguments are infrequent. 8 “Preservationism” sometimes refers to the view that memory preserves justification, as opposed to the view that it preserves content (see Lackey, 2005; Fernández, 2016; Frise, forthcoming). We are concerned here neither with this form of preservationism nor with the corresponding form of generationism.

30  Kourken Michaelian and Sarah K. Robins

References Adams, F. (2011). Husker du? Philosophical Studies, 153(1), 81–94. Bernecker, S. (2008). The metaphysics of memory. New York: Springer. Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press. Brockmeier, J. (2015). Beyond the archive: Memory, narrative, and the autobiogra­ phical process. Oxford: Oxford University Press. Byrne, A. (2010). Recollection, perception, imagination. Philosophical Studies, 148(1), 15–26. Cheng, S., & Werning, M. (2016). What is episodic memory if it is a natural kind? Synthese, 193(5), 1345–1385. Cheng, S., Werning, M., & Suddendorf, T. (2016). Dissociating memory traces and scenario construction in mental time travel. Neuroscience & Biobehavioral Reviews, 60, 82–89. De Brigard, F. (2014a). Is memory for remembering? Recollection as a form of episodic hypothetical thinking. Synthese, 191(2), 155–185. De Brigard, F. (2014b). The nature of memory traces. Philosophy Compass, 9(6), 402–414. Debus, D. (2008). Experiencing the past: A relational account of recollective memory. Dialectica, 62(4), 405–432. Debus, D. (2010). Accounting for epistemic relevance: A new problem for the causal theory of memory. American Philosophical Quarterly, 47(1), 17–29. Debus, D. (2014). “Mental time travel”: Remembering the past, imagining the future, and the particularity of events. Review of Philosophy and Psychology, 5(3), 333–350. Debus, D. (2017). Memory causation. In S. Bernecker & K. Michaelian (Eds.), Routledge handbook of philosophy of memory(pp. 63–75). London: Routledge. Deutscher, M. (2017). The trace as structural analogue. In The Routledge encyclopedia of philosophy. Taylor and Francis. Retrieved from www.rep.routledge.com/ articles/thematic/memory/v-2/sections/the-trace-as-structural-analogue Dokic, J. (2014). Feeling the past: A two-tiered account of episodic memory. Review of Philosophy and Psychology, 5(3), 413–426. Draaisma, D. (2000). Metaphors of memory: A history of ideas about the mind. Cambridge: Cambridge University Press. Fernández, J. (2016). Epistemic generation in memory. Philosophy and Phenomeno­ logical Research, 92(3), 620–644. Fernández, J. (Forthcoming). The ownership of memories. In M. García-Carpintero & M. Guillot (Eds.), The sense of mineness. Oxford: Oxford University Press. Frise, M. (Forthcoming). Preservationism in the epistemology of memory. The Philosophical Quarterly. Goldie, P. (2012). The mess inside: Narrative, emotion, and the mind. Oxford: Oxford University Press. Hamilton, A. (2003). “Scottish commonsense” about memory. Australasian Journal of Philosophy, 81(2), 229–245. Holland, R. F. (1954). The empiricist theory of memory. Mind, 63(252), 464–486. Hopkins, R. (2014). Episodic memory as representing the past to oneself. Review of Philosophy and Psychology, 5(3), 313–331.

Beyond the Causal Theory?  31 Jacoby, L. L., & Whitehouse, K. (1989). An illusion of memory: False recognition influenced by unconscious perception. Journal of Experimental Psychology: General, 118(2), 126–135. James, S. (Forthcoming). Epistemic and non-epistemic theories of remembering. Pacific Philosophical Quarterly. Klein, S. B. (2014). Autonoesis and belief in a personal past: An evolutionary theory of episodic memory indices. Review of Philosophy and Psychology, 5(3), 427–447. Klein, S. B. (2015). What memory is. Wiley Interdisciplinary Reviews: Cognitive Science, 6(1), 1–38. Klein, S. B., & Nichols, S. (2012). Memory and the sense of personal identity. Mind, 121(483), 677–702. Kolers, P. A., & Roediger, H. L. (1984). Procedures of mind. Journal of Verbal Learning and Verbal Behavior, 23(4), 425–449. Lackey, J. (2005). Memory as a generative epistemic source. Philosophy and Phenomenological Research, 70(3), 636–658. Mahr, J., & Csibra, G. (Forthcoming). Why do we remember? The communicative function of episodic memory. Behavioral and Brain Sciences. Malcolm, N. (1963). Knowledge and certainty. Englewood Cliffs, NJ.: Prentice-Hall. Malcolm, N. (1977). Memory and mind. Ithaca, NY: Cornell University Press. Martin, C. B., & Deutscher, M. (1966). Remembering. The Philosophical Review, 75(2), 161–196. Martin, M. G. (2001). Out of the past: Episodic recall as retained acquaintance. In T. McCormack & C. Hoerl (Eds.), Time and memory: Issues in philosophy and psychology (pp. 257–284). Oxford: Oxford University Press. McCarroll, C. J. (2017). Looking the past in the eye: Distortion in memory and the costs and benefits of recalling from an observer perspective. Consciousness and Cognition, 49, 322–332. Michaelian, K. (2011a). Generative memory. Philosophical Psychology, 24(3), 323–342. Michaelian, K. (2011b). Is memory a natural kind? Memory Studies, 4(2), 170–189. Michaelian, K. (2015). Opening the doors of memory: Is declarative memory a natural kind? Wiley Interdisciplinary Reviews: Cognitive Science, 6(6), 475–482. Michaelian, K. (2016a). Against discontinuism: Mental time travel and our knowledge of past and future events. In K. Michaelian, S. B. Klein, & K. K. Szpunar (Eds.), Seeing the future: Theoretical perspectives on future-oriented mental time travel (pp. 62–92). Oxford: Oxford University Press. Michaelian, K. (2016b). Confabulating, misremembering, relearning: The simulation theory of memory and unsuccessful remembering. Frontiers in Psychology, 7, 1857. Michaelian, K. (2016c). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Michaelian, K., Klein, S. B., & Szpunar, K. K. (2016). The past, the present, and the future of future-oriented mental time travel. In K. Michaelian, S. B. Klein, & K. K. Szpunar (Eds.), Seeing the future: Theoretical perspectives on future-oriented mental time travel (pp. 1–18). Oxford: Oxford University Press. Michaelian, K., & Sutton, J. (2017). Memory. In E. N. Zalta (Ed.), Stanford encyclopedia of philosophy. Retrieved fromhttps://plato.stanford.edu/archives/ sum2017/entries/memory/

32  Kourken Michaelian and Sarah K. Robins Perrin, D. (2016). Asymetries in subjective time. In K. Michaelian, S. B. Klein, & K. K. Szpunar (Eds.), Seeing the future: Theoretical perspectives on future-oriented mental time travel (pp. 38–61). Oxford: Oxford University Press. Perrin, D., & Michaelian, K. (2017). Memory as mental time travel. In S. Bernecker & K. Michaelian (Eds.), The Routledge handbook of philosophy of memory (pp. 228–239). London: Routledge. Perrin, D., & Rousset, S. (2014). The episodicity of memory. Review of Philosophy and Psychology, 5(3), 291–312. Robins, S. K. (2016a). Misremembering. Philosophical Psychology, 29(3), 432–447. Robins, S. K. (2016b). Representing the past: Memory traces and the causal theory of memory. Philosophical Studies, 173(11), 2993–3013. Robins, S. K. (Forthcoming). Confabulation and constructive memory, Synthese. Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering the past and imagining the future. Philosophical Transactions of the Royal Society B: Biological Sciences, 362(1481), 773–786. Schacter, D. L., Addis, D. R., Hassabis, D., Martin, V. C., Spreng, R. N., & Szpunar, K. K. (2012). The future of memory: Remembering, imagining, and the brain. Neuron, 76(4), 677–694. Schechtman, M. (1994). The truth about memory. Philosophical Psychology, 7, 3–18. Shanton, K., & Goldman, A. (2010). Simulation theory. Wiley Interdisciplinary Reviews: Cognitive Science, 1(4), 527–538. Shepard, R. N., & Chipman, S. (1970). Second-order isomorphism of internal representations: Shapes of states. Cognitive Psychology, 1, 1–17. Squires, R. (1969). Memory unchained. The Philosophical Review, 78(2), 178–196. Suddendorf, T., & Corballis, M. C. (1997). Mental time travel and the evolution of the human mind. Genetic, Social, and General Psychology Monographs, 123(2), 133–167. Sutton, J. (1998). Philosophy and memory traces: Descartes to connectionism. Cambridge: Cambridge University Press. Sutton, J. (2010). Memory. In E. N. Zalta (Eds.), The Stanford encyclopedia of philosophy. Retrieved from https://plato.stanford.edu/archives/spr2010/entries/ memory/ Tulving, E. (1985). Memory and consciousness. Canadian Psychology/Psychologie canadienne, 26(1), 1–12. Vosgerau, G. (2010). Memory and content. Consciousness and Cognition, 19(3), 838–846. Whittlesea, B. (1997). Production, evaluation, and preservation of experiences: Constructive processing in remembering and performance tasks. The Psychology of Learning and Motivation, 37, 211–264. Wittgenstein, L. (1953). Philosophical investigations (G. E. M. Anscombe, Ed.). Oxford: Blackwell. Zemach, E. M. (1983). Memory: What it is, and what it cannot possibly be. Philosophy and Phenomenological Research, 44(1), 31–44.

2 A Case for Procedural Causality in Episodic Recollection1 Denis PerrinA Case for Procedural Causality

Denis Perrin

1 The Causal Condition Under Scientific Threat: The Soundness Problem This chapter aims at making a case for what I propose to call “procedural causality” that is operative in episodic memory. To do so, I rely on the widely accepted notion of episodic memory understood as a mental occurrence of remembering. Indeed, among the many forms memory can take, one should distinguish declarative memory, i.e., the form of memory that provides explicit representations of personal or general facts. This mnemonic form is usually distinguished from procedural memory in the wide sense, which includes (among other things) motor and cognitive skills and does not involve any representation. Declarative memory, in turn, can come in either of two ways. On the one hand, semantic memory is the memory for facts conceptually represented, and on the other hand, episodic memory is the quasi-experiential representation of past personal episodes. Now, such episodes can be the objects of a semantic memory as well as of an episodic memory. So, a question is what makes a mental occurrence an episodic memory, rather than a semantic memory, of a personal episode experienced in the past. One can refine this question in three main subquestions (Fernandez, 2013):

• The metaphysical question: for any mental occurrence o in a subject • •

S’s mind, what does it take for o to be an episodic memory of a past perceived event e? Which actual relation between o and e is required? The intentional question: for any mental occurrence o in S’s mind, what o should be about, i.e., what should o’s content be, for o to qualify as an episodic memory of e? The phenomenological question: for any mental occurrence o in S’s mind, what should it be like for S to entertain o for o to qualify as an episodic memory of e?

Arguably, these three questions are not independent. I suggest considering the phenomenological question as primary. First, indeed, in episodic memory

34  Denis Perrin phenomenology plays a substantial role regarding the informational content. It “is not a mere phenomenological embellishment” (Dokic, 2014) of the intentional content.2 As S is episodically remembering a past personal event e, S’s phenomenal representation is of e as of a past personally experienced event. Thus, the way e is presented in episodic remembering is informative in and by itself, or: it is part and parcel of the intentional content. Second, through their informativeness phenomenological features also present why an o should be considered as an episodic memory from the metaphysical point of view, or: which metaphysical conditions o claims to satisfy so that it should be considered as an episodic memory. Let me specify the phenomenological features and the metaphysical conditions they posit. As S is episodically remembering an event e, S is conscious of e as:

• Pastness condition: having occurred in the past • Actuality condition: having actually occurred (in the past) • Personal experience condition: having been personally experienced by S •

(in the past) Causality condition: being mentally re-experienced by S because S has personally experienced it (in the past).

Among these features, the fourth is of prominent importance. Since Martin and Deutscher’s paper (1966), it has been common stand to posit a causal necessary condition among the metaphysical requirements of episodic remembering. To put it in general terms: Causal connection condition: For an occurrence o that represents an event e to qualify as an episodic memory of e, o must derive causally (in an appropriate way) from e. Note that in line with my previous remarks, this condition falls in agreement with what the phenomenology of episodic remembering requires from any (apparent) episodic memory and what it builds into its intentional content. Indeed, the causal phenomenological feature:

• represents e as mentally re-experienced by S because S has personally • •

experienced it in the past—phenomenological aspect; thereby informs S that s/he is mentally re-experiencing e because s/he has personally experienced it in the past—intentional aspect; thereby requires from o to derive causally (in an appropriate way) from S’s past experience of e for o to qualify as an episodic memory of e—metaphysical aspect.

Now, drawing on empirical results provided by the current constructivist approach to remembering (Schacter & Addis, 2007, 2012), some researchers have recently cast doubt on the necessity of including a causal condition

A Case for Procedural Causality  35 among the metaphysical features of episodic memory. Robins (2016) pinpoints a conflict between the causal theory and the currently widely endorsed view of mnemonic traces as a distributed and superpositional storage of memories, to the effect that the latter blurs the singular causal path from e to o that is claimed by the causal theory to make o a memory of e specifically. She speculates that the causal theory could not reflect our best analysis of what remembering requires. Michaelian (2016) bites the bullet and proposes to go beyond the causal theory. He argues for a simulation theory that posits no necessary causal connection for remembering. Instead, the right condition would be a properly functioning constructive episodic system, that is a system that tends to produce accurate representations of the past, whatever the causal origin of the bits of experience through which it constructs episodic simulations. If these criticisms are right in their rejection of the causal condition—and I think they are as far as a certain notion of causality is concerned—then the fourth phenomenological feature is in danger of saying something wrong about, and requiring something irrelevant from, the mental occurrence to which it belongs. Indeed, as I have suggested, episodic memory has its causal origin on its sleeves. As Dokic (2001: 228) puts it: “episodic memory gives me a reason to believe that it comes directly from my own past experience, because the fact that it does so is presented in the memory experience itself.” Again: “The thought is that the causal chain that originates in a certain event and terminates in a certain memory experience . . . is . . . what that memory experience is representing” (Fernandez, 2008b, 348). So, the fourth phenomenological feature together with its contentual and metaphysical implications appears to stand in contradiction with what scientific results reveal about episodic memory and causality. At this stage, one faces a dilemma. Either one takes seriously into account what science data suggest, but then one is doomed to consider the phenomenological features are unsound. This looks like an undesirable move, because it raises problematic questions, as (for instance) why evolution would have built a systematic lie in the very heart of episodic memory.3 So, it seems preferable to endorse a soundness constraint for any satisfying theory of episodic memory: Soundness constraint: the phenomenological features of episodic memory (the causal one, in particular) are sound to the effect that the elements they build into the intentional content of such mental occurrences as metaphysical requirements are correct.4 In other words, while the actual metaphysical features of an apparent episodic memory are not transparent to the subject, to be sure, the features it must possess for it to qualify as an actual episodic memory are presented by its phenomenology. But insisting on the soundness constraint, one is led to contradict—this is the second horn of the dilemma—what psychology has repeatedly shown over the past decades, namely: there is no causal

36  Denis Perrin connection as the one required by causalism. So, what? The way out of this dilemma favored in this chapter is to acknowledge the aforementioned criticisms of causalism, while suggesting that the actual causal connection is not the one criticized by Robins and Michaelian. If the causal condition can be rebutted, this is due to the commitment to an inadequate conception of causality. Once the right concept is promoted, the way out of the dilemma is smooth and the fourth phenomenological feature is sound. While I defend the causal theory, I thus do not propose to come back to one of its existing versions, I rather suggest there is a (still widely unexplored) causal dimension of episodic memory, namely the procedural one. In my view, this dimension puts to rest the mentioned criticisms and is the one the phenomenological feature is about. To make the case for this proceduralist account of causality in episodic memory, I proceed in two stages. In Section 2, I distinguish componential and procedural notions of causality and go on to argue that, provided a revised view of what one should expect from causality in episodic memory, a procedural causal connection avoids the difficulties faced by the componential conception. In Section 3, I account for the phenomenology of episodic memory in terms of a feeling of pastness. I show that this account neatly falls in line with the proceduralist version of causalism promoted in Section 2.

2  Promoting Procedural Causality On Martin and Deutscher’s causal theory of memory (1966), the causal connection condition must hold between a current mental occurrence o and a past experience of an event e. In their terms: S’s experience of e has to have been “operative in producing a state or successive states in him finally operative in producing his representation” (166). More specifically four criteria define the causal connection necessary for remembering, namely: (1) The operative-in criterion: the experience of e must be the cause of the state stored by the system; (2) The operative in-the-circumstances-of-remembering criterion: the state stored must be the cause of the remembering state;5 (3) The analogue-memory-trace criterion: the information retained must be a representational analogue of the experienced e.6 Note that this view of memory trace has been criticized. A much more empirically plausible (connectionist) view is that traces are distributed patterns of activation on a neural network (Sutton, 1998; Bernecker, 2010; Michaelian, 2011; De Brigard, 2014); (4) The singular-causal-signature criterion: there must be a unique causal chain running from the past experience of e to the current remembering state. Though not quite explicit, this criterion belongs to the causal theory, as emphasized by Robins (2016: 16–7).

A Case for Procedural Causality  37 Are all of these four criteria legitimate? In this section, drawing on the notion of a procedural causality, I argue that while criteria (1) and (2) should be endorsed, criteria (3) and (4) are to be rejected. Justifying this move requires introducing a new notion of causality for remembering. 2.1  The Componential vs. Procedural Causality Distinction My point is twofold here. First, I hold that different causal relations at play in episodic remembering involve different types of relata. On the one hand, the components of the perceptual and memory representations, and on the other hand, the processes that make up these representations. I will speak of componential causality about the causal relations that involve the first type of relata, and of procedural causality about the causal relations that involve the second type of relata. Second, I claim that the latter type is irreducible to the former. To make this point, it will be useful to draw on an analogy with a jigsaw puzzle. Let’s imagine you have two copies of one and the same puzzle. The pieces are exactly the same in number, forms, and pictures in each box. Obviously, however alike the pieces are, as one makes one the two puzzles one does not use the pieces of the other. In other terms, the two series of construction operations apply respectively to causally unrelated (though similar) bits of representation. But at the same time—this is crucial point to my argument—these construction operations themselves can be causally related. For instance, if you make the two puzzles one after the other, you will possibly perform the second time better than the first time. And should you repeat the operations further, the enhancement will probably get ever clearer. So arguably, an earlier series of construction operations can get a later series enhanced, while the manipulated sets of pieces are distinct and causally unrelated. Now, as repeatedly emphasized by psychological literature about perception (Neisser, 1967; O’Reagan & Noë, 2001), as one perceptually experiences a certain event, one does not merely receive sensory information from the latter in a passive way. Instead one is actively engaged in a perceptual exploration and construction of the scene, which eventually results in the perceptual experience. In other terms, some constructive processes underlie any perceptual experience. Moreover, just as causally unrelated sets of puzzle pieces can be gathered into distinct copies of the same picture by causally connected operations, causally connected constructive processes can be applied to causally unrelated bits of sensory information and bring about highly similar sensory scenes. Now the prevailing constructive episodic simulation hypothesis (Schacter & Addis, 2007; Schacter et al., 2012) strongly emphasizes the role played in episodic memory by constructive operations of piecing together the elements of an episode. Hence my proposal is to say that the constructive processes at play respectively in the remembered and the perceptual scenes can be causally connected, even though the bits of experience pieced together in the present are causally unrelated to the past episode that is being remembered.

38  Denis Perrin The suggested distinction thus permits to envision causal relations that, on the one hand, are operative at the procedural level of construction rather than at the level of the experiential componential blocks, and on the other hand, whose operativity does not presuppose the identity of the blocks to which the constructive processes are applied at each stage. My proposal is precisely to posit in episodic memory a causal relation between the processes of the construction of the initial experience of an event and the construction of the remembering state—typically, the former cause a relatively high procedural fluency of the latter—without positing any necessary causal relation between the elements manipulated by these operations. A crucial consequence of this move is that the causalist theory of memory turns out to possibly come in two different ways depending on the type of causal relata considered. Correspondingly, one can flesh out the initial Causal connection condition in two ways:

• Componential Causal Condition: For a mental occurrence o that

represents an event e to qualify as an episodic memory of e, the elements composing the scene represented by o must derive causally (in an appropriate way) from the elements composing the initial experience of e.

• Componential Causal Theory of Memory (CCTM) • Causal criterion (3) is necessary. • Procedural Causal Condition: For a mental occurrence o that represents

an event e to qualify as an episodic memory of e, the processes whose o is the outcome must be causally determined (in an appropriate way) by the processes that constructed the initial experience of e.

• Procedural Causal Theory of Memory (PCTM) • Causal criterion (3) is not necessary. In what follows, I will argue that componential causality puts no necessity constraint on remembering, in line with recent anticausalist analyses (Robins, 2016; Michaelian, 2016). But against the conclusions promoted by such criticisms, I will also argue that this does not discard every causal condition and that procedural causality does put a necessity constraint on remembering. I will thus endorse PCTM. The view just sketched out receives strong theoretical and empirical support from an important trend in psychology, though neglected by philosophers of memory, namely the attributionalist view of remembering (Kolers & Roediger, 1984; Jacoby et al., 1989; Kelley & Jacoby, 1990; Whittlesea, 1997; Leboe-McGowan & Whittlesea, 2013). Let me be more specific about this framework. The attributionalist account of episodic memory has rarely received much, let alone positive attention from philosophers. Some squarely reject it as inadequate (Hoerl, 2001), while the ones who mention it in positive terms consider attributionalism as providing merely

A Case for Procedural Causality  39 one possible element of explanation among others (Fernandez, 2008a). My suggestion is that, on the contrary, attributionalism is still a promising view. Let’s introduce it. Its core contention is that the phenomenology of episodic recollection is the result of a subpersonal causal attribution of a current mental occurrence to a past experience. For instance, as I episodically remember playing football with my son yesterday:7 (i) Because I have already constructed the experience of such an episode, when I do it again as a mental simulation the way I do it displays specific procedural features. (ii) The easiness or the fluency (e.g.) of the simulation operations are subpersonal cues available to detection. The detection of such cues is inference-grounding, that is: detecting the cues leads the cognitive system to carry out an inference about the causal origin of the cues, thereby about the status of the simulated scene—this is the attribution. For instance, the system infers that it has already constructed such an experience, thus that the currently simulated scene of playing football with my son is a past experience. (iii) The personal counterpart of this attribution is the episodic phenome­ nology under which the scene presents itself to my mind. Obviously procedural causality is crucial to attributionalism. The subper­ sonal cues that ground the process of attribution are normally the causal effects of the past construction processes. Should they actually be so, one has to do with an actual episodic memory; should they not, one has to do with a merely apparent episodic memory. So attributionalism strongly suggests one endorse PCTM about episodic recollection. Then the question is: how this attributionalist account does comply with the aforementioned causality criteria? It complies well with the first two criteria. It indeed claims that adequate procedural information—i.e., the construction of the experience of the event e—is retained in some way (criterion (1)), and that this information is operative in the circumstances of remembering (criterion (2)), provided it must be involved in them to be possibly detected. So far so good, but attributionalism does not seem to meet so easily criteria (3) and (4). To show this is no real problem, I have to be more explicit about what one should expect from the causal connection. 2.2 What One Should (and Should Not) Expect From Causality in Episodic Memory Dismissing causality about episodic memory often relies on the claim that it is unable to meet certain expectations assumed to be legitimate. So, it seems prudent to be clear about what one should legitimately expect from causality in the first place. Maybe its apparent inability results from mistaken expectations rather than from any unreliability. So, let’s question what the

40  Denis Perrin legitimate expectations with respect to causality are as for its format, on the one hand, and its role in episodic recollection, on the other hand. As for its format, consider what connectionism, invoked by many philosophers of memory nowadays (Sutton, 1998; Bernecker, 2010; Michaelian, 2011; Robins, 2016, see also McClelland, 2000), teaches us about causality in memory. My suggestion is that connectionism promotes a more radical picture of causality than usually admitted. Philosophers usually consider that connectionism substitutes a new conception of representational mnemonic traces—henceforth distributed, superpositional, and constructed—for the old one.8 I propose to go further and see connectionism as discarding the very notion of a representational trace at the neural level. According to connectionism, each connection between two neural nodes has a weight. The only reality retention has at this neural level consists of this weight. Thus, the network at a certain stage of its dynamics consists of a set of different weights of connection. Now the so configured network could yield, for a certain input stimulus, an output representation. But this in no way implies that a representation would be deposited in the network, which (to repeat) only includes weights and potential reactions to stimuli. A first, critical conclusion is that because of its strongly representational meaning, the very notion of a trace, whether local or distributed, is to be rejected. This boils down to giving up CCTM and causal criterion (3). A second, more positive conclusion is that on connectionism properly understood, what is retained is dispositions to react, that is a procedural ability to (re-)construct representations, which falls in line with the attributionalist account. As for the role of causality now, on existing causalist views, causality is expected to account for the metaphysical singularity feature of episodic recollection. As an illustration, consider two qualitatively identical events e and e’ both experienced by a subject S in her past. Most causalists say that the causal connections between S’s present remembering and his past have to be able to secure the fact that a given episodic recollection is about event e rather than e’. Now as Robins (2016) points out, if connectionism is right, storage is superpositional, which implies that the patterns of activation deposited by e and e’ are blended. Then one can hardly see how causal criterion (4) could be secured. Importantly, a procedural conception of causality faces exactly the same difficulty. The events e and e’ are likely to bring about the same procedural ability after all. As a reply, I propose to promote an alternative view of the role causality plays in episodic memory. On this view, this is an undue expectation to assign to causality the task of accounting for metaphysical singularity. Let’s consider singularity as a phenomenological feature, instead. I propose to say that this feature consists in (what I suggest to call) a presuppositional singularity. Thereby I mean that as an episodic recollection represents an event as a singular event, it takes for granted that the represented event could be assigned singular spatiotemporal coordinates. But for doing so, it has not to be able to actually assign such coordinates, and thus it has not to draw on any metaphysically unique

A Case for Procedural Causality  41 causal link to a past event. As many experimental studies make it clear, the procedural feature of fluency is able to bring about the consciousness of remembering a singular event even though there is no singularity feature intrinsic to such procedural features. This nicely falls in line with having the experience of remembering a singular event without being able to assign a singular spatiotemporal address to the event (Hoerl and McCormack, 1999, 158). So PCTM can account for the phenomenological feature of singularity without endorsing causal criterion (4) that turns out to be too strong a constraint to be satisfied by a cognitive system. On the whole, if I am on the right track, the proceduralist conception of causality provides an efficient alternative to the classical causal theory. Indeed, it avoids the difficulties recently pointed out about causal criteria (3) and (4). More specifically, it shows that even if there is no causal continuity as far as the representational aspect of remembering is concerned, there is still a causal connection as far as the procedural aspect is concerned. It also shows that even if (as documented by neuropsychology) the occurring of singular causal connections is highly unlikely, this does not prevent from accounting in causal terms for the phenomenological feature of singularity, which is all we need.

3 Deeper Into the Phenomenology of Episodic Memory: Reconsidering the Feeling of Pastness So far, after formulating the soundness constraint (Section 1), I have fleshed out and promoted a different notion of causality with a view to arguing for the necessity of the causal connection in episodic memory (Section 2). Indeed, this notion permits to preserve the soundness constraint—more specifically the soundness of its causal component—while accommodating neuropsychological data. In this section, I will argue that the proceduralist notion of causality permits not only to preserve the soundness constraint, but also to account for the very phenomenology that is declared to be sound by this constraint. I will do so by showing that the feeling of pastness (FP henceforth) that is often held to constitute the phenomenology of episodic memory has to be conceived as a metacognitive feeling. 3.1  An Attributionalist Account of the Feeling of Pastness That a FP is intrinsic to the experience of remembering has long been ­envisioned and discussed by philosophers (Russell, 1921, Lecture 9) and ­psychologists (James, 1890, ch. 16). It is still important in the study of m ­ emory, some authors being strongly reluctant to endow it with any substantial role (Teroni, 2015; Debus, 2016), while others are eager to promote it or one of its cognates in the account of remembering (Matthen, 2010; Dokic, 2014; Michaelian, 2016). But even among its defenders, there is no agreement about the way it should be understood. In particular, a strong divide runs

42  Denis Perrin between those who think that pastness is one of the elements encoded in the intentional content of an episodic memory (Perner, 2000; Fernandez, 2008b, 2013)—let’s call this line of analysis encoding intentionalism—and those who think it issues from a subpersonal detection of the way the cognitive system carries out certain tasks (Fernandez, 2008a; Matthen, 2010)— let’s call this line of analysis retrieval intentionalism.9 I definitely take side with the latter view in this chapter. But I think that even its proponents are mistaken in the way they conceive the FP or minimize its importance for episodic memory. In particular, few people see it as a metacognitive feeling, and the rare people who do so (Dokic, 2014, for instance) do not place it in the right specific category. In this regard, my claim is that the FP is not a metacognitive feeling relative to epistemic capabilities to retrieve or reconstruct a past event—it is not an episodic feeling of knowing—but a feeling relative to the causal source, as sustained by attributionalism. Studies about metacognition have repeatedly shown that some feelings are the outcomes of heuristic metacognitive mechanisms (Koriat, 2007; ArangoMunoz, 2014; about episodic memory specifically, Souchay et al., 2007). In their pioneering study, Schacter and Singer (1962) argued experimentally for a two-factor theory of emotions, on which emotions are (sometimes, at least) the affective results (i.e., feelings) of an interpretation process of physiological sensations. In an experiment, unknown to them some subjects were given a drug that produced physiological arousal in them—so was the first factor. When the reaction occurred, half of the subjects were not provided the explanation that they had been injected a drug whose effects they were presently sensing—the explanation (the second factor) was missing. As it turned out, depending on the emotional context in which they were placed, they subpersonally interpreted the physiological arousal they sensed as a manifestation of happiness or angriness. So, feelings (e.g., emotional) can be the outcomes of an interpretation of a certain functioning of the cognitive system (e.g., a physiological arousal). In Nelson and Narens’ terms (1990), feelings can be the outcomes of a meta-level evaluation (or monitoring) of happenings occurring at the information processing level. Arguably, such meta-level interpretation operations play a crucial role in remembering. In particular, attributionalists have it that the FP is one of the possible outcomes of these operations. In Whittlesea’s terms: “the feeling of pastness that separates remembering from other memory-supported activities is not a direct product of interacting with memory, but the result of evaluating that interaction” (1997, 241—see also Jacoby et al., 1989, 393 and 400). If this is right, then the FP has to be considered as a metacognitive feeling in nature and elucidated accordingly. Committing oneself to this analysis has an immediate interesting consequence relative to the phenomenology of episodic memory. It locates the attribution of the pastness feature at a subpersonal level. Indeed, the occurrence of feelings seems to require that their cause remain at a subpersonal level. In some experiments (Jacoby & Whitehouse, 1989; see also Roediger and McDermott, 1995), the perceptual

A Case for Procedural Causality  43 fluency of items was increased unknown to the subjects. In a recognition test, this increased fluency resulted in an increased number of “old” (or recall) responses. But as the source of the “old” feeling was made explicit to the subjects, the number of “old” responses dramatically decreased. Such results favor the idea that the interpretive process that causes the FP has to occur at a subpersonal level. If so, then one can neatly account for the phenomenological status of pastness. Why an episodically remembered event does appear as intrinsically past, so that one is prone to describe one’s recollective experience in terms of a mental re-experiencing of the event? Because the represented event has been assigned the pastness feature at a subpersonal level by a heuristics. This means, first, there is no deliberate control of the subject on this operation, thus the event appears as already possessing the mentioned feature as an intrinsic feature when the subject accesses its representation. Second, the attribution is no conceptual process, thus pastness has the format of a phenomenological felt feature. On the whole, the attributionalist analysis of the FP has thus strong supports. I now embark upon a critical discussion of the existing philosophical accounts of the FP to restore its due place in the theory of episodic memory along attributionalist lines. 3.2  Disentangling the Locating-in-the-Past/Feeling-as-Past Conflation It is tempting to assimilate two things in the one single notion of “being presented to [a subject] S as happening in the past” (Fernandez, 2008b, p. 336—see also Michaelian, 2016, p. 194 for this assimilation):

• Locating an event e in the past: S’s assigning a temporal address to e •

within S’s personal past. Feeling an event e as past: S’s feeling of mentally re-experiencing e.

Closely considered, these two cognitive happenings are different in various respects. To be sure, it would be simplistic to say there is no link between them. In particular, both achievements rely on the detection and interpretation of features of the processes through which the memory operation is carried out, as their fluency for the FP and the strength of the memory trace for the relative recency of an event. Strong discrepancies must be pinpointed, however. To begin with, to engage in the operation of dating an event in one’s own past does not involve but rather presupposes locating it in one’s past. In other words, assigning a date is not an attempt to answer the question of whether one is imagining something belonging to one’s own past, but to answer the question of where the event is situated within one’s own past, in particular how it relates to other events of the same subject’s past. Moreover, one can episodically remember an event while one is unable to provide any location within one’s personal past. In other words, the feeling that one is mentally re-experiencing a particular

44  Denis Perrin event does not depend on one’s ability to provide a specific location to this event in one’s past—in terms suggested previously, the particularity of episodically remembered events is presuppositional. Instead, initiating the process of determining the temporal location of an event is often contingent on having episodically remembered this event. Conversely, one can use the various mechanisms usually resorted to by cognitive subjects to determine the temporal address of their memories (Friedman, 1990, 1991) without being episodically remembering. Consider the following example due to Friedman: “Which word have you read most recently, beach or theorizing?” (1990, 27) According to the mechanism pinpointed by so-called strength model,10 one draws on the respective strengths of the trace of beach and of the trace of theorizing to determine their temporal order of appearance in one’s past. But to do so in no way supposes to episodically remember these words. The traces in question have not to be episodic traces; they can be semantic. So, one has not to feel a remembered event as past—one can merely know it is past—in order to assign it a temporal address. This is certainly why the strength theory proponents talk about the operation of “time estimation” and “dating events” (see Brown et al., 1985—in the same way, Hinrichs (1970) speaks of “the ability to judge the recency of events”), but never of “feeling” them as past. It thus seems preferable not to consider attributionalism, as far as it claims to account for the FP, as merely one among various theories that attempt to work out the mechanisms through which we locate events in our personal pasts (Fernandez, 2008a). As far as attributionalism is concerned, its specificity should be preserved to the effect that it accounts for something (the FP) for which the mentioned theories do not claim to account. 3.3  Rebutting Encoding Intentionalism on the Feeling of Pastness According to encoding intentionalism, the pastness proper to an episodically remembered event is either an intentional element included in content at encoding—call this strict encoding intentionalism (Perner, 2000)—or something derived from such an intentional element at retrieving—call this qualified encoding intentionalism (Fernandez, 2008b). Arguing for the latter, Fernandez says: “the intentional properties of memory can account for the feeling of pastness,” and: “we do not experience temporal properties of past events when those events are presented to us in memory. However, we do experience certain properties of past events that, as it were, track their property of occurring in the past when those events are presented to us in memory” (2008b, resp. 337 and 354). According to Fernandez, it is the causal status of past events that would “track their property of occurring in the past” and brings about the FP (2013, 442). The event remembered would be intentionally represented as being the cause of the current episode of remembering, and since the remembering subject knows that a cause precedes its effect, the pastness of the event would derive from its causal

A Case for Procedural Causality  45 status. On the whole, depending on the version one endorses, encoding intentionalism has it that the FP is an encoded intentional component of content or an element of content derived from such a component. At this stage, it is important to recall the categorical status of the feeling of pastness (see 3.1). According to strict encoding intentionalism, the pastness feature of episodic remembering would consist in a piece of information included among the elements of the content acquired at encoding. But if, as argued by attributionalism, pastness has to be conceived as a metacognitive feeling, then on the one hand, it is an intrinsically phenomenological feature rather than something we would first know about the event remembered, which would trigger a certain appearance added to the remembered event. On the other hand, it is something generated at remembering rather than something acquired at encoding and preserved since then. Therefore, it seems hard to maintain strict encoding intentionalism. What about ­qualified encoding intentionalism? It also faces serious difficulties. To begin with, if it were right, then things would go this way: one would first have to retrieve the causal information about the representation of the remembered event and then derivatively conclude that the event in question is a past one. But it seems as likely to say that as one is episodically remembering an event, one is conscious of the latter as an event one has experienced in the past, and for this very reason as the causal source of one’s current memory (Hoerl & McCormack, 2001, p. 209). The alleged anteriority of causality on pastness is not obvious. Moreover, if qualified intentionalism were right, one would face a difficulty faced by strict intentionalism too. As argued previously, feeling something as past in episodic recollection consists in experiencing it as having an intrinsic temporal property when it is presented to us. If qualified intentionalism were right, during the process of remembering there should be a preliminary phase at which one would be conscious of the event represented with its causal status and would conclude therefrom to its pastness. This knowing-it-to-be-past state would then have to trigger the adequate phenomenology. But one can hardly see how such a state of knowing could endow the remembering state with such phenomenology, and actually remembered events are very often immediately presented as past. These difficulties are serious enough to renounce encoding intentionalism. Moreover, they promote a much more procedural account of the feeling of pastness, as desired by attributionalism. To repeat: the intrinsic phenomenological feature of pastness can hardly be explained in terms of an encoded knowledge about the remembered event. 3.4  Rebutting the Chronesthesia Account of the Feeling of Pastness In recent psychological literature, the FP is conceived in terms of the time consciousness designated by Tulving as autonoetic consciousness (Tulving, 1985, pp. 5–6; Klein, 2015). Tulving (2002) refines this notion

46  Denis Perrin and distinguishes “autonoetic consciousness” as the consciousness of oneself as extended in time and “chronesthesia” as the consciousness of the experiential time in which one can place oneself through autonoetic consciousness. The FP would thus be the consciousness of the past time provided by chronesthesia. Importantly, chronesthesia would be the consciousness of subjective time extended toward the future as well. On the mental time travel paradigm in psychology, episodic memory system permits episodic simulations of the past (episodic recollection, in particular) as well as of the future. Thus, it seems legitimate to consider there is a feeling of futurity just in the same way as there is a FP (Michaelian, 2016), the latter being the mere symmetrical counterpart of the former. A broad disagreement between this account and proceduralism concerns the respective view of memory each one promotes. On the chronesthesia account, the subjective sense of pastness at play in episodic recollection is a given structure of the mind, while the proceduralist view explains it as the outcome of an inferential attribution process.11 Here are two arguments for the latter view. First, the symmetry between the FP and the feeling of futurity is not as neat as it should be expected if the structural account were right. As emphasized by (moderate) discontinuism on mental time travel (Perrin, 2016), the subjective sense of time in episodic recollection is not symmetrical to the subjective sense of time in episodic future thought. One striking feature is that the FP as described previously does not occur at will in episodic recollection, contrary to what happens in the future episodic thought. Indeed, one can try but fail to episodically remember an event, even though one is imagining it as located in one’s past. Instead, episodically thinking of a future event is something one actually does as soon as one simulates it. Moreover, in the case of episodic future thought, the event does not have to appear as future in the first place; it is rather assigned in a stipulative way this temporal feature by the imagining subject. Instead, in episodic recollection pastness appears as an intrinsic feature of the appearing event, which it possesses independently of any explicit assignment operation by the subject. Furthermore, the FP does not merely consist in sensing something as located in one’s past; it also involves sensing it as having actually been experienced by the subject in her past. These discrepancies—this is the second argument—are smoothly explained by proceduralism. Indeed, the latter grounds the FP on the detection of some procedural features. Now an obvious difference between backward and forward mental time travel is that in the former case, but not in the latter, something has already happened (in the past) that is thus liable to have caused a modification of the cognitive system that is in turn liable to be subpersonally detected and to automatically trigger an inferential attribution. This explains the asymmetry with the episodic future thought case. Moreover, the proceduralist account of the pastness phenomenology provides an explanation of the intrinsicness feature of pastness in episodic

A Case for Procedural Causality  47 recollection, which is not observed in episodic future thought. Provided the attribution to the past is a subpersonal and automatic cognitive process, this is no surprise that the simulated event appears as intrinsically past. These critical discussions all point toward the idea introduced previously that the FP is of a metacognitive nature. But which specific kind of metacognitive feeling is it? 3.5 Distinguishing Feeling of Pastness and Episodic Feeling of Knowing (EFOK) Some researchers—in the minority—have proposed to consider the FP as a metacognitive feeling. Dokic (2014: 10), in particular, proposes to define it as an episodic feeling of knowing, claiming that “EFOKs can be bound to fully explicit memories.” I think such a characterization is inaccurate about the metacognitive type of the FP. On a standard definition (Souchay, 2013), EFOK:

• is an affective state endowed with a content issuing from metacognitive •

monitoring—a feature shared with other metacognitive feelings (Koriat, 2007); usually occurs before any actual recollection (Paynter et al., 2009; Arango-Munoz & Michaelian, 2014: 100) and has a predictive content relative to the capacity of the subject to episodically recollect.

I think that the second, specifying feature makes it impossible to assimilate FP to EFOK. First, indeed, if EFOK and FP were the same feeling both in the predictive and the recollective situations, then we should have a FP before actually recollecting. But in the predictive situation, we very often only have a feeling concerning our epistemic capabilities relative to episodic recollection without carrying out the latter, a point strongly confirmed by the dissociation between EFOK and actual episodic performances in Alzheimer patients (Souchay et al., 2007). Moreover, on an important line of analysis (Souchay & Moulin, 2009), even as both feelings cooccur, they, however, do not confuse with each other. On the mentioned analysis, indeed, one ought rather to say that positive EFOKs are contingent on partial recollective experience, which implies that they are distinct. Second, as far as the content is concerned, EFOK is about epistemic capabilities, rather than about the causal status of an actually represented event, the latter being precisely the content of the FP. The proceduralist view here defended can account smoothly for this important difference, since it grounds FP on the detection of an apparent causal effect, which requires the construction process to have occurred for it to display the detected procedural feature. If this is right, then one should prefer to define FP as a causal-source relative metacognitive feeling, rather than as an epistemic one.

48  Denis Perrin

4 Conclusion Where does all this leave us? In what precedes, I have argued for a proceduralist view of the causal connection required by episodic recollection, rebutting the representational and metaphysical singularity requirements of the existing versions of causalism, and for a causal-source relative metacognitive view of the phenomenology of episodic recollection, framed in the attributionalist terms of a FP. By way of conclusion, I make it clear how these two threads should be pieced together.

• The feeling of pastness is efficiently grounded by procedural causality in



episodic memory. Indeed, provided it is a metacognitive feeling about the causal source of a current mental representation, procedural causality can explain the FP in terms of the interpretation of the detection of a procedural effect due to a prior encounter with the represented event. In particular, this line of analysis is well equipped to account for the intrinsicness of the pastness feature to the recollective representation and for the presuppositional singularity feature of the FP. Conversely, procedural causality is what the feeling of pastness phenomenologically represents. The latter has a complex content (not a merely time-relative content) including the parameters of the cause of the detected procedural effect: the event appears as a causal singular source actually experienced by the rememberer in her own past. These features are precisely the conditions to be satisfied for the mentioned causal effect to occur in normal circumstances.

If this is correct, an interesting consequence is that, at least as far as the information conveyed by the phenomenology of episodic memory is concerned, the encoded content should not be conceived in representational terms. It ought rather to be explained in procedural terms, as maintained by Whittlesea (1997), that is: in terms of procedural skills detected as one is remembering.

Notes 1 Thanks for feedback at the University of Otago, the Thumos seminar of the University of Geneva and Issues in the Philosophy of Memory (Cologne, 2017). Special thanks for very helpful written comments to Kourken Michaelian, Sarah Robins, and Santiago Arango-Munoz. 2 The relation between intentional and phenomenal aspects of sensory experiences has raised debates of its own. I leave them aside in this paper. 3 A thorough exploration of this point would require to discuss the epistemic innocence stand (Bortolotti, 2015). I leave it aside for today. 4 The soundness constraint should not be confused with the claim that phenomenology is always rightly informative rather than misinformative. Phenomenological information concerns the normal use of memory but excludes in no way the deviant cases; it rather explains, for such cases, why it can merely

A Case for Procedural Causality  49 seem that one is remembering. In other terms, I am not committed to the “doctrine of concordance of cognition, behaviour, and experience” discussed critically by Tulving (1989, 8). 5 These two criteria are meant to accommodate the relearning cases. 6 This criterion is intended to accommodate the suggestibility case. 7 See (Jacoby et al., 1989). Similar claims are made in source monitoring framework (Johnson et al., 1993) 8 “When an object is experienced a stimulus enters the system and gives rise to a pattern of activity over a network of neurons. This pattern of activity is taken to be the representation of the object” (Bernecker, 2010, 132). 9 Both versions of intentionalism consider that the feeling of pastness is endowed with a specific meaning, i.e., they endorse the view that “feelings are always directed toward an object or piece of information” (Arango-Munoz, 2014, 196). 10 There are others, see Friedman (1990). 11 Note that at some point (1983, 187–8) Tulving frames his account of subjective pastness in terms of feelings of pastness rather than in terms of a structural “autonoetic” mode of consciousness.

References Arango-Munoz, S. (2014). The nature of epistemic feelings. Review of Philosophy and Psychology, 27(2), 193–211. Arango-Munoz, S., & Michaelian, K. (2014). Epistemic feelings, epistemic emotions: Review and introduction to the focus section. Philosophical Inquiries, 2(1), 97–122. Bernecker, S. (2010). Memory—a philosophical study. Oxford: Oxford University Press. Bortolotti, L. (2015). The epistemic innocence of motivated delusions. Consciousness and Cognition, 33, 490–499. Brown, N. R., Rips, L. J., & Shevell, S. K. (1985). The subjective dates of natural events in very-long-term memory. Cognitive Psychology, 17, 139–177. De Brigard. (2014). The nature of memory traces. Philosophy Compass, 9(6), 402–414. Debus, D. (2016). Temporal perspectives on imagination: On the nature and value of imagining the future. In K. Michaelian, S. Klein, & K. Szpunar (Eds.), Seeing the future (pp. 217–240). Oxford: Oxford University Press. Dokic, J. (2001). Is memory purely preservative? In C. Hoerl & T. McCormack (Eds.), Time and memory: Issues in philosophy and psychology (pp. 213–232). Oxford: Clarendon Press. Dokic, J. (2014). Feeling the past: A two-tiered account of episodic memory. Review of Philosophy and Psychology, 5, 413. Fernandez, J. (2008a). Memory, past and self. Synthese, 160, 103–121. Fernandez, J. (2008b). Memory and time. Philosophical Studies, 141(3), 333–356. Fernandez, J. (2013). Memory. In H. Dyke & A. Bardon (Eds.), A companion to the philosophy of time (pp. 432–443). Oxford: Wiley-Blackwell. Friedman, W. J. (1990). About time: Inventing the fourth dimension. Cambridge, MA: MIT Press. Friedman, W. J. (1991). The development of children’s memory for the time of past events. Child Development, 62, 139–155. Hinrichs, J. V. (1970). A two-process memory-strength theory for judgment of ­recency. Psychological Review, 77(3), 223–233.

50  Denis Perrin Hoerl, C. (2001). The phenomenology of episodic recall. In C. Hoerl & T. McCormack (Eds.), Time and memory (pp. 315–335). Oxford: Oxford University Press. Hoerl, C., & McCormack, T. (1999). Memory and temporal perspective: The role of temporal frameworks in memory development. Developmental Review, 19, 154–182. Hoerl, C., & McCormack, T. (2001). The child in time: Temporal concepts and self-consciousness in the development of episodic memory. In K. Lemmon & C. Moore (Eds.), The self in time—developmental perspectives (pp. 203–227). Londres: Lawrence Erlbaum. Jacoby, L. L., Kelley, C. M., & Dywan, J. (1989). Memory attributions. In H. L. Roediger & F. I. M. Craig (Eds.), Varieties of memory and consciousness: Essays in honour of Endel Tulving (pp. 391–422). Hillsdalle: Lawrence Erlbaum. Jacoby, L. L., & Whitehouse, K. (1989). An illusion of memory: False recognition influenced by unconscious perception. Journal of Experimental Psychology: General, 118(2), 126–135. James, W. (1890). The principles of psychology. New York: H. Holt. Kelley, C. M., & Jacoby, L. L. (1990). The construction of subjective experience: Memory attributions. Mind and Language, 5(1), 49–68. Klein, S. (2015). What memory is? WIREs Cognitive Science, 6, 1–38. Kolers, P. A., & Roediger, H. L. (1984). Procedures of mind. Journal of Verbal Learning and Verbal Behavior, 23, 425–449. Koriat, A. (2007). Metacognition and consciousness. In M. Zelazo, M. Moscovitch, & E. Thompson (Eds.), Cambridge handbook of consciousness (pp. 289–325). New York: Cambridge University Press. Leboe-McGowan, J. S., & Whittlesea, B. (2013). Through the SCAPE looking glass—sources of performance and sources of attribution. In D. Reisberg (ed.), The Oxford Handbook of Cognitive Psychology (pp. 243–266). Oxford: Oxford University Press. pp. 243–266. Martin, C. B., & Deutscher, M. (1966). Remembering. Philosophical Review, 75(2), 161–196. Matthen, M. (2010). Is memory preservation? Philosophical Studies, 148, 3–14. McClelland, J. L. (2000). Connectionist models of memory. In F.I.M. Craik, & E. Tulving (Eds.). The Oxford handbook of memory (pp. 583–596). Oxford: Oxford University Press. Michaelian, K. (2011). Generative memory. Philosophical Psychology, 24, 323–342. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge: MIT Press. Neisser, U. (1967). Cognitive psychology. New York: Meredith. Nelson, T. O., & Narrens, L. (1990). Metamemory: A theoretical framework and new findings. The Psychology of Learning and Motivation, 26, 125–173. O’Regan, J. K., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and Brain Sciences, 24, 939–1031. Paynter, C. A., Reder, L. M., & Kieffaber, P. D. (2009). Knowing we know before we know: ERP correlates of initial feeling-of-knowing. Neuropsychologia, 47(3), 796–803. Perner, J. (2000). Memory and the theory of mind. In F. I. M. Craik & E. ­Tulving (eds.), The Oxford handbook of memory (pp. 297–312). Oxford: Oxford University Press.

A Case for Procedural Causality  51 Perrin, D. (2016). Asymmetries in subjective time. In K. Michaelian, S. Klein, and K. Szpunar (Eds.), Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel (pp. 39–61). Oxford: Oxford University Press. Robins, S. (2016). Representing the past: Memory traces and the causal theory of memory. Philosophical Studies, 173, 2993–3013. Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21(4), 803–814. Russell, B. (1921). The analysis of mind. London: Allen & Unwin. Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering the past and imagining the future. Philosophical Transactions of the Royal Society, Biological Sciences, 362, 773–786. Schacter, D. L., Addis, D. R., Hassabis, D., Martin, V. C., Spreng, R. N., & Szpunar, K. K. (2012). The future of memory: Remembering, imagining, and the brain. Neuron, 76, 677–694. Schacter, S., & Singer, J. E. (1962). Cognitive, social, and physiological determinants of emotional states. Psychological Review, 69(5), 379–399. Souchay, C. (2013). Métamémoire et troubles de la mémoire: l’exemple du feelingof-knowing. Revue de Neuropsychologie, 5(4), 265–272. Souchay, C., & Moulin, C. (2009). Memory and consciousness in Alzheimer’s disease. Current Alzheimer Research, 6. Souchay, C., Moulin, C. J., Clarys, D., Taconnat, L., & Isingrini, M. (2007). Diminished episodic memory awareness in older adults: Evidence from feeling-ofknowing and recollection. Consciousness and Cognition, 16, 769–784. Sutton, J. (1998). Philosophy and memory traces. Cambridge: Cambridge University Press. Teroni, F. (2014). The epistemological disunity of memory: In mind, values, and metaphysics—philosophical. In A. Reboul (Ed.), Essays in honor of K. Mulligan (Vol. 2, pp. 183–202). Dordrecht: Springer. Tulving, E. (1983). Elements of episodic memory. New York: Oxford University Press. Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26, 1–12. Tulving, E. (1989). Memory: Performance, knowledge, and experience. European Journal of Cognitive Psychology, 1(1), 3–26. Whittlesea, B. (1997). Production, evaluation, and preservation of experiences: Constructive processing in remembering and performance tasks. In D. L. Medin (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol. 37, pp. 211–264). San Diego, CA: Academic Press.

3 The Functional Character of Memory

Jordi FernándezThe Functional Character of Memory

Jordi Fernández

1 Introduction The purpose of this chapter is to determine what is to remember something, as opposed to imagining it, perceiving it, or introspecting it. What does it take for a mental state to qualify as remembering, or having a memory of, something?1 The main issue to be addressed is therefore a metaphysical one. It is the issue of determining that features those mental states that qualify as memories typically enjoy, and those states that do not qualify as such typically lack. I will proceed as follows. In Sections 2 and 3, I will discuss the two main existing conceptions of the conditions that a mental state must satisfy to count as an episode of remembering. The first of these approaches is backward-looking. It puts forward conditions that strictly concern the aetiology of the mental state. I will argue that the conditions offered by the backward-looking approach are both too strong and too weak: They rule out mental states that, intuitively, count as memories while including mental states that, intuitively, do not qualify as memories. The second approach is forward-looking. It puts forward conditions that only concern the use that the subject makes of the mental state while forming beliefs about their own life. I will argue that the conditions proposed by the forward-looking approach are both too weak and too strong as well. However, the discussion of the two approaches will allow us to extract some helpful lessons on the constraints that any proposal about the nature of remembering should respect. An alternative approach aimed at incorporating those lessons will be offered in Section 4 by drawing on the literature on functionalism. In Section 5, I will argue that this approach can, on the one hand, accommodate as memories those mental states that indicate that the backward-looking approach and the forward-looking approach are too strict while, on the other hand, excluding those mental states that suggest that the two alternative approaches are too permissive. Accordingly, I will conclude that construing memory along functionalist lines is a satisfactory approach to the nature of remembering. The scope of this project will be modest, in that I will only be concerned with a particular type of remembering. There are several forms of remembering.

The Functional Character of Memory  53 There is, for example, memory for events or states of affairs, for abilities, and for objects. Our discussion in this chapter will focus on a specific kind of memory for events and states of affairs; a form of remembering that essentially involves having a mental image but does not require having formed, at any previous moment, the belief that the relevant event, or the relevant state of affairs, obtained in the past.2 This is the sense in which one may remember, for example, that one left the door of one’s house open when one left for work in the morning. Let us call this ‘episodically remembering’ the event.3 This chapter will be concerned with the conditions that a mental state must meet to qualify as episodically remembering an event (over and above that of having a mental image of the event). Accordingly, in what follows, when I speak of remembering an event or having a memory of an event, I will refer to, respectively, episodically remembering the event and having an episodic memory of the event.

2  The Causal Theory of Memory There is a popular view about the nature of memory according to which the issue of whether a mental state qualifies as remembering or not hinges on the origin of the state. The so-called causal theory of memory (or “CTM” for short) remains the most influential version of this view. According to the classical formulation of CTM, for any subject S and event e, S remembers e just in case S is representing e, S represented e in the past, and the fact that S represented e in the past has caused S to represent e in the present.4 To clarify this formulation of CTM, it is worth elaborating on the notions of representation, causation, and event used in it. As far as CTM is concerned, the notion of representing an event should be understood as an umbrella notion that covers a number of possible ways in which a subject may experience an event. For example, if a subject perceptually experienced a worldly event in the past or introspectively experienced a mental event in their own mind, or they experienced being the agent of one of their actions, then the subject counts, on this reading of ‘representation,’ as representing each of those events. Furthermore, the notion of representation used by the advocate of CTM also includes a way in which a subject can experience an event that may not be happening in the present, namely, by visualizing the event. The reason why the advocate of CTM is employing such a broad notion of representation is that we all seem to have memories of events of very different types—events that we witnessed in the past, events of which we were aware subjectively, and events that consisted in our own actions. In all of these cases, the advocate of CTM claims, the mental states wherein we visualize those events qualify as memories of them if, and only if, they have been caused by our past representations of those events, whether those representations were perceptual experiences, episodes of introspection, or agentive experiences. The classical formulation of CTM is meant to be neutral on the precise nature of the causal relation. The thought is that whatever the correct view

54  Jordi Fernández about the nature of the causal relation turns out to be, that relation must hold between a subject’s past representation of an event and a current mental state of the subject in order for us to count the subject’s mental state as a memory of the event. The advocate of CTM is, however, committed to the view that the causal link between the past representation and the memory must be ‘nondeviant.’ What does this mean? Suppose that I am walking by a basketball court, and I see a basketball in the air on its way to the hoop. Suppose that the perception of the ball’s color makes me think of a New York Knicks jersey, which in turn triggers in me the mental image of a basketball in the air on its way to a hoop. In this case, I am visualizing the traveling of a basketball toward a hoop, and my mental image is at the end of a causal chain that originates in a recent perception of a basketball traveling toward a hoop. But intuitively enough, my mental state does not qualify as an episode of remembering. The reason seems to be that, in this case, the causal link between the basketball and my mental image is, in some sense, indirect or deviant. Thus, the advocate of CTM requires that the causal link between a memory of an event and the subject’s past representation of the event must be nondeviant. Naturally, this move places the burden of specifying exactly what counts as a deviant instance of a causal relation, and what does not, on the CTM advocate’s shoulders. However, for the purposes of our discussion of CTM, I will assume that the advocate of CTM can draw the distinction between deviant and nondeviant instances of the causal relation in a principled way. As far as I can see, none of the objections to be raised against CTM later hinge on this point. The classical formulation of CTM is not explicit on how the notion of an event should be understood. However, it does seem that one particular view about the nature of events fits most naturally with CTM, namely, that of events as property exemplifications.5 Notice that, except for rare cases of so-called eidetic memory, we do not remember all the details of those events that we have represented in the past. For example, imagine that, as a child, I once saw a horse performing the canter at some festival. Suppose that I now have a mental image wherein I visualize the horse moving with some gait during the festival performance. Let us stipulate that I have that mental image as a result of having seen the horse at the festival and that, in virtue of having that mental image, I am inclined to believe that I once saw a horse moving with some gait during a festival performance. Intuitively enough, it seems that the mental state that I now occupy is an episode of remembering. And yet, on the face of it, CTM seems to preclude us from accepting this mental state as an episode of remembering, since the event that I represented in the past (the horse performing the canter) seems to be different from the event that I am now representing in the present (the horse moving with some gait). To handle this difficulty, the advocate of CTM may individuate events as property exemplifications or more precisely, exemplifications of sets of properties by an object at a time. This allows the advocate of CTM to

The Functional Character of Memory  55 introduce a relation of inclusion between events: An event e*, the advocate of CTM might claim, is included into an event e just in case all the properties that are exemplified as part of event e* are properties exemplified as part of event e, in the same object and at the same time. Introducing this relation makes it possible for the advocate of CTM to argue that, in the past, I did represent, contrary to what it may have seemed at first glance, the very event that I am now representing when I have a mental image of a horse moving with some gait during a festival. For it seems reasonable to assume that, by representing an event, one always represents all the events included in it. And the event that consists in the horse moving with some gait is included in the event that consists in the horse performing the canter.6 Thus, provided that events are construed as property exemplifications, it seems that CTM can accommodate the intuition that we may remember an event without representing all the details of the event that we represented in the past.7 Thus, I will be assuming this conception of events for the purposes of our discussion of CTM. What considerations can be offered in support of CTM? Episodic memory seems to register and store the content of those (typically, perceptual) experiences that we had in the past by producing mental images that inherit their contents from those experiences.8 The preservative aspect of memory is, for the reasons mentioned previously, imperfect. Typically, not every detail of an event represented in the past will be preserved in our memory of it. Nevertheless, the preservative aspect of memory is of great importance to us. It provides us with the ability to navigate familiar environments by allowing us to recognize people, places, and situations that we encountered in the past.9 Now, a virtue of CTM is that it accounts for this aspect of memory straightforwardly. If CTM is correct, then it is no wonder that our memories make available to us information that was initially provided to us by some of our own past experiences. For memories must inherit their contents from those experiences in which the memories originate. After all, according to CTM, a mental state would not count as a memory in the first place if it did not represent something that the experience in which it causally originates represented in the past. CTM enjoys, therefore, a significant virtue. Unfortunately, CTM also needs to face two important difficulties. First, CTM seems to be too strict. For, even though it allows a memory experience to include less detail than the experience in which it originates, it does not allow it to include more. And yet, a memory of a past event may, intuitively enough, include details that one’s past experience of the event did not include. Let us suppose that, as a child, I enjoyed accompanying my father while he went on country walks hunting rabbits and, once, I saw him shoot a white rabbit. Suppose that, as a result of having had that perceptual experience in the past, I am now having an experience wherein I visualize the rabbit, almost exactly as the rabbit appeared to me in the past, except for the fact that I now visualize the rabbit as being black. As a result, I am

56  Jordi Fernández disposed to claim that I once saw a black rabbit being shot. According to CTM, my present experience does not constitute a memory of a black rabbit being shot, since it contains details that were never present in my original experience of the rabbit. This seems highly counterintuitive.10 After all, cases of this sort are extremely common, so ruling this case out as an episode of remembering seems to make the scope of memory unreasonably narrow.11 It seems more natural to say that this is a case in which I misremember the event. Since incorrect memories still count as memories, this means that the case should be acknowledged as an instance of remembering. Let us call cases of this type, ‘embellishment’ cases. Notice that, in the embellishment case considered earlier, the detail of the remembered event included in the memory, but not in the experience in which the memory originates, did not obtain in the past. (The rabbit was not, in fact, black.) But this is inessential to the objection being raised against CTM. Suppose that, as a result of having my perceptual experience of my father shooting a white rabbit in the past, I am now having an experience wherein I visualize the event, almost exactly as I perceptually experienced it in the past. The only difference is that I now visualize my father as wearing a belt with a silver buckle on it at the time of the shooting, a detail that (let us stipulate) I could not have perceived at the time of the shooting given my spatial position in the scene relative to my father’s. Suppose that this is a belt with which I am very familiar, and as it turns out, my father was indeed wearing it at the time of the shooting. The intuition remains that I am having a memory. And yet, according to CTM, my present experience does not constitute a memory of my father shooting the rabbit, since it contains details that were never present in my original experience of the event. Thus, it seems that embellishment cases pose a threat to CTM whether or not the details of the remembered event that, in those cases, are present in our memory of it, but not in our original experience of the event, were in fact details that the event enjoyed in the past. Let us refer to the view according to which a memory of an event cannot include details that were never present in the subject’s original experience of the event as ‘preservationism about content.’ There is a different view about memory in the philosophical literature according to which, if a subject knows some proposition on the basis of memory, it is because she knew it, at some point in the past, through some source other than memory. Let us refer to this view as ‘preservationism about knowledge.’12 It is worth pointing out that preservationism about content and preservationism about knowledge are logically independent views. The former type of preservationism does not entail the latter type, since there are, arguably, cases of knowledge generation in memory that seem to be consistent with preservationism about content.13 And the latter type of preservationism does not entail the former type, since embellishment cases of the two varieties considered previously are, arguably, counterexamples to preservationism about content, and yet they are consistent with preservationism about knowledge. For instance, in

The Functional Character of Memory  57 the case in which I visualize a black rabbit being shot, no new knowledge is generated by my memory. For if I formed, on the basis of my memory, the belief that a black rabbit was shot, I would be forming a false belief, which does not amount to knowledge. By contrast, in the case in which I visualize my father shooting a rabbit while wearing a belt with a silver buckle on it, no new knowledge is generated by my memory for a different reason. In this case, if I formed, on the basis of my memory, the belief that my father was wearing such a belt, my belief would be true, but it could have easily been false. Hence, it does not amount to knowledge either.14 Second, CTM seems to be too permissive. Imagine that I am capable of visualizing events that I witnessed in the past, but due to some substantial cognitive deficit of mine, my mental images of those events do not convey to me the sense that the relevant events did happen, and that I witnessed them in the past.15 Suppose, furthermore, that I am a painter. I decide to paint a bird landing on the roof of a house, and at the end of my work, I have painted that scene in my canvas. However, I am not inclined to think that these are a house and a bird I have ever encountered in the past. I do not think that these are, or were, a real house and a real bird. And yet, it turns out that I am wrong: The house that I have painted is a house that, in fact, I visited as a child. The bird that I have painted is a bird that, in fact, I saw during my visit. And the experience wherein I visualized the scene that I have painted in my canvas causally originates in my past perceptual experience of that bird landing on the roof of that house.16 According to CTM, in this situation, the experience wherein I visualized the house that I have painted in my canvas does qualify as an episode of remembering.17 This seems highly counterintuitive. After all, the experience I was having when I engaged in the project of painting a bird landing on the roof of a house did not convey to me the sense that this house and this bird were parts of my life in any way. And for that reason, having such an experience did not make any difference as to which propositions I was inclined to believe regarding my past and which propositions I was inclined to disbelieve. The experience made no impact on my beliefs about my past. It seems more natural to describe the case by saying that I was imagining a bird landing on the roof of a house when I engaged in the project of painting such a scene. Let us call cases of this type, cases of ‘epistemic irrelevance.’18 One might dispute this way of describing the case by proposing, instead, that I am remembering, but I do not think that I am remembering. (This is, in fact, the way in which the original version of the case was understood by Charles Martin and Max Deutscher.) As far as I can see, the choice between construing the painting case as a case in which I imagine the event, and construing it as a case in which I remember the event, involves a trade-off between two explananda about memory. One fact in need of explanation in the painting case is the remarkable similarity between the event painted by me in the present and the event witnessed by me in the past. Let us abbreviate this fact as ‘Event Similarity.’ A different fact in need of explanation is a

58  Jordi Fernández more general fact about the connection between memory and belief: All other things being equal, when a subject remembers some event, it seems that the subject will be disposed to believe that the event took place in the past. Naturally, this kind of inclination can be overridden, for example, by the acquisition of other beliefs such as the belief that the subject’s faculty of memory is not to be trusted. Nevertheless, just like prima facie we are inclined to believe that those events that we perceive are taking place in the present, we are also prima facie inclined to believe that those events that we remember took place in the past. Let us abbreviate this fact about the connection between memory and belief as ‘Epistemic Relevance.’ If we construe the painting case as a case in which I remember the bird landing on the roof of the house, then Event Similarity seems, on the one hand, easy to explain. It does not seem surprising that the content of the mental image that I am having when I paint the event is so similar to the content of the past perceptual experience in which that mental image originates. After all, given the preservative aspect of memory, this is exactly what you would expect if the mental image that I am having, when I paint the event, is a memory of that event. But Epistemic Relevance, on the other hand, seems difficult to explain on this construal of the painting case. If it is indeed possible for a subject to have a memory of some event, and yet lack the inclination to believe that the event took place in the past, then such an inclination is not essential to memory. And if it is not essential to memory, then there must be a reason why, in common instances of remembering, a subject believes that the content of their memory took place in the past. But it is hard to see what the subject’s reason for believing this might be. For example, since a subject could imagine a past event without having the inclination to believe that the event happened in the past, it seems that the reason for believing that remembered events happened in the past cannot be that remembered events are presented to us as having happened in the past. Alternatively, one might think that there is some phenomenological feature that is characteristic of memories (such as their vivacity perhaps), and we have learned that, typically, those events that are presented to us through mental images with that feature have happened in the past. But such a process of learning would be grounded on past experience. And for that reason, it presupposes (and therefore cannot explain) our inclination to believe that those events that we remember took place in the past.19 Conversely, if we construe the painting case as a case in which I imagine the bird landing on the roof of the house, then Epistemic Relevance seems, on the one hand, easy to explain. If it is impossible for a subject to have a memory of some event, and yet lack the inclination to believe that the event took place in the past, then such an inclination is constitutive of memory. Thus, the reason why a subject is inclined to believe that a remembered event happened in the past is simply that this is part of what it is for the subject to remember the event. But Event Similarity, on the other hand, seems difficult to explain on this construal of the painting case. Having said that, it does

The Functional Character of Memory  59 not seem that the similarity between the event painted by me in the present and the event witnessed by me in the past is impossible to explain. Such a similarity can still be explained by the fact that my current mental image of the painted event has been caused by my past perceptual experience of the very same event. Admittedly, there is a price to be paid for such an explanation: We must give up the idea that a causal relation between a past experience and a current mental image of the subject can only take place in memory. However, in the absence of independent considerations in support of this idea, the cost of explaining Event Similarity does not seem to be too high. It seems, therefore, that CTM captures an important property of memory, namely, its retentive aspect. But it also seems that CTM ignores two other important features of memory. Remembering, as embellishment cases illustrate, is not only a matter of preserving some information provided by our past experiences of some events, but it can also involve reconstructing that information. Furthermore, remembering, as cases of epistemic irrelevance illustrate, must have the capacity to inform our beliefs about our past. It needs to make a difference as to which things we are inclined to believe, and which we are inclined to disbelieve, regarding our past. The outcome of our discussion in this section, then, is that a successful account of the type of mental state that qualifies as an episode of remembering should accommodate these two features while, ideally, preserving the main virtue of CTM. Let us now turn to an alternative conception of remembering that emphasizes precisely those two features.

3  The Narrative Theory of Memory There is an alternative conception of memory wherein memory is not a passive device for registering and reproducing contents. It is instead a faculty akin to imagination in its creative capacity. The main tenet of this ‘narrative’ conception of memory (or ‘NTM,’ for short) is that, in memory, we are engaged in an inventive project wherein we build stories about our past by integrating content that we have acquired through our own experience with content from other sources, such as testimony, inference, and imagination. Memory, on this conception, is not meant to represent the past as we have experienced it. Instead, memory reconstructs the past in order to help us build a smooth and robust narrative of our lives. We may formulate NTM, more precisely, as the view that, for any subject S and event e, S remembers e just in case S is representing e, and S uses their representation of e as part of a narrative of S’s life.20 To clarify this formulation of the narrative view, it is worth elaborating on the notions of representation and narrative used in it. The notion of representation used in the formulation of NTM should be understood even more broadly than that used in the formulation of CTM. The reason for the broadening of this notion concerns a somewhat complicated taxonomical issue in the study of memory. NTM is not usually

60  Jordi Fernández offered as an answer to the specific question of what it takes for a mental state to qualify as an episodic memory. Instead, it is offered as an answer to the question of what it takes for a mental state to qualify as a so-called ‘autobiographical’ memory. The complication lies in the fact that the notion of autobiographical memory does not match that of episodic memory. An autobiographical memory is meant to be a memory of an event in one’s past, or in one’s life, whereas a nonautobiographical memory is supposed to be a memory of an event that does not concern one at all.21 Thus, an autobiographical memory may be episodic (such as the memory of one’s first kiss) or not (such as the memory that one was once seriously ill while being a 3-month-old baby). For this reason, when the NTM advocate claims that a mental state representing some event counts as remembering the event just in case the subject makes it part of the narrative of their life, the relevant notion of representation should be understood as covering not only experience, but also belief. For, in some cases, the mental state concerned will simply be the subject’s belief that the event was a part of their life.22 The notion of narrative being used in NTM is particularly hard to spell out. Generally speaking, a narrative of the life of a remembering subject is a story of the subject’s life. Stories, however, are more than mere collections of pieces of information. They have, for example, a temporal structure. Thus, a story of the remembering subject’s life will confer some temporal structure to the events that the subject remembers. However, it seems that this will be a loose structure. Thus, it seems plausible to suppose that, for every remembered event e in the subject’s life, a story of that life will include a series of events, ordered by the earlier-than relation, which connects e with the current event that consists in the subject remembering e.23 But it also seems plausible to suppose that it will not be the case that, for any two remembered events e and e* in the subject’s life, the story will include a series of events, ordered by the earlier-than relation, which connects e with e*. Thus, a story of my life, as I remember it, may include a number of events that were part of a family vacation that took place during my childhood, and it may include a number of events that were part of my early schooldays. But it need not be precise as to whether the events in the family vacation happened before, or after, those in my early schooldays.24 Furthermore, stories have authors, since the matter of which episodes belong to the story, and that of how those episodes are organized, are not random matters. Quite the contrary, those aspects of the story are the products of someone’s making. Thus, if NTM is right, then the fact that a subject’s memory includes some events, and not others, and the fact that it organizes them in some particular way, is something for which the remembering subject is responsible; something for which they can be held accountable. The remembering subject is, after all, the author of the narrative that includes their memories. Beyond this, it is difficult to say how much more the advocate of NTM will build into the notion of a narrative. A number of issues about this notion are open to discussion, and it seems that different positions on these issues

The Functional Character of Memory  61 will yield different versions of NTM. We may wonder, for example, about the phenomenology of narratives, and whether the affective properties of a memory of an event, when that event is included in a narrative of one’s own life, must be identical with the affective properties of the original experience wherein that event was initially presented to one.25 Similarly, we may wonder about the function of narratives. One reasonable view about this issue is that the function of including, in memory, a past event as part of the narrative of one’s own life is epistemic: Including the event in the narrative is meant to provide one with answers to the questions of who one is, and what kind of person one is. Another reasonable view is that the function of including, in memory, a past event as part of the narrative of one’s own life is normative: Including the event in the narrative is meant to provide one with a standpoint from which one can evaluate one’s own past actions, feelings, and reactions to significant events in one’s past.26 Finally, we may wonder about the intentionality of narratives, and the extent to which a narrative of one’s life must represent oneself as a character in the story. If the events in the story of one’s life are all represented, as it were, from the inside (and therefore one is never visualized as being a participant in those events when one remembers them), then is one still a character in the story being narrated? And if one is not, does the story count as a story of one’s own life?27 For the purposes of our discussion of NTM, I will assume a notion of narrative that is neutral on all of these issues. As far as I can see, none of the objections to be raised against NTM hinges on any of these points. What considerations can be offered in support of NTM? The narrative theory can accommodate our intuitions regarding embellishment cases and cases of epistemic irrelevance. Consider, first of all, the mental state wherein I visualize a black rabbit being shot by my father. If NTM is right, then my mental state counts as a state of remembering in spite of the fact that the mental state at issue does not originate in a past perception of a black rabbit being shot. The reason is that, in virtue of occupying that mental state, I am inclined to believe that this is a scene that I witnessed in the past, and it meshes well with other things I believe about my past, namely, that I used to go on country walks with my father as a child, and that he used to shoot rabbits during those walks. The fact that NTM accommodates our intuitions regarding embellishment cases illustrates that NTM captures the reconstructive aspect of memory; the capacity of memory to alter information about events in our past. It is a significant virtue of NTM that it makes room for this feature of memory. Consider, now, the mental state wherein I visualize a bird landing on the roof of a house; a mental state that I occupy while I am trying to paint that scene on my canvas. If NTM is right, then my mental state does not count as a state of remembering. This is in spite of the fact that the mental state at issue originates in my past perception of the scene. For it is not the case that, in virtue of occupying that mental state, I am inclined to believe that this is a scene that I witnessed in the past, and it does not mesh well with other

62  Jordi Fernández things I believe about my past, such as whether I have ever visited a house that looks like the house on my canvas. The fact that NTM accommodates our intuitions regarding cases of epistemic irrelevance illustrates that NTM captures the requirement that memories cannot be neutral on whether remembered events actually happened in the past or not. Memory informs our beliefs about events in our past, and it is a significant virtue of NTM that it respects this feature of memory. Unfortunately, NTM must also face two significant difficulties. First, NTM seems to be too strict. For it does not allow a mental state to qualify as a memory if the subject cannot fit the event represented by that mental state into the narrative of their life. And yet, a memory of a past event in our lives may, intuitively enough, be isolated from all other memories in our possession. Let us suppose that, as a small child, I once fell into a swimming pool, and not being able to swim, I briefly sank to the bottom of the pool before being pulled out of it by someone. Suppose that I can now visualize sinking to the bottom of the pool. Let us stipulate, furthermore, that I can visualize it as a result of having once experienced it. Suppose that the mental state wherein I visualize the event does convey the sense that this event actually happened to me. And yet, let us assume I cannot integrate this event within any account of the relevant part of my childhood: I cannot remember whether this happened at a public pool or at somebody’s house. I cannot remember whether it was an open-air pool during summer or an indoor pool during winter. I cannot remember who was present during the event, and I cannot remember who pulled me out of the pool. In fact, I do not know anything else about this event, either through memory or through any other source. (We could, in fact, stipulate that this remains the case after one has searched for information about the event, for example, by asking one’s parents about it.) Thus, I cannot place any other event in my early childhood as happening before or after this event. On NTM, then, the mental state wherein I visualize sinking to the bottom of the pool does not qualify as a memory. The reason is that, as far as my ability to tell a story of my childhood is concerned, this event is completely isolated from all other events I take to be in my past. As a result, it is not part of any narrative that I can construct about my life. And yet, it seems counterintuitive to say that, in this instance, I do not remember sinking to the bottom of the pool (even though I may very well think that I do not remember sinking to the bottom of the pool). Let us call cases of this type ‘isolation’ cases. Second, NTM seems to be too permissive. Even though NTM accom­ modates the requirement that memory must have the capacity to inform our beliefs about our past, it does not require memories to have any particular aetiology. And yet, it intuitively seems that mental states of a type that does not normally originate in our past experiences do not qualify as memories. Take, for instance, a subject with Korsakoff’s syndrome; a form of amnesia typically caused by a lifetime of heavy drinking. Patients with Korsakoff’s syndrome can produce detailed descriptions of events that are supposed to

The Functional Character of Memory  63 have happened on the previous day, even though the events that the patients are visualizing never happened to them. Furthermore, it is not unusual for patients with Korsakoff’s syndrome to construct sophisticated stories that include those events as being parts of their lives. The patient may, for example, sincerely give a detailed account of having recently gone on a trip and report having had an interesting conversation with a fellow traveler in a train, even though they have been in bed for weeks.28 Now, according to NTM, the mental state wherein the patient represents their conversation with a fellow traveler in a train counts as a memory of that conversation. The reason is that the patient has the ability to tell a story about going on a trip that includes this event as a part of it. And yet, it seems counterintuitive to say that the patient remembers having had that conversation in the train in virtue of having this ability. Instead, it seems more natural to say that the patient is imagining the conversation, and that the patient’s use of that scene while constructing a story of having recently gone on a trip is an instance of confabulation. Let us call cases such as this one ‘confabulation’ cases.29 It seems, then, that NTM captures two important properties of memory, namely, the fact that memory informs our beliefs about events in our past, and the fact that memory reconstructs the information that it conveys to us about those events. But it also seems that NTM ignores another important feature of memory. Part of what it takes for a mental state to qualify as an episode of remembering some event is, as confabulation cases illustrate, that there must be a robust, sufficiently reliable relation that holds between that type of mental state and the subject’s own past experiences of the event. The subject cannot be, as it were, making things up.30 Hence, the upshot of our discussion in this section is that a successful account of the type of mental state that qualifies as an episode of remembering should respect this feature of memory while, ideally, preserving the two virtues of NTM. Let us turn, then, to a proposal about the nature of remembering that is aimed at satisfying those constraints.

4  Functionalism About Memory At this point, it seems natural to try to include, within the conditions that a mental state must satisfy to qualify as an episode of remembering, some conditions that concern the aetiology of the mental state as well as some conditions that concern the impact that the mental state has on the subject’s beliefs. The framework of functionalism suggests a proposal about the nature of memories that allows us to incorporate conditions of precisely those two types. The main tenet of functionalism is that a state of a subject does not qualify as a mental state of theirs in virtue of the intrinsic properties of the state but in virtue of its association with a certain functional role. The functional role of a state is constituted by the causal relations in which the state tends to stand to perceptual inputs, behavioral outputs, and other

64  Jordi Fernández states of the subject. Accordingly, the different types of mental states that a subject has are, on the functionalist framework, individuated by reference to the various functional roles that a state can play in the subject’s cognitive economy. Now, what do ‘association’ and ‘by reference to’ exactly mean here? Functionalism comes in two different versions, depending on how one chooses to specify the way in which the functional role associated with a mental state is essential to the state belonging to a particular mental type. On one version of functionalism (its ‘realizer’ version), a state that a subject occupies is a mental state of a certain type because mental states of that type have a characteristic functional role, and the subject’s state has that role. Thus, on the realizer version of functionalism, the functional role associated with a mental state is essential to the state being a state of a particular mental type in the following sense: The state must have that functional role, or ‘realize’ it, to qualify as being a state of the relevant mental type.31 By contrast, on a different version of functionalism (its ‘role’ version), a state that a subject occupies is a mental state of a certain type because it is the property of being in some state or other with a characteristic functional role, and the subject is in some state with that role. Thus, on the role version of functionalism, the functional role associated with a mental state is essential to the state being a state of a particular mental type, but not in the sense that the mental state must have the functional role at issue. It is essential in the sense that some state of the subject must have that functional role for the subject to occupy the mental state.32 The difference between realizer functionalism and role functionalism, then, is that, on the former version of functionalism, a mental state is a first-order mental state (the state that actually plays such-and-such role) whereas, on the latter version, it is a higher-order state (that of being in some state or other which plays suchand-such role). What does functionalism have to do with memory? The proposal that I wish to put forward is that episodes of remembering are mental states that should be characterized functionally or more precisely, should be characterized along the lines of role functionalism. According to what we may call the functionalist theory of memory (or for short, ‘FTM’), for any subject S and event e, S remembers e just in case S has some mental image i such that i tends to cause in S a disposition to believe both that e happened and that S experienced e to happen, and i tends to be caused in S by having experienced e to happen. If we introduce the term ‘mnemonic role’ to refer to such a functional role, then we can abbreviate FTM as the thesis that remembering an event, or having a memory of it, consists in having a mental image that plays the mnemonic role for that event in the subject.33 FTM is a functionalist view because, according to it, a mental image of a subject qualifies as a memory not in virtue of the intrinsic properties of that mental image, but in virtue of the functional role that the mental image plays in the subject. And it is a role-functionalist view because, according to FTM, the subject’s state of having a memory, or remembering, is a higher-order

The Functional Character of Memory  65 state. FTM pulls apart, on the one hand, the property of having the specific mental image that the subject is having when they remember an event and on the other hand, the subject’s state of remembering the event. According to FTM, having the mental image at issue is different from remembering the event, since the subject could have had the same mental image without remembering the event. Consider, for example, a possible situation in which the mental image at issue plays a functional role in the subject that is different from the mnemonic role. In that situation, the subject does not remember the event despite the fact that the subject has the same mental image as that which, in the actual situation, they are having when they remember the event. Conversely, the subject could have remembered the event without having the mental image that, in fact, they are having when they remember the event. Consider, for example, a possible situation in which a different mental image plays the mnemonic role for that event in the subject. In that situation, the subject remembers the event despite the fact that the subject has a different mental image from that which, in the actual situation, they are having when they remember the event. Let us examine, now, what considerations can be offered in support of FTM.

5  Narratives, Causal Histories, and Functional Roles It seems that FTM is permissive enough, in the sense that it does not conflict with our intuitions about those cases that suggested that both CTM and NTM were too strict. Consider, first of all, the case in which I have a mental image of a black rabbit being shot even though, in the past, I saw a white rabbit being shot. According to FTM, as long as the mental image that I am having when I visualize the black rabbit being shot is an image that plays the mnemonic role in me, I qualify as remembering the event (falsely or not). And it seems that the mental image at issue does play the mnemonic role in me: On the one hand, my mental image tends to cause in me the belief that I once saw a black rabbit being shot, and it tends to cause in me the belief that my perceptual experience was veridical; that the event did take place in the past. On the other hand, my mental image is the type of image that would be produced in me by past perceptions of black rabbits being shot. To be sure, on this particular occasion, my mental image was not actually caused by a perception of a black rabbit being shot since, in the past, I did not see a black rabbit being shot. Nonetheless, the fact remains that my faculties of perception and memory are related in such a way that perceptual experiences of black rabbits do produce in me the type of mental image that I am currently having. Had I seen, in other words, a black rabbit being shot in the past, this is the type of mental image that I would be having now. Thus, unlike CTM, FTM acknowledges this case as an episode of remembering.34 Consider, now, the case in which I have a mental image of sinking to the bottom of a pool even though I am unable to relate this event to any

66  Jordi Fernández other event in my childhood. According to FTM, as long as the mental image that I am having when I visualize sinking to the bottom of the pool is an image that plays the mnemonic role in me, I qualify as remembering the event. And it seems that the mental image concerned does play the mnemonic role in me: It is the type of mental image that would be produced by perceptual experiences of being underwater in a pool and is the type of mental image that makes me believe that this event happened, and that I experienced it. Thus, unlike NTM, FTM acknowledges this case as an episode of remembering. Furthermore, FTM seems to be strict enough, in the sense that it does not seem to conflict with our intuitions about those cases that suggested that both CTM and NTM were too permissive. Consider, first, the case in which I have a mental image of a bird landing on the roof of a house, and this is a scene that I witnessed in the past, but due to a serious cognitive deficit of mine, mental images of scenes that I witnessed in the past do not produce in me any inclination to believe that they were real, and that I have witnessed them in the past. If my having the mental image of a bird landing on the roof of a house does not tend to produce in me the belief that this is a real event, and the belief that I perceived it in the past, then, according to FTM, I do not qualify as remembering the event. And it seems that the mental image at issue does not play that part of the mnemonic role in me: Due to my stipulated cognitive deficit, my mental image is not the type of image that would produce in me beliefs of those kinds. Other mental images of the same type would continue to have a null effect on my beliefs about the past. Thus, unlike CTM, FTM rules out this case as an episode of remembering. Consider, now, the case in which the patient with Korsakoff’s syndrome has a mental image of conversing with someone while traveling on a train days before, and the patient is able to construct a sophisticated story within which they can insert the scene that they are visualizing. If the patient’s mental image of the conversation in the train is of a kind that does not tend to be produced, in that patient, by experiences of such conversations, then, according to FTM, the patient does not qualify as remembering the event. And it seems that the mental image concerned does not play that part of the mnemonic role in the patient: Since the patient suffers a form of amnesia, the patient’s mental image of the conversation in the train is not the type of image that would be produced by having experienced such a conversation. In other words, had the patient experienced having a conversation with a fellow traveler in a train in the past, the patient would not have retained this experience in the type of mental image that they are having now. Thus, unlike NTM, FTM rules out this case as an episode of remembering.

6 Conclusion Role functionalism seems to have provided us with a satisfactory answer to the question of what remembering is: Remembering, or having a memory

The Functional Character of Memory  67 of, an event consists in having a mental image that plays a certain functional role in the subject. Such a functional role is backward-looking in that it involves a certain set of typical causes, and it is forward-looking in that it involves a certain set of typical effects. These two features of the proposal allow it to overcome the shortcomings of those proposals about the nature of remembering which are exclusively backward-looking; proposals such as the causal theory of memory. They also allow the functionalist proposal to overcome the shortcomings of those proposals that are exclusively forwardlooking; proposals such as the narrative theory of memory. In the end, the functionalist proposal is closer to the causal theory of memory than it is to the narrative theory, since it relies heavily on causal relations. The main difference with respect to both theories, however, is that the functionalist proposal manages to draw the line between those mental states that count as episodes of remembering and those that do not at the right point.35

Notes 1 I will use the expressions ‘remembering’ and ‘having a memory of’ indistinctly. For that reason, the term ‘memory’ will be used to refer to a faculty as well as to the mental states which are delivered by that faculty. Hopefully this will cause no confusion. Both ‘remembering’ and ‘having a memory of’ will be used non-factively. 2 In what follows, I will switch between talk of memory for events and talk of memory for states of affairs. Hopefully this will cause no confusion. By a ‘mental image’ of, for example, an event, I mean an experience wherein the event is presented to the subject. In the common case in which the event is visible (it could have been visually perceived), a mental image of the event is an experience wherein the subject visualizes the event. 3 The notion of episodic memory is introduced in (Tulving, 1972). 4 The classical version of CTM is due to Charles Martin and Max Deutscher in their (1966). 5 For details, see (Kim, 1993). 6 Alternatively, a more sophisticated formulation of CTM may explicitly mention relations of inclusion in the conditions for remembering. The CTM advocate may propose, right from the start, that for any subject S and event e*, S remembers e* just in case S is representing e*, and there is an event e such that e* is included in e, S represented e in the past, and the fact that S represented e in the past has caused S to represent e* in the present. A formulation of this kind for CTM is discussed in (Bernecker, 2010). Other relations between events, such as that of reduction in (Bernecker, 2008), might also be helpful to handle the complication for CTM discussed earlier. 7 An alternative conception of events is that of events as particulars. Can CTM still accommodate the same intuition if events are construed as particulars? As far as I can see, this may depend on whether the relevant particulars are individuated by their causal powers, or they are individuated by the spatio-temporal region that they occupy. In the former case, one can see how the CTM advocate may still be able to claim that the event that consists in the horse moving with some gait is included in the event that consists in the horse performing the canter. (One would expect, for example, that anything that were caused by the horse moving with some gait would also be caused by the horse performing the canter.)

68  Jordi Fernández In the latter case, however, it is hard to see how the CTM advocate could avail themselves of a relation of inclusion between events. For discussion on the individuation of events construed as particulars, see (Davidson, 2001). 8 Versions of this picture can be found, for instance, in (Aristotle, 1972, pp. 28–32), (Locke, 1975, pp. 149–153), (Hume, 2000, p. 12), (Broad, 1937, pp. 239–41), (Malcolm, 1963, p. 208), and (Shoemaker, 1984, p. 19). Thomas Reid also seems to endorse this view when he claims: ‘Things remembered must be things formerly perceived or known. I remember the transit of Venus over the sun in the year 1769. I must therefore have perceived it at the time it happened, otherwise I could not now remember it’ (Reid, 1969, p. 326). 9 This ability is, in turn, highly valuable from a survival point of view. If we want to avoid danger, for example, being able to recognize a situation in which we find ourselves as one that we have previously experienced to be dangerous will be advantageous to us. Likewise, if we want to find food, then being able to recognize a place in which we find ourselves as one where we have previously experienced that there is food will be advantageous to us as well. The preservative aspect of memory is, for this reason, a crucial feature of it. 10 To be clear, the difficulty for the causal theorist is not only that the event that I am now visualizing is different from the event that I represented through the perceptual experience in which my mental image causally originates. The difficulty is, more strongly, that the former event is not included into the latter event. 11 The view that memories may include details that were not originally experienced by the subject can be traced back to Frederic Bartlett’s work on constructive memory (Bartlett, 1932), and it currently enjoys wide acceptance within psychology. For a helpful survey of the relevant empirical literature, see (Roediger & DeSoto, 2015). The constructive aspect of memory highlights a further sense in which its preservative aspect may be considered highly imperfect. It may actually be considered so imperfect that the preservative aspect of memory might not be enough to motivate CTM (Michaelian, 2016). Nevertheless, these imperfections are not sufficient to disregard the preservative aspect of memory entirely, for the reasons mentioned in note 9. 12 On preservationism about knowledge, see, for example, (Dummett, 1994, p. 262) and (Audi, 1997, p. 410). 13 See (Lackey, 2005) for these cases. 14 The point that a belief which could have easily been false does not constitute knowledge can be motivated by using examples such as the classical ‘barn façade’ case in (Goldman, 1976). 15 It is hard to tell whether there are actual cases of this kind. The case of patient R.B., recently discussed by Shaun Nichols and Stanley Klein, might be a candidate (Klein & Nichols, 2012). Patient R.B. suffers, due to head trauma sustained during a bicycle accident, a remarkable cognitive impairment. Patient R.B. can have, it seems, accurate mental images of scenes from his past. And yet, patient R.B. claims, of some of those images, that they are not his memories. One possible explanation for claims of this sort is that the patient does not experience those scenes as having happened in the past when he has the relevant mental images. If this hypothesis is correct, then patient R.B. has the kind of deficit envisioned. However, it is debatable whether the hypothesis in question is correct. See my (forthcoming) for a defense of it, and (Klein & Nichols, 2012) for an alternative hypothesis. 16 This is a variation of a case discussed in (Martin & Deutscher, 1966). 17 This is true in the classical formulation of CTM. However, one could add an extra condition to the conditions for remembering posited in the classical

The Functional Character of Memory  69 formulation of CTM. One might require that, when the subject now represents an event, the representation needs to involve a specific type of phenomenology for it to qualify as a memory. This addition would have the effect of ruling out the mental image of the bird landing on the roof of the house. But this benefit would come at a price. The causal theorist would then be saddled with the task of specifying which traits of the phenomenology of memories are essential, and which are accidental. The task may not be insurmountable, but it seems worth exploring whether alternative theories about the nature of memory may not need to incur that cost. 18 The term is due to Dorothea Debus, who raises essentially this objection against CTM in (Debus, 2010). 19 For discussion of this point, see (Fernández, 2006). 20 Versions of this conception of memory can be found, for example, in (Schechtman, 1994), (Goldie, 2012), and (Brockmeier, 2015). 21 On autobiographical memory, see (Conway & Pleydell-Pearce, 2000) and (Barsalou, 1988). 22 If the narrative conception of memory is a conception of autobiographical memory, one might wonder, why is it relevant to our present project? The reason is that, if the lesson drawn from cases of epistemic irrelevance in Section 2 is correct, then episodic memories are autobiographical memories. For episodic memories must dispose us to believe that remembered events are part of our past, or belong to our lives. This means that, when the advocate of the narrative conception proposes a criterion for a mental state to qualify as a memory, the criterion being proposed will apply to all episodic memories, since it applies, more broadly, to autobiographical memories. 23 How many events should a series include in order for the series to be a component in the story of someone’s life? Strictly speaking, nothing in our pretheoretical notion of a story seems to rule out the possibility that such a series might only include two events: an event e and the subject having a mental image of e. However, it is hard to see how NTM could get any traction by presupposing such a weak reading of ‘a series of events.’ In what follows, therefore, I will assume that NTM presupposes a more robust notion of a series of events as a component of the story of someone’s life. 24 The thought here is that the temporal structure of memory, if memory is conceived as a narrative, will be a tree structure. The idea that the temporal structure of memory is a tree structure is discussed, independently of the narrative theory of memory, in (Campbell, 1997). 25 Richard Wollheim seems to endorse this view in (1984) whereas Peter Goldie rejects it in (2012). 26 The epistemic view about the function of narratives is discussed in (Schechtman, 1994) whereas the normative view is discussed in (Goldie, 2012). 27 Peter Goldie, for example, is sympathetic to the idea that a narrative of one’s life may include memories wherein the subject visualizes herself, as it were, from the outside in (2012). But it is unclear whether, according to Goldie, such memories are required for a narrative to include the subject as a character in the story. 28 For details of this phenomenon in Korsakoff’s syndrome, see (Talland, 1961). 29 Confabulation is not exclusive to Korsakoff’s syndrome. It is debatable, however, whether confabulation can take place without amnesia. For a comprehensive discussion of confabulation, see (Hirstein, 2006). 30 The intuition that the subject cannot be making things up can be captured by positing some sort of causal relation between memories and past experiences of the subject, but it does not need to be captured in this way. After all, the only thing that confabulation cases show is that the connection between the type of

70  Jordi Fernández memory of some event that the subject is having, and the type of perceptual experience of the event that the subject would have in the past cannot be random. For a noncausal proposal about such a connection which captures our intuitions about confabulation cases, see (Michaelian, 2016). 31 For an example of this variety of functionalism, see (Lewis, 1980). 32 Role functionalism is discussed, for example, in (Block, 1978). 33 Could the mnemonic role associated with a memory of some event involve, so to speak, the input side of that memory only? As far as I can see, this would reduce FTM to a version of CTM. More importantly, it would dispossess FTM of the necessary resources to accommodate, as we will see in Section 5, our intuitions about cases of epistemic irrelevance. 34 Analogous considerations apply to the case in which the extra details were in fact enjoyed by the event in the past. According to FTM, if the mental image that I am having when I visualize my father shooting while wearing a belt with a silver buckle is an image that plays the mnemonic role in me, I qualify as remembering the event. And it seems that the mental image at issue does play the mnemonic role in me: On the one hand, my mental image tends to cause in me the belief that I once saw my father shooting while wearing such a belt, and it tends to cause in me the belief that my perceptual experience was veridical; that my father was wearing that belt while he was shooting the rabbit. On the other hand, my mental image is the type of image that would be produced in me by past perceptions of silver belt buckles. To be sure, on this particular occasion, my mental image was not actually caused by a perception of a silver belt buckle since, in the past, I did not see my father’s belt at the time of the shooting. Nonetheless, the fact remains that my faculties of perception and memory are related in such a way that perceptual experiences of objects of that type do produce in me the type of mental image that I am currently having. Had I been, in other words, in a position to see the belt that my father was wearing at the time of the shooting, this is the type of mental image that I would be having now. Thus, unlike CTM, FTM acknowledges this case as an episode of remembering. 35 Versions of this chapter were presented at the 2016 International Conference on Memory (ICOM) in Budapest, the New Directions in the Philosophy of Memory 2016 conference at the University of Otago, and the Philosophy Work in Progress Colloquium at the University of Adelaide. I am grateful to the audiences at those talks for their feedback on the material presented there. For comments on a previous draft of this chapter, I am very grateful to Kourken Michaelian and Denis Perrin.

References Aristotle. (1972). De Memoria. In R. Sorabji (Ed.), Aristotle on memory. London: Duckworth. Audi, R. (1997). The place of testimony in the fabric of knowledge and justification. American Philosophical Quarterly, 34, 405–422. Barsalou, L. (1988). The content and organisation of autobiographical memories. In U. Neisser & E. Winograd (Eds.), Remembering reconsidered: Ecological and traditional approaches to the study of memory (pp. 193–243). Cambridge: Cambridge University Press. Bartlett, F. C. (1932). Remembering: A study in experimental and social psychology. Cambridge: Cambridge University Press. Bernecker, S. (2008). The metaphysics of memory. New York: Springer.

The Functional Character of Memory  71 Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press. Block, N. (1978). Troubles with functionalism. Minnesota Studies in the Philosophy of Science, 9, 261–325. Broad, C. (1937). Mind and its place in nature. London: Routledge. Brockmeier, J. (2015). Beyond the archive: Memory, narrative, and the autobiographical process. Oxford: Oxford University Press. Campbell, J. (1997). The structure of time in autobiographical memory. European Journal of Philosophy, 5, 105–118. Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobiographical memories in the self memory system. Psychological Review, 107, 261–288. Davidson, D. (2001). The individuation of events. In Essays on actions and events: Philosophical essays (Vol. 1, pp. 163–181). Oxford: Oxford University Press. Debus, D. (2010). Accounting for epistemic relevance. A new problem for the causal theory of memory. American Philosophical Quarterly, 47, 17–29. Dummett, M. (1994). Testimony and memory. In B. Matilal & A. Chakrabarti (Eds.), Knowing from words. Dordrecht: Kluwer Academic Publishers. Fernández, J. (2006). The intentionality of memory. Australasian Journal of Philosophy, 84, 39–57. Fernández, J. (Forthcoming). The ownership of memories. In M. García-Carpintero & M. Guillot (Eds.), The sense of mineness. Oxford: Oxford University Press. Goldie, P. (2012). The mess inside: Narrative, emotion, and the mind. Oxford: Oxford University Press. Goldman, A. (1976). Discrimination and perceptual knowledge. The Journal of Philosophy, 73, 771–791. Hirstein, W. (2006). Brain fiction: Self-deception and the riddle of confabulation. Cambridge, MA: MIT Press. Hume, D. (2000). A treatise of human nature (D. F. Norton & M. J. Norton, Eds.). Oxford: Oxford University Press. Kim, J. (1993). Events as property exemplifications. In Supervenience and mind (pp. 33–53). Cambridge: Cambridge University Press. Klein, S., & Nichols, S. (2012). Memory and the sense of personal identity. Mind, 121, 677–702. Lackey, J. (2005). Memory as a generative epistemic source. Philosophy and Phenomenological Research, 70, 636–658. Lewis, D. (1980). Mad pain and martian pain. In N. Block (Ed.), Readings in the philosophy of psychology (pp. 216–222). Cambridge, MA: Harvard University Press Locke, J. (1975). Essay concerning human understanding (P. Nidditch, Ed.). Oxford: Oxford University Press. Malcolm, N. (1963). Knowledge and certainty. Englewood Cliffs, NJ: Prentice Hall. Martin, C. B., & Deutscher, M. (1966). Remembering. Philosophical Review, 75, 161–196. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Reid, T. (1969). Essays on the intellectual powers of man (B. Brody, Ed.). Cambridge, MA: MIT Press. Roediger, H. L., & DeSoto, K. (2015). Psychology of reconstructive memory. In J. D. Wright (Ed.), International encyclopedia of the social and behavioral sciences (pp. 50–55). Amsterdam: Elsevier.

72  Jordi Fernández Schechtman, M. (1994). The truth about memory. Philosophical Psychology, 7, 3–18. Shoemaker, S. (1984). Identity, cause and mind. Cambridge: Cambridge University Press. Talland, G. A. (1961). Confabulation in the Wernicke-Korsakoff syndrome. The Journal of Nervous and Mental Disease, 132, 361–381. Tulving, E. (1972). Episodic and semantic memory. In W. Donaldson & E. Tulving (Eds.), Organisation of memory (pp. 381–403). New York: Academic Press. Wollheim, R. (1984). The thread of life. Cambridge: Cambridge University Press.

Part II

Activity and Passivity in Remembering

4 Remembering as a Mental Action*

Santiago Arango-Muñoz and Juan Pablo BermúdezRemembering as a Mental Action

Santiago Arango-Muñoz and Juan Pablo Bermúdez

1 Introduction There has been a recent shift in our understanding of episodic memory. Traditionally conceived as the capacity to store and retrieve information from our personal past, episodic memory is now thought of as a particular form of a broader cognitive function: that of imagining, or mentally simulating, egocentric events, whether they belong to the past or the future, whether actual, hypothetical, or counterfactual. Understanding episodic memory as part of this more general capacity for ‘mental time-travel’ allows us to account for why it often produces not a reliable reconstruction of the specific details of the past, but rather a bare-bones reconstruction of past situations that diverge from the remembered events. In this chapter we seek neither to defend this reconstructive conception of episodic memory, nor to argue that it is empirically better grounded than the traditional, preservative account (These tasks have been performed by De Brigard (2014a), Michaelian (2016), and Schacter and Addis (2007), and we refer the interested reader to these works). There are at least two ways to understand the constructive theory of memory: as episodic hypothetical thought (De Brigard, 2014a), and as episodic imagination (Michaelian, 2016), but we need not decide among them here. Rather, we assume the imaginative reconstructive conception (but analogous considerations should apply to the other version), and ask a further philosophical question: does remembering constitute a mental action? Is the imaginative reconstruction, or mental simulation, involved in episodic memory, an action? If so, what kind of agentive processes and mechanisms are at stake in remembering? We argue that in some cases remembering is indeed a mental action. But in order to argue for this we have to face objections from thinkers who worry that the “ballistic” nature of mental content production leaves no room for agency and control in imaginative processes. In particular, Strawson (2003) has argued that although there are some cases of mental actions, they take place prior to, and merely ‘set the stage’ for, the emergence of mental content. I may agentively try to remember the first time I read The Myth of Sisyphus, and these efforts to bring a certain mnemonic content to mind may count as

76  Santiago Arango-Muñoz and Juan Pablo Bermúdez mental actions, but the actual imaginative reconstruction of the situation in which I first encountered that text (images of the place, the time, the people around me, etc.) is not under my agentive control. It occurs spontaneously, in the sense of ‘involuntarily’; its production mechanisms are outside of my awareness; I could not control them, as demonstrated by the fact that if I intend to not remember the first time I read Camus’ text, I would still end up bringing (at least some of) these mental contents to mind. This is the objection from mental ballistics: imagination is a ballistic mental process, and given that episodic memory is a kind of imagination, episodic remembering is also a ballistic process that cannot be a mental action. This chapter argues against the mental-ballistics interpretation of reconstructive episodic memory. We claim that the mnemonic processes that constitute episodic reconstruction are not necessarily ballistic but can count as instances of mental action. To show this, (2) we distinguish between two kinds of agentive control: reflective and automatic. The latter is often taken to be impossible, because automaticity and control are often considered opposites; but we show that there are cases of automatic and controlled mental processes. Then we argue that (3) there are two levels of metacognition that correspond to the reflective and automatic control processes distinguished previously, and that automatic, feeling-based metacognition is a control process that can be applied to mental processes like remembering. Last, (4) by looking at the evidence for remembering as an imaginative reconstruction, we argue that this process is often characterized by particular forms of feeling-based metacognitive control, and that the presence of this control process reveals the agentive nature of episodic memory. Remembering is often a mental action, because the imaginative reconstruction processes that constitute it are imperfect, and therefore need some kind of control. To fill this need, our agentive involvement is frequently needed to manage and structure the reconstructive processes, and thus guarantee that we produce good memories as an outcome of the episodic reconstruction. When this happens, episodic reconstruction is constrained and guided by a process of feeling-based metacognitive control.

2  Mental Ballistics, Automaticity, and Control The idea of memory as reconstruction seems to imply some kind of activity: remembering requires bringing the multiple traces of the remembered experience together in a mental simulation of the event (De Brigard, 2014b; Michaelian, 2016). Although this looks prima facie like an agential process, there is a difficulty: a number of lines of cognitive science research have shown that most of the mental processes that shape our daily lives take place automatically; they are fast, associative, working-memory-independent, and therefore apparently produced outside of effortful cognitive control (for general overviews see Evans (2010b) and Kahneman (2011)). Thus, although the reconstructive account suggests that memory is an agentive process, it still has to confront the “threat of automaticity” (Wu,

Remembering as a Mental Action  77 2013). Although remembering is a reconstruction, this reconstruction seems to take place automatically without the subject’s agential intervention. Wayne Wu remarks that “automaticity is what makes decisions about mental agency controversial” (Wu, 2013, p. 244). Most philosophers consider memory to be a more or less automatic capacity. In a recent paper, for example, Andy Clark claims that “ordinary biological memory, for the most part, functions in a kind of automatic, subterranean way” (Clark, 2015). If this is so, then calling ‘remembering’ an action is highly controversial. The core philosophical issue here is that automatic processes seem to be by their very nature nonagential. After all, we tend to call ‘automatic’ those processes that are invariably triggered by the same kind of stimuli and respond to them in systematically the same way. They are rigid, difficult or impossible to correct, and insensitive to novel evidence. In a word, automatic mental processes seem to be ballistic: they “run to completion once triggered and cannot be aborted in midcourse” (Stanovich, 2004, p. 39). Add to this what Wu calls the “simple connection” between automaticity and control: “automaticity implies the absence of control . . . by the subject” (2013, p. 246). This traditional view has been questioned empirically, but it retains a powerful hold over our theoretical intuitions. The simple connection entails that agentive control consists in the deployment of top-down attention and cognitive effort toward the attainment of a goal, and that automatic processes, being fast and effortless, are devoid of agentive control and can participate in it only to the extent that they are subsumed under higherorder, cognitively effortful mental processes. According to this view, if remembering was to be considered an action, the subject would have to concentrate, pay attention, avoid distracting elements, try mnemonic strategies intentionally, gather cues . . . and all this would imply a certain amount of nonautomatic attention, consciousness, and mental effort, even if not particularly “strenuous” (Mele, 1997). But even if this happens, even if the preparatory ground-setting steps are cognitively effortful and controlled, the very process of imaginative reconstruction seems to work spontaneously and subpersonally: when I remember what I had for dinner last night, I cannot really try to remember: the memory traces are assembled and presented as if by magic; all I can do at the personal level is try to bring it about that I remember, conjuring the content, and wait for it to arrive (Mele, 2009).1 Thus, if automatic processes are ballistic, there is nothing particularly agential about them. 2.1  The Mental Ballistics Argument: From Imagination to Memory Given this view of automatic processes as ballistic, Strawson concludes that the space for action in the mental is rather small: “the role of genuine action in thought is at best indirect. It is entirely prefatory, it is essentially— merely—catalytic” (2003, p. 236). The mental phenomena that can be properly called actions are those that ‘set the stage’ for the emergence of mental content, but not the content’s emergence itself. Setting myself

78  Santiago Arango-Muñoz and Juan Pablo Bermúdez up for tackling a problem may be a mental action; the actual solution’s appearance in my mind is not. Bringing my attention back to a task after I have been distracted may be a mental action; the subsequent steps involved in completing the task probably are not. This also applies in the case of imagination. If I ask you to imagine a pink elephant gracefully walking on a rainbow, you can do this immediately and intuitively, in a way that suggests agentive control: you can picture it big or small, you can make it dance, smile, or wave its hat. This sense of agentive control over imagination is, however, set in a different light if I ask you not to imagine a pink elephant at all. Now it is harder to make imagination comply: you may try to bring it about that you do not imagine the elephant (by occupying your mind with something else), but you cannot directly try to not imagine the elephant. This reveals that the actual imaginary content production is a rather ballistic and rigid process that takes one input and automatically produces content in a way that is not responsive to the agent’s intention, structure, and content (perhaps the experience of control in the first case was due to the ease with which the content was produced, rather than with proper intentional control). Thus, the imaginative production of mental content seems not to be a case of mental action. If so, the only agentive part of the process would be the initial mental stage-setting. Applied to imagistic production, the argument from mental ballistics goes like this: i) If M is a mental action, then I can control the occurrence of M by both intentionally trying to perform M and intentionally trying not to perform M. ii) I cannot control the occurrence of P (the imaginative production of a mental content) by intentionally trying not to perform P. iii) Therefore, P is not a mental action. Now, if episodic memory is a kind of imagination, would the same argument apply to it? If we ask you to remember what your childhood bedroom looked like, it seems like you can exert swift, even effortless control over this process, focusing on different aspects of the room (the floor, the positions of the bed and other furniture, the lighting. . .). But if we ask you not to remember your childhood bedroom, the recollection process becomes harder to control. Mnemonic contents, like imaginary ones, seem to be produced automatically after the reception of a relevant stimulus, and therefore, ballistically and outside of the subject’s agentive control. 2.2  The Simple-Connection Reply to the Mental Ballistics Argument Here is a way to reply to these arguments: content generation may be automatic and ballistic, but it is nevertheless a constitutive part of mental action whenever the automatically generated content is triggered by,

Remembering as a Mental Action  79 and responsive to, the agent’s occurrent intention. Call this the ‘simpleconnection reply,’ since it argues for imagination, deliberation, and the like being mental actions, but does so without putting the simple connection (i.e., automatic processes are not controlled) into question. Briefly, the strategy is as follows (Wu, 2013): mental action requires the production of a specific mental content out of many possible ones (in trying to remember my childhood bedroom, I could end up generating images of my current bedroom, other rooms of my childhood home, someone else’s childhood bedroom, and so on); and so mental action requires selecting the right content to be generated. Thus, a certain mental process counts as an action if the produced mental content corresponds to the intention’s representational content, since this implies that the agent has selected the right content, i.e., has directed her attention to the proper content. This can occur only if the agent’s occurrent intention plays a top-down causal role in directing attention toward the right content. And to the extent that her intention plays this causal role, this process is an instance of agentive control, regardless of whether some of its subprocesses are automatic and ballistic. We think, however, that this simple-connection strategy falls short of fully answering the problem of mental ballistics, for two reasons. First, although we agree that intentions often play this top-down role, the account so far does not explain how intentions do precisely that. How does intention’s capacity to coordinate and structure automatic processes work? In other words, what makes automatic processes—otherwise blind, ballistic, and unintelligent—responsive to the intention’s content, and able to yield adequate results? Although the top-down causal role of intentions should certainly be a part of the story of mental agency, there is another part of it that is missing, i.e., the part that makes automatic mental content-generation processes susceptible to being recruited by higher-order processes. And second, if we have no story to tell about the agent’s role in guiding her automatic processes, then we must still accept Strawson’s view that the agent generates an intention, holds it in mind, and then simply waits for the automatic processes—over which she has no control—to generate the relevant content. This fails to agree with our phenomenology of mental action: not only do we experience that we can bring the relevant mnemonic content to mind; we also feel like remembering is sometimes easier, sometimes harder; we feel that a given memory is more or less accurate; sometimes we feel that we can remember something if we try harder, whereas other times we simply know that we will not be able to remember, no matter how hard we try. All of these phenomena suggest that there is more to agentive mental control than simply holding an intention in mind and waiting for the content. They suggest that we control not only the top-down initiation of remembering, but also (aspects of) the bottom-up production of mental content. If this is right, and if content production is an automatic process, then there may be a more thorough reply to the mental-ballistics argument than

80  Santiago Arango-Muñoz and Juan Pablo Bermúdez the simple-connection account. And we think that we should let go of the simple connection because there is empirical evidence that some automatic mental processes are also controlled. 2.3  Can Automatic Processes Be Controlled? Psychological research has traditionally considered the concepts of automaticity and control as opposites.2 One of the strongest traditional pieces of evidence for the uncontrollability of automatic processes comes from Stroop-type effects, where unattended dimensions of a stimulus interfere with the attention-demanding task that subjects attempt to perform (MacLeod, 1991; Stroop, 1935). Findings like these suggest that automatic processes are ballistic: the relevant input triggers them almost invariably, and once activated, they run rigidly to completion. If automatic processes are ballistic, they are not in themselves controlled or controllable. We present now some evidence that challenges the ballistic interpretation, showing that some automatic processes do not respond invariably to the same stimulus, or run rigidly to completion once triggered, but rather display contextual and normative sensitivity. 2.3.1  Automatic Processes and Context Sensitivity The traditional view that automatic processes are reflexlike, activated by the mere presence of a given input, carries its influence to this day (Bargh, Chen, & Burrows, 1996; Gendler, 2008a).3 And yet, no sensible account, no matter how mechanistic, could deny the power of context. Stroop effects themselves have been found to be modulated by contextual features, like how many dissonantly colored characters each word contains, the participant’s direction of attention, and the goals of the task. This suggests that the context that triggers the activation of a given automatic process can include not only external features of the environment, but also the agent’s own cognitive dispositional and occurrent states. In fact, there is ample evidence that, contrary to the ballistic interpretation, automatic process activation is affected by “where the current focus of conscious attention is, what the individual was recently thinking, or what the individual’s current intentions or goals are” (Bargh, 1997, p. 3).4 The conditionality of automaticity is so broad in scope that conceiving automatic processes as ballistic is a misdescription. If reflexes are indeed triggered by the mere presence of the relevant stimulus, then this is a way in which (at least some) automatic processes are unlike reflexes, since in many cases there is nothing like a clearly identifiable stimulus whose mere presence almost invariably activates the same automatic process. The latter, on the contrary, are sensitive not only to a specific triggering condition, but also to the agent’s motivational states and occurrent goals.

Remembering as a Mental Action  81 2.3.2  Automatic Error Detection and Correction Several researchers have proposed that automaticity’s core trait is its being unstoppable once initiated. This seems phenomenologically correct: when we see a familiar word, we cannot help but read it, and when we see a person’s face, we cannot but recognize it as such.5 It thus seems that automatic processes are recalcitrant to available evidence, so that their behavioral output cannot be cancelled or modified even when we have clear evidence for its inappropriateness. Gendler (2008a, 2008b) mentions examples like the fear that even wise persons feel when hanging from a precipice despite being certain that the cage they are suspended in is completely safe, and the unavoidable disgust caused by the prospect of eating delicious, yet feces-shaped, chocolate fudge. Such epistemic insensitivity is yet another reason to consider automatic processes uncontrollable, since they cannot be corrected mid-performance on the basis of updated information. Yet phenomenology also suggests that automatic processes are often selfcorrecting. When you over-squeeze a plastic cup, you immediately readjust the strength applied by your hand. As you walk, run, skate, cycle, etc., you automatically adjust your posture on the basis of visual and vestibular cues, in ways so complex that explicit calculation would be unable to specify, relying instead on automatic processes acquired through practice. In fact, automatic error-correction is so fast and efficient that oftentimes we are completely unaware of its occurrence. Koch and Crick (2001, p. 893) discuss a study in which participants must move their eyes and fingers rapidly toward an appearing light at their visual field’s periphery. They reliably do this even if the light moves a bit to the left or right as their eyes move toward it, and interestingly they do not report the light having moved. Thus, automatic processes can display an ability to adapt to enviromnental changes, to detect and resolve tensions present within the dynamic stream of associative activity (Brownstein & Madva, 2012; Rietveld, 2008). You probably just performed an automatic error correction in reading this paragraph, by reading ‘environmental’ instead of ‘enviromnental.’ This automatic normativity is not proper to bodily action: it is also present in mental action. In Walsh and Anderson’s (2009) paradigm, subjects were confronted with multiplication problems they could solve using one of two strategies: mentally calculating or using a calculator machine. In selecting the strategy subjects had also to consider a delay of the calculator machine (4 seconds). There were three types of problems: easy, intermediate, and difficult; and two conditions: with calculator delay and without delay. The rate time, the accuracy, and the cursor movement from the starting place to the calculator or to the answering box were recorded. Participants quickly initiated a movement corresponding to an initially favored strategy, and then decided whether to complete the problem using that strategy (or shift to

82  Santiago Arango-Muñoz and Juan Pablo Bermúdez another) by conducting a more thorough evaluation while moving the cursor from one place of the screen to another. They often redirected the initial movement of the cursor from one strategy to the other showing “imperfect sensitivity to the current problem”; but then the “commitment to a specific strategy, which occurred later, reflected nearly perfect sensitivity to the profitability of mental and calculator solutions” (Walsh & Anderson, 2009, p. 345). Participants showed an adaptive behavior by performing mental calculations less frequently as the difficulty increased, and more frequently in difficult cases when the calculator was delayed. Strategy selections were extremely fast and depended on the interaction between problem difficulty and calculator responsiveness. Moreover, there’s also fast automatic error detection in mental action. In a recent study, Fernández, Arango-Muñoz, and Volz (2016) designed an experiment to test subjects’ awareness of their errors in fast mental calculation. Participants were presented with a triplet, and they had to estimate whether the number in the middle was the arithmetic mean of the two other numbers (e.g., 2 4 8) by a Yes/No answer. Participants were instructed to press the Yes/No buttons as fast as possible and then report whether they had a feeling of error as fast as possible using again the Yes/ No buttons. Time pressure was exerted to prevent reflective control and analytical thought. Interestingly, the feeling of error reports was strongly correlated with actual arithmetic errors; in other words, subjects reported having a feeling of error mainly when they had actually committed an error. Additionally, the experimenters tested participants’ confidence in their answers when they did not report feelings of error. Surprisingly, in these cases participants reported less confidence for wrong answers than for right ones, suggesting that in cases where subjects did not have a reportable feeling they still had a lower-level awareness of their errors (the feeling of error was in ‘fringe consciousness’ (Norman, Price, & Duff, 2010)). So we have automatic error-detection systems, which sometimes produce an explicit feeling of error. Add to this that we are able to not only detect, but also correct errors automatically. In a study that compared people’s abilities to correct, report, and recall errors (Rabbitt, 2002), participants were instructed to look at a screen split into four squares, and press the corresponding button when a dot appeared in each square. Group 1 was instructed to immediately correct the mistakes they made during the task; group 2 was instructed to press a fifth button every time they had a feeling of error; group 3 was randomly interrupted and asked whether they remembered having made a mistake in their last three responses; and group 4 was told to simply ignore all errors. The stimulus’ duration (the Response Signal Interval, or RSI) varied randomly throughout the task between 150ms and 1s. The first relevant finding was that for all participants the response that followed a mistake was slower than the previous ones. This slowing occurred even when they were unable to report or recall the errors. Further, participants

Remembering as a Mental Action  83 in group 1 were remarkably fast and accurate in correcting their errors: throughout all RSI’s, errors were on average faster than correct responses, and error corrections were on average even faster than errors. Additionally, participants displayed much more accuracy in error correction than in error report or error recall. Across all conditions, it took participants much longer to report an error than to correct it, and failed to report errors much more often than they failed to correct them. Given that error correction can occur notably quickly (as quickly as 40ms after the error is performed (Rabbitt, 1966a, 1966b)), error correction cannot depend on reflective recognition and rectification processes; it must therefore be automatic. Crucially, these automatic error corrections always increased response accuracy, and never turned a right response into a wrong one (Rabbitt, 2002, p. 1082). Thus, automatic processes (both mental and bodily) can be sensitive to error, and able to perform quick and efficient self-correction without the intervention of slower and coarser-grained higher-order processes.6 More specifically, the aforementioned evidence suggests that we can perform fast and intuitive error-detection and -correction processes that are independent from the effortful and cognitively effortful processing that relies on working memory. Following dual-process accounts of human cognition, we distinguish here between automatic or intuitive processes that can be performed independently from working memory, and reflective processes whose performance requires the use of working memory (Evans, 2010a; Nagel, 2014).7 Here ‘working memory’ (Baddeley, 2007) refers to the set of higher-order cognitive capacities that allow for the mental manipulation of task-oriented representations. Thus, there are two different ways of exerting cognitive control over our behavior: one of them is reflective control—the often slow, effortful top-down control that we exert by recruiting working memory in novel or attention-demanding tasks. The other is intuitive or automatic control— the fast, rather effortless, and intuitive control that we exert automatically, without the intervention of working memory. The cases discussed previously reveal that automatic processes are not (at least not necessarily) ballistic, evidence-insensitive phenomena. It thus seems too harsh to say that automatic processes are “unable to keep pace with variation in the world or with norm-world discrepancies” (Gendler, 2008b, p. 570), even if they do behave that way in some cases, or with respect to certain kinds of evidence—particularly when it is in the form of novel propositional knowledge. We should rather agree with Brownstein and Madva in claiming that automatic processes “can be norm-sensitive in virtue of their responsiveness to affective states of disequilibrium. Responsiveness to such affective states is flexible, self-modifying, and . . . a genuinely normative phenomenon” (Brownstein & Madva, 2012, p. 428). Those affective states of disequilibrium can, of course, be misguided with respect to the overall available evidence (as in the case of the trembling wise man hanging over a precipice from a perfectly safe cage), but need not

84  Santiago Arango-Muñoz and Juan Pablo Bermúdez be so: affective states of “felt tension” or “directed discontent” can be a part of reliable dynamic online error-detection and -correction systems. If we use such mechanisms to control mental action, they deserve the name ‘automatic metacognition.’

3  Automatic Metacognition 3.1  Control as the Mark of Action Inspired by neuroscientific accounts of motor control (Jeannerod, 2006; Wolpert & Ghahramani, 2000; Wolpert, Ghahramani, & Jordan, 1995), recent theorists of action have centered their definitions of action on processes of control (Bermúdez, 2017; Hopkins, 2014; Mossel, 2005; Pacherie, 2008; Proust, 2005; Shepherd, 2014; Wu, 2016). From this viewpoint, performing an action requires not just consciously intending or trying to do something, but also doing it in a controlled way, exerting control over the production of the bodily or mental events.8 This perspective on action nicely dovetails with the data on automatic control summarized so far. In the cited examples, the subject adjusts, corrects, modifies, or simply acts in a given way while exerting an automatic form of control over performance. Even though the subject controls her behavior without a need for reflection, second-order thoughts, and/or metarepresentational awareness, her action is sensitive to some normative constraints. Therefore, given the norm-sensitivity of these behaviors, the agent can be considered to be exerting agentive automatic control while performing them. So, the next question that we should confront is whether there is a similar form of automatic control over mental actions, particularly over remembering. 3.2  Mental Action and Metacognition Although the motor control theory works well in the case of bodily actions, we should resist the temptation of claiming that the control of mental action is carried out by the motor system, as some philosophers have suggested (Campbell, 1999; Peacocke, 2007). It is unlikely that you have to rely on motor images about your own body dispositions, as the motor systems does, in order to control your mental performance (Carruthers, 2009a; Proust, 2009). No bodily movement can fulfil a mental action’s satisfaction conditions. For example, your mental attempt to retrieve a telephone number does not correspond to any bodily movement or to any physical event, because you can mentally retrieve the telephone number without moving any muscle or changing anything in the world. Therefore, the direction of fit of remembering is not world-to-mind, as in the case of bodily intentions (Searle, 1983), since it is not necessary that it should involve a change in the

Remembering as a Mental Action  85 world (outside your mind) to be satisfied. This fact is basically explained by the nature of mental actions in contrast with bodily actions: the former aims at producing an epistemic, emotional, attentional, or motivational change in the mind, whereas the latter aims at producing a change in the body and/or in the world (Kirsh & Maglio, 1994). Thus, a first distinction between bodily action and mental action refers to their goal and the kind of control the agent must exert to perform it. Whereas in bodily action we aim at changing our body posture and/or the world, and use mainly motor control, in mental action we aim at changing the mind and use metacognition.9 Metacognition is the capacity to monitor and control mental processes and dispositions (Proust, 2013). 3.3  Two Levels or Types of Metacognition The classic view is that metacognition is thinking about thinking, i.e., forming metarepresentations, or second-order thoughts, about first-order mental states (Flavell, 1979; Nelson & Narens, 1990). Accordingly, metacognition is just turning the mindreading capacity toward oneself, and deploying mental concepts to produce self-ascriptions (Carruthers, 2009b, 2011). This type of metacognition has been called “theory-based metacognition” (Koriat, 2000), “high-level metacognition” (Arango-Muñoz, 2011), or “system 2 metacognition” (Proust, 2013; Shea et al., 2014) in analogy with system-2 control processes. In contrast, recent studies have proposed that there is a leaner form of metacognition that does not require consciousness, theory of mind, or mental concepts, and operates implicitly: “Much cognitive control takes place outside system 2” (Shea et al., 2014, p. 188). Metacognition includes the mental capacity to monitor and control mental processes implicitly (Shea et al., 2014), or by means of metacognitive feelings (Arango-Muñoz, 2011, 2014; Proust, 2009, 2013).10 This type of metacognition has been called “experience-based metacognition” (Koriat, 2000) or “system-1 metacognition” (Proust, 2013; Shea et al., 2014).11 “System-2 metacognition” is widely accepted, so we will only briefly discuss evidence for “system-1 metacognition,” which comes mainly from three domains: 1) comparative psychology has shown that animals devoid of mindreading capacity and mental concepts pass metacognitive tasks: Rhesus monkeys and bottlenose dolphins are able to monitor and control their perceptual and memory capacities (Hampton, 2001; Smith, 2009; Smithet al., 2006; Smith & Washburn, 2003). 2) Psychologists have found that subjects very often monitor their mental processes based on metacognitive feelings and heuristics, and not necessarily on theoretical information (Koriat, 2000). A growing literature shows the important role of metacognitive feelings in the control of mental processes (Fernández Cruz, Arango-Muñoz, & Volz, 2016; Koriat, 2000; Schwartz & Metcalfe, 2010). Finally, 3) there seem to be differences in the neural activity related to mindreading (system-2

86  Santiago Arango-Muñoz and Juan Pablo Bermúdez metacognition) and system-1 metacognition (Proust, 2012; Schnyer et al., 2004). According to some studies, mindreading is correlated with neural activity in the right temporal-parietal junction, the prefrontal antero-medial cortex, and anterior temporal cortex (Del Culet al., 2009; Perner & Aichorn, 2008); whereas system-1 metacognition is correlated with neural activity in the dorsolateral prefrontal cortex, the ventro-medial prefrontal cortex, and the anterior cingulate cortex (Schnyer et al., 2004). As argued before, whenever we exert agentive control over our cognitive processes through metacognition, we perform mental actions. So, if remembering is an action, we should exert control over our episodic reconstruction processes by means of one of these two kinds of metacognition. One can control one’s remembering either by system-2 metacognition (i.e., thinking about what one is remembering and making some metacognitive inferences about it), or by system-1 metacognition (i.e., monitoring and controlling the process of remembering by means of metacognitive feelings). However, since system-2 metacognition is cognitively very demanding and costly, most of the metacognitive control is feeling-based. That is the reason why we will focus on system-1. In what follows, we argue that system-1, feelingbased metacognition allows us to exert agentive control over the contentproduction aspect of episodic memory processes.

4  Feeling-Based Metacognition and Episodic Memory 4.1 Episodic Memory as the Imaginative Reconstruction of the Personal Past The reconstructive theory of memory claims that memory is not the passive retrieval of stored representations, but rather the active reconstruction of a representation of a past episode: “Remembering is a matter of imagining or simulating the past” (Michaelian, 2016, p. 60). However, it is also worth noting that reconstruction takes place not only during retrieval, but also during the encoding stage of the process, in which memory selects what to retain and how to retain it, often encoding only a gist of the remembered event (Schacter & Addis, 2007). If remembering is the action of imaginative reconstruction of a past experience, this opens the door for many mnemonic errors such as false recognition (Roediger & McDermott, 1995), boundary extension (Intraub, Bender, & Mangels, 1992), the superportrait phenomenon (Rhodes, 1996), and confabulation (Michaelian, 2011). Misremembering occurs when a reliable mnemonic system produces a false or inaccurate representation of the past. Reconstruction theorists highlight that, given the constructive character of memory, misremembering is a systematic and ordinary occurrence in our daily lives (De Brigard, 2014a). Given the imperfect character of memory outputs, some kind of control is needed to ensure the reliability of memory. This is not the preparatory,

Remembering as a Mental Action  87 stage-setting control of mental action discussed previously (trying to bring it about that I remember X); rather, what is required here is control over the automatic imaginative content-production processes and its reliability (trying to remember correctly). Metacognition, the control of mental processes and dispositions, achieves this largely by means of metacognitive feelings. 4.2  How Metacognitive Feelings Guide Episodic Reconstruction As we said earlier, memory has a rich phenomenology: recollection is often not an immediate mental event, but a lengthy process, and while it is taking place we experience not only that we can bring the relevant mnemonic content to mind; we also feel like remembering is sometimes easier, sometimes harder; we feel that a given memory is more or less accurate; we feel that we will be able to remember something if we try harder (e.g., in the tip-of-the-tongue phenomenon), or we simply know that we will not be able to remember, no matter how hard we try; sometimes we feel that we are forgetting something we should remember; and so on. All these phenomena are metacognitive experiences that convey some information about the way the episodic reconstruction process is going, whether it is running smoothly or it has found some obstacles, and how serious these obstacles are. As mentioned previously, these phenomena suggest that there is more to episodic memory control than simply holding an intention in mind and waiting for the right content to come. On this respect, Souchay et al. remark that “presumably on-line feelings and thoughts generated during retrieval by the object level are monitored by the meta-level, leading to the implementation of mnemonic strategies, termination of search, and so on” (2013, p. 1). These phenomena suggest that we not only have control over the global act of remembering, but also that, with the guidance of metacognitive feelings, agents have some level of control over the reconstruction process. Thus, metacognitive feelings enable the agent’s strategic involvement in the mnemonic process, bolstering the reliability of the mental content produced. The feeling of knowing (FOK) is one of the most interesting and puzzling experiences related to memory. Sometimes when you are faced with a memory problem, even before trying to retrieve the solution, you already feel that you know it, that you will be able to reconstruct the memory. In the domain of semantic memory, it has been shown that FOKs play a central role in deciding whether to remember a piece of information or try another strategy to retrieve the information (Arango-Muñoz, 2013; Paynter, Reder, & Kieffaber, 2009; Reder, 1987). Although FOK has been less studied in the domain of episodic memory,12 it seems likely that this experience provides the subjects with a sense of whether they would be able to reconstruct a piece of information or not. When asked whether she will be able to remember the events that took place at her school graduation, the feeling of knowing will motivate the subject to give a positive answer and to start a reconstruction attempt.

88  Santiago Arango-Muñoz and Juan Pablo Bermúdez Subjects also feel that remembering is sometimes easier or harder. The feeling of ease has been shown to play a fundamental role in the monitoring and control of memory retrieval. According to Koriat’s FOK accessibility model (1993, 2000), the accessibility of partial or contextual information relevant to the memory target gives rise to this feeling. For example, immediate or quick access to a song or anthem’s first lines gives rise to the feeling of knowing it. Think about what happens when you want to tell people a joke you heard earlier: you mentally rehearse the first few lines and the punchline, and if they come to mind easily, this feeling may motivate you to start telling the joke, with the (often confirmed) certainty that you will be able to reconstruct the rest. But if these traces do not come to mind quickly, you delay the start of the joke until you successfully retrieve enough of them. (This is a skill that must be learned, which explains why children way too often ruin jokes by starting to tell them, and only later realizing they have forgotten the punchline.) Conversely, a person may have a feeling of uncertainty, of not knowing or forgetting, which may motivate her to give up on the mnemonic reconstruction rapidly. We also feel that a given memory trace is more or less accurate, which leads us to structure the recollection process in different ways. Some remembered information feels right, and some feels wrong. It has been found that the fluency with which a memory is reconstructed is a key determinant of the feeling of rightness. “Fluency” refers to the ease with which information is reconstructed and thus has been mainly measured through reaction time, i.e., the speed of the mnemonic reconstruction. A memory that is reconstructed fluently (i.e., quickly) feels right, whereas a reconstruction that isn’t fluent feels dubious (Benjamin, Bjork, & Schwartz, 1998; Kelley & Lindsay, 1993; Whittlesea & Leboe, 2003). When episodic retrieval is accompanied by a feeling of wrongness or doubt, this tends to motivate the agent to revise the outcome until the outcome feels more accurate (Gallo & Lampinen, 2015). The feeling of forgetting also plays an important role in remembering. When we try to mentally reconstruct lists of items, scenes, or to-do lists, we often have a feeling that we have forgotten something (Halamish, McGillivray, & Castel, 2011). This feeling motivates the subject to revise the mnemonic reconstruction, to seek confirmatory cues and traces, and to check if something is missing; this feeling also casts doubt on the memory’s integrity. All these metacognitive feelings are of vital importance in the agent’s production and evaluation of the mnemonic reconstruction, and her further decision of whether to endorse the reconstructed information (Michaelian, 2012). Our claim is that when a subject is remembering guided by these metacognitive feelings, she is actually exerting agentive automatic control over her mental mnemonic processes, since said feelings motivate the subject to resolve felt tensions and guide her mnemonic reconstruction toward reliability. Thus, given the centrality of agentive control for action, when the agent controls her mental processes of episodic reconstruction by means

Remembering as a Mental Action  89 of metacognitive feelings, remembering counts as a mental action. Further, in feeling-based remembering the metacognitive and the reconstructive processes jointly constitute the action, since it is through metacognition that the agent manages to structure the automatic reconstructive processes toward the production of a reliable memory that she can then endorse.

5 Conclusion The main reason for some philosophers to consider that memory is just a passive information retention and retrieval capacity is its automaticity. The apparent “ballistic” nature of the production of mental content seems to leave no room for agency and control. In opposition to this view, we have defended an active account of memory according to which remembering can count as a mental action. Our claim differs from a “simple-connection” view because we hold that agents are able not only to control the top-down initiation of remembering, but also to structure the content-production process, monitoring and guiding it by means of intuitive metacognitive feelings. Memory, as an imaginative reconstruction of past experience, frequently requires some kind of metacognitive control to ensure its reliability. Drawing from recent work on the normativity of automaticity and automatic control, we distinguish two kinds of metacognitive control: top-down, reflective control, and automatic, intuitive, feeling-based control. We propose that whenever agents control their mental remembering processes by means of intuitive or feeling-based metacognition, remembering is an action. Before ending, it is worth making some clarifications. First, our view does not entail that metacognitive feelings are perfect guides to reliable memories. We often misattribute reliability levels to our memories, and some people are surely better than others at producing reliable memories; but to the extent that we can intentionally improve the reliability of our recollections through purely internal mental processes, it is largely thanks to our metacognitive feelings. Second, we do not want to exclude high-level metacognition as another source of mnemonic agency: there certainly are some high-level inferential processes at play in some of our remembering routines (like the ones present in source monitoring). However, given our tendency to minimize executive control processes, most everyday mnemonic control processes surely are system-1 processes (Shea et al., 2014, p. 188), so low-level metacognition seems to be more common than high-level metacognition, covering a wider scope of cases. Third, we do not claim that all memory-like episodes are mental actions: there are episodic mental happenings or events that are not agentive, like the sudden involuntary rehearsal of the song I heard yesterday. Our claim is that those mental events that are monitored and controlled by agents via feeling-based metacognition are actions; certainly higher-order metacognition can also guarantee that some other mental processes are actions too.

90  Santiago Arango-Muñoz and Juan Pablo Bermúdez

Acknowledgments The authors would like to express their gratitude to Felipe de Brigard, Kirk Michaelian, Markus Werning, Denis Perrin, Wayne Wu, Michael Brent, Flavia Felletti, Santiago Amaya, Sam Murray, and two anonymous reviewers for their thoughtful, often very detailed, and tremendously helpful comments and suggestions.

Notes * This research was supported by the “Metacognitive Feelings and the Cognitive Phenomenology Thesis” research project funded by the CODI, Cód. 20167047, Acta 01 de 2016, Universidad de Antioquia, and by a research fellowship awarded within the framework of the “Getting Better at Simple Things: Understanding and improving vigilant control” project funded by the John Templeton Foundation. 1 This distinction between “trying to X” and “trying to bring it about that X,” plus the claim that one can do only the latter with respect to memory, led Mele to revise his earlier view (Mele, 1997) and conclude that remembering is never a mental action (Mele, 2009). 2 The picture has become much more complicated than a duality of opposites. However, the original description of the concepts as opposites (Schneider & Shiffrin, 1977; Shiffrin & Schneider, 1977) still bears influence on contemporary research, and is therefore worth taking as a starting point for conceptual analysis. 3 Schneider and Shiffrin defined an automatic process as an activation of a series of neural nodes in which “[t]he sequence of nodes (nearly) always becomes active in response to a particular input configuration, where the inputs may be externally or internally generated and include the general situational context” (Schneider & Shiffrin, 1977, p. 2). The view has remained influential: “The essence of TASS [=intuitive] subprocesses is that they trigger whenever their appropriate stimuli are detected, that they cannot be selectively ‘turned off’ ” (Stanovich, 2004, p. 52). In a review of literature on automaticity in social behavior, Bargh et al. (1996, p. 252) find that “[r]ecent research has shown that attitudes and other affective reactions can be triggered automatically by the mere presence of relevant objects and events, so that evaluation and emotion join perception in the realm of direct, unmediated psychological effects of the environment.”—a view that Gendler (2008a, p. 644) quotes approvingly. 4 For the context-dependence of Stroop effects, see Francolini & Egeth (1980), Kahneman & Henik (1981), and Bresner & Stolz (1999). These studies contributed to the rejection of the traditional view that automaticity is attentionindependent (LaBerge & Samuels, 1974; Posner & Snyder, 1975; Schneider & Shiffrin, 1977; Shiffrin & Schneider, 1977). 5 For Shiffrin & Schneider, “[s]ome automatic processes may be initiated under subject control, but once initiated all automatic processes run to completion ­ ­automatically” (Shiffrin & Schneider, 1977, p. 160). Norman and Shallice (1986) consider that, since automatic processes run to completion once triggered, any kind of error-correction requires deliberate attentional control. Bargh refers to this as automaticity’s “autonomy” (1992, p. 186), and suggests that it is its one core feature. For Stanovich, “TASS [i.e., intuitive] processes cannot be turned off or interfered with by central systems. Their operation is obligatory when triggered by the relevant stimuli. . . . TASS processes tend to be ballistic—they run to completion once triggered and cannot be aborted in midcourse” (Stanovich, 2004, p. 39). (See also Logan & Cowan, 1984; Moors & De Houwer, 2006, pp. 301–302).

Remembering as a Mental Action  91 6 For other cases of automatic error detection and correction in sensorimotor processes, see Gordon & Soechting (1995), Rabbitt (1990), and Logan & Crump (2010). 7 Notice that this dual-process approach is significantly different from coarser dual-system approaches. (On the general outlook, and the difference between dual-system and dual-process accounts, see Evans (2008, 2010b); Evans & ­Stanovich (2013)). 8 Hopkins (2014) has recently appealed to the notion of control to claim that remembering is a mental action. However, we are not using the same notion of control. His notion of control refers to a type of external causation or determination of the external past events upon the remembered content (“the past must somehow control the way I now represent it. (. . .) So my representing must be directly determined by the episode: how I represent it is causally controlled by how it was, and not via the mediation of any other conscious state” p. 324). Our notion of control refers to metacognitive control, and in this sense, is internal. 9 This is obviously an oversimplification: one may intend to stay still, and in this case, there would not be any bodily movement involved. 10 The implicit and the feeling-based views of the low-level metacognition are different. The former considers that there can be control in the total absence of consciousness, whereas the latter proposes that feelings are the outputs of metacognition and influence control processes. The feeling-based view is especially relevant for discussions of mental action. It is these feelings which allow us to claim that automatic error-detection and -correction processes, like the ones discussed in Section 2, are at work in many situations and help the subject to detect and correct her errors: just like you are able to correct your posture when you ride your bicycle through vestibular feelings of balance, similarly you are able to correct your recollection process when you try to remember through metacognitive feelings of knowing, of forgetting, of fluency, and so on (more about this in part 4). 11 The literature employs a distinction between ‘system 1’ and ‘system 2’ metacognition, so we use those terms here, although generally we favor a distinction in terms of process types, consistently with our preference for dualprocess over dual-system accounts of cognition (Evans & Stanovich, 2013). 12 Some researchers consider that studying and recalling word lists or sentences is a means to study episodic memory and not only semantic. Although we agree that studying and recalling word lists may indirectly inform our knowledge of episodic memory, it is not the most straightforward way to study this type of memory.

References Arango-Muñoz, S. (2011). Two levels of metacognition. Philosophia, 39, 71–82. https://doi.org/10.1007/s11406-010-9279-0 Arango-Muñoz, S. (2013). Scaffolded memory and metacognitive feelings. Review of Philosophy and Psychology, 4(1), 135–152. https://doi.org/10.1007/ s13164-012-0124-1 Arango-Muñoz, S. (2014). The nature of epistemic feelings. Philosophical Psychology, 27(2), 1–19. https://doi.org/10.1080/09515089.2012.732002 Baddeley, A. (2007). Working memory, thought, and action. Oxford: Oxford University Press. Bargh, J. A. (1992). The ecology of automaticity: Toward establishing the conditions needed to produce automatic processing effects. The American Journal of Psychology, 181–199.

92  Santiago Arango-Muñoz and Juan Pablo Bermúdez Bargh, J. A. (1997). The automaticity of everyday life. In R. S. Wyer (Ed.), Advances in social cognition (pp. 1–62). Mahwah, NJ: Lawrence Erlbaum. Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait construct and stereotype activation on action. Journal of Personality and Social Psychology, 71(2), 230–244. Benjamin, A. S., Bjork, R. A., & Schwartz, B. (1998). The mismeasure of memory: When retrieval fluency is misleading as a metamnemonic index. Journal of Experimental Psychology: General, 127, 55–68. Bermúdez, J. P. (2017). Do we reflect while performing skillful actions? Automaticity, control, and the perils of distraction. Philosophical Psychology, 30(7), 896–924. Bresner, D., & Stolz, J. A. (1999). What kind of attention modulates the Stroop effect? Psychonomic Bulletin & Review, 6(1), 99–104. Brownstein, M., & Madva, A. (2012). The normativity of automaticity. Mind and Language, 27(4), 410–434. Campbell, J. (1999). Schizophrenia, the space of reasons and thinking as a motor process. The Monist, 82(4), 609–625. Carruthers, P. (2009a). Action-awareness and the active mind. Philosophical Papers, 38(2), 133–156. https://doi.org/10.1080/05568640903146443 Carruthers, P. (2009b). How we know our own minds: The relationship between mindreading and metacognition. Behavioral and Brain Sciences, 32(2), 121–138. https://doi.org/10.1017/S0140525X09000545 Carruthers, P. (2011). The opacity of mind: An integrative theory of self-knowledge. Oxford: Oxford University Press. Clark, A. (2015). What “extended me” knows. Synthese, 192(11), 3757–3775. https://doi.org/10.1007/s11229-015-0719-z De Brigard, F. (2014a). Is memory for remembering? Recollection as a form of episodic hypothetical thinking. Synthese, 19(2), 155–185. https://doi.org/10.1007/ s11229-013-0247-7 De Brigard, F. (2014b). The nature of memory traces. Philosophy Compass, 9(6), 402–414. Del Cul, A., Dehaene, S., Reyes, P., Bravo, E., & Slachevsky, A. (2009). Causal role of prefrontal cortex in the threshold for access to consciousness. Brain, 132(9), 2531–2540. Evans, J. S. B. T. (2008). Dual-processing accounts of reasoning, judgment, and social cognition. Annual Review of Psychology, 59, 255–278. Evans, J. S. B. T. (2010a). Intuition and reasoning: A dual-process perspective. Psychological Inquiry, 21(4), 313–326. Evans, J. S. B. T. (2010b). Thinking twice: Two minds in One Brain. Oxford: Oxford University Press. Evans, J. S. B. T., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate. Perspectives on Psychological Science, 8(3), 223–241. Fernández Cruz, A. L., Arango-Muñoz, S., & Volz, K. (2016). Oops, scratch that! Monitoring one’s own errors during mental calculation. Cognition, 146, 110– 120. https://doi.org/10.1016/j.cognition.2015.09.005 Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. American Psychologist, 34(10), 906–911. https://doi.org/10.1037//0003-066x.34.10.906

Remembering as a Mental Action  93 Francolini, C. M., & Egeth, H. (1980). On the nonautomaticity of ‘automatic’ activation: Evidence of selective seeing. Perception and Psychophysics, 27, 331–342. Gallo, D. A., & Lampinen, J. M. (2015). Three pillars of false memory prevention: Orientation, evaluation and corroboration. In J. Dunlosky & S. K. Tauber (Eds.), The Oxford handbook of metamemory. Retrieved from https://doi.org/10.1093/ oxfordhb/9780199336746.013.11 Gendler, T. S. (2008a). Alief and belief. Journal of Philosophy, 105(10), 634–663. Gendler, T. S. (2008b). Alief in action (and reaction). Mind and Language, 25(5), 552–585. Gordon, A. M., & Soechting, J. F. (1995). Use of tactile afferent information in sequential finger movements. Experimental Brain Research, 107(2), 281–292. Halamish, V., McGillivray, S., & Castel, A. D. (2011). Monitoring one’s own forgetting in younger and older adults. Psychology and Aging, 26(3), 631–635. https://doi.org/10.1037/a0022852 Hampton, R. R. (2001). Rhesus monkeys know when they remember. Proceedings of the National Academy of Sciences U.S.A, 98, 5359–5362. Hopkins, R. (2014). Episodic memory as representing the past to oneself. Review of Philosophy and Psychology, 5(3), 313–331. https://doi.org/10.1007/ s13164-014-0184-5 Intraub, H., Bender, R. S., & Mangels, J. A. (1992). Looking at pictures but remembering scenes. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 180–191. Jeannerod, M. (2006). Motor cognition. Oxford: Oxford University Press. Kahneman, D. (2011). Thinking, fast and slow. Farrar: Straus and Giroux. Retrieved from http://doi.org/10.1007/s13398-014-0173-7.2 Kahneman, D., & Henik, A. (1981). Perceptual organization and attention. In M. Kubovy & J. Pomerantz (Eds.), Perceptual organization and attention (pp. 181– 211). Mahwah, NJ: Lawrence Erlbaum. Kelley, C. M., & Lindsay, S. D. (1993). Remembering mistaken for knowing: Ease of retrieval as a basis for confidence in answers to general knowledge. Journal of Memory and Language, 32, 1–24. Kirsh, D., & Maglio, P. (1994). On distinguishing epistemic from pragmatic action. Cognitive Science, 18, 513–549. Koch, C., & Crick, F. (2001). On the zombie within. Nature, 41(6840), 893–893. Koriat, A. (1993). How do we know that we know? The accessibility model of the feeling of knowing. Psychological Review, 100(4), 609–639. Koriat, A. (2000). The feeling of knowing: Some metatheoretical implications for consciousness and control. Consciousness and Cognition, 9(2), 149–171. LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6(2), 293–323. Logan, G. D., & Cowan, W. B. (1984). On the ability to inhibit thought and action: A theory of an act of control. Psychological Review, 91(3), 295. Logan, G. D., & Crump, M. J. C. (2010). Cognitive illusions of authorship reveal hierarchical error detection in skilled typists. Science, 330, 683–686. https://doi. org/10.1126/science.1190483 MacLeod, C. M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109(2), 163–203. Mele, A. R. (1997). Agency and mental action. Nous, 31, 231–249.

94  Santiago Arango-Muñoz and Juan Pablo Bermúdez Mele, A. R. (2009). Mental action: A case study. In L. O’Brien & M. Soteriou (Eds.), Mental actions (pp. 17–37). Oxford: Oxford University Press. Michaelian, K. (2011). Generative memory. Philosophical Psychology, 24(3), 323–342. Michaelian, K. (2012). Metacognition and endorsement. Mind and Language, 27(3), 284–307. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Moors, A., & De Houwer, J. (2006). Automaticity: A theoretical and conceptual analysis. Psychological Bulletin, 132(2), 297—326. Mossel, B. (2005). Action, control and sensation of acting. Philosophical Studies, 124, 129–180. Nagel, J. (2014). Intuition, reflection, and the command of knowledge. Proceedings of the Aristotelian Society, 88, 217–239. Nelson, T. O., & Narens, L. (1990). Metamemory: A theoretical framework and new findings. The Psychology of Learning and Motivation, 26, 125–173. Norman, E., Price, M. C., & Duff, S. C. (2010). Fringe consciousness: A useful framework for clarifying the nature of experience-based metacognitive feelings. In A. Efklides & P. Misailidi (Eds.), Trends and prospects in metacognition research (pp. 63–80). Springer. Norman, D. A., & Shallice, T. (1986). Attention to action: Willed and automatic control of behavior. In R. J. Davidson, G. E. Schwartz, & D. Shapiro (Eds.), Consciousness and self-regulation (pp. 1–18). New York: Springer. Pacherie, E. (2008). The phenomenology of action: A conceptual framework. Cognition, 107, 179–217. https://doi.org/doi:10.1016/j.cognition.2007.09.003 Paynter, C. A., Reder, L. M., & Kieffaber, P. D. (2009). Knowing we know before we know: ERP correlates of initial feeling-of-knowing. Neuropsychologia, 47(3), 796–803. https://doi.org/10.1016/j.neuropsychologia.2008.12.009 Peacocke, C. (2007). Mental action and self-awareness (I). In B. McLaughlin & J. Cohen (Eds.), Contemporary debates in the philosophy of mind (pp. 358–376). Oxford: Blackwell. Perner, J., & Aichorn. A. (2008). Theory of mind, language and the temporo-parietal junction mystery. Trends in Cognitive Sciences, 12(4), 123–126. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. Solso (Ed.), Information processing and cognition: The Loyola symposium (pp. 55–85). Mahwah, NJ: Lawrence Erlbaum. Proust, J. (2005). La nature de la volonté. Paris: Folio-Gallimard. Proust, J. (2009). Is there a sense of agency of thought? In L. O’Brien & M. Soteriou (Eds.), Mental actions and agency (pp. 253–279). Oxford: Oxford University Press. Proust, J. (2012). Metacognition and mindreading: One or two functions? In M. Beran, J. Brandl, J. Perner, & J. Proust (Eds.), The foundations of metacognition (pp. 234–250). Oxford: Oxford University Press. Proust, J. (2013). The philosophy of metacognition: Mental agency and selfawareness. Oxford: Oxford University Press. Rabbitt, P. (1966a). Error correction time without external error signals. Nature, 212(5060), 438–438. Rabbitt, P. (1966b). Errors and error correction in choice-response tasks. Journal of Experimental Psychology, 71(2), 264–272.

Remembering as a Mental Action  95 Rabbitt, P. (1990). Age, IQ and awareness of errors. Ergonomics, 33(10), 1291–1305. Rabbitt, P. (2002). Consciousness is slower than you think. The Quarterly Journal of Experimental Psychology, 55(4), 1081–1092. Reder, L. M. (1987). Strategy selection in question answering. Cognitive Psychology, 19(1), 90–138. https://doi.org/10.1016/0010-0285(87)90005-3 Rhodes, G. (1996). Superportraits: Caricatures and recognition. Hove: Psychology Press. Rietveld, E. (2008). Situated normativity: The normative aspect of embodied cognition in unreflective action. Mind, 117(468), 973–1001. Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 803–814. Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering the past and imagining the future. Philosophical Transactions of the Royal Society B, 262, 773–786. https://doi.org/10.1098/ rstb.2007.2087 Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human information processing: I. Detection, search, and attention. Psychological Review, 84(1), 1–66. Schnyer, D. M., Verfaellie, M., Alexander, M. P., LaFleche, G., Nichols, L., & Kaszniak, A. W. (2004). A role for right medial prefrontal cortex in accurate feeling-of-knowing judgments: Evidence from patients with lesions to frontal cortex. Neuropsychologia, 42, 957–966. Schwartz, B., & Metcalfe, J. (2010). Tip-of-the-tongue (TOT) states: Retrieval, behavior, and experience. Memory and Cognition, 39, 737–749. Searle, J. (1983). Intentionality. Cambridge: Cambridge University Press. Shea, N., Boldt, A., Bang, D., Yeung, N., Heyes, C., & Frith, C. D. (2014). Suprapersonal cognitive control and metacognition. Trends in Cognitive Sciences, 18(4), 186–193. https://doi.org/http://dx.doi.org/10.1016/j.tics.2014.01.006 Shepherd, J. (2014). The contours of control. Philosophical Studies, 170(3), 395– 411. https://doi.org/10.1007/s11098-013-0236-1 Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human information processing: II. Perceptual learning, automatic attending and a general theory. Psychological Review, 84(2), 127–194. Smith, J. D. (2009). The study of animal metacognition. Trends in Cognitive Sciences, 13(9), 389–396. https://doi.org/10.1016/j.tics.2009.06.009 Smith, J. D., Beran, M. J., Redford, J. S., & Washburn, D. A. (2006). Dissociating uncertainty responses and reinforcement signals in the comparative study of uncertainty monitoring. Journal of Experimental Psychology: General, 135, 282–297. Smith, J. D., & Washburn, D. A. (2003). The comparative psychology of uncertainty monitoring and metacognition. Behavioral and Brain Sciences, 26, 317–373. Souchay, C., Guillery-Girard, B., Pauly-Takacs, K., Wojcik, D. Z., & Eustache, F. (2013). Subjective experience of episodic memory and metacognition: A neurodevelopmental approach. Frontiers in Behavioral Neuroscience, 7, 1–16. https://doi.org/10.3389/fnbeh.2013.00212 Stanovich, K. E. (2004). The Robot’s rebellion: Finding meaning in the age of Darwin. Chicago: University of Chicago Press. Retrieved from http://doi.org/10.1007/ s13398-014-0173-7.2

96  Santiago Arango-Muñoz and Juan Pablo Bermúdez Strawson, G. (2003). Mental ballistics or the involuntariness of spontaneity. Proceedings of the Aristotelian Society, New Series, 103, 227–256. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18(6), 643–662. https://doi.org/10.1037/h0054651 Walsh, M. M., & Anderson, J. R. (2009). The strategic nature of changing your mind. Cognitive Psychology, 58, 416–440. Whittlesea, B. W. A., & Leboe, J. P. (2003). Two fluency heuristics (and how to tell them apart). Journal of Memory and Language, 49, 62–79. Wolpert, D. M., & Ghahramani, Z. (2000). Computational principles of movement neuroscience. Nature, 3, 1212–1217. Wolpert, D. M., Ghahramani, Z., & Jordan, M. I. (1995). An internal model for sensorimotor integration. Science, 269, 1880–1882. Wu, W. (2013). Mental action and the threat of automaticity. In A. Clark, J. Kiverstein, & T. Vierkant (Eds.), Decomposing the will (pp. 244–261). Oxford: Oxford University Press. Wu, W. (2016). Experts and deviants: The story of agentive control. Philosophy and Phenomenological Research, 93(1), 101–126. https://doi.org/10.1111/phpr.12170

5 The Roots of Remembering

Daniel D. Hutto and Anco PeetersThe Roots of Remembering

Radically Enactive Recollecting Daniel D. Hutto and Anco Peeters

Our understanding of what lies at the roots of remembering needs to evolve. Recent years have witnessed some daring new thinking about this topic in response to empirical findings from three main sources. Two of these sources paint a picture of remembering as transactional and extraindividual, on the one hand, and reconstructive and re-creative, on the other. The third set of findings challenge standard assumptions about the fundamentally representational character of remembering. Going wide, a body of empirical work shows that—at least sometimes— successful acts of remembering require heavy scaffolding by the ­environment or by other individuals (Ren & Argote, 2011; Sparrow, Liu, & Wegner, 2011). For example, there are well-documented cases in which interactions between romantic partners enable couples to remember things together that, as individuals, they cannot (Wegner, Erber, & Raymond, 1991). Such findings have inspired the idea that remembering can be a widely distributed and interactive process that draws on extraindividual resources. In this vein, there has been an explosion of theoretical work that seeks to recast how we conceive of remembering in extended cognition terms (Sutton et al., 2010; Tollefsen, Dale, & Paxton, 2013; Kirchhoff, 2016; Huebner, 2016; Heersmink, 2017a, 2017b). No doubt, these intellectual efforts are, as Skorburg (2017) observes, spurred on by the fact that transactional remembering is “low-hanging fruit for extended cognition theorists” (p. 473). Looking inward, dramatic new proposals about memory ask us to reconceive what goes on in the heads of individuals when they remember. Novel scientific investigations concerning mental time travel have repeatedly confirmed the existence of strong similarities in the patterns of neural activity that enable us to recall past happenings and those that enable us to imagine possible futures. There is convincing empirical support that acts of memory and acts of imagining have a common basis (Szpunar, Watson, & McDermott, 2007; Schacter et al. 2007; Schacter & Addis, 2009; Mullally & Maguire, 2014). These findings encourage some theorists to propose that remembering and imagining are either identical or at least intimately related. Accordingly, remembering is deemed to be fundamentally

98  Daniel D. Hutto and Anco Peeters creative, imaginative, and dynamic in character (Michaelian, 2016; Clark, 2016). Remembering thus conceived is not a matter of passive recollection, as some traditional cognitivist theories would have it. Relatedly, a wealth of empirical findings has put pressure on the traditional assumption that the primary function of remembering is to accurately represent past happenings. Against the idea that acts of memory involve the straightforward recovery or replay of past experiences, it has been shown that we shift between field and observer perspectives during recall (Nigro & Neisser, 1983). Against the idea that our memories are built for accuracy, what we remember is usually riddled with distortions. We regularly experience telescoping effects such that recent events are perceived as occurring in the more distant past than they did and vice versa, of more distantly occurring events being perceived as being more recent than they were (Neter & Waksberg, 1964; Thompson et al., 1996; Janssen et al., 2006). Also, we typically extend the boundary of scenes. For example, we remember more of the spatial boundary of a visual scene than was actually experienced (Intraub & Richardson, 1989). Finally, it is well known that memories can be easily and significantly corrupted or wholly implanted by various means (Loftus, Miller, & Burns, 1978; Loftus & Pickrell, 1995; Roediger and McDermott, 1995; Garry et al. 1996; Wade et al. 2002; Lindsay et al. 2004; Loftus, 2005). All told, these combined scientific findings demonstrate that our natural memory capacities are fragile, easily open to outside influence and ill-suited to reliably reproduce undistorted, accurate representations. The amassed empirical evidence reveals that we get by when remembering in everyday contexts, doing well enough, without having to accurately represent past happenings. Indeed, stressing this, De Brigard (2014) concludes that, in the end, “it is a mistake to think of memory as [a] system that is uniquely— or even primarily—dedicated to reproducing the contents of previous experiences” (p. 177). We propose an enactivist account of remembering that casts it as creative, dynamic, and wide-reaching. Such an account can more easily accommodate the aforementioned empirical findings than cognitivist accounts that conceive of remembering as always involving passive recollections that occur wholly and solely inside heads. According to our enactivist proposal, the roots of remembering can be understood in line with the Equal Partner Principle that holds that invoking neural, bodily, and environmental factors can all make equally important contributions when it comes to explaining and characterizing cognitive activity. Augmenting this proposal, pivotally, it is argued that we achieve a stronger and more elegant account of the basis of remembering by going radically enactivist—viz., by abandoning the widely held assumption that remembering always and everywhere involves the retrieval of stored information or content in order to represent past events. Seeking to understand the roots of remembering in a radical, content-free way is independently motivated by the fact that, should such an approach

The Roots of Remembering  99 to memory prove tenable, it would avoid seemingly intractable theoretical problems that arise when we try to account for information, content, and their causal powers within a naturalistic framework. We currently lack workable naturalistic theories of information and content that would allow us to explain how information gets encoded, processed, and retrieved in order to generate remembered contents. As such we are motivated to explore and develop an alternative—extensively enactive—characterization of the roots of remembering that makes no such commitments. The action of the chapter is as follows. Section 1 reviews a range of competing theoretical proposals about the nature of the contributions that individual and supraindividual resources make to everyday acts of remembering. We demonstrate that whatever stance is taken on the extended memory debate, the common denominator in all of the standard offerings is endorsement of the thesis that remembering necessarily involves having access to remembered content of some kind. In Section 2, we propose an enactive account of the roots of remembering that surrenders this Remembered Content Assumption, or RCA. We propose an enactive alternative that can explain even semantically rich forms of memory retrieval in an empirically adequate manner, while avoiding the deep theoretical problems of accounting for the relevant notions of information and content naturalistically. We demonstrate these attractive features of our radically enactive account of remembering, showing how it is able to successfully handle classic cases from the extended mind literature. Section 3 demonstrates that a radically enactive memory account also has the resources to explain individuals’ on-board, experientially rich forms of episodic memory. Even on the assumption that episodic memories take the form of or strongly depend on dynamic, re-creative acts of simulative imagining, we argue that such imaginings themselves are not best understood as inherently contentful nor are they best explained by the recovery of remembered content. We conclude, in Section 4, that a radical enactive account of memory is able to understand the roots of remembering—even its dynamic and reconstructing character—in terms of on-board biological capacities that are sometimes supported by environmental and social resources, without assuming access to or retrieval of content plays any part in this process.

1 Remembered Contents and Questions of Extended Memory There is an important assumption that goes unquestioned in much contemporary theorizing about memory: it is that acts of remembering primarily involve access to and recovery of content of some kind. Call this the Remembered Content Assumption, or RCA. The RCA is foundational in prominent debates about how to characterize the type of contribution that

100  Daniel D. Hutto and Anco Peeters external resources make in enabling everyday acts of recall. This is because the RCA supports a second important assumption: call it the Contentful Contribution Assumption, or CCA. According to the CCA, the contributions that external resources make to cognitive processes only count as cognitive contributions if they are informational or contentful contributions.1 Deciding whether external resources satisfy the criterion set out in the CCA has occupied the attention for those on both sides of the extended memory issue. For example, internalists who are skeptical of the very idea of extended memory wear their commitment to RCA and CCA on their theoretical sleeves. Pivotally, attempting to draw a familiar distinction, they insist that the use of external resources, even if crucial for completing certain memory tasks, can make only causal, and not properly cognitive, contributions to the process of remembering. The standard way internalists attempt to draw the line between merely causal and properly cognitive contributions is to appeal to the further distinction between intrinsic and derived content. The former is thought to be a feature of certain kinds of mental states, where such mental states are taken to be the ultimate and original source of any and all content. The latter is borrowed from such mental states and assigned to artifacts such as “traffic lights, gas gauges and flags” (Aizawa & Adams, 2005, p. 662). Accordingly, any properly cognitive contributions are isolated to and “constituted by certain sorts of causal processes that involve nonderived content” (Adams & Aizawa, 2010, p. 68). Consequently, any content that an external resource—say, a repository, such as a notebook—may be said to bear could, at best, be derived content. Taking this internalist ruling at its word raises the question of whether it makes sense to think of external resources as literally containing, or really being imbued with, any kind of content of their own. Certainly, external resources—so conceived—even if they are not simply dead signs, can make no contentful contribution to remembering that is independent of the memorizer. Hence such resources can add nothing cognitive into the memory mix, since the resources themselves are only assigned content by beings with contentful mental states. Clark and Chalmers (1998) provide a philosophically famous thought experiment that compares the cognitive feats of Inga, who relies on her onboard biological capacities in order to remember, and Otto, an Alzheimer’s sufferer who relies on an external notebook in order to do so. Internalists deny Otto’s notebook makes a cognitive contribution because it lacks content of the right kind. The chief spokespersons for this view, Adams and Aizawa (2001), hold that the crucial difference is that only biological brains have nonderived content whereas the symbols of Otto’s notebook have merely derived content. For this reason, they conclude that Otto’s accessing of the content of his notes is a noncognitive process and that the content of his notes do not “constitute beliefs or memories” (p. 55, emphasis added, see also Adams & Aizawa, 2010, p. 70).

The Roots of Remembering  101 External aids, on this view, help individuals to recover whatever contents their native biological memories are already assumed to have. Hence, use of external supports would at best instrumentally aid individuals’ capacities for recall of relevant events or facts, by helping them to construct the content of their memories rather than actually contributing any content to this process. Going the opposite way on this issue, the exciting idea that launched a thousand papers on the extended mind hypothesis is precisely that, sometimes, memories can be found outside of people’s heads. Or more precisely, those attracted to the original, first wave version of the extended mind hypothesis assume that the information that constitutes beliefs r­ elevant to our memories is normally located somewhere within our biological brains. But such information need not be, and in fact is not always, internally located. On this way of understanding what is essential for having a memory it is possible to imagine, for example, that the informational content of a memory might be located either in, say, one’s hippocampus or possibly in an artificial hard drive in one’s head.2 It is but a short step from this assumption to the idea that the content of one’s memory could also be externally located, say, in a hard drive outside of one’s head. This is precisely what is being assumed when, in making their classic comparison, Clark and Chalmers (1998) tell us that Otto’s notebook and Inga’s hippocampus play the same roles in their respective rememberings. On their account, “the information in the notebook functions just like the information constituting an ordinary non-occurrent belief; it just happens that this information lies beyond the skin” (Clark & Chalmers, 1998, p. 13, see Rowlands, 1999, p. 122, 142). Thus for Inga, the content of her “belief was sitting somewhere in memory waiting to be accessed” (Clark & Chalmers, 1998, p. 12). The difference is that the informational basis of Otto’s belief resides in his notebook, not in his wetware. Thus, a­ ccording to the original version of the parity principle, advanced by Clark and Chalmers, we ought to conclude that the information in Otto’s notebook is part of his extended mind. This is because—on this way of telling the story—the information in Otto’s notebook functions in the same way as the information in Inga’s brain does in their respective acts of remembering. Thus, in advancing their classic, first wave argument for the extended mind thesis, Clark and Chalmers (1998) present an objectified account of the contents of memory, according to which remembered contents are conceived of as accessible, objective commodities (see Loader, 2013, p. 167). Indeed, it is precisely because these first wave extended mind theorists have a commodity conception of the contents of biological memory that the idea of transposing the content of memories to external artifacts is such “an easy move for them” (Loader, 2013, p. 177). The RCA and associated idea that memories have locatable contents is so familiar in the analytic tradition as to go unquestioned, even by those who deny the CCA. Thus new school internalists—those who defend the

102  Daniel D. Hutto and Anco Peeters idea that only that which is within occurrent consciousness properly counts as mental—do not bat an eye at the possibility that memory contents might be located outside the head (Gertler, 2007). Defenders of the idea that consciousness is what really matters for mentality do not take a stand on whether contents of memories are located internally or externally. For in their view, what really matters to mentality is only whether or not one actively and consciously entertains such contents, not the location of such contents. Thus, on this way of construing things, “Otto would . . . only have the content that MoMA is on 53rd Street ‘in mind’ when he reads his notebook and consciously entertains it” (Coleman, 2011, p. 105). Setting out from the idea that conscious activity is all that matters to mindedness, Coleman (2011) offers a different parity principle—which states: “If, as we confront some task, a part of the head functions as a process which, were it to go on in the world, we would have no hesitation in rejecting as part of the cognitive/mental process, then that part of the head is (for that time) not part of the cognitive/mental process” (p. 105). According to Coleman’s (2011) revised parity principle, it follows that any contents that we might imagine to be stored in organisms will not be part of cognitive processes on the assumption that we do regard any contents stored in the environment as part of cognitive processes. This will be so, in either case, whether they prove accessible to consciousness or not. All in all, what the analysis of all three of these options reveals is that there is a pervasive tendency to assume a “reified conception of memory” (Loader, 2013, p. 170). As such, to think of information and content in this way encourages us to think of memory as a kind of storehouse: it treats the notions of information and content interchangeably and regards both as commodities to which we can have access. Importantly, although they hold on to the idea that information and ­contents are stored, second wave extended mind theorists break faith with the idea that biological and external memory are alike in how they get their work done. Sutton (1998) warns against modeling the way brains store information on the way computers store information. The crucial difference is that nonbiological memory systems retain information statically such that it is “unchanged unless manipulated” (p. 4). In contrast, biological memory is fluid, reconstructive, and shifting (Sutton, 2010, p. 206). Similarly, based on a wide-ranging review of the empirical literature, Michaelian (2012) argues that items stored in biological memory are unlike external records in that they are not discrete, stable items that are readily endorsed on retrieval. Forgetting is a feature, not a bug of biological memory. Hence, when it comes to understanding the dynamics of biological memory, Michaelian (2012) gives reasons for moving beyond the “simple, preservative picture,” adding that, “if memory is a container it is a rather leaky one” (p. 1156). Bearing these considerations in mind, second wave extended mind ­theorists recognize that even though internal and external resources—engrams and

The Roots of Remembering  103 exograms—have vastly different properties and profiles, they can still cooperate because they can serve the same overall cognitive ends in enabling acts of remembering. Shifting the emphasis away from parity arguments, in line with this observation, second wave theorists instead aim to clarify what individual and external resources, respectively, bring to the table in acts of remembering. Second wave theorists seek to understand how memories are constituted and sustained by the integration of complementary contributions of internal and external resources involving “patterns of information-sharing and transmission” (Barnier & Sutton, 2008, p. 178). Importantly, once again, for those who are attracted to this sort of view, talk of processed information is treated as interchangeable with talk of content.3 In line with the CCA, there can still be a definitively cognitive contribution to the overall process if, for example, the external resources make a contentful contribution, say, by filling in the ‘gaps’ in the contents of already, partially recovered memories (Barnier & Sutton, 2008, p. 179). Focusing on the case of collective remembering, Barnier and Sutton (2008) are interested in addressing the question: “how do groups operate to process information” (p. 179). Here they emphasize the need to look for complementary processes both within and across individuals, precisely because they see no good scientific reason to adopt either of two extremes: thinking of social factors as mere external triggers for internal acts of remembering or thinking of acts of remembering as occurring entirely extra-individually (p. 177). Thus, on their view, information is exchanged between individuals when memories are transactionally scaffolded by others. Crucially, for second wave extended mind theorists, memory picks out a diverse set of “cognitive capacities by which we retain information and reconstruct past experiences, usually for present purposes” (Sutton, 2012).4 Thus despite Sutton’s warning about not modelling the human mind on computers, it is fair to say that the ‘archival caricature’ still captures something important at the heart of the way certain contemporary philosophers of mind and cognitive scientists understand memory. Despite acknowledging that the putative informational contents of our memories are stored in quite different ways in biological memory as opposed to external devices, second wave extended mind theorists remain committed to the RCA. Thus, they remain committed to a content-based view of remembering to the extent that they assume that there are contents to be stored at all—viz. to the extent that they retain the idea that remembering essentially involves receiving, encoding, and retrieving stored contentful information about particular events or episodes. Those anticipating a third wave of extended mind go a step further. Whereas second wave extended mind theorists understand the parity principle of the first wave as being a special case of complementarity between inner and outer resources (Sutton, 2010, p. 206), it is suggested

104  Daniel D. Hutto and Anco Peeters that the third wave might altogether do away with “distinct inner and outer realms of engrams and exograms” (p. 213). Consequently, it is imagined that there could be a “deterritorialized cognitive science which deals with the propagation of deformed and reformatted representations, and which dissolves individuals into peculiar loci of coordination and coalescence among multiple structured media” (p. 213). Although third wave extended mind theorists are prepared to imagine a more fluid and dynamic vision of cognitive science, as Sutton’s (2010) talk of creatively constructed representations reveals, it would seem that even such a dramatic shift in thinking would still retain the fundamental idea of RCA—namely, that memory depends upon manipulation of content of some kind.

2  Extensively Enactive Remembering A more daring move, in thinking about what and how individual and supraindividual resources contribute to acts of remembering, is to abandon the RCA—at least in thinking about basic memory processes—and to let go of the idea that such processes necessarily involve the production, sharing, and transmission of any contents whatsoever. The more radical forms of extensive enactivism propose just such a rethink of the nature of the most basic biological forms of individual memory and how they combine with external resources to enable many mundane acts of recall. Radical enactivists about memory agree with internalists in holding that there is no informational content literally contained in Otto’s notebook that is accessed by him when he remembers. Yet, unlike most other theorists, they also deny that Inga’s biological brain is an interestingly different position. Radical enactivists pronounce a plague on both of these RCA houses: That is their version of parity. Fundamentally, radical enactivists hold that the activity of remembering is not best understood in terms of the retrieval of remembered contents (see Loader, 2013; Hutto & Myin, 2017). Taking a radical line, enactivists of this stripe hold that memories come into being “on the spot” during acts of recall while abandoning the idea that “ongoing behaviour in such cases is explained by appeal to identifiable inner content bearers” (Clark, 1998, p. 100). Going the radically enactivist way is to conceive of r­emembering as a matter of ‘on-the-fly’ construction that can be grounded in ­structural synaptic changes in the brain as well as other structural changes in the environment without assuming the existence of stored and retrieved contents. Scientific research on memory is rife with talk of “memory traces,” of “encoded and retrieved information,” and of “the storage and retrieval of information and representations.” Despite the popularity of this talk and associated metaphors, close inspection of how these notions operate in science reveals them to have serious limitations—limitations that make them

The Roots of Remembering  105 prime candidates for theoretical explication or elimination (see Roediger, 1980). As De Brigard (2014) observes: “Storing” is a rather misleading term. What seems to occur when we encode information is the strengthening of neural connections due to the co-activation of different regions of the brain, particularly in the sensory cortices, the medial temporal lobe, the superior parietal cortex, and the lateral prefrontal cortex. During encoding, each of these regions performs a different function depending on the moment in which the information gets processed. A memory trace is the dispositional property these regions have to re-activate, when triggered by the right cue, in roughly the same pattern of activation they underwent during encoding. (p. 169) Radical enactivists assume that De Brigard’s analysis is mostly correct, modulo any residual commitment in the previous claims that information or content is actually encoded and processed. Rather, radicals seek to explain basic forms of learning and memory entirely in terms of re-enacted know-how. Focusing solely on the simpler kinds of procedural memory widespread in the animal kingdom, remembering can be understood as the capacity to re-enact embodied procedures—often prompted and supported by patterns of response that are triggered by external phenomena. Memory of this sort entails knowing what to do in familiar circumstances. It is surely not necessary to posit stored mental contents in order to explain the dispositional basis of such capacities (Ramsey, 2007, Chapter 5). The brain’s underlying contribution to such capacities “turns out to be just a matter of either organizing extant synaptic circuits in new wiring patterns or switching on genes in neurons that produce new synapses. . . . The brain does everything without thinking about anything at all” (Rosenberg, 2014, pp. 26–27). Importantly, purely embodied know-how is not grounded in or mediated by any kind of stored information or knowledge; rather it can be understood as the overall responsiveness of a complex system as shaped through habit and past experience (Barandiaran and Di Paolo, 2014). This radically enactive account of basic memory can be provided, without gaps, so long as no appeal is made to the encoding and processing of information or representations. Radically enactive accounts of memory suspect that an information-processing story adds nothing of explanatory value to our understanding of how experience modulates neural connections and weights that support acts of remembering. If they are right, appeals to stored information, however popular, are superfluous. Moreover, not only is it questionable whether talk of stored information and retrieved content adds any explanatory value, it is difficult to understand—ultimately—how such explanations would go even if it was thought that they are needed.

106  Daniel D. Hutto and Anco Peeters Here it is important to acknowledge a deep theoretical problem—which Hutto and Myin (2013) dub the Hard Problem of Content, or HPC. The HPC highlights a problem in making easy appeal to the idea that brains s­ omehow literally process and store informational or other kinds of remembered ­content. The HPC arises from the fact that the notion of information that can be most easily called on to do serious explanatory work in the sciences of the mind—a notion that might possibly fill in the scientifically respectable details of the information processing story—is the notion of informationas-covariance. According to that notion of information, a state-of-affairs is said to carry information about another state-of-affairs if and only if it lawfully covaries with that other state-of-affairs, to some specified degree. The parade example is that of the age of a tree covarying with the number of its rings. Information in this sense is a perfectly objective and utterly ­ubiquitous commodity—it literally litters the streets. Moreover, this notion of information has impeccable naturalistic credentials: it is used in many sciences. Thus, it can clearly serve the needs of a cognitive science with ­explanatorily naturalistic ambitions. Yet cognitivists face a dilemma if they try to tell the story of remembered contents using a naturalistically respectable notion of information. They might, on the one hand, opt for an information-as-covariance notion, or, on the other, they might try to call on some other naturalistic notion of information to understand memory in terms of content encoding and ­information processing. Going either way has its problems. In dealing with the first horn of the dilemma, they can try to give a naturalistically respectable explanation of information encoding and processing by appeal to the notion of information-as-covariance. Yet if that is the only notion of information in play in cognitivist theorizing about remembering, then it is difficult to understand what it could possibly mean for information to be literally encoded. How can relations that hold between covarying states of affairs be literally “extracted” and “picked up” from the environment so as to be “encoded” within minds? Perhaps it will be objected that what should be focused on here is not the medium but the message. Sometimes the story about remembered contents is told in quasi-communicative terms of signaling and receiving messages. Yet how seriously should we take these analogies and the talk of encoding and decoding “messages” in the brain? Once again there are grounds for caution. Despite the widespread popularity of talk of encoded signals and messages, attempts to seriously explicate the nature of neural or mental “codes” and their alleged encoded content are few and far between. Goldman (2012, p. 73) gives a frank appraisal of the current situation: “There is no generally accepted treatment of what it is to be . . . a mental code, and little if anything has been written about the criteria of sameness or difference for such codes. Nonetheless, it’s a very appealing idea, to which many cognitive scientists subscribe.”

The Roots of Remembering  107 Even more worryingly, how should we understand the nature and source of any such putatively contentful messages? As information-as-covariance is not to be understood as the transmission of information from one system to another, this notion is surely not able to help us to understand how sense perception supplies the mind with contentful messages—messages that supply contents that can be encoded and decoded by biological brains. At any rate, as things currently stand, we lack a naturalistic theory of content to provide us with a substantive explanatory account of how to understand what contents are or how they might be stored and retrieved. Can cognitivists deal more successfully with the second horn of the dilemma? They might try to call on some other naturalistically respectable notion of information that will enable them to tell content encoding and processing stories in full detail. Telling those tales in a different way requires identifying an alternative notion of information with sound naturalistic credentials that can do the relevant explanatory heavy lifting. Yet, at least as things currently stand, it is unclear if there is an alternative, scientifically respectable notion of information available has the right characteristics to do such explanatory work. In light of this analysis, it becomes clear that the “storage” metaphor is not the only, or even the most, problematic card in the cognitivist deck, pace De Brigard (2014). The very idea that remembered contents might reside in biological brains invokes quite serious scientific mysteries. Such mysteries arise for anyone who takes seriously the idea that information is some kind of accessible, content-bearing commodity. These mysteries want dispelling, one way or another—they want explaining or explaining away. Yet what is the alternative? One possibility is that “memory might not be a store at all, and that remembering might be active in a stronger sense than a reconstructive account of memory by itself allows—i.e., that it might be fruitfully understood as a type of action” (Loader, 2013, p. 173). A radically enactive take on memory “would focus not on access to the contents of a store but on remembering as a type of action” (Loader, 2013, p. 168). As Loader (2013) observes, an enactive account of memory is one that, following an attempted pattern of explanation used for understanding a range of other cognitive phenomena, eschews “representation in favour of action” (p. 175).5 The core idea of an enactivist account of memory, so construed, is captured in the slogans advanced by Stern (1991)—“memory is an ability to think and act in certain ways” (p. 203)—and Neisser (1996)—“­remembering is a kind of doing” (p. 203). Indeed, the elements of the enactive view can be found in a number of accounts of memory: those “which reject representationalism (e.g., Ryle, 1949; Malcolm, 1977; Wittgenstein, 1967; Stern, 1991; Neisser, 1996; Shanon, 1998) and those which emphasize the active nature of memory in ways which go beyond mere reconstructivism (Stern, 1991; Neisser, 1996; Toth & Hunt, 1999)” (Loader, 2013, p. 174). Procedural memory—remembering how to execute certain types of action—lends itself to a radically enactive treatment. This is because

108  Daniel D. Hutto and Anco Peeters remembering of the procedural kind requires nothing more than reinitiating a familiar pattern of prompted response, albeit with adjustments that are dynamically sensitive to changes in circumstance and context (Sutton & Williamson, 2014). Pivotally, for our purposes, purely embodied acts of remembering are marked by the fact that they do not require representing any specific past happening or happenings, and especially not representing these as past happenings. Remembering of this sort can be best characterized as enactive precisely because it is a matter of re-enactment that does not involve representation (Casey, 1987). Embodied or enactive forms of remembering, as exemplified by procedural memory, do “not store representations of external states of the world” (Schacter & Tulving, 1994, p. 26, see also Michaelian, 2016, p. 26ff.). Yet even though we should not expect a contentless account of enactive memory to explain all features of every kind of remembering, if the approach is to be explanatorily interesting it would need to apply go beyond providing an account of procedural memory only. For, as Loader (2013) recognizes, there would be nothing “particularly surprising or useful in baptising procedural memory only as ‘enactive’ ” (p. 174). There are sophisticated, situated forms of remembering involving semantic recall that can be adequately explained in terms of individuals directly interacting with environmental resources, without assuming the need for their accessing remembered contents. Take Otto, for example. He is clearly capable of forming contentful beliefs about the world. The reminders penned in his notebook prompt him to judge—reliably and correctly—that the MoMA is on 53rd Street. This is so even though it is not plausible that he is retrieving that remembered content from either the notebook itself—as it only contains pen marks—or from his degraded biological memory. The point is that in Otto’s case an appropriate content is reliably formed—he forms an accurate semantic memory about a particular place. This is a declarative, not just a procedural memory, but arguably, the content of Otto’s memory is not accessed or recovered from any kind of memory store. Or take another case that has impressed second wave extended mind theorists. Tribble (2005) offers an insightful account of how Shakespearean actors in Elizabethan and Jacobean times used their theatre environment to a­ ugment and support their biological memories. Their memory tasks could be formidable: some leading actors “had to secure and retain command of about ­seventy-one different roles, of which number fifty-two or fifty-three were newly learned” (Beckerman, 1962, p. 9, as cited by Tribble, 2005, p. 136). On Tribble’s (2005) analysis, the actors managed to remember their lines and parts by making special use of “immediate physical and social resources outside the person” (p. 140). Part of her explanation for their incredible memory feats is the fact that they relied on plots—folio-sized sheets of paper containing specific instructions on when and where to enter and exit, and

The Roots of Remembering  109 sound and music cues (p. 144). Yet even though these devices would have helped the actors grasp the large, skeletal action of the play—the ebb and flow of the scenes—they were “impossibly scanty. . . [containing] only the barest of cues, sometimes just a word or two” (p. 151). As such, the actors would have used well-rehearsed techniques of rote memory and the hooks provided by iambic pentameter, in conjunction with other cues and prompts, given by both the other actors and the structure of the playhouse to produce their performances. In sum, they actively used their environment so as to ‘under-load’ their memories. The answer then, according to Tribble (2005, p. 151), lies in the actors exploiting opportunities for stripping away any “superfluous information” and off-loading “information” into the ambient environmental and social structures: the playhouse itself and other players. Tribble and others attempt to explain these memory feats in terms of offloading information, but the details of this part of their explanation are brief and unclear. Which information, exactly, would have been off-loaded? How is this achieved? And once off-loaded, how does the informational content make a difference to the acts of remembering? A simpler and sufficient explanation is that the actors, through repeated practice in their particular environments, were able to use individual techniques, supported by resources in the environment, including one another, in order to reliably generate their performances. Once again, even though this complex activity does not reduce to mere procedural remembering, accessing remembered contents need play no part in explaining how the actors manage to generate their familiar pattern of lines. Perhaps it will be conceded that access to, or the processing of, contents need play no part in explaining the acts of remembering that typify how Otto uses his notebook or how Shakespearean players used local cues to remember their lines while walking the boards. Still, it might be thought that even if radically enactive accounts can handle such cases of nonprocedural remembering, they cannot go further. Hence, it might be conceded that radical enactivists can explain the kinds of remembering that are environmentally and socially scaffolded—those that are driven by direct engagement with external resources—without the need to posit stored and retrieved contents. Yet, even if this proves so, it might be thought that an enactive account of memory is still unable to tell the whole story of basic biological remembering. In particular, an enactive account may be deemed to lack the resources needed for explaining the sort of reconstructive imaginative acts that constitute experientially rich episodic forms of remembering. For many, it is difficult to imagine such a possibility.

3  Episodic Remembering as Enactive Imagining A growing number of theorists have proposed that episodic remembering centrally involves or just simply is a form of re-creative or simulative

110  Daniel D. Hutto and Anco Peeters imagining—a kind of imagining that enables us to construct and entertain possible episodes that may have occurred (Gerrans & Kennett, 2010, De Brigard, 2014, Michaelian, 2016). The assumption here is that episodic memory, here, “refers, roughly, to the form of memory responsible for allowing us to revisit specific episodes or events from the personal past” (Michaelian, 2016, p. 5). Crucially, the kind of recollective recall associated with episodic memory is taken to be essentially experiential in character; it has a distinctive ­phenomenology (Debus, 2008, pp. 407–408).6 Many in the field assume that such episodic experiences are inherently representational. Bernecker (2010) exemplifies: Experiential memory has two characteristics. First, one can experientially remember only what one has personally experienced. Experiential memory is restricted to cases in which the claim to remember something incorporates the claim to have experienced it for oneself. Second, experiential memory represents the remembered content from the first-person perspective—from ‘within’—and involves qualitative experiences (qualia) and imagery. Experiential memory consists in the evocation of parts of the original experience in imagination, allowing one to relive or re-experience the original situation and going over what it was like . . . To experientially remember something one must not only remember what happened but also remember what it was like. (pp. 13–14, final emphases added) As the previous passage highlights, when we consider cases of episodic remembering in adult humans it can seem as if representational content is somehow built into re-lived experiences. This is because, inevitably, in reflecting on such cases we contentfully represent and make claims about those relived experiences. Arguably, in general, the phenomenal character and representational content can, and sometimes do, come apart (see Block, 1990, Hutto, 2009). If we assume that this sort of separation can occur in episodic remembering then it follows that the phenomenal character of episodic remembering does not, by itself, entail or suffice for any representational content that may be connected with acts of episodic remembering. Indeed, being able to makes claims about one’s personal past—to have a truth-conditional take on it, as in cases of autobiographical memory that are distinguished from instances of pure forms episodic memory—is a special achievement, requiring the acquisition of additional cognitive machinery (Hutto, 2017). In particular, to make claims with truth conditional content requires the mastery of distinctive sociocultural practices (Hutto & Satne, 2015; Hutto & Myin, 2017). It might be thought that if episodic memory just is, or centrally involves, simulative imagining then it must be representational. This line of reasoning can seem compelling since imagining, by its very nature, deals with presence

The Roots of Remembering  111 in absence. If all imagining is necessarily representational then defeat for radically enactive accounts of imagination and memory follows briskly. However, on close inspection, there are strong reasons to doubt that basic forms of sensory imagination actually are, or even could be, representational (Medina, 2013; Hutto, 2015). A major reason for doubting that simulative imaginings are representational at all is that unlike other familiar cognitive attitudes—such as beliefs and desires—they appear to lack any inherent correctness or congruence conditions. For example, as Gerrans (2014) maintains, “qua simulations imaginative states do not have congruence conditions” (p. 105, see also p. 18). Or as Langland-Hassan (2015) tells us, “Much of what has been said about sensory imagination conflicts with the idea that imaginings have substantive correctness (or veridicality, or accuracy) conditions at all” (p. 665). Turning the tables on the usual line of argument, if it turns out that imaginings lack any kind of correctness conditions essentially, then we have grounds for thinking that pure forms of episodic remembering are necessarily not content-involving. This is because—at least in the ­analytic tradition—having some kind of correctness conditions is taken to be a ­necessary condition for being in a state of mind with representational ­content. We will not attempt to settle these important debates about the nature of sensory imaginings here. It suffices for our purposes to note that whatever their ultimate outcome, it is surely conceptually possible that imaginings, though they make a cognitive difference, not only lack the sort of content enjoyed by other canonical mental attitudes, but lack any kind of content whatsoever. Still, even if this is accepted, it might be thought that episodic remembering is content-involving in a different way and for a different reason. Coming at the question from the bottom up, so to speak, it might be held that imaginative simulation—even if it is not inherently contentful in character in-and-of-itself—nevertheless depends on and is best explained in terms of so-called subpersonal processes that involve the acquisition, storage, and processing of informational contents. For example, Michaelian (2016) holds that the best explanations of episodic remembering, understood as constructive and simulative, need to “assign an important role to information storage” (p. 8). By his lights, it is only by positing processes that access stored information that it is possible to explain how we imagine past episodes. Assuming tight links between remembering, imagining, and perceiving—in line with predictive processing accounts of these cognitive processes—it is assumed that what is cognized in such cases is a composite of content from multiple sources. Some content is taken to be supplied by low-level perceptual representations while other content is taken to be supplied by mental simulations, fueled in part by stored information. Clark (2016) offers a similar account. Conceiving of perceiving, imagining, and remembering as a package deal, he offers us a similar prediction-based

112  Daniel D. Hutto and Anco Peeters account of episodic memory that freely employs talk of the mixing of high and low-level contents that stem from incoming information and stored representations (p. 102). Like Michaelian’s (2016) characterization, Clark’s account of episodic remembering is dripping in cognitivist assumptions and vocabulary. Yet, despite the popularity of cognitivist characterizations of episodic rememberings, non-representational renderings are not only possible, they are arguably preferable. Why so? Consider that with respect to procedural memory, there is nothing to declare and nothing that is declared. For this very reason, many are prepared to agree with radical enactivists that there is no explanatory gain in assuming that rules for the relevant procedures are somehow captured in information that is stored and encoded within cognitive systems (see Loader, 2013; Sutton & Williamson, 2014; Michaelian, 2016). Still, some balk at thinking of declarative forms of memory—episodic and semantic—in the same vein. It is thought that the latter phenomenon somehow depends upon recovering information or content. Thus when we recall particular episodes in our lives or recall semantic facts we somehow retrieve information stored in memory traces, making it available to the organism again (see Michaelian, 2016, p. 26). Allegedly, stored information does special causal, explanatory work in declarative forms of remembering— work of a sort that there is simply no need for in nondeclarative forms of remembering. Illustrating this point by means of example, Michaelian (2016) tells us: Declarative memories are available to consciousness and thus can affect the activities of other cognitive systems. I remember that it has been sunny every day for the past week; I infer that it will be sunny today; I therefore leave home without bringing an umbrella. An appeal to stored information is essential to the explanation of my behavior in this case, for the memory does not cause the behavior directly, but only via an intermediate process of reasoning. Conscious access to stored contents allows them to enter into the causation of behavior in various indirect ways, via reasoning, imagination, and other routes; if the subject were not to have such access, declarative memory would be unable to provide inputs to other cognitive processes. Thus, were we to suppose that declarative memory does not store information, we would be unable to explain certain modifications in the behavior of the organism. (pp. 27–28, emphases added) Here lie philosophical monsters. To tell this story in a fully naturalistic register would require not only dealing with the hard problem of content, but also the hard problem of consciousness, as well as the problem of mental causation. And despite what Michaelian (2016) implies in his chapter when

The Roots of Remembering  113 giving his modally ‘musty’ assessment, a content-involving story about the roots of remembering is not, in fact, the only possible explanation available. Rather than consciously accessing the stored content “It has been sunny every day for the past week,” I perhaps form the judgment with that content on the basis of executing a few episodic simulations. I might re-enact something I was doing on each of the various days to reach a conclusion about the weather for the week. If so, so long as the set of simulations that ground such declarative remembering are not themselves content-involving, even though a contentful memory is formed, it will not have been formed by consciously accessing any stored contents. Episodic re-creative simulative imaginings could explain how we come to remember that it has been sunny every day for the past week without our having to posit any stored informational contents at all. And there seems to be no obvious reason why individual re-enactments themselves—even in full phenomenal glory—need involve the retrieval of any stored information. This is the crux. Why should we assume that the brain must make information available to organisms in order to explain how simulation works in terms of re-creative re-enactment? There appears to be no more reason to assume that stored information and contents must play a part in simulatively re-enacting an experience than there is for thinking they need play a part in re-creating the repeated acts involved in procedural remembering. Our recommendation is to think of episodic memory as strongly reconstructive (see Loader, 2013, p. 172). Radical enactivists see no reason at all to suppose that what lies at the roots of declarative memory differs from what lies at the roots of nondeclarative memory in this key respect. On the ground floor, neither form of memory gets its work done by making information or content stored in the brain available to the organism. If so, the capacities for experiential re-enactment—those that feature in pure episodic remembering and that are tied to be having certain kinds of phenomenal experience—need not make stored information available to the organism, as opposed to simply, partially re-creating patterns of activity in neural pathways in response to particular cues and prompts.

4 Conclusion Taken together, the considerations in this chapter give us grounds for maintaining that the roots of remembering need not involve and may not be best explained in terms of storing and retrieving any kind of remembered contents. If the analyses of this chapter prove sound, this will be true not only of procedural forms of remembering, but also of sophisticated kinds of remembering that benefit from environmental and social scaffolding, as well as reconstructive forms of experientially rich episodic remembering. The account we propose indeed understands the relevant kinds of remembering as the contentless activity of enactively constructing memories using biological, environmental, and social resources.

114  Daniel D. Hutto and Anco Peeters We do not expect that this will be the last word on these issues. In particular, extended mind theorists (especially those arriving on third wave) and those who favor reconstructive accounts of episodic remembering may be able to marshal responses to the challenges raised previously. It will be interesting to see, in the long run, to what extent a commitment to contentful representations and information-processing needs to be retained in our best explanations of what lies at the roots of remembering. To the extent that the next wave of extended mind theorists opt for more bidirectional, dynamic, and interactive visions of what remembering involves, a further question will be to determine to what extent we will need to retain a commitment to a familiar functionalist framework for thinking about minds (see Hutto, Peeters, & Segundo-Ortin, 2017). Rethinking how we manage to think about the past may indeed help to radically shift how we think about thinking in the future.

Notes 1 Notably, a commitment to neither the RCA nor the CCA is necessary for extended mind theorists. They may choose, for example, to endorse a non-representationalist functionalism. Nevertheless, most extended mind theorists do, de facto, make both commitments. For a discussion of a related point see Wheeler (2017, p. 460). 2 Clark and Chalmers (1998) rarely mention content but instead talk of the beliefs that comprise the having of certain memories as being constituted by information. On the standard assumption that the beliefs relevant to having certain memories— e.g., the belief that MoMA is on 53rd Street—are contentful it follows that, for Clark and Chalmers, information is either interchangeable with or entails content on their analysis of extended minds. Thus, as Coleman (2011) observes, in their seminal paper Clark and Chalmers “compare the storage of endorsed content in a notebook with the storage of endorsed content in the cranial biological memory” (p. 101). 3 For example, Barnier and Sutton (2008) tell us that “Just as different forms of memory within the individual can operate on the same information as it is transformed or abstracted or conventionalized (Toth and Hunt, 1999), so the same content can be transmitted across individuals, with its fate shaped by the available social resources and dynamics (Sperber, 1996)” (p. 179, emphasis added). 4 In later writings Sutton restricts the need to use information to reconstruct past experience to declarative forms of episodic memory (see, e.g., Sutton and Williamson, 2014). 5 Any enactive account of memory that eschews the idea that memory is fundamentally in the business of accessing and manipulating representational or informational contents, would count as a radically enactive account of memory. Following Loader’s usage, we treat enactive of memory as co-extensive with radically enactive accounts of memory. 6 As Debus (2007) illustrates “when you remember the last dinner party you went to, you might have an experience as if seeing again the person who sat opposite at the dinner. Or you might have an experience as if hearing once more some particular sounds or noises—the sudden bang emanating from the kitchen at some point, or a new tune played toward the end of the party. Indeed, in an attempt to describe those occurrences, we might say that ‘you see the person again in front of your mind’s eye,’ that you can ‘hear the tune in your head,’ and so on for the other senses” (p. 175).

The Roots of Remembering  115

References Adams, F., & Aizawa, K. (2001). The bounds of cognition. Philosophical Psychology, 14(1), 43–64. Adams, F., & Aizawa, K. (2010). Defending the bounds of cognition. In R. Menary (Ed.), The extended mind (pp. 67–80). Cambridge, MA: MIT Press. Aizawa, K., & Adams, F. (2005). Defending non-derived content. Philosophical Psychology, 18(6), 661–669. Barandiaran, X. E., & Di Paolo, E. A. (2014). A genealogical map of the concept of habit. Frontiers in Human Neuroscience, 8. Barnier. A. J., & Sutton, J. (2008). From individual to collective memory: Theoretical and empirical perspectives. Memory, 16(3), 177–182. Beckerman, B. (1962). Shakespeare at the Globe. New York: Palgrave Macmillan. Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press. Block, N. (1990). Inverted earth. Philosophical Perspectives, 4, 53–79. Casey, E. (1987). Remembering: A phenomenological study. Bloomington, IN: Indiana University Press. Clark, A. (1998). Contribution to: Hooker, C. A., O’Brien, G., Quinn, N. et al., Being there: Review symposia. Metascience, 7, 95–104. Clark, A. (2016). Surfing uncertainty: Prediction, action, and the embodied mind. Oxford: Oxford University Press. Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19. Coleman, S. (2011). There is no argument that the mind extends. Journal of Philosophy, 108(2), 100–108. De Brigard, F. (2014). Is memory for remembering? Recollection as a form of episodic hypothetical thinking. Synthese, 191(2), 155–185. Debus, D. (2007). Perspectives on the past: A study of the spatial perspectival characteristics of recollective memories. Mind & Language, 22(2), 173–206. Debus, D. (2008). Experiencing the past: A relational account of recollective memory. Dialectica, 62(4), 405–432. Garry, M., Manning, C. G., Loftus, E. F., & Sherman, S. J. (1996). Imagination inflation: Imagining a childhood event inflates confidence that it occurred. Psychonomic Bulletin and Review, 3, 208–214. Gerrans, P. (2014). The measure of madness. Cambridge, MA: MIT Press. Gerrans, P., & Kennett, J. (2010). Neurosentimentalism and moral agency. Mind, 119(475), 585–614. Gertler, B. (2007). Overextending the mind? In B. Gertler & L. Shapiro (Eds.), Arguing about the mind (pp. 192–206). New York: Routledge. Goldman, A. I. (2012). A moderate approach to embodied cognitive science. Review of Philosophy and Psychology, 3(1), 71–88. Heersmink, R. (2017a). Distributed selves: Personal identity and extended memory systems. Synthese, 194(8), 3135–3151. Heersmink, R. (2017b). The narrative self, distributed memory, and evocative objects. Philosophical Studies, 1–21. Huebner, B. (2016). Transactive memory reconstructed: Rethinking Wegner’s research program. The Southern Journal of Philosophy, 54, 48–69. Hutto, D. D. (2009). Mental representation and consciousness. In W. Banks (Ed.), Encyclopedia of consciousness. New York: Elsevier. Hutto, D. D. (2015). Overly enactive imagination? Radically re-imagining imagining. Southern Journal of Philosophy, 53(S1), 68–89.

116  Daniel D. Hutto and Anco Peeters Hutto, D. D. (2017). Memory and narrativity. In S. Bernecker & K. Michaelian (Eds), The handbook of philosophy of memory (pp. 192–204). London: Routledge. Hutto, D. D., & Myin, E. (2013). Radicalizing enactivism. Cambridge, MA: MIT Press. Hutto, D. D., & Myin, E. (2017). Evolving enactivism. Cambridge MA: MIT Press. Hutto, D. D., Peeters, A., & Segundo-Ortin, M. (2017). Cognitive ontology in flux: The possibility of protean brains. Philosophical Explorations, 20(2), 209–223. Hutto, D. D., & Satne, G. (2015). The natural origins of content. Philosophia, 43(3), 521–536. Intraub, H., & Richardson, M. (1989). Wide-angle memories of close-up scenes. Journal of Experimental Psychology: Learning, Memory and Cognition, 15, 179–187. Janssen, S M. J., Chessa, A G., & Murre, J. M. J. (2006). Memory for time: How people date events. Memory and Cognition, 34, 138–147. Kirchhoff, M. (2016). Composition and transactive memory systems. Philosophical Explorations, 19, 59–77. Langland-Hassan, P. (2015). Imaginative attitudes. Philosophy and Phenomenological Research, 90(3), 664–686. Lindsay, D. S., Hagen, L., Read, J. D., Wade, K. A., & Garry, M. (2004). True photographs and false memories. Psychological Science, 15, 149–154. Loader, P. (2013). Is my memory an extended notebook? Review of Philosophy and Psychology, 4, 167–184. Loftus, E. F. (2005). Planting misinformation in the human mind: A 30-year investigation of the malleability of memory. Learning and Memory, 12(4), 361–366. Loftus, E. F., Miller, D. G., & Burns, H. J. (1978). Semantic integration of verbal information into a visual memory. Journal of Experimental Psychology: Human Learning and Memory, 4, 19–31. Loftus, E. F., & Pickrell, J. E. (1995). The formation of false memories. Psychiatric Annals, 25, 720–725. Medina, J. (2013). An enactivist approach to the imagination: Embodied enactments and fictional emotions. American Philosophical Quarterly, 50(3), 317–335. Michaelian, K. (2012). Is external memory memory? Biological memory and extended mind. Consciousness and Cognition, 21, 1154–1165. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Mullally, S. L., & Maguire, E. A. (2014). Memory, imagination, and predicting the future: A common brain mechanism? The Neuroscientist, 20(3), 220–234. Neisser, U. (1996). Remembering as doing. Behavioral and Brain Sciences, 19, 203–204. Neter, J., & Waksberg, J. (1964). A study of response errors in expenditures data from household interviews. American Statistical Association Journal, 59, 18–55. Nigro, G., & Neisser, U. (1983). Point of view in personal memories. Cognitive Psychology, 15, 467–482. Ramsey, W. M. (2007). Representation reconsidered. Cambridge: Cambridge University Press. Ren, Y., & Argote, L. (2011). Transactive memory systems 1985–2010: An integrative framework of key dimensions, antecedents, and consequences. The Academy of Management Annals, 5, 189–229.

The Roots of Remembering  117 Roediger III, H. L. (1980). Memory metaphors in cognitive psychology. Memory & Cognition, 8(3), 231–246. Roediger III, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 803–814. Rosenberg, A. (2014). Disenchanted naturalism. In B. Bashour & H. D. Muller (Ed.), Contemporary philosophical naturalism and its implications London: Routledge. Rowlands, M. (1999). The body in mind: Understanding cognitive processes. Cambridge: Cambridge University Press. Schacter, D. L., & Addis, D. R. (2009). On the nature of medial temporal lobe contributions to the constructive simulation of future events. Philosophical Transaction of the Royal Society B: Biological Sciences, 364, 1245–1253. Schacter, D. L., Addis, D. R., & Buckner, R. L. (2007). Remembering the past to imagine the future: The prospective brain. Nature Reviews Neuroscience, 8, 657–661. Schacter, D. L., & Tulving, E. (1994). What are the memory systems of 1994? In D. L. Schacter & E. Tulving (Eds.), Memory systems (pp. 1–38). Cambridge, MA: MIT Press. Shanon, B. (1998). Metaphorical pluralism—not on the substantive level. Behavioral and Brain Sciences, 21, 164–165. Skorburg, J. A. (2017). Lessons and new directions for extended cognition from social and personality psychology. Philosophical Psychology, 30(4), 458–480. Sparrow, B., Liu, J., & Wegner, D. M. (2011). Google effects on memory: Cognitive consequences of having information at our fingertips. Science, 333(6043), 776–778. Stern, D. G. (1991). Models of memory: Wittgenstein and cognitive science. Philosophical Psychology, 4(2), 203–218. Sutton, J. (1998). Philosophy and memory traces: Descartes to connectionism. Cambridge: Cambridge University Press. Sutton, J. (2010). Exograms and interdisciplinarity: History, the extended mind, and the civilizing process. In R. Menary (Ed.), The extended mind, (pp. 189–225). Cambridge: MIT Press. Sutton, J. (2012). Memory. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Winter 2012 ed.). Retrieved from http://plato.stanford.edu/archives/ win2012/entries/memory/ Sutton, J., Harris, C. B., Keil, P., & Barnier, A. J. (2010). The psychology of memory, extended cognition, and socially distributed remembering. Phenomenology and the Cognitive Sciences, 9, 521–560. Sutton, J., & Williamson, K. (2014). Embodied remembering. In L. Shapiro (Ed.), The Routledge handbook of embodied cognition. London: Routledge. Szpunar, K. K., Watson, J. M., & McDermott, K. B. (2007). Neural substrates of envisioning the future. Proceedings of the National Academy of Science USA, 104, 642–647. Thompson, C. P., Skowronski, J. J., Larsen, S. F., & Betz, A. L. (1996). Autobiographical memory: Remembering what and remembering when. ­Mahwah, NJ: Lawrence Erlbaum. Tollefsen, D., Dale, R., & Paxton, A. (2013). Alignment, transactive memory, and collective cognitive systems. Review of Philosophy and Psychology, 4, 49–64. Tribble, E. B. (2005). Distributing cognition in the Globe. Shakespeare Quarterly, 56(2), 135–155.

118  Daniel D. Hutto and Anco Peeters Wade, K. A., Garry, M., Read, D., & Lindsay, D. S. (2002). A picture is worth a thousand lies: Using false photographs to create false childhood memories. Psychonomic Bulletin & Review, 9(3), 597–603. Wegner, D., Erber, R., & Raymond, P. (1991). Transactive memory in close relationships. Journal of Personality and Social Psychology, 61, 923–929. Wheeler, M. (2017). The revolution will not be optimised: Radical enactivism, extended functionalism and the extensive mind. Topoi, 36(3), 457–472.

6 Handle With Care

Dorothea DebusHandle With Care

Activity, Passivity, and the Epistemological Role of Recollective Memories Dorothea Debus 1  Setting the Scene The present chapter is concerned with one particular type of memory, namely with memories that have experiential characteristics. Such ‘experiential’ or ‘recollective’ memories are common, everyday occurrences. For example, try to remember what you did last Sunday. Chances are that when you do so, you’ll have at least some ‘recollective’ memories of some of the events that occurred in your life last Sunday. Recollective memories (or ‘R-memories’) are memories that have experiential characteristics. They are those cases of remembering that characteristically “correspond to our use of the distinct senses” (Martin, 2002, p. 403)—that is, they are those memories that are ‘as if’ seeing, hearing, touching, tasting, or smelling the remembered event (or process, or state of affairs) again.1 In our everyday mental lives, R-memories play an important epistemological role: When making judgments about the past, we often rely on our R-memories of relevant past events. It also seems that compared to other kinds of memories, we take R-memories especially seriously and give them special weight and particular attention when making judgments about the past. Thus, (Epistemological Role Claim) R-memories play a characteristic (and important) epistemological role, that is, they play a characteristic (and important) role with respect to a subject’s knowledge about the past. At the same time, it also seems true that (Activity and Passivity Claim) R-memories have characteristic features of activity as well as characteristic features of passivity: A subject who experiences an R-memory is characteristically passive with respect to the occurrence of the R-memory itself, but subjects nevertheless also can be, and often are, actively involved with respect to their R-memories. There are various ways in which we are actively involved with respect to our R-memories. For example, when you just tried, a moment ago, to remember what you did last Sunday, you quite actively did something (namely: think about what you did last Sunday), which in turn (probably) led to your having

120  Dorothea Debus various R-memories of the events that occurred in your life last Sunday. On the other hand, it also seems plausible to accept that whether or not relevant R-memories do occur is in some sense beyond our control; the relevant R-memories themselves do just ‘come to us’ (and sometimes they don’t); we are the passive ‘recipients’ of relevant experiences, and in this sense R-memories are clearly passive. Thus, it seems that both the Epistemological Role Claim as well as the Activity and Passivity Claim have some initial intuitive appeal and seem worth further exploration. What is more, there are important links between the epistemological role R-memories play on the one hand, and our R-memories’ characteristic features of passivity and activity on the other, and we can understand both these aspects of the phenomenon of R-memory better by setting out to understand them together. Thus, the following chapter aims to develop both the Epistemological Role Claim and the Activity and Passivity Claim in tandem, and it thereby aims to further our understanding of the phenomenon of R-memory as a whole.

2 R-Memories: Sometimes a Source and Ground of New Knowledge? What might the epistemological role of R-memories be? Which role do R-memories play with respect to our knowledge about the past? With respect to the phenomenon of memory as a whole, philosophers often hold that memories are a particular form of knowledge, namely retained knowledge. On this view, to remember something simply is to have knowledge that was acquired in the past and that has been retained.2 For example, Sally remembers that Charlemagne was crowned emperor of the Romans in the year 800, and Peter remembers that the first bike he had as a child was red. It seems quite plausible to hold that these memories should count as pieces of retained knowledge; the subject’s memory of the relevant fact is a piece of knowledge the subject acquired in the past and that she has retained. However, it is sometimes suggested that not all memories are cases of retained knowledge. Rather, so it is said, some memories are, for the remembering subject, a source and ground of new knowledge about the past.3 I will here understand these claims as follows: (Source) A mental occurrence is a source of new knowledge for the subject iff the mental occurrence plays an important causal role in bringing about the relevant piece of new knowledge. (Ground) A mental occurrence is a ground of new knowledge for the subject iff the mental occurrence provides the subject with a reason to endorse the relevant new belief, a reason which in turn is needed for the relevant new belief about the past to count as a piece of knowledge.

Handle With Care  121 In an attempt to develop the claim that some memories are, for the remembering subject, a source and ground of new knowledge about the past, it seems that from among the various different kinds of memories, R-memories might be a particularly good candidate, that is, it might seem especially plausible to say that R-memories sometimes can be a source and ground of new knowledge about the past. For example, (Anna’s R-memory of yesterday’s lecture) Mary asks Anna whether Paul attended her lecture yesterday afternoon. Anna does not know how to answer this question, so in an attempt to find an answer to Mary’s question, Anna tries to visually remember the audience during yesterday’s lecture; her attempt is successful, she has a visual experience as of Paul in the left back corner of the lecture room, and on the basis of her visual memory of the audience during yesterday’s lecture, Anna comes to judge that Paul did indeed attend her lecture yesterday afternoon.4 It seems plausible to hold that in cases like Anna’s R-memory of yesterday’s lecture, an R-memory is the source of new knowledge about the past for the remembering subject—that is, the R-memory plays a causal role in bringing about a relevant new piece of knowledge about the past—and the R-memory also is a ground of new knowledge about the past for the remembering subject—that is, the R-memory provides the subject with a reason to endorse the relevant new belief, a reason that in turn is needed for the relevant new belief about the past to count as a piece of knowledge. Thus, in an attempt to develop the Epistemological Role Claim, it might seem quite plausible to accept that (Source-and-Ground-Claim) at least sometimes, R-memories can be a source and ground of new knowledge about the past for the remembering subject. And indeed, this claim can in turn be supported further with the help of some considerations about the feature of passivity that, as highlighted by the ‘Activity and Passivity Claim,’ is a characteristic feature of our everyday R-memories.

3 Being Passive: Experiencing the Past and Gaining New Knowledge About the Past When a subject experiences an R-memory, the subject is in some important sense passive in the face of the occurrence of the R-memory itself. Sometimes, R-memories just ‘come to us’—they simply occur completely out of the blue; at other times, we are thinking about something we did in the past, and in that context various thoughts, and various R-memories, occur; at yet other times, we might quite explicitly try to remember something about the

122  Dorothea Debus past, and in an attempt to do so, we ask ourselves questions about the past (e.g., “What did I do last Sunday?”) and relevant R-memories sometimes then occur ‘in response’ to those questions; but whether or not relevant R-memories do occur ‘in response’ to such questions is not under the subject’s control. Thus, it seems that in the cases just sketched, the subject herself is passive with respect to the actual occurrence (or nonoccurrence) of the relevant R-memory. Whether or not the relevant R-memory occurs is beyond the subject’s control. This feature of passivity, which seems to characterize most (indeed, maybe all) R-memories, might in turn suggest an analogy between R-memories on the one hand and perceptual experiences on the other. For, just like R-memories, perceptual experiences also seem to be characterized by the subject’s being passive in some way with respect to the relevant experience.5 In the case of a perceptual experience, it is usually agreed by all that the characteristic passivity of the subject in the face of the perceptual experience’s occurrence is due to the subject’s present environment ‘impinging’ in some way upon the subject at the time at which the perceptual experience occurs. Quite analogously, so we might say, in the case of an R-memory, the characteristic passivity of the subject in the face of the R-memory’s occurrence is due to the subject’s past environment ‘impinging’ upon the subject in some way at the time at which the R-memory occurs.6 Just as perceptual experience is in some sense passive because in perceptual experience we are receptive to our present environment, so R-memory is in some sense passive because in R-memory we are receptive to our past. Something of the past is ‘given’ to us in R-memory—a subject who R-remembers a certain past event is presented with the relevant past event in experience.7 I have suggested elsewhere that we can develop this view further with the help of a ‘Relational Account’ of R-memory (see Debus, 2008). According to a ‘Relational Account’ of R-memory, a subject who R-remembers a past event is directly aware of the relevant past event in R-memory, that is, the subject stands in a direct experiential relation to the relevant past event.8 If the past ‘impinges’ on us when we R-remember a certain past event, that is, if something of the past is ‘given’ to us in R-memory, it seems very plausible to hold that R-memories can be a source and ground of knowledge about the past for the R-remembering subject. Thus, if we accept that the characteristic passivity of a subject in the face of an R-memory’s occurrence is indicative of, and indeed due to, the subject’s relevant past environment ‘impinging’ upon the subject at the time at which the R-memory occurs, that something of the past is ‘given’ to the subject in R-memory, then it also seems plausible to accept the Source-and-Ground Claim.

4  Indicative, but Neither Necessary nor Sufficient However, it is important to note that while the feature of passivity that R-memories have is indicative of their being a source and ground of

Handle With Care  123 knowledge about the past, a relevant feature of passivity is neither necessary nor sufficient for an experience to be a source or ground of knowledge about the past. One might gain knowledge about the past by actively setting out to imagine the past, but relevant imaginary experiences are in important respects characterized by some sort of activity. Thus, it is not necessary that a relevant experience have characteristic features of passivity in order for a subject to gain new knowledge about the past from it. One might also find oneself with experiences that are somehow characteristically passive—vivid daydreams, for example, which, just like R-memories, might be characterized by some sort of passivity—but that should most definitely not count as sources of knowledge. Thus, it is not sufficient that a relevant experience have characteristic features of passivity for a subject to gain new knowledge about the past from it either. We find that an experience’s passivity is neither necessary nor sufficient for the relevant experience to count as a source and ground of new knowledge for the experiencing subject. Nevertheless, it is plausible to think of the relevant passivity in the case of R-memories as indicative of the subject’s past environment ‘impinging’ upon the subject at the time at which the R-memory occurs. Thus, the fact that R-memories do have a characteristic feature of passivity does give us good reason to hold that (Source-and-Ground-Claim) at least sometimes, R-memories can be a source and ground of new knowledge about the past for the remembering subject.

5  Dummett’s Objection However, not everybody agrees with this conclusion. Indeed, Michael Dummett states his opposition quite unequivocally when he says that “[m]emory is not a source, still less a ground, of knowledge: it is the maintenance of knowledge formerly acquired by whatever means” (Dummett, 1996, p. 420f.). The argument Dummett offers in support of this opposing view might be reconstructed as follows: Dummett’s Objection (i) If memory was not knowledge retained, but was instead a source or ground of new knowledge about the past, all knowledge about the past would have to be inferential knowledge. (ii) If all knowledge about the past was inferential knowledge, we could not possibly have any knowledge about the past at all. (iii) But then, we do have knowledge about the past. (C) Hence, “memory is not a source, still less a ground, of knowledge; it is the maintenance of knowledge formerly acquired by whatever means” (Dummett, 1996, p. 420f.).

124  Dorothea Debus In response, we should consider the objection’s premise (i) in greater detail. It seems that premise (i) tacitly assumes that we can ascribe one and the same epistemological role to all memories. As we will see later, this is a problematic assumption. But to begin with, I suggest we assess premise (i) as it stands more carefully and try to formulate it as precisely as possible. Indeed, given that the consequent of premise (i) speaks about all knowledge about the past, the antecedent of premise (i) should be taken to be referring implicitly to all cases of memory. Thus, a more precise formulation of premise (i) should run as follows: (i*) If it was the case that, instead of being knowledge retained, all memories were a source or ground of new knowledge about the past, all knowledge about the past would have to be inferential knowledge. Next, we might ask why precisely we should accept this conditional. Clearly, a defender of premise (i*) assumes that in order for a mental occurrence, such as a memory, to count as a source and ground of new knowledge, the subject needs to engage in some inferential reasoning to gain the relevant piece of knowledge on the basis of the relevant mental occurrence. This might be contentious. Indeed, one might (in my view, quite plausibly) deny this and argue instead that certain mental occurrences, such as R-memories or perceptual experiences, should count as a source and ground of new knowledge while the relevant newly acquired pieces of knowledge do not depend on any inferential reasoning at all and thus should not count as inferential knowledge. Thus, one way of resisting the conclusion of the present objection would be to deny the truth of its premise (i*). However, for present purposes it might be easier to grant our opponent premise (i*), and to focus instead on the structure of the argument as it stands. Indeed, we find that even if we accept premise (i*) for the sake of the argument, once we have clarified the actual meaning of premise (i) with the help of premise (i*), it turns out that Dummett’s Objection poses no threat to the Source-and-Ground Claim. For, with the relevant quantifiers clearly in place, we can formulate a more precise version of Dummett’s Objection as follows: Dummett’s Objection* (i*) If it was the case that, instead of being knowledge retained, all memories were a source or ground of new knowledge about the past, all knowledge about the past would have to be inferential knowledge. (ii) If all knowledge about the past was inferential knowledge, we could not possibly have any knowledge about the past at all. (iii) But then, we do have knowledge about the past. (C*) Hence, it is not the case that all memories are a source, or a ground, of new knowledge about the past; at least some memories must be “the

Handle With Care  125 maintenance of knowledge formerly acquired by whatever means” (Dummett, 1996, p. 420f.). At first sight, conclusion (C*) seems to differ importantly from conclusion (C), but if one assumes that all memories are mental occurrences of the same kind, and thus that all memories play one and the same epistemological role, then conclusion (C*) is equivalent to conclusion (C).9 However, it seems plausible to accept that the various everyday occurrences we subsume under the label ‘memory’ are of importantly different kinds, and it seems plausible to assume that those different kinds of memory also do play different epistemological roles. Indeed, it seems quite plausible to hold that some of our memories—namely factual memories—are cases of retained knowledge (and thus are not a source or ground of knowledge), while some other memories—namely R-memories—might, at least sometimes, very well be the source and ground of new knowledge for the remembering subject (and so are not, or at least not always, retained knowledge). This view is perfectly compatible with conclusion (C*) of the revised version of Dummett’s Objection. Thus, as long as we do (as seems plausible) grant that different kinds of memories might play different epistemological roles, Dummett’s Objection does not pose any threat to the Source-andGround-Claim at all.

6  The Previous Awareness Condition Nevertheless, an opponent might still want to resist the Source-and-GroundClaim; indeed, so the opponent might say, the Source-and-Ground-Claim is implausible because it seems unclear how it might be compatible with an important condition that needs to be met in order for any R-memory to count as a memory, a condition that, following Shoemaker (1970), is usually called the ‘previous awareness condition.’ In order for someone to recollectively remember a past event, so it seems plausible to accept, it is necessary that the relevant subject was in some way aware of the relevant event when it originally occurred; or, in Shoemaker’s words, “it is a necessary condition of its being true that a person remembers a given past event that [s]he, that same person, should have observed or experienced the event, or known of it in some other direct way, at the time of its occurrence” (Shoemaker, 1970, p. 269). But then, so an opponent to the Source-and-Ground-Claim will say, if this condition needs to be met for a mental occurrence to count as an R-memory at all, it could not possibly be true that at least sometimes, R-memories are a source and ground of new knowledge about the past for the remembering subject; for if it is necessary that a subject who now R-remembers a certain event has “observed or experienced the relevant event, or known of it in some other direct way” in the past, then whatever knowledge the relevant R-memory might provide the subject with now could not possibly be new

126  Dorothea Debus knowledge for the subject, given that the subject did already know about the relevant event in the past. Accordingly, so our opponent concludes, the Source-and-Ground-Claim must be false.10 To assess the present objection, we should consider the ‘previous awareness condition’ a little more carefully, and ask what precisely the ‘previous awareness condition’ might plausibly demand. It seems eminently reasonable to accept that in order for someone to recollectively remember a past event (or a past state of affairs), it is necessary that they did in some way witness it when it occurred (or obtained)—that is, that they were in some way aware of the event (or state of affairs), which they presently remember when it originally occurred (or obtained); this much seems to be a conceptual truth—it captures how we use the word ‘memory’ in relevant cases—a truth the previous awareness condition aims to capture. Indeed, it seems to be a part of our concept of memory that for a subject to be said to remember anything now, she needs to have earlier been in some sort of contact with the information that her memory now presents her with.11 However, what seems less clear is what kind of ‘awareness’ of the original event at the time at which it occurs is required in order for the subject to be said to remember the relevant event. More specifically, we should ask whether the kind of previous ‘awareness’ that is required for R-memory is such that it should always already count as a form of knowledge, as Shoemaker seems to imply when he holds that “it is a necessary condition of its being true that a person remembers a given past event that [s]he, that same person, should have observed or experienced the event, or known of it in some other direct way, at the time of its occurrence” (Shoemaker, 1970, p. 269). Indeed, consideration of cases like Anna’s as described earlier might suggest otherwise; when Mary asks Anna whether Paul attended her lecture yesterday afternoon, so we said earlier, Anna does not know how to answer this question, but once she has visually remembered Paul in the left back corner of the lecture room during yesterday’s lecture, Anna comes to judge, and does come to know, that Paul attended her lecture yesterday afternoon. Taking the phenomenology of cases like the present one at face value, it would seem plausible to hold that while Anna might have been aware of Paul’s presence in some sense at the time of the lecture, we probably should not say of her that at the time she knew that Paul was present.12 Rather, it seems plausible to hold that her awareness of Paul’s presence at the time, on which her present R-memory of Paul’s presence is grounded, is of a much more basic form; we might say that she was ‘aware’ of Paul’s presence during the original event in so far as her visual system somehow ‘registered’ Paul’s presence at the time; it might be possible for a subject’s visual system to ‘register’ relevant information without the subject herself at all being conscious of the relevant information at the time; alternatively, the relevant information might be conscious in only a very fleeting way, with the subject not paying any attention to it at all. So, as far as Anna’s case is concerned, maybe Anna’s visual system ‘registered’ Paul’s presence during the lecture, but Anna was not really conscious of Paul’s presence at the time, or else

Handle With Care  127 maybe Anna was in some sense conscious of Paul’s presence at the time, but only in a very fleeting way; in either case, given the situation as described earlier, it seems intuitively plausible to hold that at the time of the lecture, Anna did not have any knowledge about Paul’s presence or absence, and that only now (at the time of remembering yesterday’s lecture) does she come to know that Paul did attend yesterday’s lecture on the basis of her present R-memory of the event. This seems true to the phenomenology—subjects in a situation like Anna’s often seem to experience their own situation as one in which they gain new knowledge about the past on the basis of their current R-memory—and it also seems compatible with the previous awareness condition, once we realize that the relevant kind of ‘previous awareness’ that is required for an R-memory might be awareness of a very basic form indeed. Cases like Anna’s case as described do occur in ordinary, everyday contexts, and it seems that our description of the case captures some salient features of the relevant everyday phenomenology. What is more, under the description we have offered, it seems uncontroversial to say that Anna does R-remember Paul’s presence during her lecture. But then, assuming that R-memory presupposes ‘previous awareness,’ we will also have to hold that Anna does meet the ‘previous awareness condition’ in a relevant sense, which in turn gives us reason to hold that a subject can meet the ‘previous awareness condition’ without, in doing so, acquiring any relevant knowledge at the time. This in turn has two implications: First, we find that a plausible reading of the ‘previous awareness condition’ will not demand that in order for a subject to be said to R-remember anything at all, it is necessary that she have previously acquired any knowledge about what she is now said to remember; Shoemaker’s formulation of the previous awareness condition therefore is too strong. In order for someone to recollectively remember a past event, so it seems plausible to accept, it is necessary that the relevant subject was in some way aware of the relevant event when it originally occurred, but it is not necessary that the subject’s relevant previous awareness entailed any form of knowledge on the part of the subject at the time. Second and more importantly for present purposes, given that a plausible formulation of the awareness condition does not require the subject to have acquired any knowledge at the time of previous awareness in order for her later to be said to recollectively remember a relevant event or state of affairs, we also find that we can continue to endorse both the Source-and-Ground-Claim as well as the previous awareness condition.13 Thus, our opponent’s present objection is unsuccessful; we find that, against the opponent’s assumptions, the Source-and-Ground-Claim is compatible with the previous awareness condition, once it is formulated appropriately. Hence, we can continue to hold that (Source-and-Ground-Claim) at least sometimes, R-memories can be a source and ground of new knowledge about the past for the remembering subject,

128  Dorothea Debus and the feature of passivity that is characteristic of everyday R-memories is indicative of this special epistemological role that R-memories can play in our everyday mental lives.

7  Active Involvement (e.g., Switching Perspectives) However, while everyday R-memories are characterized by a feature of passivity, we also are, as we saw earlier, actively involved with our everyday R-memories, and this in turn might throw doubt on whether R-memories ultimately can play the epistemological role described by the Source-andGround-Claim. In some contexts, a subject’s active involvement with her everyday R-memories might well seem epistemologically ‘harmless,’ and indeed might be thought of as epistemologically valuable. For example, a subject might ask herself a question about a particular past event, and an R-memory might occur to her ‘in response’ to the relevant question. In those cases, a subject’s active involvement seems epistemologically valuable because it is only in the context of such active involvement that the relevant R-memory (which might in turn be a source and ground of new knowledge) does occur first of all. However, in other contexts the possibility of our being actively involved with our everyday R-memories might seem to pose a problem for anybody who wants to hold that R-memories do play the epistemological role set out by the Source-and-Ground-Claim. For example, some people can ‘switch perspectives’ when R-remembering past events. At first, they might R-remember a particular past event from the perspective from which they originally witnessed it. (In the empirical literature, this perspective is usually called a ‘field-perspective.’) They are then asked to ‘switch perspectives’ and try to experience the same particular past event from an ‘observerperspective,’ a perspective that was not their own perspective at the time at which they originally witnessed the relevant event. And indeed, some subjects can successfully manipulate their R-memories in this way: While they first R-remember a particular past event from a field-perspective, having manipulated that experience according to the instructions they have been given, they then do bring it about that they have a new and different experience, namely an experience as of the very same past event, but now from an observer-perspective (see e.g., Rice & Rubin, 2009).14 What are we to say about the epistemological status of experiences that are the result of such active and willful manipulation on the part of the subject?

8  An Epistemological Worry There might be reason to doubt whether any experience that is the result of such an active intervention on the part of the experiencing subject could nevertheless still count as the source and ground of new knowledge

Handle With Care  129 about the past for the subject. For as we saw earlier, R-memories are (at least sometimes) a source and ground of new knowledge because the past ‘impinges’ on the R-remembering subject as a passive recipient. An R-memory’s epistemological value thus seems bound up in some way with the R-memory’s characteristics of passivity, and any active intervention on the part of the experiencing subject might well seem to ‘spoil’ those passive characteristics. Indeed, a subject’s active manipulation of an R-memory might accordingly be thought of as rendering the R-memory epistemologically worthless. More systematically, relevant doubts might be supported by something like the following train of thought:

Epistemological Worry (i)

(Causal Condition) In order for an experience to count as a source and ground of new knowledge about a particular past event (or process, or state of affairs) for a subject, it is necessary that the experiencing subject stand in an appropriate causal relation to the relevant past event (or process, or state of affairs) about which the experience is said to provide her with new knowledge.15 (ii) But then, an active intervention on the part of the subject with respect to a particular R-memory will, at least frequently (and indeed, maybe always), change the causal relation that obtains between the R-remembered past event (or process, or state of affairs) and the R-remembering subject.16 (iii) An intervention that changes the causal relation that obtains between the R-remembered past event (or process, or state of affairs) and the R-remembering subject might render the relevant causal relation inappropriate for the experience that is the result of the intervention to count as a source or ground of new knowledge about the past for the subject any further. (C) Thus, an active intervention on the part of the subject with respect to a particular R-memory might potentially render the resulting experience epistemologically useless. Whether the potential problem expressed by the conclusion of this Epistemological Worry should ultimately be considered to be an actual problem will importantly depend on what we think about premise (iii) of the present train of thought. Premise (iii) presupposes a distinction between ‘appropriate’ and ‘inappropriate’ causal relations between relevant past events and subjects who presently have experiences as of those past events. But then, what are the features that make a causal relation appropriate, and what are the features that make a causal relation inappropriate for epistemological purposes in the present context? How should we draw the distinction between ‘appropriate’ and ‘inappropriate’ causal relations here?

130  Dorothea Debus

9  Deviant Causal Chains17 In other contexts in which analogous questions arise (e.g., in the philosophy of action, or the philosophy of perception), philosophers usually tackle those questions by offering exemplary cases of causal relations that should count as inappropriate because they are in some way ‘deviant.’ One feature the various exemplary cases of ‘deviant causal chains’ have in common is that the relevant ‘deviant’ element in the causal chain is not a very ‘reliable’ causal element but seems to occur by chance somehow, so that the occurrence of the event for which the causal chain is important is itself in an important way due to chance. This seems true of Davidson’s mountaineer (as described in Davidson, 2001, 79), as well as of Lewis’ cases of ‘veridical hallucination’ (see e.g., Lewis, 1980), and in the present context, we might describe an analogous case of ‘veridical memory-hallucination,’ as follows: (Crazy Neuroscientist) A crazy neuroscientist has disabled her subject’s neurophysiological ‘recollective apparatus,’ but she brings it about that her subject has experiences of exactly the kind he would have if he was in a position to R-remember various past events. There is a causal chain which links relevant past events, via the neuroscientist, to the subject; but as soon as the crazy neuroscientist’s mood changes, the neuroscientist will make it the case that none of the experiences the subject has are of the kind he would have if he was in a position to R-remember his past at all. It seems plausible to hold that the causal chain that obtains, via the neuroscientist, between relevant past events and the subject who presently has experiences as of these past events is a deviant causal chain. Most people agree that the cases that are said to be characterized by ‘deviant causal chains’ are cases in which the relevant causal link is inappropriate, while they also agree that in other, ordinary cases relevant causal links should count as appropriate, and this in turn supports the view that a distinction between appropriate and inappropriate causal links should be drawn. (However, at the same time it is also widely accepted that no one has so far been able to offer a clear answer to the question as to how precisely that distinction is to be drawn.)

10  Adhering to Epistemological Norms How do cases in which a subject is actively involved with her own R-memories compare to such cases of deviant causal chains? It certainly seems plausible to hold that at least in some (and indeed, probably many) cases in which a subject is actively involved with her R-memories, relevant causal links should not count as ‘deviant causal chains’ in the sense just described. For example, when a subject who R-remembers a past event manipulates her

Handle With Care  131 R-memory with respect to its perspective, the resulting causal chain certainly does not seem to entail any elements of chance whatsoever. Quite on the contrary, in ‘switching perspectives,’ the subject adheres to certain norms—namely, norms regulating the spatial manipulation of mental imagery that are in turn designed to preserve information accurately during the process of intervention. But then, as long as a subject does adhere to such ‘information-preserving’ norms while manipulating an R-memory, it seems that the experience that is the result of the relevant intervention should be of the same epistemological value as the original R-memory on which the subject performed the relevant intervention first of all. This in turn should help us to assess premise (iii) of the Epistemological Worry: We might endorse that premise, but we have now also seen that at least some active interventions on R-memories on the part of the R-remembering subject (such as switching the perspective of an R-memory) will not render the relevant causal relation inappropriate. Thus, at least with respect to some cases of active interventions on R-memories on the part of the R-remembering subject, the Epistemological Worry can be put to rest. More generally, present considerations suggest that relevant interventions will not pose a threat to the epistemological status of R-memories as a source and ground of new knowledge about the past as long as a subject’s active involvement with respect to her R-memories adheres to relevant epistemological norms. Indeed, as long as a relevant intervention is governed by relevant epistemological norms, an intervention on an R-memory does not render the causal relation that obtains between the R-remembered past event (or process, or state of affairs) and the R-remembering subject inappropriate for the experience that is the result of the intervention to count as a source or ground of new knowledge about the past for the subject.

11  Taking Responsibility The observation that we can, but might not, follow certain epistemological norms when being actively involved with our R-memories in turn highlights the fact that each of us might well also bear some responsibility for the ways in which we are, and are not, actively involved in our mental lives quite generally, and with our R-memories more specifically. A subject who is actively involved with her R-memories might adhere to relevant norms, or she might not, and as long as it is in some way up to the subject whether she does or does not adhere to relevant norms while manipulating her R-memories, the subject might also bear responsibility for relevant interventions. For example, consider Matt: (Matt’s case) Matt has very low self-esteem and feels especially insecure in the presence of his boss, and he (wrongly) thinks that his boss thinks poorly of his work. Matt presently R-remembers a meeting he had with

132  Dorothea Debus his boss last week, and while Matt at first accurately R-remembers his boss’ broad and enthusiastic smile in reaction to some of his suggestions, dwelling on the encounter further he manipulates his original R-memory and ‘super-imposes’ imaginary experiences as of his boss looking at him rather disappointedly. The resulting experience—an experience as of his boss’ office, with both of them seated at her desk, and her looking at him disappointedly—is accurate in most respects, but not with respect to the boss’ facial expressions. We can easily offer psychological explanations as to why Matt manipulates his R-memory in the way as described—we might say that it is Matt’s low self-esteem that prompts him to search out, and if unavailable, to fabricate (apparent) evidence in support of his belief that his boss is very critical of him, and so it is his low self-esteem that prompts him to imagine his boss looking at him disappointedly, even though she was in fact smiling enthusiastically at him at the time. However, it also seems plausible to accept that it is epistemologically irresponsible of Matt to succumb to the temptation of manipulating his R-memory in this way. To make epistemological use of our R-memories, we need to make sure that our R-memories remain accurate. One of the norms by which we ordinarily regulate our active involvement with our R-memories therefore is a norm which says that one must not insert falsehoods by means of imagination into one’s R-memories. Matt does violate this norm, which in turn gives us reason to hold that his active involvement with his R-memory is epistemologically irresponsible.18 More generally, present considerations suggest that, while R-memories can be a source and ground of new knowledge about the past for the R-remembering subject, memories quite generally, and R-memories in particular, are also very fragile (Schacter, 1996). Thus, in order for R-memories to provide us with a source and ground of new knowledge about the past, when being actively involved with our R-memories we have to show great sensitivity in the face of their fragility. In order to make any epistemological use of our R-memories at all, it is important not to engage in retrospective imaginary activities that ‘block out’ the past from our present awareness. Instead, we have to make sure that we remain ‘receptive’ to the past in R-memory, and that we let the past ‘impinge’ upon us when we R-remember past events (or processes or states of affairs). The question as to how exactly subjects are able to show this sensitivity, and to bear relevant epistemological responsibility, when being actively involved with their R-memories (and their wider mental lives) will have to wait for another occasion. For the time being, we can conclude that R-memories’ characteristic feature of passivity gives us reason to hold that R-memories (sometimes) can be a source and ground of new knowledge about the past; subjects are also actively involved with their R-memories, but as R-memories are fragile, it is important that subjects adhere to relevant epistemological norms and take epistemological responsibility when they are actively involved with

Handle With Care  133 their R-memories, so that relevant R-memories can continue to play their epistemological roles.19

Notes 1 In order to clarify which cases of remembering should count as ‘R-­memories,’ it might also be useful to relate the phenomenon of R-memory to other ­classifications of the various phenomena of remembering that are offered in the recent philosophical and psychological literature (cf. e.g., Shoemaker [1967], Tulving [1983], and Martin [2001: 258ff.]). There is no time to do so here, but I do so in Debus (2007), Section 1. 2 In the relevant literature this position is sometimes called ‘preservationism’; for a helpful summary of the view see Bernecker and Grundmann (2017), who also offer a useful list of relevant contributions to the recent literature on the topic (see their fn.1). 3 See Bernecker and Grundmann (2017: fn. 2) for a list of authors who have recently considered and endorsed various different versions of the view that memory is not just the preservation of knowledge, but might sometimes also ‘generate’ new knowledge. In the relevant literature this view is sometimes called ‘generativism,’ and in their paper, Bernecker and Grundmann (2017) offer their own novel form of ‘generativism’ about memory. As will become clear in a moment, my own contribution to the debate here focuses on one particular kind of memory, namely R-memory, and the suggestion that R-memories might sometimes be a source and ground of new knowledge about the past for the remembering subject. 4 A detailed discussion of a case similar to the one described here is offered in Lackey (2005: 650–653). 5 See also Eilan (1998), who offers an interesting discussion of a similar analogy. 6 The claim that something ‘impinges’ on the subject is clearly a little metaphorical. For the purposes of developing the relevant train of thought in the main text, I think we can operate at this metaphorical level, but for anybody who seeks greater precision it might be useful to add that I think that we could, and should, develop the metaphor used here in causal terms: Just as in perception, the subject’s present environment might be said to ‘impinge’ on the subject because it causally affects the subject, so in R-memory the subject’s past environment might be said to ‘impinge’ on the subject because it causally affects the subject. (I offer some more detailed considerations of this issue in Debus [2008] and Debus [2017].) Indeed, this seems the most plausible way of accounting for the phenomenology here. Of course, those (e.g., Michaelian [2016]; for relevant discussion see also Perrin [this volume]) who hold that no relevant causal relation obtains between a past event and a present R-memory of the relevant past event at all will have to find other ways of accounting for the phenomenological feature which I have here described by saying that something ‘impinges’ on the subject; there is no room to engage with that debate in the present context, but I think it is clear that the causal suggestion sketched here will offer the most straightforward (and presumably also the most immediately plausible) account of the phenomenology. (Thanks to Denis Perrin and Kirk Michaelian for prompting the present comment.) 7 The suggestion that something is ‘given’ to us in R-memory is here to be ­understood in a perfectly innocent way; there is no danger here of a ‘myth of the given,’ as discussed by Sellars (1997). 8 An attentive reader might worry that the claim that the past ‘impinges’ on the subject in R-memory, that is, that the past in some way causally affects the

134  Dorothea Debus subject at the time at which the R-memory occurs, is incompatible with the claim that in R-remembering a relevant past event, the subject is directly aware of the relevant event. I address this worry in Debus (2008), where I show that both claims are ultimately compatible. (Thanks to Kirk Michaelian for prompting the present comment.) 9 This might be spelled out in greater detail as follows: If all memories play the same epistemological role, and if (as conclusion (C*) says) it is the epistemological role of at least some memories to be “the maintenance of knowledge formerly acquired by whatever means,” then it must also be true that “memory [quite generally, i.e., all memory] is the maintenance of knowledge formerly acquired by whatever means,” as conclusion (C) has it. Thus, if one assumes that all memories are mental occurrences of the same kind, and thus that all memories play one and the same epistemological role, then (C*) is equivalent to (C). (Thanks to Denis Perrin for prompting the present clarification.) 10 I am grateful to Christoph Hoerl for suggesting in conversation that I consider the present train of thought more carefully. 11 All this seems rather plausible, but is not uncontested; indeed, Michaelian (2016) attempts to develop an account of memory that does not include a previous awareness condition. 12 Similar examples have been used in the literature to make the same point, e.g., by Lehrer and Richard (1975: 122), who in turn acknowledge Thomas Reid as the original source of the example they use. 13 Of course, someone might oppose the present train of thought by arguing that relevant states of previous awareness of the kind just described should, against first appearances, count as cases of knowledge after all. However, I doubt that anybody would want to suggest that calling relevant states of previous awareness cases of knowledge is the most intuitive way to describe relevant situations. Accordingly, anybody who wants to defend this view will have other (theorydependent) reasons as to why it would seem important to show that, against first appearances, relevant states of previous awareness should count as cases of knowledge. Presumably, one reason for trying to defend this claim would be a wish to defend a theory of memory according to which all cases of memory are cases of retained knowledge. However, it seems clear that any argument relying on the claim that all memory is retained knowledge would be begging the question in the present context, so in the present context a defender of the claim that relevant states of previous awareness should, against first appearances, count as cases of knowledge would have to find an argument that does not presuppose the truth of the claim that all memory is retained knowledge, and it might be difficult to do this. But be this as it may, all I aim to show in the present context is that the previous awareness condition itself does not entail the claim that memory could not possibly be a source and ground of new knowledge, and this has, as far as I can see, been successfully established in the main text. (Thanks to Denis Perrin for prompting the present comment; for related discussion see also Bernecker (2010: ch.3) 14 Some people think that ‘observer-memories’ should not count as cases of ­memory at all; I consider the nature of relevant mental occurrences in detail and show that and why we can and should take relevant cases to be cases of memory in Debus (2007). The phenomenon of ‘observer memory’ has recently also found some wider philosophical attention—see e.g., Sutton (2010) and McCarroll (2017). 15 I discuss this causal condition in greater detail in Debus (2017). 16 To say that an intervention might ‘change’ the causal relation that obtains between the R-remembered past event and the R-remembering subject is simply

Handle With Care  135 meant to indicate here that because of a relevant intervention, the relevant causal relation might be different from what it would have been like had the relevant intervention not occurred. (Thanks to Kirk Michaelian for prompting the present clarification.) 17 In the following, we will focus on one issue related to the role which causation might play in memory, namely the question how to distinguish between ‘appropriate’ and ‘inappropriate’ causal relations in this context. However, departing from Martin and Deutscher’s (1966) seminal work, the more general question as to which role causation should play in an account of memory has been rather prominent in the philosophy of memory during the last fifty years or so. I offer a survey in Debus (2017); further relevant work is offered by Michaelian and Robins (this volume). 18 On most views of responsibility, ascribing responsibility to Matt does p ­ resuppose that Matt can be, and is, in some way ‘actively involved’ with how his own ­mental life develops in the relevant respect; and indeed, under the assumption that Matt is a healthy mature human being, I think this presupposition is reasonable. There is no room to say more about this here, but I consider our ability to be actively involved in our own mental lives more generally in much greater detail in Debus (2016). (Thanks to Denis Perrin for prompting the present comment.) 19 My work on the material presented here was made possible by a generous grant from the ‘Philosophy and Science of Self-Control’ project at Florida State University, sponsored by the John Templeton Foundation and directed by Al Mele; the views expressed here are my own. I am very grateful to the John Templeton Foundation, and especially to Al Mele, for their support. I presented versions of this paper at workshops and conferences in Grenoble, Otago, Cambridge, Cologne, and Rome, and I am grateful for the many helpful questions and comments from members of the audience on those occasions. My heartfelt thanks also go to the Alexander von Humboldt Foundation who provided me with a fellowship during which I was able to prepare the final draft of this paper.

References Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press. Bernecker, S., & Grundmann, T. (2017). Knowledge from forgetting. Philosophy and phenomenological research. doi: 10.1111/phpr.12469 Davidson, D. (2001). Freedom to act. In Essays on actions and events (pp. 63–81). Oxford: Clarendon Press. Debus, D. (2007). Perspectives on the past: A study of the spatial perspectival characteristics of recollective memories. Mind and Language, 22, 173–206. Debus, D. (2008). Experiencing the past: A relational account of recollective memory. Dialectica, 62, 405–432. Debus, D. (2016). Shaping our mental lives: On the possibility of mental selfregulation. Proceedings of the Aristotelian Society, 116, 341–365. Debus, D. (2017). Memory causation. In S. Bernecker and K. Michaelian (Eds.), The Routledge handbook of the philosophy of memory (pp. 63–75). London: Routledge. Dummett, M. (1996). Testimony and memory. In The seas of language (pp. 411– 428). Oxford: Oxford University Press. Eilan, N. (1998). “Perceptual intentionality, attention and consciousness.” In A. O’Hear (Ed.), Current issues in philosophy of mind (pp. 181–202). Cambridge: Cambridge University Press.

136  Dorothea Debus Lackey, J. (2005). Memory as a generative epistemic source. Philosophy and Phenomenological Research, 70(3), 636–658. Lehrer, K., & Richard, J. (1975). Remembering without knowing. Grazer Philosophische Studien, 1, 121–126. Lewis, D. (1980). Veridical hallucination and prosthetic vision. Australasian Journal of Philosophy, 58, 239–249. Martin, C. B. and Deutscher, M. (1966). Remembering. Philosophical Review 75, 161–196. Martin, M. G. F. (2001). Out of the past: Episodic recall as retained acquaintance. In C. Hoerl and T. McCormack (Eds.), Time and memory. Issues in philosophy and psychology (pp. 257–284). Oxford: Oxford University Press. Martin, M. G. F. (2002). The transparency of experience. Mind and Language, 17, 376–425. McCarroll, C. J. (2017). Looking the past in the eye: Distortion in memory and the costs and benefits of recalling from an observer perspective. Consciousness and Cognition, 49, 322–332. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Rice, H. J., & Rubin, D. C. (2009). I can see it both ways: First- and third-person visual perspectives at retrieval. Consciousness and Cognition, 18, 877–890. Schacter, D. L. (1996). Searching for memory. New York: Basic Books. Sellars, W. (1997). Empiricism and the philosophy of mind. Cambridge, MA: Harvard University Press. Shoemaker, S. (1967). Memory. In P. Edwards (Ed.), The encyclopedia of philosophy (pp. 265–274). New York & London: Palgrave Macmillan. Shoemaker, S. (1970). Persons and their pasts. American Philosophical Quarterly, 7, 269–285. Sutton, J. (2010). Observer perspective and acentred memory: Some puzzles about point of view in personal memory. Philosophical Studies, 148, 27–37. Tulving, E. (1983). Elements of episodic memory. Oxford: Clarendon Press.

Part III

The Affective Dimension of Memory

7 Affective Memory

Margherita Arcangeli and Jérôme DokicAffective Memory

A Little Help From Our Imagination1 Margherita Arcangeli and Jérôme Dokic

1  Introduction: The Puzzle of Affective Memory It often happens that when we remember a past situation, the emotional import of the latter transpires in a modified form at the phenomenological level of our present memory. When it does, we experience what is sometimes called an “affective memory.” Theorists of memory have disagreed about the status of affective memories.2 Defenders of the existence of affective memories claim that there is a special class of emotional phenomena that can be involved in remembering. On this view, there is an emotional way of remembering a past situation, whether or not the latter actually elicited (or is remembered to have elicited) an emotional reaction on the subject’s part. In contrast, sceptics about affective memories argue that the relationship between memory and emotion can only be of two types: either the memory is about a past emotion, or it causes a present emotion (see Debus, 2007). In the former case, the emotion is part of what is remembered. For instance, Maria remembers having been sad, or that she was sad, yesterday. In the latter case, the emotion is a separable effect of the memory. For instance, Maria remembers her root canal operation yesterday and is now relieved that the operation is over. Her memory of the operation causes a separate positive emotion in her. In this picture, there is no room for the category of affective memories. This chapter is a defense of affective memory as a genuine mental phenomenon. Our main claim is that affective memory is a case of episodic memory in which the past situation is represented through a special use of the imagination. When we remember a past situation, we often imaginatively re-live the situation as if it were present. In other words, we represent the situation from a specific mental perspective, which may differ from our actual present perspective. Drawing from Peter Goldie’s notion of mental narratives, which covers at least memories and imaginings, we show that three levels of perspective should be distinguished (Section 2): the perspective of the represented subject (the character, if there is one), the perspective of the representing subject (the author), and the intermediary perspective of the narrator. In many cases, the narrator remains virtual; what is at stake is only a specific way of representing the situation. For instance, different

140  Margherita Arcangeli and Jérôme Dokic narratives can truly represent the same situation in different emotional ways. As Wittgenstein put it, the world of the happy man is different from the world of the unhappy man (Wittgenstein, 1961, §6.43). Once the three-way perspectival character of mental narratives is acknowledged, several mental phenomena can be described in terms of dependence relations between perspectives. What we call “strong immersion” involves the alignment of all three perspectives. In contrast, “weak immersion” refers to cases in which the narrator’s and the character’s perspectives are aligned. There is emotional contagion when the author “catches” the narrator’s emotion, while imaginative resistance shows the opposite direction of influence, i.e., the author’s perspective limits the narrator’s perspective (Section 3). However, alignment is only one modality of dependence. Sometimes, one perspective can be coupled with another perspective even though they are not aligned. When we use our imagination to figure out what is possible or not in a given situation (for instance, whether our new sofa would fit into our living-room), the author’s perspective is constrained by the narrator’s perspective (Section 4). Affective memories are precisely cases in which the narrator’s emotional perspective has direct implications for the author’s emotional perspective, even if the former typically differs from the latter (Section 5). Episodic future thinking, i.e., the ability to imagine in detail subjectively probable future situations, also involves constrained imaginings in this sense (Section 6). Whatever explanation is given for the apparent asymmetry between episodic memory and episodic future thinking, our main point still holds, namely that affective memories are genuine and cannot be reduced to either memories about past emotions or memories causing present emotions.

2  Perspectives in Narratives: A Tripartite Account 2.1  The Notion of Mental Perspective Any mental state with representational content introduces a mental perspective on the represented situation. The notion of mental perspective, or perspective for short, can refer to at least three types of features (Goldie, 2003; 2012). First, a perspective can be conceptual, in the sense that the situation is represented by means of more or less sophisticated concepts. For instance, we can think of a given geometrical figure as a chiliagon or merely as a complex polygon. Second, a perspective can be epistemic, in the sense that representing a situation from a perspective may require some knowledge-conducing access to the situation. Visual perspectives are epistemic, because visual perception is knowledge-conducing. It enables knowledge and not mere representation of the situation. Finally, a perspective can be emotional, in the sense that it reflects a specific affective evaluation of the situation. For instance, we feel sad about a situation that other subjects may feel neutral or even happy about. Features of a perspective are determined either by the mode of the mental state (the psychological attitude) or by its representational content. The

Affective Memory  141 conceptual feature is clearly determined by the content, whose constituents are or involve the relevant concepts. In contrast, the other features, epistemic and emotional, are determined by the mode. Our epistemic perspective is different when we know or merely conjecture about a state of affairs, but the representation involved can be the same. Similarly, the emotional perspective is different whether one is sad or happy about a given represented situation. Mental perspectives can be either real or simulated. When Maria feels happy about her success in the tennis match, the emotional perspective on the represented situation is determined by her real state of happiness. Now a type of mental state can also be simulated or “re-created” via imagination (see Currie & Ravenscroft, 2002, and for a taxonomical exploration of the types of re-created mental states in imagination, Dokic and Arcangeli, 2015a). As an opponent to Maria, Sam does not feel happy about her success in the match, but he can imagine Maria being happy. Via imagination, Sam re-creates a state of happiness, which he attributes (rightly, in this case) to Maria. This simulated state involves a specific emotional perspective, which is different from the emotional perspective introduced by Sam’s actual state of mind, which can be, say, anger at having lost the match. Note that Sam’s emotions have different cognitive bases: his simulated happiness arises from his imagining the world according to Maria, whereas his real anger is triggered by his belief that he has lost the match. Even among philosophers sympathetic to the general view that mental states can be re-created through imagination, there is a specific issue about the nature of re-created emotions. Some argue that re-created emotions are ordinary emotions (Carroll, 1990; Currie & Ravenscroft, 2002; Weinberg & Meskin, 2006), whereas others claim that they are intrinsically imaginative or “as if” emotions (Goldman, 2006; Velleman, 2000; Walton, 1994). We can remain neutral on this issue here. If re-created emotions are ordinary emotions, then we should allow for a rational subject to be in several emotional states at once (provided that their cognitive bases are distinct), or at least successively in the course of a dynamic imaginative narrative. 2.2  External vs. Internal Perspectives To the extent that they are perspectival, representational mental states can be described as mental narratives (Goldie, 2003; 2012). A mental narrative involves some perspective on what is represented, but it need not be narrated—if it is, the result will be a linguistic narrative, i.e., a story about the represented world. Unlike linguistic narratives, though, which are propositions or statements, mental narratives need not have linguistic vehicles or conceptual contents.3 Peter Goldie has pointed out that all narratives involve two kinds of perspective, which he calls “external” and “internal”: A narrative, whether or not narrated, always involves an external perspective, and any external perspective is necessarily distinct from

142  Margherita Arcangeli and Jérôme Dokic any internal perspective. This is so even where, as in autobiographical remembering, the two perspectives are those of one and the same person. (Goldie, 2003, pp. 310–311) The notion of internal perspective is reasonably clear. Some narratives introduce characters who are represented as having various mental states, for instance as being happy. The mental perspectives that the characters have on the represented situation are internal perspectives. In contrast, Goldie’s notion of external perspective is trickier. It seems to refer to the perspective of the author of the narrative. If Sam remembers a past situation in which he was happy, the external perspective is the present perspective of the remembering subject. It differs from the internal perspective that Sam occupied in the past. For instance, Sam may be presently nostalgic in remembering his happiness as a young adult. These emotional perspectives differ even if, as Goldie points out, they are occupied by the same person, namely Sam. Sam is both the author and the character of his memory. So far so good, but Goldie also defines the external perspective as “a perspective that is external to the actions and events that are narrated” (Goldie, 2003, p. 302). Now we want to suggest that the author’s perspective is not the only external perspective in this sense. Consider observer memories (as opposed to field memories; see Section 3.1, and Nigro & Neisser, 1983; Sutton, 2010). An observer memory represents the past situation from a point of view that differs from and encompasses the past subject’s point of view. For instance, Sam remembers the last time he climbed his favorite mountain, by adopting the point of view of a virtual observer, who can see him climbing from sideways or somewhat behind. In this case, three epistemic perspectives should be distinguished. First, there is the internal perspective of the narrative, namely the visual perspective that Sam occupied in the past situation. Second, there is the visual, sideways perspective of the narrator, which differs from the internal perspective. Finally, there is Sam’s (the author’s) present epistemic perspective, which is obviously different from both the character’s and the narrator’s perspectives. The author’s and the character’s perspectives are necessarily occupied, both by Sam in this example. What about the narrator’s perspective? Although this is controversial, we are sympathetic to the view that the narrator’s perspective can be empty. Sam’s memory involves the re-creation of a visual experience that someone could have enjoyed in the past, but this visual experience need not be attributed to any real person in the past or in the present. As an analogy, consider Berkeley’s famous example in which a subject tries to visually imagine an unseen tree. In both cases (pace Berkeley himself), the situation is represented from a visual perspective that nobody occupies.4 Moreover, the fact that the narrator’s perspective in Sam’s memory is empty shows that observer memories need not be distorted memories (pace Fernández, 2015). The relevant visual perspective itself might have been available in the past situation, even if no actual observer

Affective Memory  143 has had the relevant experience. In principle observer memories can be fully veridical (for a similar point, see McCarroll, 2017). The example of observer memories shows that we should go beyond Goldie and acknowledge two “external” perspectives beside the internal perspective of the narrative. The narrator’s perspective is external relative to the character, but it is still internal, so to speak, to the narrative itself. Now there is a perspective that is external to both the narrator and the character, namely the author’s perspective. For the sake of commodity, let us give a label to these three perspectives:5 P1. The perspective of the character (the represented subject) P2. The perspective of the narrator (the way of representing) P3. The perspective of the author (the representing subject) Although the author’s and the narrator’s perspectives are at least notionally distinct, they do not always differ. A visual experience or a belief, for instance, does not involve the perspective of a narrator distinct from the author. We suggest that the narrator’s perspective emerges as a distinct perspective only when the author re-creates a mental state that is not attributed to the character. It follows that P2 differs from P3 only when the mental state is an imagining, or involves an imagining. Although we do not think that memories are imaginings (pace Hopkins, forthcoming), some memories clearly involve or depend on imagination. Sam’s observer memory is a case in point. It involves imagining Maria from a visual perspective that also encompasses Sam himself. Of course, the example of Sam’s observer memory only shows that a single narrative can involve three distinct epistemic perspectives. We are going to suggest that an ambitious notion of affective memory goes hand in hand with the possibility that a memory involves three distinct emotional perspectives. Thus, you might accept our description of Sam’s observer memory but insist that no example can be given of a single narrative involving three distinct emotional perspectives. For instance, you might want to argue that the narrator’s emotional perspective can be reduced to either the author’s or the character’s emotional perspective. When a memory concerns a happy situation, the narrator’s happiness may just belong to the character, i.e., the subject need not be presently happy but can represent the situation from her own affective point of view in the past. When the narrator’s happiness does not belong to the character (because the latter was not happy in the past situation), then it may be nothing but the author’s happiness, i.e., the subject is happy in remembering the past situation. Admittedly, the idea of a three-way emotional perspectival gap is controversial, and needs to be defended carefully. In the next sections we shall discuss cases of mental narratives involving various alignments or other types of dependence relations between emotional perspectives. Then we shall come back to the notion of affective memory and suggest a definition

144  Margherita Arcangeli and Jérôme Dokic in terms of a dependence, but not an alignment, between the narrator’s and the author’s emotional perspectives.

3 Imaginative Immersion and Other Perspectival Alignments 3.1  Strong vs. Weak Immersion The most radical form of alignment is what we call “strong immersion.” In strong immersion, all three perspectives (P1, P2, and P3) are aligned, so that the subject is apparently living through the represented situation. In the field of memory, so-called “flashbulb memories” might illustrate such an alignment (Conway, 1995). When one has a flashbulb memory of a past scene (often emotionally loaded), one has the impression of actually reliving the scene—the past is falsely represented as present, so to speak. In other words, the subject is temporarily deluded in actually having the re-created experience. A more modest form of alignment concerns only the narrator’s (P2) and the character’s (P1) perspectives. The situation is represented exclusively from the point of the view of the character, while the author’s perspective remains distinctive. When there is such an alignment of perspectives without strong immersion, we shall say that the representing subject is weakly immersed in the character’s perspective. The notion of weak immersion at stake here can be illustrated by means of the notion of field memories (Nigro & Neisser, 1983; Sutton, 2010). Unlike observer memories, field memories are cases in which the past situation is remembered from the subject’s original point of view. As we saw in the last section, observer memories illustrate the epistemic gap between the narrator’s and the character’s perspectives: the narrator’s perspective (P2) is that of the virtual observer, and it is distinct from the character’s perspective (P1), which is the perspective that the subject has actually occupied in the past. In contrast, field memories are characterized by an alignment of these perspectives: the past situation is represented from the point of view of its main character, the subject herself. There is not strong immersion, because the author’s perspective is not aligned with the perspective shared by the narrator and the character. The remembering subject is only weakly immersed in her past self’s perspective. Beside strong and weak immersion, two other examples of perspectival alignments are worth discussing, namely emotional contagion and imaginative resistance, which can also be described in terms of our tripartite account. 3.2  Emotional Contagion6 In our account, there is emotional contagion when the representing subject irrationally “catches” the narrator’s virtual emotion, i.e., where P2

Affective Memory  145 becomes P3 through a confusion between the subject’s represented situation and the actual (or present) situation. A famous and dramatic example of emotional contagion is provided by Gustav Schwab’s popular story of the Horseman and the Lake of Constance (later mentioned by Freud in a letter to Minna Bernays).7 A rider crosses the frozen Lake of Constance by night without knowing it. When he reaches the opposite shore, and is told by a villager whence he comes, he dies of terror. In this story, the rider starts by remembering (or perhaps just imagining) a past situation from the point of view of a virtual narrator terrorized by the fact that what appears as land covered with snow is in fact a lake with a thin layer of ice. Then, the rider actually experiences the fear of the narrator. Such emotional contagion is of course irrational or at least inappropriate because the rider is now on safe grounds, so that there is presently nothing to fear about. Note that P3 is contaminated by P2 and not P1: the perspective of the character himself is not terror, but focus and determination to find a place to rest. One might wonder whether other cases of emotional contagion involve a contamination of P3 by P1. For instance, Maria becomes sad just by imagining someone (maybe herself) sad. However, we surmise that emotional contagion is always mediated by the perspective of the virtual narrator, namely P2. In the latter scenario, Maria becomes sad by imagining someone sad only because the virtual narrator paints the imagined situation black, so that P3 is contaminated by P2 after all, which in the case in point is emotionally commensurate with P1. What is the relationship between emotional contagion and the two kinds of immersion introduced earlier? To the extent that emotional contagion does not directly depend on the character’s perspective (P1), it is compatible with, but does not entail, strong immersion. For instance, emotionally loaded flashbulb memories illustrate both emotional contagion and strong immersion. In contrast, emotional contagion is trivially incompatible with weak immersion since their combination would imply strong immersion (a collapse of all three perspectives). 3.3  Imaginative Resistance In Of the Standard of Taste (1757), Hume famously observed that it is subjectively hard to imagine situations incompatible with one’s actual moral perspective. For instance, we may feel uneasy to imagine a world in which infanticide is morally permissible (Walton, 1994; Moran, 1994; Gendler Szabó, 2000). Like emotional contagion, we surmise that imaginative resistance involves a porosity between the perspective of the representing subject (P3) and the narrator’s perspective (P2). However, there are two important differences. First, the direction of influence is different. While emotional contagion involves a contamination of P3 by P2, imaginative resistance involves a contamination of P2 by P3. The imagining subject’s moral beliefs influence the narrator’s way of representing the imagined

146  Margherita Arcangeli and Jérôme Dokic Table 7.1  Varieties of perspectival alignments (in grey) P1 Character

P2 Narrator

P3 Author

Strong immersion Weak immersion Emotional contagion



Imaginative resistance



situation. The imagined situation is easier to represent if the narrator’s epistemic and emotional perspectives are closer to the imagining subject’s actual perspectives. (In this sense, as many have observed, imaginative resistance is not restricted to moral beliefs.) Second, it is not obvious that imaginative resistance is like emotional contagion in being an irrational phenomenon. Imaginative resistance prevents us from constructing a world in imagination that reflects morally unacceptable beliefs. This indicates a cognitive limitation on our part, but not necessarily a failure of rationality. Imaginative resistance can prevent both weak and strong immersion. Weak immersion is compromised if the character’s perspective (P1) involves morally unacceptable beliefs. In such a case, the narrator’s perspective (P2) cannot easily be aligned with the character’s perspective. For instance, many readers might have experienced unease while reading novels such as American Psycho by Bret Easton Ellis, which is told exclusively from the point of view of a (fictional) sociopath. As a consequence, the world as represented in such novels is morally so distant from our world that immersion in fiction becomes hard, if not impossible, to achieve. A fortiori strong immersion, which requires the alignment of all three perspectives, is compromised as well. (Table 7.1 summarizes the different cases of alignment just discussed.)

4  Free vs. Constrained Imagination Let us say that our imagination is free if the narrator’s perspective (P2) can vary freely without much affecting the imagining subject’s perspective (P3). In contrast, our imagination is constrained if P2 directly affects P3, i.e., the imagining subject spontaneously or unreflectively updates her actual model of reality on the basis of what she is imagining. When imagination is constrained, a question arises as to the nature of the relevant constraint. The latter can be merely causal, or it can reflect the subject’s rational sensitivity to the actual world. Emotional contagion, where the representing subject’s perspective blindly takes on the narrator’s perspective, clearly involves a mere causal, irrational constraint. If I am

Affective Memory  147 terrorized just by imagining a random nonactual terrorizing event, my emotion is inappropriate. It is also inappropriate as a direct response to my remembering a past terrorizing event, although of course my memory can make it appropriate for me to fear similar present dangers. In contrast, epistemic uses of imagination involve the narrator’s perspective rationally constraining the subject’s perspective. Suppose that Maria wants to buy a new sofa for her living room. She goes to the furniture store and spots a nice sofa. She is now using her imagination to determine whether the sofa would fit in her living room. She imagines the sofa with its actual shape and size at a particular place in the living room. Maria’s use of her imagination is epistemically constrained since what she imagines has immediate consequences on what she believes, for instance, that a particular sofa would not fit her living room. Moreover, the beliefs that Maria acquires through imagination can amount to knowledge or at least be justified. Thus, the dependence of her present perspective on the narrator’s perspective is rational.8 Episodic memory typically involves an epistemic use of imagination. When we use our memory and mentally re-live an episode of our own past, the narrator’s perspective represents the episode as present but we form past-tense beliefs whose contents systematically depend on the contents of the narrator’s re-created mental states. Since some of these beliefs are justified and can amount to knowledge (episodic memory, even when it involves imagination, is a source of knowledge about one’s own past), the dependence of the subject’s perspective on the narrator’s perspective can be rational. In the remainder of this chapter, we shall focus on the affective dimension of episodic memory. We are interested in cases where the narrator’s emotional perspective directly influences the remembering subject’s emotional perspective, even when these perspectives are distinct. In these cases, the narrator’s emotional way of representing the past situation is not neutral for the remembering subject, and modulates her actual emotional state. We argue that the subject then has a genuine affective memory.

5  Affective Memory We are now in the position to see how the tripartite account of the perspectival character of mental narratives leads to a clarification of the notion of affective memory. In a nutshell, affective memory is a case of episodic memory that hinges on emotionally constrained imagination. Consider the following example of an affective memory. Last night Maria was driving on the highway heading home. As she was very tired, she fell asleep on the wheel for only a few seconds. As a consequence, her car was about to hit the crash barrier, but fortunately she was soon back in control and her trajectory adjusted at the last minute. When Maria now remembers the scene, she represents it as an extremely dangerous episode. However,

148  Margherita Arcangeli and Jérôme Dokic fear was not part of her original experience. On the contrary, she remembers having been relaxed and sleepy, not afraid of hitting the crash barrier. Fear is not part of her present experience either. Maria is not now afraid of hitting the crash barrier. On the contrary, she is relieved by the fact that a serious accident has been avoided. Thus, the present example clearly differs from examples involving emotional contagion, where the author’s perspective is aligned with the narrator’s perspective. Here, all three emotional perspectives seem to be distinct. The author’s perspective (P3) involves relief, the virtual narrator’s perspective (P2) involves fear, and the character’s perspective (P1) involves calmness and peace of mind. In parallel with Sam’s observer memory (see Section 2), Maria’s memory seems to involve a perspective on the represented situation that remains unoccupied: nobody is or was afraid of hitting the crash barrier. The idea of an empty emotional perspective is perhaps harder to envisage than that of an empty visual perspective. We believe that it is possible to represent certain values, such as danger, from emotional perspectives that are merely re-created, and that re-created mental states in general need not be attributed to anyone in the real or represented situation. Consider a character-free imagining; for instance, Sam imagines what the earth would be like after the next world war. Sam might construct his imagined world in such a way that humanity is extinct. No emotion can be part of the content of his imagining. Nevertheless, Sam might paint the situation in a gloomy or depressing way. The narrator’s emotional perspective makes a phenomenological contribution to the imagining, and it might differ from Sam’s present emotional perspective. Although Sam might himself be gloomy or depressed just by imagining this possible future situation, he might also be more sedate, perhaps because he is too optimistic to consider the situation as a probable future one. In any case, if we are wrong and the narrator’s perspective must be occupied in the represented situation, then affective memories will necessarily be distorted. On this view, Maria’s memory is not fully veridical because it brings in an imaginary observer who is afraid of events and actions in the represented situation. From an epistemological point of view, this would be an unfortunate consequence, but it would leave intact the claim that there are genuine affective memories. Note that the dangerousness of the situation is represented experientially and more specifically emotionally. The fact that the car was about to hit the crash barrier is presented as fearsome, rather than in more neutral terms. If we leave out this emotional dimension of the memory, Maria’s present emotional response would be hardly intelligible: Maria feels immediate relief at least partly because the situation is represented as fearsome. In contrast, the move from re-creating the emotion of fear to Maria’s present relief makes perfectly sense, even though Maria does not, indeed cannot remember having been afraid at the time. In other words, the subject’s perspective (P3) and the narrator’s perspective (P2) are coupled at the

Affective Memory  149 emotional level. Maria’s relief is a rational response to the virtual narrator’s fear. What makes Maria’s memory affective is the fact that it involves emotionally constrained imagination. Here is another example of affective memory. Sam feels presently nostalgic in remembering his life as a young adult, which he now pictures in a more happy and favorable light than he would have admitted at the time. In this example, the subject’s emotional perspective (P3) is nostalgia, the narrator’s emotional perspective (P2) is happiness or joy about the events and actions in the past situation, while the character’s perspective (P1) is more akin to recklessness or unconcern. Again, Sam’s memory does not involve either weak immersion or emotional contagion, and the narrator’s perspective seems to make a special contribution to its phenomenology. On the present account, a given memory can be affective even if the remembered situation did not involve the character’s having an emotion. What makes a memory affective is the dependence of the author’s emotional perspective on the narrator’s (not the character’s) emotional perspective. Now, in some cases there will be in addition an alignment between the narrator’s and the character’s perspectives. In other words, affective memories can be either field or observer memories, depending on whether the narrator’s perspective is aligned on the character’s perspective. What is the nature of the dependence involved in affective memory? Maria is not merely imagining the event but also remembers it. Thus, the narrator’s perspective is controlled by episodic information. The extent to which Maria’s relief is an appropriate response and reflects her rational sensitivity to the remembered event depends on the existence of principles of emotional rationality. As a first approximation, her present emotion of relief is appropriate only if, in the situation remembered, the emotion of fear had been appropriate (irrespectively of whether that emotion was actually felt then), which in turn is a matter of whether what happened then actually put the subject in danger.9 Independently of the latter issue, we can now see what is wrong with the argument of the sceptics about affective memory. Once the existence of three separable perspectives is properly acknowledged, there is room for the category of affective memories. When the narrator’s perspective (P2) is emotional and constrains the subject’s emotional perspective (P3), there is a special emotional way of remembering an event. On this view, memories are affective when the narrator’s perspective is emotionally loaded and directly affects the subject’s emotional perspective, independently of whether a past emotion is part of what is remembered. One might object that on our account, affective memories are not genuine memories but rather combinations of memories and imaginings. Affective memories are just ordinary memories combined with affective imagination. Correspondingly, one might acknowledge cases of imagination that involve three-way emotional perspectives, but argue that memories do not exhibit the same structure.

150  Margherita Arcangeli and Jérôme Dokic In answer to this objection, we agree that we have not shown that ­affective memories are genuine kinds of memory, rather than memories that are only contingently associated with imaginings. This is not a problem, because we want to account for a phenomenologically salient mental phenomenon that may or may not correspond to a mental essence. Still, what the objection underestimates is the nature of the relationship between affective memories and the imaginings they are based on. Affective memories are not just ­memories that happen to be accompanied by imaginings. Rather, the ­relevant imaginings are constrained by episodic information possessed by the subject. When we have an affective memory, we may enjoy a unified conscious experience, at least if our possession of the relevant episodic information is not manifested in a separate, memory experience, but is used only in the course of our imagining the past from a certain (narrator’s) perspective. What makes the subject’s whole experience a memory experience is that her imagining (and more precisely the narrator’s perspective) is controlled by episodic information, in a way that enables rational constraining of the subject’s perspective.10

6  Mental Time Travel As we have seen, episodic memory typically involves the imaginative ­capacity to re-live an episode of one’s own past. According to recent work in cognitive science, the same imaginative capacity is also at work when one mentally pre-lives an episode of one’s anticipated future (Schacter & Addis, 2007; Michaelian, 2016a). In other words, a single mental capacity labelled “mental time travel” has both past-oriented and future-oriented exercises. Episodic memory involves the former exercises, while the latter are cases of so-called “episodic future thinking” (Atance and O’Neill, 2001). Episodic future thinking is not merely a case of imagining a situation that might be our future situation. When mental time travel is impaired, as in certain forms of amnesia, the patient is still left with the ability to imagine general facts about the future. What is not preserved is the ability to imagine detailed episodes of one’s subjectively probable future. Episodic future thinking thus involves a specific way of representing one’s own future. In our view, episodic future thinking involves a constrained use of the imagination. When the relevant constraint is emotional, episodic future thinking directly modulates our present emotional states. Just like episodic memory, episodic future thinking can be and is often affective. Suppose that Maria anticipates an important examination that she must take next week. She believes that she is well prepared for this, and imagines herself answering an apparently simple question with a strong feeling of confidence (P1). However, Maria imagines such a situation from the perspective of a virtual narrator who knows that Maria is overconfident, and gives the wrong answer. The narrator colors the imagined future situation with shame and embarrassment (P2). As a consequence, Maria presently feels anxious about

Affective Memory  151 the prospects of a not so well-prepared examination (P3). What makes this example a case of episodic future thinking in contrast to other cases of imagining the future is the emotional dependence of the subject’s perspective on the narrator’s perspective. What Maria imagines has direct consequences on her present emotional state, without the mediation of beliefs or other conceptual representations.11 Let’s contrast the examination example with another one involving emotionally free imagination. This time, Maria imagines herself being a candidate for presidency and waiting for the results of the elections. In real life, Maria is not motivated whatsoever to become the president; she imagines such a situation merely for the fun of it. However, she can imagine herself being motivated to be the president, so that she is highly anxious about the results. This is the internal perspective of the mental narrative (P1). Now the virtual narrator can represent Maria as being in a favorable situation, i.e., as having won the elections (even though she does not know the results yet). In other words, the narrator’s virtual emotion (P2) is happiness or satisfaction, which colors the imagined scene. P2 is clearly distinct from P1, at least epistemically and emotionally. Emphatically, P2 is also not the perspective of the imagining subject, who is completely indifferent to the issue about being the president. The emotional perspective of the imagining subject (P3), is indifference or perhaps amusement, while the narrator’s perspective is happiness and the character’s perspective is anxiety. Now Maria’s imagining is free, since it is done for the mere fun of it, and does not lead to any adjustment in her actual view of reality. What she imagines has no or low impact on what she believes or feels about the actual world. An interesting question is whether the dependence of emotional perspectives in episodic future thinking is rational or not. As in the case of episodic memory, the answer depends on the existence of principles of emotional rationality. For instance, de Sousa (2007, 2008) discusses what he calls “Philebus Principle,” according to which a pleasure taken in anticipation of a future event should be proportional to the pleasure that will be actually yielded when experiencing the event. On this view, “it is irrational to take great pleasure in the expectation of something that one knows will bring little or no pleasure when it comes. Conversely, it seems quite reasonable to dread the prospect of future suffering” (2007, p. 151). Arguably, ­Maria’s present anxiety is a rational response to the narrator’s perspective in the future. It seems to be an emotion that is in a relevant sense proportional to the emotion experienced by the (virtual) narrator.12 There is still an important issue left, which concerns the precise relationship between episodic memory and mental time travel. On the ­ continuist hypothesis, episodic memory and episodic future thinking are nothing but exercises of the same psychological natural kind, namely ­mental time travel. Both past-oriented and future-oriented mental time travel are controlled by episodic information, and such information is always the ­result of a construction by the brain. We make up our own past just as we

152  Margherita Arcangeli and Jérôme Dokic do for our own future. On the alternative, discontinuist hypothesis, episodic memory involves more than just mental time travel. Episodic memory is tightly controlled by reliable episodic information about the past. Futureoriented mental time travel is still sensitive to episodic information (since the imaginative construction of detailed situations is known to involve episodic information), but only loosely so, by reflecting subjective probabilities about future events.13 Although we can remain neutral on this issue here, we tend to favor discontinuism. If discontinuism is true, we might wonder whether there are past-oriented exercises of mental time travel that do not involve episodic memory. If there are, then the case for discontinuism would be strengthened. Past-oriented constrained imaginings that are not memories of the past situation seem to be available. We can represent a past situation in a way that epistemically and emotionally constrains our present perspective, not because we remember the past situation but because we picture it as having very probably occurred given our present worldview. For instance, Sam imagines Maria cheating on him while he was away last weekend. The narrator’s perspective might be sensitive to episodic information in a way that constrains Sam’s present emotional response. Sam might then be jealous just by imagining a highly probable past situation, even though he was not around to witness it and cannot remember it. To sum up, whatever the fate of the debate between continuism and discontinuism, we have taken a stance on the nature of mental time travel itself. Mental time travel is nothing but the ability to engage in exercises of constrained imagination. On the one hand, episodic future thinking is itself a case of constrained imagination; the subject is directly affected by her future-oriented imaginings. On the other hand, the examples of episodic memory discussed here at least involve constrained imagination; the subject is directly affected by her past-oriented imaginings. Note that we go beyond the claim that mental time travel is mere imagination, oriented toward the past or the future. Mental time travel involves a special use of the imagination, which is characterized by a dependence of the subject’s perspective on the narrator’s perspective. (See Table 7.2 for a summary of the discussion.)

Table 7.2  Varieties of free and constrained imagination

Free imagination Episodic memory Episodic future thinking Unconscious recollection14

P3 constrained by P2

P2 controlled by episodic information

– (not much) ++ ++ – (not much)

– (not much) ++ +/– ++

Affective Memory  153

7 Conclusion The puzzle of affective memory arises because it may seem hard to understand how memory can be emotionally loaded if the relevant emotion is neither the object nor the separate effect of the memory. We can remember our past emotions, and our memories can cause various emotions in us, but neither phenomenon gives justice to the idea that there is an emotional way of remembering the past. In this chapter, we have suggested a way out of this puzzle. When we mentally re-live a past situation from memory, we imagine the situation from the perspective of a virtual narrator. This perspective is neither the present nor the past perspective of the subject. Once the existence and the emotional dimension of such intermediary perspective are properly acknowledged, affective memory can be considered a genuine phenomenon. Affective memory involves a dependence (but not an alignment) between the narrator’s and the subject’s perspectives; more precisely, the latter is emotionally constrained by the former, in a way that can reflect the subject’s rational sensitivity to the world. Constrained imagination is what is going on in mental time travel. We have suggested that both past-oriented and future-oriented exercises of mental time travel involve an epistemic and emotional dependence of the subject’s perspective on the narrator’s perspective. Our account of mental time travel is relatively neutral with respect to the question of whether episodic memory and episodic future thinking belong to a single psychological kind, which would be the ability to mentally travel in time. Continuists give an affirmative answer to this question, while discontinuists claim that there is more to episodic memory than mental time travel. In our view, the continuism vs. discontinuism debate boils down to the question of whether episodic memory as a psychological kind can be identified with past-oriented constrained imagination. To sum up, the separate existence of the narrator’s perspective and its direct emotional impact on the subject’s perspective are conditions of possibility of affective memory. Moreover, our tripartite account of imaginings places affective memory within a logical space that includes other (more or less rational) mental phenomena, such as episodic future thinking, but also field and observer memories, flashbulb memories, emotional contagion, and imaginative resistance.

Notes 1 We thank Christoph Hoerl and an anonymous referee for very helpful comments on an earlier version of this chapter. Margherita Arcangeli’s funding has been provided by the Alexander von Humboldt Foundation. 2 The existence of genuine affective memories was at the heart of hot debates in philosophy and psychology at the turn of the twentieth century. See Trakas (2015, Ch. 5) for a helpful historical and theoretical account. As she points out, Théodule Ribot, Marcel Mauxion, François Pillon, Frédéric Paulhan, Ludovic

154  Margherita Arcangeli and Jérôme Dokic Dugas, Paul Sollier, and Ernst Heinrich Weber were among the defenders of the existence of affective memories, while sceptics included William James, Harald Höffding, Edward Bradford Titchener, and Édouard Claparède. Of course, we do not want to suggest that all these authors were focusing on a single phenomenon. The more general issue was about the nature of the relationship between memory and emotion. 3 The explanatory or at least clarificatory role of the notion of narration in an account of representational mental states has been acknowledged by several authors, including Ricoeur (1983) and Wollheim (1984). 4 For a defense of the possibility of unoccupied visual perspectives in imagination, see Williams (1976), Noordhof (2002), and Dokic and Arcangeli (2015b). For critical discussion, see Peacocke (1985) and Martin (2002). 5 We have offered a preliminary account of these perspectives in Arcangeli and Dokic (2015). 6 We focus on cases of emotional contagion in which what is emoted originates from the imagination (which may or may not be associated with a memory). Whether emotional contagion always involves the imagination (in contrast to, say, perception) is something that we can leave open here. 7 Freud’s letter is reproduced in Longfellow (2011)’s anthology. 8 See Kind (2016) and other essays in Kind and Kung (2016). Other examples of rational constraints in imagination include imagination-based modal reasoning (see, e.g., Gendler Szabó and Hawthorne, 2002) and mental simulation (see, e.g., Goldman, 2006). 9 Thanks to Christoph Hoerl for suggesting this formulation. The general claim that there are principles of emotional rationality is controversial but (we think) plausible. 10 Our account is compatible with a view defended by one of us (Dokic, 2014), according to which a memory experience typically involves in addition a metamemory feeling that tags the relevant contents as coming directly from one’s past experience. 11 The claim that episodic future thinking involves constrained (and not free) imagination has affinities with Gerrans and Mulligan (2013)’s account of what they call “hybrid imagination.” They suggest that in episodic future thinking, “we try to work out what might happen by imagining the future within contextual constraints supplied by propositional knowledge” (p. 254). However, they seem to reduce hybrid imagination to de re imagination, and we surmise that not all cases of imagination de re have to be constrained in our sense. 12 See also Boyer (2008) for an account of mental time travel according to which it “may be functional to the extent that it provides emotions that bypass current goals as well as time discounting and, therefore, provide us with immediate counter-rewards against opportunistic motivation” (p. 22). 13 We borrow this terminology from Michaelian (2016b) and Perrin (2016), who favor continuism and discontinuism, respectively. One argument frequently invoked in favor of discontinuism is epistemological: whereas episodic memory is a source of knowledge about the represented past situation, episodic future thinking cannot be a source of knowledge about the represented future situation (although it can yield new knowledge, for instance about Maria’s present state of preparation for her forthcoming examination). 14 Here we think of Martin and Deutscher’s (1966) famous character, who paints a house that unknowingly to him perfectly matches his childhood house. The painter seems to freely imagine a house, but the narrator’s perspective is in fact controlled by episodic information.

Affective Memory  155

References Arcangeli, M., & Dokic, J. (2015). Voyage mental dans le temps: quand l’imagination nous engage. In A. Berthoz and C. Debru (Eds.), Anticipation et Prédiction. Du geste au voyage mental (pp. 233–246). Paris: Odile Jacob. Atance, C. M., & O’Neill, D. K. (2001). Episodic future thinking. Trends in Cognitive Sciences, 5(12), 533–539. Boyer, P. (2008). Evolutionary economics of mental time travel? Trends in Cognitive Sciences, 12(6), 219–224. Carroll, N. (1990). The philosophy of horror. New York: Routledge. Conway, M. (1995). Flashbulb memories (essays in cognitive psychology). Hove: Lawrence Erlbaum. Currie, G., & Ravenscroft, I. (2002). Recreative minds: Imagination in philosophy and psychology. Oxford: Oxford University Press. de Sousa, R. (2007). Why think? Evolution and the rational mind. Oxford: Oxford University Press. de Sousa, R. (2008). Inference and epistemic feelings. In G. Brun, U. Doguoglu, & D. Kuenzle (Eds.), Epistemology and emotions (pp. 185–204). Ashgate: Aldershot. Debus, D. (2007). Being emotional about the past: On the nature and role of pastdirected emotions. Noûs, 41(4), 758–779. Dokic, J. (2014). Feeling the past: A two-tiered account of episodic memory. Review of Philosophy and Psychology, 5(3), 413–426. Dokic, J., & Arcangeli, M. (2015a). The heterogeneity of experiential imagination. In T. Metzinger and J. M. Wind (Eds.), Open MIND: 11(T). Frankfurt am Main: MIND Group Retrieved from http://open-mind.net/papers/the-heterogeneityof-experiential-imagination [published in 2016 in T. Metzinger and J. M. Wind (Eds.), Open MIND: Philosophy and the mind sciences in the 21st century (pp. 431–450). Cambridge, MA: MIT Press.] Dokic, J., & Arcangeli, M. (2015b). The importance of being neutral: More on the phenomenology and metaphysics of imagination. Reply to Brüggen. Metzinger, T., & Wind, J. M. (Eds.). Open MIND: 11(T). Frankfurt am Main: MIND Group Retrieved from http://open-mind.net/papers/the-importance-of-being-neutralmore-on-the-phenomenology-and-metaphysics-of-imagination2014a-reply-toanne-sophie-brueggen [published in 2016 in Metzinger, T., & Wind, J. M. (Eds.), Open MIND: Philosophy and the mind sciences in the 21st century (pp. 461– 465). Cambridge, MA: MIT Press.] Fernández, J. (2015). What are the benefits of memory distortion? Consciousness and Cognition, 33, 536–547. Gendler Szabó, T. (2000). The puzzle of imaginative resistance. Journal of Philosophy, 97(2), 55–81. Gendler Szabó, T., & Hawthorne, J. (Eds.). (2002). Conceivability and possibility. Oxford: Oxford University Press. Gerrans, P., & Mulligan, K. (2013). Imagination, default thinking and integration. Rivista di estetica, 54(3), 239–271. Goldie, P. (2003). One’s remembered past: Narrative thinking, emotion, and the external perspective. Philosophical Papers, 32(3), 301–319. Goldie, P. (2012). The mess inside: Narrative, emotion, and the mind. Oxford: Oxford University Press.

156  Margherita Arcangeli and Jérôme Dokic Goldman, A. (2006). Simulating minds: The philosophy, psychology, and neuroscience of mindreading. Oxford: Oxford University Press. Hopkins, R. (forthcoming). Imagining the past: On the nature of episodic memory. In F. Dorsch & F. MacPherson (Eds.), Perceptual imagination and perceptual memory. Oxford: Oxford University Press. Kind, A. (2016). Imagining under constraints. In A. Kind & P. Kung (Eds.), Knowledge through imagination (pp. 145–158). Oxford: Oxford University Press. Longfellow, H. W. (Ed.). (2011). Poems of places: An anthology in 31 volumes. Switzerland and Austria: Vol. XVI. 1876–79. New York: Bartleby. Martin, C. B., & Deutscher, M. (1966). Remembering. Philosophical Review, 75, 161–196. Martin, M. G. M. (2002). The transparency of experience. Mind and Language, 17(4), 376–425. McCarroll, C. J. (2017). Looking the past in the eye: Distortion in memory and the costs and benefits of recalling from an observer perspective. Consciousness and Cognition, 49, 322–332. Michaelian, K. (2016a). Mental time travel. Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Michaelian, K. (2016b). Against discontinuism: Mental time travel and our knowledge of past and future events. In K. Michaelian, S. B. Klein, & K. K. Szpunar (Eds.), Seeing the future. Theoretical perspectives on future-oriented mental time travel (pp. 62–92). Oxford: Oxford University Press. Moran, R. (1994). The expression of feeling in imagination. The Philosophical Review, 103(1), 75–106. Nigro, G., & Neisser, U. (1983). Point of view in personal memories. Cognitive Psychology, 15, 467–482. Noordhof, P. (2002). Imagining objects and imagining experiences. Mind and Language, 17(4), 426–455. Peacocke, C. (1985). Imagination, possibility and experience: A Berkeleian view defended. In J. Foster & H. Robinson (Eds.), Essays on Berkeley (pp. 19–35). Oxford: Oxford University Press. Perrin, D. (2016). Asymmetries in subjective time. In K. Michaelian, S. B. Klein, & K. K. Szpunar (Eds.), Seeing the future. Theoretical perspectives on future-oriented mental time travel (pp. 39–61). Oxford: Oxford University Press. Ricoeur, P. (1983). Temps et Récit. Tome I: L’intrigue et le récit historique. Paris: Éditions du Seuil. Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering the past and imagining the future. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 362(1481), 773–786. Sutton, J. (2010). Observer perspective and acentred memory: Some puzzles about point of view in personal memory. Philosophical Studies, 148(1), 27–37. Trakas, M. (2015). Personal memories. PhD dissertation. Paris: École des Hautes Études en Sciences Sociales (Doctoral School 286). Velleman, J. D. (2000). On the aim of belief. In The possibility of practical reason. Oxford: Oxford University Press. Walton, K. (1994). Morals in fiction and fictional morality. Proceedings of the Aristotelian Society, Supplementary Volumes 68, 27–50.

Affective Memory  157 Weinberg, J. M., & Meskin, A. (2006). Puzzling over the imagination: Philosophical problems. In S. Nichols (Ed.), The architecture of the imagination: News essays on pretence, possibility, and fiction. Oxford: Oxford University Press. Williams, B. (1976). Problems of the self: Philosophical papers 1956–1972. Cambridge: Cambridge University Press. Wittgenstein, L. (1961). Tractatus Logico-Philosophicus (D. Pears & B. McGuiness, trans.). London: Routledge. Wollheim, R. (1984). The thread of life. Cambridge: Cambridge University Press.

8 Painful Memories

Philip GerransPainful Memories

Philip Gerrans

1  Introduction: Mineness and Memory Recent discussions of episodic memory converge on the idea that it involves “unique awareness of re-experiencing here and now something that happened before, at another time and in another place” (Klein & Nichols, 2012). The phenomenology of subjective reacquaintance with past events is baptized autonoesis, a term introduced by Endel Tulving (Wheeler, Stuss, & Tulving, 1997; Tulving, 2002; Tulving, 2005). It refers to the experience of being the subject of experience. This experience is subtle and hard to describe and as with some other subtle phenomenologies its nature and role come more into focus when it is selectively impaired in neuropsychological and psychiatric disorders. Autonoetic experience is, as Jordi Fernandez (in press) observes, a form of “mineness,” a term of art developed by philosophers to analyze the structure of self-awareness. Marie Guillot distinguishes mineness from two other forms of self-awareness (Guillot, 2017). The first and most basic is minimal self-awareness. Self-awareness here is “a mere dative” as Zahavi once described it, a product of the fact that experience entails a subject of experience (Gallagher & Zahavi, 2005). Another form of self-awareness involves a representation of the self as the entity having the experience. The best analogy here is the way the mind represents objects as the entities in which features cohere. As a result perceptual experience is of objects not merely coherent collections of properties. Similarly, the mind represents regularities and continuities in subjective experience by treating them as occurring in a continuing entity: Me. Guillot calls the resultant form of self-awareness “Me-ness.” Mineness is a term she reserves for the type of experience that informs the subject (Me) how things matter, given her history, goals, and interests. As she says the content “resonates in a particular way for subjects” and the experience of that resonance is the feeling of mineness. On some views Me-ness and Mineness are not really distinct forms of representation but express different aspects of self-awareness. One reason for distinguishing mere dative self-awareness and Me-ness from Mine-ness is psychiatric and neuropsychological conditions in which

Painful Memories  159 subjects report that an experience does not feel like it is theirs, or that they are detached from the experience, while not denying that they are having the experience in a minimal sense, or that they are the entity/self having the experience. In some such cases it seems that what Guillot calls Mineness is selectively impaired leaving Me-ness and minimal self-awareness intact. Indeed, one argument for the relational account of autonoesis proposed by Stanley Klein is a neuropsychological case study of patient R.B. whose “impairment selectively targeted autonoesis while leaving stored content unscathed.” Thus Klein concludes that autonoesis is associated with content experienced in different modalities rather than intrinsic to the modal experience itself and, under unusual conditions, can dissociate from that content. In this respect it contrasts (phenomenologically at least) with the dative form of self-awareness, which seems intrinsic to any experience. The basis for this conclusion is R.B.’s testimony quoted in Klein 2012: When I remember the scene with my friends, studying, I remember myself walking into the room . . . and . . . other things I did and felt . . . But it feels like something I didn’t experience . . . (something I) was told about by someone else. It’s all quite puzzling. (my italics) He continued: I can see the scene in my head . . . I’m studying with friends in the lounge in the residence hall. But it doesn’t feel like it’s mine . . . that I own it. It’s like imagining the experience, but it was described by someone else. So R.B. is clear that he was the subject of the recollected episode and that he is the person recollecting it, but it does not feel as if it is his. Jordi Fernandez (in press) points out that R.B.’s loss of autonoesis for memory is just one instance of loss of the sense of mineness for experience and suggests that such dissociations are an avenue to developing a general account of what is involved in experiencing a mental content as one’s own. As he says perhaps there are other pathological conditions in which the subject disowns some of their phenomenal states; conditions to which we could appeal in order to vindicate the sense of mineness. I think that is correct. In particular depersonalization/derealization experience offers a very close parallel with R.B.’s experience. I think that the best explanation of R.B.’s condition is that he experiences a case of depersonalization for memory. To defend the idea that R.B.’s ‘loss of mineness for memory is a case of depersonalization I discuss some of the empirical/philosophical literature on depersonalization/derealization. I will

160  Philip Gerrans argue that Mineness as defined by Guillot, and autonoesis by Tulving is a case of what is called “subjective presence” in that literature. Each term refers to the form of self-awareness selectively impaired in pathological cases like that of R.B. It seems that R.B. has lost subjective presence for memory but there are many other cases where subjective presence is lost or diminished. Pain asymbolia is a striking example. In this case subjects will say that they are experiencing pain, but it does not matter, or that it feels as if it is not happening to them. Colin Klein made an insightful interpretation of these cases, comparing them to depersonalization the phenomenology of asymbolia might resemble a kind of depersonalization syndrome. . . . The asymbolic, and the depersonalized more generally, feel sensations that they are estranged from—that they do not take to be theirs in the sense that we normally do. (Klein, 2015) The idea here is that asymbolics have experience representing bodily damage (i.e., with nociceptive content) but do not associate that experience with mineness. This suggests a general strategy for understanding mineness. Namely investigating the way it is produced and lost for modal specific contents (i.e., memory, pain, perception, nociception, interoception, and so on) as well as in severe cases of depersonalization disorder (DPD) in which people say that it feels as if they are not present in all experience. To fully understand R.B. we need to understand how subjective presence for memory is produced and lost. Much of my account is based on consideration of the cognitive properties of the neural mechanisms involved. Stanley Klein thinks that focusing on the mechanisms that produce autonoetic experience and associate it with recollective experience is a category mistake: “it confuses memory with its preconditions” as he puts it. While I do not dispute the conceptual point, that phenomenology and underlying mechanisms are distinct I do think that mechanistic explanation of autonoetic phenomenology is highly informative. It tells us that autonoetic content is basically affective; it tells us how and why autonoetic content is detachable from noetic; it tells why affective content has the property of mineness/autonoesis; it suggests that R.B is undergoing a form of depersonalization for memory. I will argue that it also fits well with a constructive account of memory (Michaelian, 2016) (rejected by Klein) and explains crucial feature of prospection or future-directed mental time travel. Namely we need to feel that the represented content of memory and imagination is ours if we are to use it as a basis for decision. As a way into the issues and before introducing my account I first consider a proposal about the nature of mineness made by Jordi Fernandez.

Painful Memories  161

2  Fernandez’s Endorsement Account of Mineness Fernandez notes that R.B.’s case raises questions about the nature of mineness that have been answered in different ways, which he calls ‘identification’ and ‘endorsement.’ On the identification model the experience lacked by R.B. is the sense of being numerically identical to the person who is the subject of the recollected experience. However, as Fernandez notes, that cannot be quite right since R.B. does not doubt on the basis of his experience that he is the same person who had the recollected experience. Rather R.B reports feeling “as if” the experience is not his own, “as if” he is imagining it in response to an instruction, rather than reliving it. when I remember scenes from before the injury, they do not feel as if they happened to me—though intellectually I know that they did—they felt as if they happened to someone else. (my italics) The “as if” locution employed by R.B is quite important for two reasons. The first is that “as if” experiences have played a role in the literature that treats delusions as rationalizations of anomalous “as if” experiences. For example a patient with misidentification delusion consequent on damage to the face recognition system may feel “as if” she is looking at a stranger when she sees her husband. The delusion that her husband has been replaced by an identical-looking imposter is ex hypothesi a rationalization of that “as if” experience (Bayne & Pacherie, 2005). The second is that the “as if” locution typically arises in cases of derealization and depersonalization, local and transient (as in misidentification syndromes, deja and jamais vu) or global and/or sustained as in depersonalization and derealization disorders (Phillips et al., 2001; Baker, Hunter et al., 2003; Medford, Brierley et al., 2006; Medford, 2012; Billon, in press). Thus there is some motivation for investigating the idea that R.B. is experiencing a form of depersonalization/derealization for memory. Before pursuing that idea, however, we can set the scene by looking at the way Fernandez uses the concept of endorsement to explain R.B.’s loss of mineness for his memories. Endorsement accounts were developed to explain delusions of misidentification, such as Capgras delusion, in which patients believe that familiar people have been replaced by perceptually indistinguishable doubles or imposters. A standard explanation is that patient has damage to circuitry that links facial recognition to affective response, with the consequence that the seen person appears identical to the familiar, but the patient does not feel an anticipated affect. The seen person looks familiar, but the subject feels as “as if” she is seeing a stranger. Note that the patient does not say this. She says

162  Philip Gerrans “sees an imposter.” The idea that the experience that prompts the delusion is an “as if” experience is an inference based on a theory of the relationship between affective and face processing (Stone & Young, 1997). The endorsement model of delusional cognition is imported from philosophical accounts of empirical belief fixation. Experience is treated as evidence, warrant, or grounds for an empirical belief, given background beliefs about the likelihood of the experiential content being veridical. Endorsement accounts of delusional belief fixation suggest that delusions arise where the subject endorses an experiential content: that is takes it to provide support for an empirical belief that preserves the content of the experience. An important branch of this debate notes that in some cases of misidentification experience patients do not proceed to a delusion (i.e., they do not endorse their experience). Rather they report that the relevant person looks “as if” they are an impostor or double. So the endorsement account is an inference that in the case of delusion, a patient proceeds from an “as if” experience to an empirical belief that endorses that experience. If the experience is not endorsed the subject reports feeling “as if” her experience is of seeing a stranger who looks indistinguishable from a familiar. Much of the initial philosophical discussion of misidentification delusions was centered on whether, and in what sense, it is rational of a deluded subject to take her anomalous experience as warranting a belief that preserves its content. Extending the idea to the case of R.B. the endorsement account suggests that R.B. has an experience of mnemonic content that lacks mineness: it feels “as if” the experience is not his, while in all other respects it accurately represents his past. Interestingly, however, R.B. does not develop the delusional belief that his memories are not his. Nor does he misidentify them as memories. He says they are memories that feel “as if” they are imaginings. Thus he does not endorse his experience. Fernandez analyzes R.B.’s loss of mineness in essentially epistemic terms. He suggests that the experience of mineness lacked by R.B. consists in “an experience wherein some phenomenal state is presented to the subject as fitting, merited or appropriate.” Talk of fittingness suggests that the process involved is one in which experience is playing a justifying or epistemic role. This is confirmed when Fernandez develops the idea: “the experience of finding reasons for regarding them as being appropriate.” In support of the epistemic interpretation Fernandez suggests that patients with delusions of alien control (in which they claim that their actions are the product of other people’s intentions) are the result of passivity experience that results from a loss of the sense of “mineness” for actions. I agree that these patients have lost the sense that actions are their own. However, most accounts describe the problem as loss of a “sense of agency” for the action. That loss is a result of hyperactivity in (particularly) the inferior parietal lobule, a node in a circuit that plays a key role in action monitoring and control (Spence, 2001; Blakemore, Oakley, & Frith, 2003; Jardri,

Painful Memories  163 Delevoye-Turrell et al. 2009; Jardri, Pins et al. 2011). That circuit is intact and functioning normally in R.B.’s case and indeed in many other cases of loss of mineness. For that reason delusions of alien control are not the best evidence for a general theory of mineness and its loss. Rather they are evidence that the parietal systems play a crucial role in action monitoring by comparing actual to intended bodily movement. The patient says the movement is controlled by another’s intentions because, due to malfunction in the motor control system, he is unaware of the causal role of his own intention. At a high level of abstraction every instance of an alienated content can of course be unified in epistemic terms as grounds for a belief that the experience is “mine” or “not mine” because it is appropriate/inappropriate for that context. However, the price will be loss of explanatory power for specific phenomena. In particular it will detach phenomenological and philosophical investigation from empirical. I think it matters significantly that the mechanisms are different in delusions of alien control and R.B.’s case. That still leaves the question of why R.B. has lost his sense of mineness for memory. And a further question. Exactly what is “mineness,” and how is it lost?

3  Mineness, Depersonalization, and Derealization Guillot gives a hint when she says mineness is a matter of experience “resonating in a particular way” for the subject. Of course that is a metaphor but it does suggest that what is at issue is the significance of the content for the subject. A second clue is given by the “as if” locution. One reason this is interesting is that such locutions are characteristic of disorders of depersonalization and derealization. Fernandez actually lays the groundwork for pursuing analogies with these disorders in his discussion of R.B. s memories. He says, “They do not feel as if they had ‘happened to him.’ That seems to be a natural way of expressing the odd feeling that the scenes are not real” (my italics). The American Psychiatric Association describes Depersonalization Disorder (DPD) this way: “alteration in the perception or experience of the self so that one feels detached from, and as if one is an outside observer of, one’s own mental processes” (my italics). Alexandre Billon has argued that DPD is the result of a sustained and global loss of the sense of mineness. As he says, “Depersonalization consists in the lack of a phenomenal feature that marks my experiences as mine, which is usually called ‘mineness,’ and . . . the study of depersonalization constitute[s] an incomparable empirical probe to assess the scope, role, and even the nature of mineness” (Billon in press). After surveying a large body of clinical and historical examples from over a century he divides depersonalization into different forms according

164  Philip Gerrans the process involved in producing the relevant content. Dementalization is loss of mineness for psychological states, desomatization for bodily states, deagentivation for agency, and in the most extreme global case Death and Inexistence “a state of nothingness . . . as if I were dead.” Billon also emphasizes that depersonalization is almost always expressed by the “as if” locution.” He quotes a patient saying, “It feels as if it were someone else’s pain” for example. This contrasts with delusional cases in which people believe that someone else is feeling the pain, having the relevant thoughts or memories or in extreme cases that they have ceased to exist. An important aspect of Billon’s theory is that, like most who work in the area, he treats depersonalization and derealization as aspects of the same phenomenon. They reflect loss of mineness for different cognitive phenomena: representations of the external world (derealization) and of the internal states of the subject respectively. As Critchley and Garfinkel put it. “DPD can be summarized as a psychiatric condition marked by the selective diminution of the subjective reality of the self and world” (Garfinkel & Critchley, 2013). Subjective reality is an important term, because in DPD the world, and the patient’s body and inner states are veridically represented. But the subject’s experience nonetheless feels strange or unreal. In Billon’s terms R.B.’s account looks to be a paradigm case of dementalization: he has lost the sense of mineness for memory at the same time as he accurately represents the relevant episodes as occurring in his past. His memories are not “subjectively real” for him. Why?

4 What Is Mineness? At this point it is helpful to examine the neural mechanisms of depersonalization/derealization. The first point to note is that cognition is otherwise unimpaired. The world and the subject are cognized as normal but their representational content is experienced as though the subject is absent from the experience. This is the basic datum for relational accounts of mineness. It looks as if mineness can dissociate from cognition. The neural correlate of loss of mineness is hypoactivity or damage to the anterior insula cortex (AIC) (Sridharan, Levitin, & Menon, 2008; Craig, 2009; Craig, 2009; Singer, Critchley, & Preuschoff, 2009; Medford & Critchley, 2010; Terasawa, Shibata et al., 2012; Terasawa, Shibata et al., 2013; Gasquoine, 2014; Moayedi, 2014). The AIC is an enigmatic structure involved in many aspects of cognition but there is emerging consensus that one of its main roles is at the apex of the so-called Salience network, which signals the relevance of information to the organism A short way to explain this role is to note that it is necessary but not sufficient for a mind to represent states of the world, the subject’s body, and cognitive functioning. This is the role of perceptual and cognitive systems, which are intact in depersonalization/derealization. However even

Painful Memories  165 a veridical representation of the world and its own states is insufficient for the organism’s needs. The organism also needs to represent the significance of that represented information given the current context, its goals and their relative precedence. An elevation in blood pressure matters differently if it is an adaptive response to climbing stairs, a response to an insult in an argument, or a sudden spike during a period of rest. And experience reflects those differences. In the first case it is not sensed as emotionally significant and is represented by the mind within the context of bodily regulation, in the second it is part of the phenomenological signature of anger and in the third it might be felt as a worrying form of overarousal, perhaps a symptom. The AIC is the system that allows us to feel the significance of information represented across the mind. It thus is a hub of distributed circuits that collate, integrate, and interpret information relevant to the subject’s prospects. It plays this role by integrating representations of body state (e.g., blood pressure) produced by interoceptive systems, with representations of the world and organismic goals to determine the relevance of such body states in context. Naturally enough it is highly connected to the hierarchy of emotional processing systems, which orchestrate the processing of information relevant to the organism’s goals, which range from very basic (homeostatic maintenance, injury prevention) to high level (prevention of injury to reputation). Because goals are various and their precedence is context sensitive, emotional processing involves integrating disparate systems that operate at different levels of cognition. The result of this integrative processing is signaled by activation in the AIC. In effect it surveys the first order representation of body states and evaluates their significance in the cognitive context. As a result the most salient information is not just prioritized but also affectively marked and surfaces in consciousness as a feeling annexed to the relevant cognitive content. For example, when we see a familiar face smiling at us we experience a brief sense of familiarity and positive affect. Another way to put this is to say that the AIC allows representations of relevance to enter the global workspace in the form of feelings that inform reflective processing (Sridharan, Levitin et al. 2008; Craig, 2009; Füstös, Gramann et al. 2013). Critchley summarizes the idea that the AIC allows us to feel emotional significance. evidence from a variety of sources converges to suggest a representation of autonomic and visceral responses within anterior insula cortex, where, particularly on the right side, this information is accessible to conscious awareness, influencing emotional feelings. (Critchley, 2005) In discussing the role of AIC hypoactivity in pain asymbolia (loss of mineness for pain) Wiech et. al make the same point. If the AIC is not performing its proper function then nociceptive content will still be sensed, i.e., the

166  Philip Gerrans subject will be aware of bodily damage, but the significance of that damage in context will not be sensed as a distressing affect. Because, as they say The anterior insula ensures that salient stimuli such as painful stimuli will have preferential access to the brain’s attentional and working memory resources. (Wiech & Tracey, 2013) This explains why depersonalization is essentially a disorder of “de-affectualization” as Medford describes it. When the AIC is hypoactive, we do not feel the emotional salience of information represented by the mind. But to explain why “de-affectualization” leads to depersonalization and derealization we need another element. After all subjects do not just say that the world seems drained of affect (although they do say that) but that it seems unreal and/or that they are not part of the experience. Why does loss of affective response lead to depersonalization/derealization? The short answer is that the mind uses predictive models of the world and the subject’s body to construct its representations (Hohwy, 2013; Barrett & Simmons, 2015). Those models predict not just states of the world and states of the body, but the affective responses evoked by worldly encounters. In fact those affective predictions serve as constraints and interpretative heuristics for cognition. For example when we see a familiar face, affective processing feeds back into the recognition process modulating attention and face scanning. The constant modulation of affect as we interact with the actual, imagined, or recalled world produces changes in the quality of feeling, providing the subject with a sense of she is faring in the world. It is important to note that this experience is so automatic and constant and so closely integrated with cognition that almost every experience is affectively inflected. It is only when that intimate connection disappears that we become aware of its role. An affectively inert world and/or body in a situation where affective response is predicted is highly anomalous. A classic example is postpartum depressive psychosis in which patients may say that their baby is a doll (derealization), or that they feel they have disappeared (depersonalization). A contributing factor is a high degree of expectation (implicit and explicit) of positive emotional response to childbirth (Brockington & Kumar, 1982). When that expectation is confounded by absence of affect when the baby is presented (due perhaps to postnatal depression or hormonal disruption) the mother can feel a sense of estrangement. The idea that the flow of affect represents changes to a persisting entity, a subject, is a natural “inferred representation of the self”(Moutoussis, Fearon et al. 2014) in just the same way as postulating a physical object makes sense of the flow of sensory information. It is important to note that this “inference” is not always conscious and explicit, any more than the inference that perceptual properties are instantiated in objects is an explicit inference.

Painful Memories  167 Seth puts it this way: emotion and embodied selfhood are grounded in active inference of those signals most likely to be ‘me’ across interoceptive and exteroceptive domains. (Seth, 2013) To summarize: the mind models and predicts fluctuations of affect by attributing them to a continuing self. That “self model” allows us to experience not just the way things are, but the way they matter to us given our history, goals, and concerns (Gerrans, 2013; Hohwy & Michael, 2017; Gerrans & Letheby, 2017). When this integrative process fails (due to hypoactivity in the AIC or systems that link affective processing to cognition), but the world and body are being otherwise accurately represented, the subject feels that something is wrong. In such cases the subject feels as though cognition is intact and her consequent experience is as predicted (for example R.B. knows that his memories are accurate). But the self is not responding as it should. The inference made by the mind is that the normal subject of experience: the entity to which things are significant is absent. At the same time she knows intellectually that she is present and her cognition seems veridical. Thus she is forced to the locution it seems “as if” the experience is not mine.

5  Mental Time Travel and Empathy for Oneself In this section I apply these ideas to finesse the understanding of Mental Time Travel (MTT) as it is known. MTT is the term given to the use of memory and imagination to represent actual past, and possible future, episodes of autobiographical experience in planning and decision-making. The intuitive idea is that we use knowledge of our past history to imagine possible alternative futures for ourselves. The metaphor is that we “travel in time,” projecting ourselves back and forward to enable us to make adaptive personal decisions by re- and pre-experiencing actual and potential autobiographical episodes (Michaelian, 2016). Klein and Nichols make the point in their discussion of R.B. that autonoesis is essential if such experiences are to play that role in deliberation. Without it the mnemonic content is not experienced as personally relevant and memory (and imagination) lose their motivating force. In fact autonoesis for personal memory solves the same problem as empathy. Our concern for other people and motivation to act for their benefit is enhanced if we can share some of their affective experience. The same is true for our past and future selves: we care more if we share their experience. If I feel my future distress in anticipation I am more motivated to avoid danger than if I simply know that a course of action is dangerous.

168  Philip Gerrans Similarly, recollecting a previous dangerous episode will not help me behave adaptively in future unless I also experience a shudder of fear. This is one reason why people like R.B., whose autonoetic capacities are impaired, are also impaired decision makers. Interestingly Klein does not endorse the full constructive theory of mental time travel because he favors the retrieval theory of memory on conceptual grounds. His discussion of autonoesis is part of a conceptual argument that the term memory should be applied only to retrieval of stored content associated with feelings of autonoesis. Klein also argues that memory should not be characterized in terms of the processes that give rise to it. This view of memory runs counter to currently influential theories of memory that (i) situate its explanation in a wider context of planning and decision (ii) treat it as a constructive process, in effect arguing that memory is imagination of the past, rather than retrieval of stored experiential content (iii) can be categorized in terms of underlying processes. Together these are the three conceptual pillars of the mental time travel theory of memory. The constructive theory of memory depends heavily on the nature of the mechanisms involved. The fact that episodic memory for, and imagination of, the same experience share neural substrates suggests that the content of the experience is constructed by a system designed to simulate experiences in the absence of an eliciting stimulus “the medial temporal lobe system which has long been considered to be crucial for remembering the past might actually gain adaptive value through its ability to provide details that serve as building blocks of future event simulation”(Schacter, Addis et al. 2008). This idea is reinforced by findings that damage to this circuitry affects, not only episodic memory, but also prospection and a range of cognitive processes characterized by stimulus independence and self-representation (Hassabis, Kumaran et al. 2007a, Hassabis, Kumaran et al. 2007b; Hassabis & Maguire, 2007; Hassabis & Maguire, 2009). Rumination, planning, social prospection, empathetic imagination, daydreaming, moral cognition, and dreaming all recruit this system. These findings lead to the idea that at the level of neural computation the folk distinction between episodic memory and prospective imagination does not exist. Rather both processes are essentially simulatory: personal imagination and episodic memory are cases in which the mind constructs representations of actual and potential episodes of the subject’s life in the absence of actual perceptual elicitation. The overlap between memory and imagination results from the construction of imagery by the same neural machinery (Michaelian, 2016). That such constructive processes produce autonoetic episodes is an important part of the theory. The systems involved are usually described as “self-referential” (Gusnard, Akbudak et al., 2001; Spreng, Mar, & Kim, 2009). However the nature of such self-referential processing and the relation between mineness, affective processing, and self-referential mnemonic content is underexplored. The ability to “re-experience” and

Painful Memories  169 “pre-experience” episodes of emotion depends on a complicated neural architecture. Consider a typical case of prospection, in which we evaluate alternative future actions in terms of their potential outcomes for us. When I crash on an icy slope while skiing recklessly I feel pain and distress. Years later when I sit on the balcony of the chalet and contemplate skiing that run again on a foggy, icy, day I do not reactivate those sensory channels. I feel no broken bones or torn muscles and ligaments. But I still feel a shudder of fear when imagine myself accelerating uncontrollably and remember the chatter of the skis on the ice before they irrevocably lost contact. And I decide to stay in the chalet. Clearly I am not in actual pain when sitting on the balcony. But I feel highly aversively motivated as I contemplate the possibility of another accident. And such aversive feelings are vital: if we did not feel fear when contemplating future danger then we would not be motivated to avoid it. It is not enough to know that the piste is icy and steep. We need to feel it. How do we generate these aversive feelings if there is no perceptual stimulus to drive affective processing? The answer is that in such cases we evoke the affective content of remembered experience in order to evaluate its significance. And to do that we activate our AIC, the system that allows us to feel the significance of represented content. This suggestion about the role of the AIC in MTT is supported by an interesting finding about voluntary imagination of sensory states. In such cases, people do not simulate nociception (of representation of bodily damage) but simulate its affective consequences, thereby making the experience subjectively relevant. This explains the finding that: For the visualization of internal state sensations this meant increased activity in areas of interoceptive sensory processing, including the mid and anterior insula in the right hemisphere. This is a critical finding, as it suggests that primary interoceptive cortex, located in the posterior insula, was not significantly involved in the imagery of internal state sensations. (Bennett & Baird, 2009) Cases like this suggest that when we imagine or reflect on an experience we do not re-create a complete representation of the experience. Rather we (re)construct the affective response to that experience. In other words we represent not what happens to us but how it matters to us. We empathize with our past or future self by activating circuitry that represents not body state per se, but the significance of body state. To do this we need to be able to control activity in the AIC independently of perceptual/sensory elicitation. In order to do this, we exploit an adaptation that allows us to regulate AIC activity independently of its normal interoceptive (bodily) afferents.

170  Philip Gerrans This adaptation is evidenced in the management of pain responses and is part of the explanation of pain asymbolia (here I simply endorse Klein’s idea that asymbolia is a form of depersonalization for pain). We can reduce the distress response to pain even if we cannot change the bodily damage and its nociceptive representation. Thus one would expect that in such cases activity in the AIC would be down regulated while activity in those systems such as the posterior insula cortex (PIC) that monitor body state per se would be unaffected. We would in effect slightly depersonalize the experience. In fact this seems to be the case in reports of pain experience under opioid analgesia, in which patients report that the pain is not extinguished but “no longer matters.” A key finding here is that that opioids target, not only the PIC as one might expect, but the AIC and related limbic structures involved in emotional processing. The AIC in fact is even more responsive than PIC to low doses of opioids. This is presumably an adaptation. It is easier for an organism to regulate initial emotional response to bodily damage than to repair bodily damage. Thus, in contexts where the organism cannot devote resources to repair, it inhibits the system that produces negative affect in response to bodily damage and thereby stops pain from drawing attention away from other relevant activities. Opioids exploit or mimic this adaptation, down regulating the AIC, reducing the felt significance of pain. the FMRI data suggest that opioid analgesics can directly influence emotional responses at low doses that do not alter sensory aspects of pain. (Lee, Wanigasekera, & Tracey, 2014) Similarly pain asymbolics no longer assign emotional significance to bodily damage in virtue of hypoactivity in their AIC. For this reason pain asymbolia has been correctly characterized as a form of depersonalization disorder. The point holds not just for simulation of our own pain but also for empathetic simulation of the pain of others. When we empathize with another’s pain we do not re-create the full matrix of pain activation in our own brain. Rather we activate our AIC, which informs us that the pain is terrible. Empathy involves the simulation of felt emotional significance, not body state per se. We don’t imaginatively “feel their pain” any more than we feel our own future or past pain. This idea is supported by an ingenious study of empathy for pain experience by Rutgen and collaborators (2015) who used the effects of analgesics to probe the nature of empathy for pain. They were testing the hypothesis that “if empathy relies on the recruitment of the representations and neural processes engaged by first-hand pain, then experimentally changing these representations will also affect empathy for pain” (Rütgen, Seidel et al. 2015). They did so using placebo analgesia. The prediction was that subjects who experienced placebo analgesia should feel less empathy for pain in virtue of the diminished intensity of their own painful feelings. The placebo

Painful Memories  171 was then reversed by administration of naloxone: an opiate antagonist with the consequent prediction of increased empathy (“normalization”) of empathy for pain. The results were as predicted. Placebo analgesia reduced empathy for pain and reversal restored empathy. The interesting feature is that the target circuitry here was anterior insula and mid cingulate cortex “key areas of the network of areas activated by pain, and their activity has been directly related to the affective-motivational component of pain” (Rütgen, Seidel et al. 2015). Another way to put this is to say that to empathize with another’s pain (or even one’s own past or future pain) we do not re-create the actual bodily damage or representation thereof. Rather we re-create affective experience. This is why pain asymbolia and opioid analgesia, which deactivate the AIC, reduce distress for pain rather than nociception. The general point here is that personal significance of represented content depends on its association with AIC activity. That activity can vary independently of represented content as regulatory adaptation. But that independence means that AIC activity can dissociate from other contents. Where the AIC is damaged or involuntarily deactivated, represented contents feel “as if” they are not part of the subject’s experience. The aim of this section was to show that mental time travel requires the ability to generate and maintain affect in the absence of an eliciting stimulus in order that experience has autonoetic as well as cognitive content. The general role of the AIC at the apex of the salience network; its specific role in pain regulation, independent of nociception, together with its hypoactivity in depersonalization/derealization suggests it is the crucial substrate of autonoetic experience

6  Roger’s Version I have argued that affective response attributed to a continuing entity creates the feeling of being involved in experience baptized autonoesis or mineness. I also claimed that autonoesis understood in this way was the essential component of mental time travel, in effect allowing us to empathize with past or future selves. A large part of that argument was based on consideration of the mechanisms involved. Convergent evidence implicates the AIC and also suggests that the ability to activate the AIC in concert with memory or imagination generates the feeling of mineness essential to mental time travel. In the mechanistic spirit of this chapter, I should respond to a challenge to this view raised by discussion of the case of Roger, a lesion patient with extensive damage to the insular as well as the amygdala, ventromedial prefrontal, and anterior cingulate cortices. Roger, has a range of cognitive and affective deficits including dense anterograde amnesia. On my view Roger should have severe deficiencies in “mineness” “autonoesis” “subjective presence,” whichever term of art is preferred. The structures essential to creating perceptually driven affective response

172  Philip Gerrans (amygdala) and mnemonic and imaginatively driven affective response (ventromedial prefrontal cortex) and allowing us to feel those responses in cognitive context (AIC) are damaged or destroyed. Yet when Philippi et. al. tested Roger on a range of standard tests of self awareness (SA) they found “R is a conscious, self-aware, and sentient human being despite the widespread destruction of cortical regions purported to play a critical role in SA, namely the insula, anterior cingulate cortex, and medial prefrontal cortex”(Philippi, Feinstein et al. 2012). The experimenters concluded that that Roger was self-aware in virtue of his ability to integrate very basic bodily signals arising in the brainstem with his declarative and semantic knowledge about his life and personality traits and found “little support for the hypotheses that implicate the insular cortex as critical to all aspects of SA.” Not only that but Roger’s aversive response to painful stimuli was not only intact but amplified. Unlike the typical subject he does not seem to habituate to a series of painful stimuli but finds each occasion equally and newly painful. Thus it is very tempting to conclude that Roger has intact affective responses to pain. An obvious response is to claim that Roger has compensated for his loss, and in fact this is one interpretation of Roger’s symptoms his intact affective experience of pain is due to plasticity . . . the adaptive role of pain affect is so essential that the brain may automatically rewire in service of self-preservation . . . [suggesting that] . . . emotional experience of pain can be instantiated by brain structures outside of those traditionally presumed to be critical for pain affect. (Feinstein, Khalsa et al. 2015) Alternatively, Roger’s may compensate for his deficits by using alternative cognitive strategies. For example, his intact ability to sense his heartbeat is very likely not dependent on interoception but via the skin surface of his chest. When this surface was anaesthetized Roger’s ability declined, unlike that of controls who presumably exploited the intact interoceptive pathway (Philippi, Feinstein et al. 2012). If this is the case then the basic structure of my account can be preserved: Roger has reinstalled the basic cognitive architecture of affective response to pain through a combination of neural plasticity, redeployment, and adaptation. However, the idea that one’s whole emotional/cognitive/affective architecture could be seamlessly reinstalled seems unlikely. More likely is a complex combination of compensation and deficit. And there is a more complicated interpretation that suggests that Roger in fact has some serious affective deficits that impair his sense of the emotional significance of pain. This interpretation starts from the fact that Roger has exaggerated and unmodulated pain behavior. He does not habituate to continued painful

Painful Memories  173 trials. In this respect, he resembles some patients with ventromedial lesions who do not learn from negative experience (unsurprisingly since Roger has ventromedial lesions). His pain responses, behavioral and verbal, are like startle responses. Recall one important role of the anterior insula is to make affective feelings available in the context of reflective thought, to allow us to determine significance of bodily states. Where appropriate we can then redirect or focus attention, inhibit or amplify the affective response. We can decide to finish a task before consulting the doctor, take an aspirin, and lie down or rush to hospital according to the intensity of the pain and its relevance to emotionally significant goals. To a surprising degree (clearly not in extremely severe injury) we can independently regulate, automatically or under high-level control, the affective component of pain experience. Roger has none of this equipment available: he just has nociception, consequent stimulus driven behavior, and higher order cognition. My suggestion is that Roger is not in fact emotionally evaluating his pain, even at low levels. He is simply experiencing the automatic transition from nociception to behavioral expression. Thus pain is painful for Roger but not emotionally contextualized. This interpretation is in fact suggested by the experimenters data suggest that the limbic structures commonly associated with pain may play a fundamental role in pain regulation. Under this view, the missing regions in Roger’s brain would impair his ability to control and downregulate his pain responses. (my italics) (Feinstein, Khalsa et al. 2015) This suggests that in fact Roger may be missing precisely the affective capacity for which feeling is required. Regulation and control. This lack of ability to sense the significance his states rather than just respond reflexively should on my account be reflected in the loss of affective response to pain. His condition is consistent with such a loss despite his intact semantic knowledge about his condition. Unfortunately both lower order and higher order centers (amygdala and ventromedial prefrontal cortex) of Roger’s emotional processing system are lesioned and he is densely amnesic as well as having “myopia for the future” in Antoine Becchara’s nice phrase. Thus he cannot perform mental time travel even though he can reason about the future and past. This explains his constant happiness. Roger appears remarkably unconcerned by his condition. He hardly ever complains, and in general, shows little worry for anything in life. Both of his parents and his sister fervently claim that “Roger is always happy,” an observation that is consistent with our own impression.

174  Philip Gerrans Moreover, based on his family’s report, Roger is paradoxically happier now than he was before his brain damage. (Feinstein et. al. 2010) Roger does not in fact use affectively modulated mental time travel to navigate his world. When we focus on an affectively tinged sense of self, existing through time it seems that Roger not only lacks it, but does not feel the loss of that lack. He has lived in the same unmodulated emotional world for thirty years with one default setting that does not depend on the normal functioning of a mental time travel system that imbues simulation with affect. It is interesting to compare Roger’s case another case of episodic amnesia. KC has extensive hippocampal knowledge that also impairs his ability to imagine the future (aphantasia). A striking thing about KC is a lack of affective reaction to emotionally salient events. He does not recall emotional details of family tragedies and personal accidents for example.(Rosenbaum, McKinnon et al. 2004) Nonetheless KC performed adequately on a task designed to test his valuation of future rewards. This was despite the fact that “K.C. reported a “blank” state of mind when asked to construct ways in which he might use future rewards that he had chosen over immediate rewards”(Kwan, Craver et al. 2012). In some ways, this result runs counter to the hypothesis that mental time travel is an adaptation for adaptive decisionmaking. The ability to vividly inhabit the future allows us to feel as well as know about future prospects. Consistent with this hypothesis most subjects in this situation rely on a mixture of semantic knowledge and episodic prospection (e.g., “I thought about how I might spend the money in my retirement”) and nonepisodic (e.g., “I estimated accumulated interest”) future-oriented constructions. KC of course cannot do this, but he nonetheless reported relying on a “gut feel” to rank his preferences. I suggest that this is due to his intact amygdala function, which in effect allows him to evaluate the options as though they were presently available. Thus a larger reward later was ranked higher than a lower current reward because the decision has no temporal dimension for him. It becomes a matter of calculation supplemented with a gut feeling. He combined this affect with his intact semantic knowledge and reasoning ability to perform the task. Although the experimenters concluded that mental time travel was not necessary to avoid future discounting their conclusion is consistent with the idea that mental time travel involving affect dissociates from semantic memory “The present results suggest that in the absence of episodic memory, decision-making about future events can be based on semantic memory. These findings support the distinction between imagining and knowing about the future.”

Painful Memories  175

7 Conclusion Autonoesis is essentially a complex affective phenomenon. The self we feel in episodes of autonoesis is an entity inferred by the mind to predict and interpret the way the world makes us feel. Convergent evidence suggests that its neural substrate is the anterior insula cortex. Thus in cases like that of R.B., we would predict that either the AIC is damaged or hypoactive or disconnected from circuitry (particularly VMPFC), which coordinates the linking of affect to other representations, which comprise scenarios rehearsed in mental time travel.

References Baker, D., Hunter, E., Lawrence, E., Medford, N., Patel, M., Senior, C., Sierra, M., Lambert, M. V., Phillips, M. L., & David, A. S. (2003). Depersonalisation disorder: Clinical features of 204 cases. The British Journal of Psychiatry, 182(5), 428–433. Barrett, L. F., & Simmons, W. K. (2015). Interoceptive predictions in the brain. Nature Reviews Neuroscience, 16(7), 419–429. Bayne, T., & Pacherie, E. (2005). In defence of the doxastic conception of delusions. Mind & Language, 20(2), 163–188. Bennett, C. M., & Baird, A. A. (2009). The processing of internally-generated interoceptive sensation. Neuroimage, 47(1), S84. Billon, A. (in press). What is it like to lack “mineness.” Depersonalisation as a probe for the scope, nature and role of mineness. Blakemore, S.-J., Oakley, D. A., & Frith, C. D. (2003). Delusions of alien control in the normal brain. Neuropsychologia, 41(8), 1058–1067. Brockington, I. F., & Kumar, R. (1982). Motherhood and mental illness. London: Academic Press. Craig, A. (2009). Emotional moments across time: A possible neural basis for time perception in the anterior insula. Philosophical Transactions: Biological Sciences, 1933–1942. Craig, A. D. (2009). How do you feel—now? The anterior insula and human awareness. Nature Reviews Neuroscience, 10(1), 59–70. Critchley, H. D. (2005). Neural mechanisms of autonomic, affective, and cognitive integration. Journal of Comparative Neurology, 493(1): 154–166. Feinstein, J. S., Khalsa, S. S., Salomons, T. V., Prkachin, K. M., Frey-Law, L. A., Lee, J. E., . . . Rudrauf, D. (2015). Preserved emotional awareness of pain in a patient with extensive bilateral damage to the insula, anterior cingulate, and amygdala. Brain Structure and Function, 1–13. Feinstein, J. S., Rudrauf, D., Khalsa, S. S., Cassell, M. D., Bruss, J., Grabowski, T. J., & Tranel, D. (2010). Bilateral limbic system destruction in man. Journal of clinical and experimental neuropsychology, 32(1), 88–106. Fernandez J. (in press). The ownership of memories. In M. García-Carpintero & M. Guillot (Eds.), The Sense of Mineness. Oxford: Oxford University Press. Füstös, J., Gramann, K., Herbert, B. M., & Pollatos, O. (2013). On the embodiment of emotion regulation: interoceptive awareness facilitates reappraisal. Social Cognitive and Affective Neuroscience, 8(8), 911–917.

176  Philip Gerrans Gallagher, S. (2005). Phenomenological approaches to self-consciousness. In S. Gallagher & D. Zahavi (Eds.), Blackwell companion to consciousness. Oxford: Blackwell. Garfinkel, S. N., & Critchley, H. D. (2013). Interoception, emotion and brain: New insights link internal physiology to social behavior. Social Cognitive and Affective Neuroscience, 8(3), 231–234. Gasquoine, P. G. (2014). Contributions of the insula to cognition and emotion. Neuropsychology Review, 24(2), 77–87. Gerrans, P. (2013). Delusional attitudes and default thinking. Mind & Language, 28(1): 83–102. Gerrans, P., & Letheby, C. (2017). The self unbound: Ego dissolution in psychedelic experience. Neuroscience of Consciousness, 3(1): nix016 Guillot, M. (2017). I me mine: On a confusion concerning the subjective character of experience. Review of Philosophy and Psychology, 8(1), 23–53. Gusnard, D. A., Akbudak, E., Shulman, G. L., & Raichle, M. E. (2001). Medial prefrontal cortex and self-referential mental activity: Relation to a default mode of brain function. Proceedings of the National Academy of Sciences of the United States of America, 98(7), 4259–4264. Hassabis, D., Kumaran, D., & Maguire, E. A. (2007a). Using imagination to understand the neural basis of episodic memory. Journal of Neuroscience, 27(52), 14365–14374. Hassabis, D., Kumaran, D., Vann, S. D., & Maguire, E. A. (2007b). Patients with hippocampal amnesia cannot imagine new experiences. Proceedings of the National Academy of Sciences of the United States of America, 104(5), 1726–1731. Hassabis, D., & Maguire, E. A. (2007). Deconstructing episodic memory with construction. Trends in Cognitive Sciences, 11(7), 299–306. Hassabis, D., & Maguire, E. A. (2009). The construction system of the brain. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 364(1521), 1263–1271. Hohwy, J. (2013). The predictive mind. Oxford: Oxford University Press. Hohwy, J., & Michael, J. (2017). Why should any body have a self? In The subject’s matter: Self-consciousness and the body. F. Vignemont, & A. Alsmith, A. (Eds.). Cambridge, MA: MIT Press. Jardri, R., Pins, D., Lafargue, G., Very, E., Ameller, A., Delmaire, C., & Thomas, P. (2011). Increased overlap between the brain areas involved in self-other distinction in schizophrenia. PloS One, 6(3), e17500. Jardri, R., Delevoye-Turrell, Y., Lucas, B., Pins, D., Bulot, V., . . . Goeb, J.-L. (2009). Clinical practice of rTMS reveals a functional dissociation between agency and hallucinations in schizophrenia. Neuropsychologia, 47(1), 132–138. Klein, C. (2015). What pain asymbolia really shows. Mind, 124(494), 493–516. Klein, S. B., & Nichols, S. (2012). Memory and the sense of personal identity. Mind, 121(483), 677–702. Kwan, D., Craver, C. F., Green, L., Myerson, J., Boyer, P., & Rosenbaum, R. S. (2012). Future decision-making without episodic mental time travel. Hippocampus, 22(6), 1215–1219. Lee, M. C., Wanigasekera, V., & Tracey, I. (2014). Imaging opioid analgesia in the human brain and its potential relevance for understanding opioid use in chronic pain. Neuropharmacology, 84, 123–130.

Painful Memories  177 MCA, D., Phillips, M., Medford, N., Senior, C., Bullmore, E., Suckling, J., . . . Williams, S. (2001). Depersonalization disorder: Thinking without feeling. Psychiatry Research: Neuroimaging, 108, 14560. Medford, N. (2012). Emotion and the unreal self: Depersonalization disorder and de-affectualization. Emotion Review, 4(2), 139–144. Medford, N., Brierley, B., Brammer, M., Bullmore, E. T., David, A. S., & Phillips, M. L. (2006). Emotional memory in depersonalization disorder: A functional MRI study. Psychiatry Research: Neuroimaging, 148(2), 93–102. Medford, N., & Critchley, H. D. (2010). Conjoint activity of anterior insular and anterior cingulate cortex: Awareness and response. Brain Structure and Function, 214(5–6), 535–549. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge: MIT Press. Moayedi, M. (2014). All roads lead to the insula. Pain, 155(10), 1920–1921. Moutoussis, M., Fearon, P., El-Deredy, W., Dolan, R. J., & Friston, K. J. (2014). Bayesian inferences about the self (and others): A review. Consciousness and cognition, 25, 67–76. Philippi, C. L., Feinstein, J. S., Khalsa, S. S., Damasio, A., Tranel, D., Landini, G., . . . Rudrauf, D. (2012). Preserved self-awareness following extensive bilateral brain damage to the insula, anterior cingulate, and medial prefrontal cortices. PloS One, 7(8), e38413. Rosenbaum, R. S., McKinnon, M. C., Levine, B., & Moscovitch, M. (2004). Visual imagery deficits, impaired strategic retrieval, or memory loss: Disentangling the nature of an amnesic person’s autobiographical memory deficit. Neuropsychologia, 42(12), 1619–1635. Rütgen, M., Seidel, E.-M., Silani, G., Riečanský, I., Hummer, A., Windischberger, C., . . . & Lamm, C. (2015). Placebo analgesia and its opioidergic regulation suggest that empathy for pain is grounded in self pain. Proceedings of the National Academy of Sciences, 112(41), E5638–E5646. Schacter, D. L., Addis, D. R., & Buckner, V. (2008). Episodic simulation of future events. Annals of the New York Academy of Sciences, 1124(1), 39–60. Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences, 17(11), 565–573. Singer, T., Critchley, H. D., & Preuschoff, K. (2009). A common role of insula in feelings, empathy and uncertainty. Trends in Cognitive Sciences, 13(8), 334–340. Spence, S. (2001). Alien control: From phenomenology to cognitive neurobiology. Philosophy, Psychiatry, and Psychology, 8(2–3), 163–172. Spreng, R. N., Mar, R. A., & Kim, A. S. N. (2009). The common neural basis of autobiographical memory, prospection, navigation, theory of mind, and the default mode: A quantitative meta-analysis. Journal of Cognitive Neuroscience, 21(3), 489–510. Sridharan, D., Levitin, D. J., & Menon, V. (2008). A critical role for the right frontoinsular cortex in switching between central-executive and default-mode networks. Proceedings of the National Academy of Sciences, 105(34), 12569–12574. Stone, T., & Young, A. (1997). Delusions and brain injury: The philosophy and psychology of belief. Mind and Language, 12, 327–364. Terasawa, Y., Shibata, M., Moriguchi, Y., & Umeda, S. (2012). Anterior insular cortex mediates bodily sensibility and social anxiety. Social Cognitive and Affective Neuroscience, nss108.

178  Philip Gerrans Terasawa, Y., Shibata, M., Moriguchi, Y., & Umeda, S. (2013). Anterior insular cortex mediates bodily sensibility and social anxiety. Social Cognitive and Affective Neuroscience, 8(3), 259–266. Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53(1), 1–25. Tulving, E. (2005). Episodic memory and autonoesis: Uniquely human. The Missing Link in Cognition: Origins of Self-Reflective Consciousness, 3–56. Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic memory: The frontal lobes and autonoetic consciousness. Psychological Bulletin, 121(3), 331. Wiech, K., & Tracey, I. (2013). Pain, decisions, and actions: A motivational perspective. Front Neurosci, 7, 46.

Part IV

Memory in Groups

9 Shared Remembering and Distributed Affect John SuttonShared Remembering and Distributed Affect

Varieties of Psychological Interdependence John Sutton One significant feature of human life is our psychological interdependence. To greater or lesser extents, and across diverse cultural contexts, our cognitive and affective states are related to those of others around us. We act alongside and share experiences with partners, family members, friends, workmates, and other people with whom we are connected in our daily lives. And as a result, what each of us feels and remembers, what matters to each of us about the present and the past, and the way we imagine and plan for the future, can be influenced by what those others feel, remember, and care about. This occurs in the moment, when my emotions or moods, my decisions or thoughts, are modulated by the actions or reactions, judgments, or evaluations of someone close to me. But it also happens over time, and in many cases over years, decades, or lifetimes. Such interdependence does not mean that we think, remember, or feel the same way about things. In many cases, it matters greatly to me when the emotions or memories of someone I care about differ from my own. Our psychological lives can in certain circumstances be interdependent and mutually influencing, to different degrees and in different ways integrated with each other, whether or not the precise content or style of our thoughts, memories, and feelings happens to match. This chapter integrates four recent trends in philosophy of memory and philosophy of cognitive science, all addressing such phenomena of psychological interdependence. First, the ways that remembering is typically integrated or entwined with other cognitive and affective processes, with imagining and feeling, are acknowledged and highlighted (Sutton, 2009; Keightley & Pickering, 2012; Goldie, 2012). Second, social aspects of memory are seen as potentially beneficial: other people are not only sources of error or misinformation, but can in certain circumstances support and collaboratively structure the form and content of recall (Campbell, 2008; Sutton, 2008). Third, memory is a test case for claims that cognition can be extended or (better) distributed across an array of heterogeneous cognitive ecologies, spanning neural, bodily, social, and environmental resources (Sutton et al., 2010; Tollefsen, Dale, & Paxton, 2013), and relatedly for ideas about collective intentionality (Michaelian & Sutton, 2017). Fourth, in a

182  John Sutton more recent literature on which I focus, emotions too are seen as potentially distributed across body and world as well as brain (Colombetti & Krueger, 2015). I aim to identify tight and underappreciated links between these four points, especially between the social nature of remembering and the distribution of affective phenomena. Because memory is often in use when it is not explicitly in question, theorists whose primary attention is on another domain may not see just how heavily it is implicated. Just as plausible accounts of decision-making in group agency place demands on mechanisms of group memory to keep track of and use history effectively (Sutton, 2008), so a range of interacting forms of remembering are involved in the phenomena of ‘distributed affectivity’ (Slaby, 2016). I make this case in Section 2 by picking out four relevant features of distributed affectivity. These are features of interest in their own right that collectively confirm the close links in these contexts between emotion and memory. I then home in, in Section 3, on a specific question about what exactly is shared in shared remembering, in such socially distributed systems: again considering emotion and memory together, I argue that complementary relations between different people are often more significant than convergence or synchrony across interacting individuals.

1  Distributed Affectivity Some rather abstract philosophical discussions of the metaphysics of extended and distributed cognition risk losing sight of the real-world contexts that motivated these approaches (Hutchins, 1995; Clark, 1997; Michaelian & Sutton, 2013). In contrast, innovative recent work in the field arises from a shift to address affective phenomena—emotions, sentiments, moods, and the like—in specific personal and social contexts. If the processes of believing, remembering, and decision-making in certain circumstances spread across neural, bodily, and environmental resources, the same may hold for grieving, loving, and other kinds of feeling. This seems plausible, to many of us, both because emotional and cognitive processes are often tightly entangled, and in considering emotional lives in their own right, unfolding over multiple timescales. As theorists in this field suggest, a case can be made for the distributed nature of many affective phenomena, both occurrent and dispositional, from momentary emotion episodes to sentiments, temperaments, character traits, and moods (Colombetti & Roberts, 2015); I follow them in using the umbrella label ‘distributed affectivity’ to signal this broad scope (Slaby, 2016; Candiotto, 2016). My exposition of promising strands of recent work follows, in particular, lines of thought developed by Joel Krueger, Giovanna Colombetti, and their colleagues in powerful papers examining distributed affectivity in concrete settings—emotional interactions between parents and infants, depression and grief, music and dancing, and more (Krueger, 2014a, 2015; Varga & Krueger, 2013; Colombetti & Krueger, 2015; Krueger &

Shared Remembering and Distributed Affect  183 Szanto, 2016; Colombetti, 2015, 2017; Colombetti & Roberts, 2015; see also Griffiths & Scarantino, 2009; Stephan, Walter, & Wilutzky, 2014; Mühlhoff, 2015; Greenwood, 2013, 2015; León, Szanto, & Zahavi, 2017). After briefly characterizing the approach, I extract four features that together build my case that distributed affectivity also implicates memory, and in particular shared remembering, in surprisingly direct ways. To make the case for distributed affectivity, Krueger and colleagues focus on the processes of emotion regulation. Krueger notes that accounts of emotion regulation in mainstream cognitive, clinical, and social psychology (e.g., Gross, 1998) acknowledge the diverse, often embodied ways in which we can—fallibly but often effectively—redeploy attention, modulate affective response, or shift our appraisal and our experience of emotionally salient events and situations. But over longer timescales we also engage in sustained ‘ongoing manipulations’ of our worlds that have specifically emotional functions, distributing the mechanisms of emotion regulation across our social, physical, and cultural environments. We design our homes and rooms, our offices or cars, our devices and our networks in part to carry, amplify, or adjust, our emotions. We habitually go to a specific park or café, choose particular clothes or accessories, listen to certain kinds of music, or speak to particular friends or family members, because through repeated, sedimented experience we have built strong affective habits and norms around these external resources (Colombetti & Krueger, 2015; Krueger, 2015). Over time, we thus construct, inhabit, and modify persisting integrated affective systems incorporating both social and environmental resources, involving the kinds of rich iterative coupling that signals genuinely distributed processes. By offloading, manipulating, and otherwise regulating our emotional lives in these externally looped ways over time, we sometimes access emotions that are richer or subtler than we might otherwise experience, or at least find our affective worlds shaped or transformed differently as a result. Engaging deeply with music, for example, may for some people over time sculpt an ‘expanded phenomenological repertoire’ in that they thus “gain access to an expanded realm of feeling states and modes of expression largely inaccessible outside of a musical context . . . because music is constituted by expressive dynamics that are more agile, evocative, and nuanced than are their behavioral counterparts” (Krueger, 2014b, p. 209). Parallel affective transformations can develop in socially distributed forms of emotion regulation: sometimes we could not feel, or deal with, certain emotion experiences in the absence of particular significant others (Varga & Krueger, 2013). These arguments for distributed affectivity emerge from and sit most naturally with one particular way of developing the distributed cognition approach. If distributed emotion regulation spans and incorporates neural, bodily, social, technological, and environmental resources, then there are many significant differences across the component parts of the relevant distributed systems. This heterogeneity, as Krueger notes, is a virtue

184  John Sutton insofar as such disparate resources, wherever they happen to be located and whatever their formats and properties, have complementary features that in certain circumstances support a capacity to mesh or integrate into larger systems with novel or different properties (Krueger, 2014a, p. 538; cf Greenwood, 2013). I return to this point later: but for now, we have enough of a sketch in place to push on, fleshing out the approach by extracting four features of the phenomena of distributed affectivity.

2  Four Features of Distributed Affectivity By identifying four key characteristics of distributed affectivity, I build up a richer picture of the phenomena, and demonstrate how thoroughly they involve memory in meeting the demands of keeping track of interactions across integrated affective systems over time. 2.1  Distributed Affectivity Involves Embodied Remembering We engage in ongoing embodied emotion regulation at a number of levels. Emotion experience involves, and can often be affected by, modulation of the expressive and physiological dimensions of emotion, and by reshaping or just attending to posture or gesture, movement or breath (Krueger, 2015; and on moods Colombetti, 2017). These bodily dimensions of emotion experience, and our ongoing fallible and partial but genuine capacities to access and influence them, are sometimes coupled with environmental settings. It is the embodied feel or comfort of certain rooms or views, pieces of furniture or corners of familiar cafes, like the one in which I write these words, which grounds emotional ease. In contrast, negotiating unfamiliar places, devices, or situations often involves jarring bodily unease as well as, or as the basis of, unpleasant emotions and uncertain cognitions. An initial sense in which distributed affectivity involves embodied remembering is shown by such patterns of bodily and affective responses over time. As phenomenological philosophers argue, the smooth or disrupted feel of being in the world, the fluent or dysfluent nature of embodied coping, or the interactional fit between body and setting all involve holistic, prereflective history-dependent processes implicating the bodily sedimentation of experience (Behnke, 1997; Casey, 2000; Fuchs, 2012). Such lived associations, intuitive responses, and affective habits of engagement with the world implicate forms of embodied remembering much more flexible and idiosyncratic than reflexes. Although, as I will argue, the bodily dimensions of distributed affectivity are not restricted to these more tacit forms of embodied knowhow, we can already see that these distributed emotion systems are fundamentally diachronic, and that embodied memory is one key basis by which history animates emotion experience. The notions of ‘embodied’ and ‘procedural’ remembering cover a wide range of phenomena, and in using them we need

Shared Remembering and Distributed Affect  185 to avoid any implication that other forms of remembering are somehow not embodied (Sheets-Johnstone, 2009; Sutton & Williamson, 2014). Further, in noting the centrality of embodied remembering to distributed affectivity, I am not suggesting that specific past embodied and emotional experience is carried ‘in’ the body or ‘stored’ in discrete form. Its effects and influences are typically more cumulative and holistic, and (unlike at least core forms of personal or episodic remembering) need not be subjectively attributable to the past: from our embedded point of view, embodied emotion experience just takes the forms it does without directing us to or being about its sources in the past. The embodied aspects of distributed affectivity, finally, are not exclusively individual or bounded. This is perhaps clearest in considering intimate relationships and close friendships, where touch, embodied copresence, and joint activity can mediate or constitute love or trust. But other small groups have characteristic bodily affective styles too (Slaby, 2016). The fact that people have engaged in certain practices together many times before can shape how it feels to do so together again. Again, socially as well as environmentally distributed remembering is tightly interanimated with embodied interaction and embodied memory. 2.2  Distributed Affectivity Involves Active Self-Scaffolding Some of the processes of distributed affectivity operate pre-reflectively, forming a world of embodied emotional experience that can be taken for granted until it is disrupted. But these more tacit or sedimented characteristics by no means exhaust the phenomena. As Krueger and his colleagues argue, both individually and collectively we engage in regular and insistent active modification of affective infrastructures. We redesign our desks, repack our bags, rework our habits, reconsider our friendships, rethink our schedules. When we puzzle over, seek advice on, or google for new ways of doing things, we do so not merely for instrumental reasons and to improve efficiency or productivity, but also so that things will feel better when we turn on our devices, enter our favorite rooms, or fit an exercise session into our day. Adjusting daily practices and re-engineering features of our familiar environments are two central forms of the active self-scaffolding of distributed affect. Just as our first-order selection of certain artifacts or pieces of music or actions is a mundane means of emotional self-stimulation, as we select and manipulate situations ‘to modulate our emotional phenomenology,’ so we also engage over time in sustained second-order reworking of our environments that “loop back onto us in complex ways and shape what we feel and how we feel it.” By thus setting up features of our familiar inhabited world, we can use them over time “to grant access to kinds of experiences we couldn’t otherwise have without their regulatory input” (Krueger, 2015, p. 266). This kind of ‘self-scaffolding’ (Bickhard, 2005) has been plausibly identified as a key sign of extended or distributed cognition: Colombetti and

186  John Sutton Roberts (2015), for example, base their case for fully distributed affectivity on cases in which “a system is coupled to an environmental item through which the system loops some kind of self-stimulating activity, and this selfstimulating activity in particular has been set in place and maintained over time” to play specific affective roles within the larger system (compare Clark, 2005). Such phenomena challenge any neat distinction between active and passive emotional experiences in the way that ideas of extended and distributed cognition should. On the one hand, our affective responses are elicited or transformed by interacting with certain features of our environments. On the other hand, we have often actively and iteratively created, modified, cared for, assessed, and reassessed just those features and just those interactions. Such active self-scaffolding imposes further demands on memory. As individuals or in small groups, we must over time track the operations of our artifacts, routines, and social interactions, identifying and evaluating patterns of affective response. Given what we know of memory’s constructive nature, it is just as well that such tracking need not be anything like complete: the point is that we are typically not starting from scratch in designing, or working out how to operate within our emotional worlds, and are thus leaning on our abilities to recall at least some prior relevant interactions. A number of forms of memory are involved: both personal and embodied remembering are often needed to make adjustments within a distributed affective system on the basis of past experience. 2.3  Distributed Affectivity Is Diachronic The phenomena of distributed affectivity are typically temporally extended. The emotions or moods in question are not isolated token events existing only at or for a moment: they take time. This is true of occurrent, episode-like emotion events and experiences with a unique qualitative feel (Colombetti & Roberts, 2015, pp. 1256–1260), as well as of enduring emotional dispositions. In most cases at least, when grieving and despairing, being angry and being joyful are distributed across bodily, environmental, and social as well as neural resources, they are complex or systemic phenomena incorporating disparate states and processes that endure over time. And at longer timescales, the kind of iterative scaffolding of affective environments I described previously involves cycles of mutual modulation as embodied subjects adjust their emotional worlds and in turn dynamically attune to the worlds they thereby change. In three respects, the point that distributed affectivity is diachronic could be pushed in more metaphysically ambitious directions. In each case, what I say here is compatible with the stronger view, but does not require it: the dialectical aim is to make the case for distributed affectivity as broadly acceptable as possible. First, I need not argue that there cannot be purely synchronic or momentary emotions, only that core cases of distributed affect

Shared Remembering and Distributed Affect  187 do not take that form. Second, I need not take such affective phenomena to be essentially diachronic, in the way that Peter Goldie treats grief as essentially a temporally extended process (2011, especially 125–6, 137). Finally, I need not adopt an exclusively diachronic approach to all extended and distributed cognition in general, as in Michael Kirchhoff’s attack on the possibility of any merely synchronic cases of cognitive extension (Kirchhoff, 2015). Again, the point is just that the bulk of plausible cases that can be realistically seen as exemplifying distributed affectivity are temporally extended. This suffices to give memory a key role in grounding and supporting the ways in which we design, coopt, lean on, and adjust our distributed emotionregulation systems. Our bodily responses in affectively salient situations are forms of embodied remembering. The tracking of particular interactions is episodic remembering, typically embedded in the fuller life narratives or themes that structure autobiographical memory. Semantic memory in the form of our background knowledge, both general and personally relevant, informs our active self-scaffolding. Another pervasive form of distributed affectivity involves prospective remembering, in that emotions and moods are often relevant as we track and revise our intentions for future actions, as well as in thinking about the future more generally. But crucially, distributed affectivity does not merely implicate all these distinct forms of memory separately. Past experiences typically shape emotional ecologies and responses in many ways at once. By first distinguishing among these forms of remembering, we can then focus inquiry on the many ways in which they interact or interanimate. When I feel pleasure or emotional comfort at returning to a much-loved place or room, for example, I may be remembering specific experiences there, feeling an embodied ease that derives from many repeated visits to the place, enjoying vivid flashes of sensory imagery or memory, and drawing on or updating the beliefs or general attitudes that have formed over years. Multiple and often multimodal forms of remembering, in other words, may overlap and entwine when my emotional experience loops out through the world in situations like this. History permeates the memory systems and processes of brain, body, and world in interlocking ways. 2.4  Distributed Affectivity Involves Shared Remembering The final feature of distributed affectivity for consideration here falls naturally out of the discussion, and occupies the rest of this chapter. This is the core point that distributed affectivity is deeply entwined with memory: both individual memory in all of its forms, as in many of the cases discussed previously, and shared remembering, especially or in the strongest cases shared remembering of shared experiences or joint actions (Sutton, 2008). Another direct approach to this point can be drawn from Colombetti and Krueger’s discussion (2015) of trust as one key interpersonal affective

188  John Sutton scaffold. They note that in this realm trust in a resource (an artifact or another person) relates not to its reliability in telling the truth about the world, as Clark and Chalmers (1998) and Sterelny (2010) stress for the cognitive realm. Rather, trust is confidence in the reliability of certain effects on our affective states: we may have, for example, an “expectation that others will have a certain modulatory impact on our affective life” (2015, pp. 1162, 1166). Colombetti and Krueger continue: Part of the reason we experience intimacy with family and friends is because we know what sort of affective feedback we can expect from them. For example, based on previous interactions, we know which member of the family to turn to for the affirmation needed to elevate our mood, or which friend to call if we need a good laugh. (2015, p. 1167) This catches some of the subtle ways we navigate the social world. Note how centrally memory is implicated: we carry, and act on the basis of, the history of these ‘previous interactions.’ Typically, this is all taken for granted: memory is in use though not explicitly in question, though when things change or expectations are thwarted we may recall the past events and experiences that gave rise to those expectations, perhaps re-evaluating them on the fly as we go. Again, a number of interacting timescales can be in play at once: long-established narratives about family members or friends, which can endure for and beyond individual lifetimes, both shape and are shaped by our affective assessments of exchanges operating over days and weeks, within the standard space of interpersonal reasons, which themselves are grounded in the faster processes of alignment and embodied interaction that may escape awareness (Bietti & Sutton, 2015). Crucially, these are typically mutual relations: our family members or friends remember our own patterns of response and the shared history between us, just as we do ourselves: “we feel at home in these relationships because we have, to a certain extent, individualized them . . . we thus play an active role in shaping the way that our interpersonal relationships function as reliable affective scaffoldings” (Colombetti & Krueger, 2015, p. 1170). In small groups of this kind, each person tends to know what each other person typically feels and how they may respond in particular circumstances, and how those responses are in turn likely to affect others. This is the affective dimension of what Wegner (1987) called a ‘transactive memory system,’ in which through long experience and effective communication each member can map and access the distinctive memories of the other members as needed. Emotional interaction influences distributed memory processes, and shared remembering influences distributed affect. This opens up a range of fascinating and under-researched issues: the rest of this chapter takes some early steps.

Shared Remembering and Distributed Affect  189

3 Convergence or Complementarity? What Is Shared in Distributed Systems? I have been discussing the distribution of emotions and memories over more than one person, and using the language of ‘sharing’ to describe the relation of group members to particular emotions or memories. But what does it mean to ‘share’ cognitive or affective states or processes in these contexts? Just as there are different ways in which people can share material items, so psychological sharing comes in many forms. It is easier to get a grip on the notion of shared perceptual experiences. When two people are jointly attending to the same object, they can establish and maintain a perceptual common ground. Even though you and I now look at the candle on the table from different visuospatial perspectives, we are looking at it together, and each of us knows this: such experiences are joint in that they make knowledge available not only about the external world, but also about each other’s epistemic state (Seemann, 2011, 2017). For different reasons, in memory and emotion the same kind of relation to present environmental affordances may be missing. Reminiscing or feeling ‘jointly’ are therefore likely to be less common phenomena, and to take considerably longer to become established developmentally (Hoerl & McCormack, 2005). Here I critically consider one approach to these issues for a putative account of shared remembering and distributed affectivity. This is the idea, prevalent in some research on both memory and emotion, that the primary or most significant form of interdependence—the best ‘mark’ of genuinely shared remembering or feeling—is convergence or similarity across the members of the group in question. For memories or emotions to be genuinely distributed across two or more people, do these people need to be remembering or feeling the same things, in more or less the same ways? To put the issue another way, we want to know what has to be true of the individual members of a dyad or small group in which affectivity (or memory) is distributed. Or, what is likely to happen to those members as the distributed interactions continue? The first approach I consider treats convergence as the central form of sharing in memory and emotion. Then I pan back to theoretical considerations from the debate on distributed and extended cognition, suggesting that reason to doubt convergence can be found in the independently plausible ‘second-wave’ approach that stresses the complementarity of heterogeneous resources in distributed cognitive systems. I put that line of thought to work in arguing that asymmetric forms of interdependence are more characteristic of and more important in shared remembering and distributed affectivity. 3.1  Convergence, Synchrony, and Merger To take memory first, it is sometimes claimed that convergence is the signature of a genuinely collective process. For the larger-scale forms of

190  John Sutton collective memory studied in the social sciences, a classic statement of this view is by Peter Novick. Aiming analytically “to separate ephemeral and relatively inconsequential memories from those that endure and shape consciousness,” and contrasting memory sharply with historical consciousness, Novick writes that “collective memory simplifies; sees events from a single, committed perspective; is impatient with ambiguities of any kind; reduces events to mythic archetypes” (Novick, 1999, p. 4). In cognitive psychology, a powerful strand of the recent move to take collective and collaborative memory seriously treats convergence as a necessary condition: for Hirst and Manier, “a collective memory can only be said to form if a community converges on a shared rendering” of the past (2008, 193). When Hirst notes that a “common understanding of the past” can arise when social influences sculpt “our recollections so that we all remember the same details,” this point is the only beneficial effect of social influences mentioned in a recent popular article on “ways that other people can warp your memory” (Robson, 2016). Hirst’s flourishing research program studies “the emergence of shared mnemonic representations that preserve group membership and group identity,” seeking to identify ways that “conversational remembering leads to increased mnemonic convergence” (Coman & Hirst, 2015; Yamashiro & Hirst, 2014). Likewise, in considering again recent research on shared and distributed emotion, we find a focus on the construction and maintenance of affective synchrony across individuals. This is due in part to reliance on studies of coordinated joint actions, where synchronous movement or behavior, or the induction of entrainment across interacting individuals “on multiple levels from physiology to syntax,” has repeatedly been shown to have systematic and often positive effects on cognition and emotion in groups (von Zimmerman & Richardson, 2016). The concept of synchrony is used in a number of ways in these literatures. In a narrower usage, synchronic interactions are contrasted with cases of complementary interactions in which people realize a shared goal by each engaging in different actions (Dale et al., 2013; Skewes et al., 2015). In this usage, synchrony is defined as involving “the matching of actions in time with others” (Mogan, Fischer, & Bulbulia, 2017), and is “a specific case of coordination” in which interacting individuals “tend to exhibit the same behavior at the same time” (Paxton & Dale, 2017). Alternatively, a broader notion of synchrony can be invoked to encompass more diverse forms of coordination or interdependence. This is what the theorists of distributed affectivity have in mind, for example, when referring to phenomena of “early dyadic parent—infant affective coordination” that are “characterized by the core feature of synchrony” (Varga & Krueger, 2013, pp. 272–273). Arguing that effective or disrupted early synchronic interaction between infants and caregivers constitutes a form of “dyadically distributed emotion regulation” with long-lasting cognitive and affective results, Varga and Krueger suggest that “synchrony in distributed emotion regulation” is also found in some

Shared Remembering and Distributed Affect  191 adult relationships. Also citing research on joint action, they point to forms of bodily and emotional interactional synchrony both in relations between therapist and patient and in intimate romantic relationships (2013, 286–7). As long as the full range of phenomena of interdependence across these domains is acknowledged, the choice of terminology for the different varieties need not cause conflict. My preference is to reserve ‘synchrony’ for cases closer to convergence, matching, or identity, and thus to see it as just one specific kind of coordination. We want to distinguish different forms of interdependence, and to treat them differently across the domains of action, emotion, and memory for example, to ask novel questions about relations between these distinct forms and across these domains. There are indeed cases in which people become more alike, at least in certain respects, in certain contexts. Whether converging on a shared rendering of the past, moving together in time, coregulating positive affect, or incorporating aspects of a lover’s identity (cf Hofstadter, 2007, pp. 233–235), people experience powerful forms of togetherness. None of what follows is intended to deny or belittle such phenomena. But convergence or synchrony, understood in this narrower sense, is not the only or the most important way that either memories or emotions are shared. To prepare the specific point that there are other kinds of cognitive and affective interdependence in these domains, I first return briefly to the larger case for distributed cognition and affect. 3.2 Complementarity I noted previously that theorists of distributed affectivity acknowledge the diversity of the resources that we incorporate into larger systems of moodmanagement and emotion-regulation. Music, clothes, rooms, cafes, habits, devices, and so on operate in ways unlike the neural processes that play their own unique and essential roles in our affective lives. The formats, mechanisms, and rates of change of these external system components differ dramatically from those of the in-skull components. Similarity between internal and external parts of the boundary-spanning affective system is not required: rather, what matters is that and how the relations and interactions between these disparate elements “yield integrated larger systems” (Clark, 1998, p. 99; 1997, p. 220; Sutton, 2010). This stress on the complementarity of inner and outer resources is a different emphasis in or route toward distributed cognition and affect from the idea that we need to establish parity between inner and outer resources. What’s important is not any intuition about similarities between our internal resources and relevant external systems, but the nature and intensity of integration across these heterogeneous resources: complementarity “thus subsumes and takes precedence over parity” (Sutton, 2010, p. 206; Sutton et al., 2010, pp. 524–527; cf Krueger, 2014a). To assess which interactions between complementary and disparate internal and external resources might be signs of distributed remembering

192  John Sutton or affectivity, we study the specific relations in each case. The relevant dimensions, depending on context and explanatory project, may include Clark and Chalmers’ original criteria of trust and glue (1998), and perhaps additional dimensions relating to the intensity, uniqueness, and form of the interactive coupling in question (Wilson & Clark, 2009; Sterelny, 2010; Heersmink, 2015). Such dimensions are, crucially, matters of degree, which is just what we want. Both memory and affectivity are in practice more or less distributed: it depends. We want to construct a multidimensional space for enquiry into such more-or-less distributed systems, rather than seeking a sharp distinction: we want to examine, within the same framework, cases of more shielded or insulated memory and emotion as well as those that are more distributed (Sutton et al., 2010, 534–8). When we are dealing not with relations between agents and artifacts, but as we have been throughout this chapter with cases of putatively socially distributed cognition or emotion, we might wonder if a tweaked “social parity principle” might operate, as suggested in productive work by Tollefsen and others (Tollefsen, 2006; Theiner, 2013; cf Gallagher & Crisafi, 2009). The point would be that when the relevant ‘external resource’ is another person rather than an artifact, there are no deep differences of kind, mechanism, and process between the parts of the distributed system: my partner and my therapist are, fortunately, more like me in relevant respects than my smartphone or my handbag. This might then seem to support the idea that convergence, synchrony, or even merger is the right model for distributed affectivity and shared remembering. I think that is not the right lesson to take, and that even in the case of socially distributed cognition, complementarity rather than parity is the right route, the more perfect wave. Looking in turn at action, affect, and memory, I can make the case that convergence is only one form of or model for cognitive-affective interdependence, and in many ways not the most relevant or interesting one. The general response is not original. In the case of emotion, for example, Rainer Mühlhoff argues in developing the notion of ‘affective resonance’ that in rich face-to-face social interaction “the interactants’ behaviors and affective experiences may not necessarily resemble each other, but yet they are a jointly created dynamic, shaped within the relational interplay” (2015, 2): instead, as he puts it, resonance “is less about similarity in behavior than about co-constitution of (potentially even divergent) behavior within a joint dynamic interplay” (2015, 14). I want to underline and extend just this line of thought to the range of domains we have been considering. 3.3  Mesh and Asymmetric Interdependence Many joint actions require different contributions from the people working together. Just as a smooth and flowing conversation can occur with participants taking different roles and contributing quite different numbers and

Shared Remembering and Distributed Affect  193 lengths of conversational turns, so many physical activities and cultural practices involve different behavior from each participant across extended temporal sequences (Böckler, Knoblich, & Sebanz, 2010; Dale et al., 2013). In most genres of music and sport, successful or interesting performance arises when band or team members do their own thing—exercise their own skill set—in ways that best complement what everyone else is doing (Sutton & Tribble, 2014; Williamson & Sutton, 2014). At higher levels there is much that the participants share—performance goals, aesthetic or strategic plans and visions, for example—but in many complex actions the way that individuals’ subplans will best ‘mesh,’ as Michael Bratman puts it, will not involve them doing the same things as each other (Bratman, 2013; Pacherie, 2014). Synchrony does not exhaust the forms of coordination in joint action. Nor do kinds of entrainment and alignment in which some formally specifiable dynamic drives the interlocking actions of two or more participants. Especially when we look at connections across distinctive timescales of joint action, we find a range of richly collaborative processes that do not involve such simple or literal forms of coordination (Bietti & Sutton, 2015). Turning from action to affect, likewise, there is no reason to think that the social distribution of emotion must involve similarity, synchrony, or symmetry across interacting participants. As Krueger himself underlines elsewhere, rich affective interdependence between infant and caregiver may in fact rest on the differences between the two. Even when their actions are entrained within or across modalities in more simple ways, the adult participant’s emotional states and processes complement or mesh with, rather than matching or converging with those of the infant (Krueger, 2013; Greenwood, 2013). Likewise in psychotherapy: Varga and Krueger are aware that capacity in the therapeutic situation or system for “dyadically distributed emotional regulation and interactive repair” (2013, 287) may depend precisely on the therapist’s ability to maintain and manage quite different affective states and processes from those of the patient. Even when the therapist deploys certain forms of synchronous behavioral matching and nonverbal convergence as part of the process, she is simultaneously reflecting on and working with bodily, affective, and cognitive responses that may diverge dramatically (for a striking example see Straker 2007). Yes, the therapist must be richly and subtly responsive to the patient, maintaining an intense kind of cognitive-affective interdependence or coupling in which what each party says, does, and feels makes an ongoing and dynamically sensitive difference to what the other says, does, and feels. But the medium of such interdependence need not be convergence. We can make the next link, from emotion to memory again, by considering intimate relations between friends or lovers. Douglas Hofstadter considers the utility of thinking of a close couple as “a higher-order individual made of two ordinary individuals” (2007, 222), and points to the iterative, intensifying, ongoing ways in which each partner can come to ‘think with’ some of the other’s attitudes, tastes, beliefs, and values, and even adopt some of

194  John Sutton the other’s memories vicariously as their own. But he is not suggesting that this involves merger, or denying the residual and irretrievable differences or gulfs between two embodied selves. Of course some relationships involve more codependence than others: but this can take many different forms. There is an unresolved empirical debate in work on life satisfaction in couples between ‘emotional convergence theory,’ which suggests that similarity between partners predicts long-term satisfaction, and approaches based on ‘complementarity,’ which “suggest that the longer partners are together, the more dissimilar they become” (Schade et al., 2016). The kinds of ‘high dependency’ or ‘merger’ in relationships that some clinical psychologists see as one risk factor for ‘complicated’ grief after bereavement (Johnson et al., 2007) are sometimes seen as polar opposites of ‘independence’ (Maccallum & Bryant, 2013, p. 719). But cognitive and affective interdependence does not have to be merger, and need not exclude substantial independence. We can see this by returning to the notion of transactive memory (Wegner, 1987; Gupta & Hollingshead, 2010). Different close relationships, whether among intimate partners or in experienced work teams, can involve quite different patterns or distributions of information. In the case of memories, sometimes this first-order spread will involve high differentiation or specialization. In some couples, one partner remembers all about the car or all the events that happened on holiday, while the other remembers everything important about the house, or about the grandchildren; in others, there are no particular areas of separate expertise and the whole shared history is more equally sampled by both partners. In some work teams, a division of labor is rigorously enforced so that there is no doubling-up or redundancy: in others, overlapping expertise is encouraged, and it is easier for members to fill in for each other. What matters— what makes either kind of system a genuinely transactive memory system—is not the particular way in which the first-order memories are distributed, but rather the shared understanding among all parties of that pattern of distribution, supporting mutual access to relevant information as required. Significant similarity, overlap, or convergence in memory is therefore only one limiting case of transactive or shared memory. More differentiated memory systems are both common and effective. As Wegner stresses, there is an intrinsic instability or volatility to differentiated or specialized organizations of memory: if the right kinds of labelling, accessing, and retrieval systems are in place, some information that was once held by only one party will as required be made available to all, in the processes driving the dynamics of remembering in the group. But relatively differentiated categories and skills can remain in place, if for example the dyad or group has a clear shared identity and shared goals, engages in a range of collaborative actions, and when the members are aware of the group’s range and internal heterogeneity. There will be definite limits to this heterogeneity, in both memory and emotion. If there is extreme disagreement about the nature of certain past events, for example, or about their emotional significance, it is unlikely that

Shared Remembering and Distributed Affect  195 we are dealing with an ongoing group in which the notions of distributed memory or affectivity really get a grip. But within those limits, there is no reason to accept that either convergence or agreement is essential. An intimate and flourishing romantic couple may have differing accounts of their first meetings and of the emotionally significant features of their early days together: but as long as they share higher-order understanding and commitment, the fact that their renderings of the salient past are not precisely shared in terms of content need not matter. Likewise at larger levels, a group can encompass and tolerate—within limits—a range of distinct attitudes to past events, with neither simplification nor convergence essential for the continuing existence of the broader collective or distributed memory system. In human beings, socially distributed cognitive and affective systems are typically not cases of ‘swarm intelligence,’ in which the aggregative interaction of more or less homogeneous component elements, each following relatively simple procedures, produces emergent larger-scale outcomes (Sutton & Tribble, 2014). First, the members of such human groups are quite different from each other in psychologically relevant respects, bringing distinctive properties, capacities, and skills to the interactive situation. Second, what the individual members do in their groups is very often considerably more complicated and differentiated, and more directly reliant on shared history, than the typically implicit responses of undifferentiated flocking or swarming units. Third, some of the interactive processes in question can be more mindful, open to personal and interpersonal influence, sometimes explicitly involving active and self-reflective consideration and reconsideration of the nature of the group and its strategies, memories, and affective evaluations. I have argued, first, that distributed affectivity involves shared remembering in a number of forms and for a number of reasons. In going on to assess what exactly is ‘shared’ in each case, I have suggested that what is ‘shared’ can remain at a level higher than that of either process or content. Rich cognitive and affective interdependence can take the form of asymmetric coregulation without similarity or symmetry. Doing, feeling, thinking, or remembering the same thing as each other is only one of many ways in which we mutually influence each other. Shared strategies may specifically leave these differences in place, or even encourage and exacerbate them: effective, successful, pleasurable interdependence between partners, friends, or colleagues may often involve the complementary meshing of asymmetric capacities, skills, memories, and emotions. In this chapter I hope to have pointed to some issues about the relations between cognitive and affective interdependence, and about the range of ways in which we can share memories and emotions, which are worth further theoretical, experimental, and ethnographic investigation.

Acknowledgments I am grateful to the editors for their patience and assistance, and to two anonymous reviewers for helpful comments. Earlier versions of parts of this

196  John Sutton chapter were presented at workshops in Dunedin and Cambridge: thanks to Kirk Michaelian, Kathy Puddifoot, and Lisa Bortolotti for organizing these events, and to the audiences for helpful suggestions. Though the influence of many of my wonderful collaborators on this work is strong—especially that of Amanda Barnier, Celia Harris, Richard Heersmink, Doris McIlwain, Kirk Michaelian, and Lyn Tribble—they do not necessarily agree with much of this. Other help along the way that I appreciate greatly has come from Giovanna Colombetti, Dorothea Debus, Tim Fawns, Graeme Friedman, Christine Harris-Smyth, Joel Krueger, and McArthur Mingon.

References Behnke, E. A. (1997). Ghost gestures: Phenomenological investigations of bodily micromovements and their intercorporeal implications. Human Studies, 20(2), 181–201. Bickhard, M. H. (2005). Functional scaffolding and self-scaffolding. New Ideas in Psychology, 23, 166–173. Bietti, L. M., & Sutton, J. (2015). Interacting to remember at multiple timescales: Coordination, collaboration, cooperation and culture in joint remembering. Interaction Studies, 16(3), 419–450. Böckler, A., Knoblich, G., & Sebanz, N. (2010). Socializing cognition. In B. Glatzeder (Ed.), Towards a theory of thinking (pp. 233–250). Springer: Berlin. Bratman, M. E. (2013). Shared agency: A planning theory of acting together. Oxford: Oxford University Press. Campbell, S. (2008). The second voice. Memory Studies, 1(1), 41–48. Candiotto, L. (2016). Extended affectivity as the cognition of primary intersubjectivity. Phenomenology and Mind, 11, 232–241. Casey, E. (2000). The ghost of embodiment: On bodily habitudes and schemata. In D. Welton (Ed.), Body and flesh (pp. 207–225). Oxford: Blackwell. Clark, A. (1997). Being there: Putting brain, body, and world together again. Cambridge, MA: MIT Press. Clark, A. (1998). Author’s response: Review symposium on Being There. Meta­ science, 7, 95–103. Clark, A. (2005). Word, niche, and super-niche: How language makes minds matter more. Theoria, 20(3), 255–268. Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19. Colombetti, G. (2015). Enactive affectivity, extended. Topoi, 36(3), 1–11. Colombetti, G. (2017). The embodied and situated nature of moods. Philosophia, 45(4), 1437–1451. Colombetti, G., & Krueger, J. (2015). Scaffoldings of the affective mind. Philosophical Psychology, 28(8), 1157–1176. Colombetti, G., & Roberts, T. (2015). Extending the extended mind: The case for extended affectivity. Philosophical Studies, 172(5), 1243–1263. Coman, A., & Hirst, W. (2015). Social identity and socially shared retrieval-induced forgetting: The effects of group membership. Journal of Experimental Psychology: General, 144(4), 717–722. Dale, R., Fusaroli, R., Håkonsson, D. D., Healey, P., Mønster, D., . . . Tylén, K. (2013). Beyond synchrony: Complementarity and asynchrony in joint action. Proceedings of the 35th Annual Meeting of the Cognitive Science Society, 79–80.

Shared Remembering and Distributed Affect  197 Fuchs, T. (2012). The phenomenology of body memory. In S. C. Koch, T. Fuchs, M. Summa, & C. Müller (Eds.), Body memory, metaphor and movement (pp. 9–22). Amsterdam: John Benjamins. Gallagher, S., & Crisafi, A. (2009). Mental institutions. Topoi, 28(1), 45–51. Goldie, P. (2011). Grief: A narrative account. Ratio, 24(2), 119–137. Goldie, P. (2012). The mess inside: Narrative, emotion, and the human mind. Oxford: Oxford University Press. Greenwood, J. (2013). Contingent transcranialism and deep functional cognitive integration: The case of human emotional ontogenesis. Philosophical Psychology, 26(3), 420–436. Greenwood, J. (2015). Becoming human: The ontogenesis, metaphysics, and expression of human emotionality. Cambridge, MA: MIT Press. Griffiths, P., & Scarantino, A. (2009). Emotions in the wild. In M. Aydede & P. Robbins (Eds.), The Cambridge handbook of situated cognition (pp. 437–453). Cambridge: Cambridge University Press. Gross, J. J. (1998). The emerging field of emotion regulation: An integrative review. Review of General Psychology, 2(3), 271–299. Gupta, N., & Hollingshead, A. B. (2010). Differentiated versus integrated transactive memory effectiveness: It depends on the task. Group Dynamics: Theory, Research, and Practice, 14(4), 384–398. Heersmink, R. (2015). Dimensions of integration in embedded and extended cognitive systems. Phenomenology and the Cognitive Sciences, 14(3), 577–598. Hirst, W., & Manier, D. (2008). Towards a psychology of collective memory. Memory, 16(3), 183–200. Hoerl, C., & McCormack, T. (2005). Joint reminiscing as joint attention to the past. In N. Eilan, C. Hoerl, T. McCormack, & J. Roessler (Eds.), Joint attention, communication, and other minds: Issues in philosophy and psychology (pp. 260– 286). Oxford: Oxford University Press. Hofstadter, D. (2007). I am a strange loop. New York: Basic Books. Hutchins, E. (1995). Cognition in the wild. Cambridge, MA: MIT Press. Johnson, J. G., Zhang, B., Greer, J. A., & Prigerson, H. G. (2007). Parental control, partner dependency, and complicated grief among widowed adults in the community. Journal of Nervous and Mental Disease, 195(1), 26–30. Keightley, E., & Pickering, M. (2012). The mnemonic imagination. London: Palgrave Macmillan. Kirchhoff, M. D. (2015). Extended cognition and the causal-constitutive fallacy: In search for a diachronic and dynamical conception of constitution. Philosophy and Phenomenological Research, 90(2), 320–360. Krueger, J. (2013). Ontogenesis of the socially extended mind. Cognitive Systems Research, 25–26, 40–46. Krueger, J. (2014a). Varieties of extended emotions. Phenomenology and the Cognitive Sciences, 13(4), 533–555. Krueger, J. (2014b). Musical manipulations and the emotionally extended mind. Empirical Musicology Review, 9(3–4), 208–212. Krueger, J. (2015). The affective “we”: Self-regulation and shared emotions. In T. Szanto & D. Moran (Eds.), The phenomenology of sociality: Discovering the “We” (pp. 263–280). London: Routledge. Krueger, J., & Szanto, T. (2016). Extended emotions. Philosophy Compass, 11(12), 863–878.

198  John Sutton León, F., Szanto, T., & Zahavi, D. (2017). Emotional sharing and the extended mind. Synthese. Maccallum, F., &. Bryant, R. A. (2013). A cognitive attachment model of prolonged grief: Integrating attachments, memory, and identity. Clinical Psychology Review, 33(6), 713–727. Michaelian, K., & Sutton, J. (2013). Distributed cognition and memory research: History and current directions. Review of Philosophy and Psychology, 4(1), 1–24. Michaelian, K., & Sutton, J. (2017). Collective memory. In K. Ludwig & M. Jankovic (Eds.), The Routledge handbook of collective intentionality (pp. 140– 150). London: Routledge. Mogan, R., Fischer, R., & Bulbulia, J. A. (2017). To be in synchrony or not? A metaanalysis of synchrony’s effects on behavior, perception, cognition and affect. Journal of Experimental Social Psychology, 72, 13–20. Mühlhoff, R. (2015). Affective resonance and social interaction. Phenomenology and the Cognitive Sciences, 14(4), 1001–1019. Novick, P. (1999). The Holocaust in American life. New York: Houghton Mifflin Harcourt. Pacherie, E. (2014). How does it feel to act together? Phenomenology and the Cognitive Sciences, 13(1), 25–46. Paxton, A., & Dale, R. (2017). Interpersonal movement synchrony responds to high- and low-level conversational constraints. Frontiers in Psychology, 8(1135). Robson, D. (2016, September 20). Four ways that other people can warp your memory. BBC Future. Retrieved from www.bbc.com/future/story/20160920-fourways-that-other-people-can-warp-your-memory Schade, H. M., Hülür, G., Infurna, F. J., Hoppmann, C. A., & Gerstorf, D. (2016). Partner dissimilarity in life satisfaction: Stability and change, correlates, and outcomes. Psychology and Aging, 31(4), 327–339. Seemann, A. (2011). The role of joint experience in historical narratives. Journal of the Philosophy of History, 5(2), 201–229. Seemann, A. (2017). Reminiscing together: Joint experiences, epistemic groups, and sense of self. Synthese, 1–16. Sheets-Johnstone, M. (2009). Animation: The fundamental, essential, and properly descriptive concept. Continental Philosophy Review, 42, 375–400. Skewes, J. C., Skewes, L., Michael, J., & Konvalinka, I. (2015). Synchronised and complementary coordination mechanisms in an asymmetric joint aiming task. Experimental Brain Research, 233(2), 551–565. Slaby, J. (2016). Mind invasion: Situated affectivity and the corporate life hack. Frontiers in Psychology, 7. Stephan, A., Walter, S., & Wilutzky, W. (2014). Emotions beyond brain and body. Philosophical Psychology, 27(1), 65–81. Sterelny, K. (2010). Minds: Extended or scaffolded? Phenomenology and the Cognitive Sciences, 9(4), 465–481. Straker, G. (2007). A crisis in the subjectivity of the analyst: The trauma of morality. Psychoanalytic Dialogues, 17(2), 153–164. Sutton, J. (2008). Between individual and collective memory: Coordination, interaction, distribution. Social Research, 75(1), 23–48. Sutton, J. (2009). Looking beyond memory studies: Comparisons and integrations. Memory Studies, 2(3), 299–302.

Shared Remembering and Distributed Affect  199 Sutton, J. (2010). Exograms and interdisciplinarity: History, the extended mind, and the civilizing process. In R. Menary (Ed.), The extended mind (pp. 189–225). Cambridge, MA: MIT Press. Sutton, J., Harris, C. B., Keil, P. G., & Barnier, A. J. (2010). The psychology of memory, extended cognition, and socially distributed remembering. Phenomenology and the Cognitive Sciences, 9(4), 521–560. Sutton, J., & Tribble, E. B. (2014). The creation of space: Narrative strategies, group agency, and skill in Lloyd Jones’s The Book of Fame. In C. Danta & H. Groth (Eds.), Mindful aesthetics: Literature and the sciences of mind (pp. 141–160). London: Bloomsbury/ Continuum. Sutton, J., & Williamson, K. (2014). Embodied remembering. In L. Shapiro (Ed.), The Routledge handbook of embodied cognition (pp. 315–325). London: Routledge. Theiner, G. (2013). Onwards and upwards with the extended mind: From scaffolding individual cognition to scaffolded group cognition. In L. Caporael, J. Griesemer, & W. Wimsatt (Eds.), Developing scaffolds in evolution, culture, and cognition (pp. 191–208). Cambridge, MA: MIT Press. Tollefsen, D. P. (2006). From extended mind to collective mind. Cognitive Systems Research, 7(2), 140–150. Tollefsen, D. P., Dale, R., & Paxton, A. (2013). Alignment, transactive memory, and collective cognitive systems. Review of Philosophy and Psychology, 4(1), 49–64. Varga, S., & Krueger, J. (2013). Background emotions, proximity and distributed emotion regulation. Review of Philosophy and Psychology, 4(2), 271–292. von Zimmermann, J., & Richardson, D. C. (2016). Verbal synchrony and action dynamics in large groups. Frontiers in Psychology, 7. Wegner, D. M. (1987). Transactive memory: A contemporary analysis of the group mind. In B. Mullen & G. R. Goethals (Eds.), Theories of group behavior (pp. 185– 208). New York: Springer. Williamson, K., & Sutton, J. (2014). Embodied collaboration in small groups. In C. T. Wolfe (Ed.), Brain theory: Essays in critical neurophilosophy (pp. 107–133). London: Palgrave Macmillan. Wilson, R. A., & Clark, A. (2009). How to situate cognition: Letting nature take its course. In P. Robbins & M. Aydede (Eds.), The Cambridge handbook of situated cognition (pp. 55–77). Cambridge: Cambridge University Press. Yamashiro, J. K., & Hirst, W. (2014). Mnemonic convergence in a social network: Collective memory and extended influence. Journal of Applied Research in Memory and Cognition, 3(4), 272–279.

10 Memory, Attention, and Joint Reminiscing Felipe De BrigardMemory, Attention, and Joint Reminiscing

Felipe De Brigard

1 Introduction Since grammar school was far away from home, I often endured long bus rides in heavy traffic. To avoid boredom, my friends and I used to play “Veo Veo”—“I See, I See”—the Colombian equivalent of “I spy.” One of us, gazing through the window of the bus, would glance over the busy scenery of the city. Meanwhile, everybody else would keep their eyes closed. Eventually, the kid surveying the scene would single out a particular object and would say “Veo Veo.” That was the sign for the rest of us to open our eyes and ask “¿Qué ves?”—“What do you see?” He would then give us a clue, a particular feature of the selected object, and we would then try to guess the object he had in mind. We could ask up to five questions of the form “Does it have an X?” where X was a property of the object we thought the kid was attending to. If the kid said “no,” that meant we were focused on the wrong object, so we would have to attend to a different one. If the kid said “yes,” then one could either keep asking—to make sure one had the right object in mind—or one could try to guess what the object was. If you were wrong, you were out. But if you guessed correctly, you’d get to pick the next object. The point of the game was to be the first one to attend to the same object as the kid who got to pick it. What we were doing was an exquisite exercise in what psychologists call joint attention: our capacity to attend to the same external target and realize that the other person is attending to the same one (Moore & Dunham, 1995). Consider the moment in which the kid who had mentally selected the object in his visual field realizes that another kid has guessed correctly. How does the former know that his thought refers to the same object the latter has in mind? First, both of them need to have the selected object in their visual fields. This, however, is not enough, as likely other kids, at some point, might had the selected object in their visual fields. Additionally, they both needed to single out that object from its surroundings; both of them must have selectively attended to it. But again, this isn’t enough. Another kid, whether playing or not, may have been attending to that very object, at that precise moment, without realizing that the object of his or her attention

Memory, Attention, and Joint Reminiscing  201 was the object chosen by the kid who was picking it out. What it is required, then, is a kind of attentive triangulation, whereby both kids are aware of the object and of each other with the recognition that each other knows that the object they are attending to is the chosen one. According to Campbell (2002), this attentive coordination makes the other subject, as well as the object, constituents of the content of their joint mental state. Now suppose that we want to play a different game, one that we may call “I Remember, I Remember.” It is just like “I see, I see” except that, in this version, one of the participants remembers a particular object while the others try to guess what she has in mind. It sounds much harder, doesn’t it? After all, unlike the case of perceptual joint attention, the alleged constitutive relation between the perceivers and the object of attention cannot be met. In the case of joint reminiscing, the intentional object is not present—it may not even exist. Moreover, unlike perceptual joint attention, it isn’t required that both subjects were ever at the same time in direct contact with the object of their memories. For instance, one can jointly reminisce about an old professor with another alumnus of the same school one just met. Of course, it may be possible that, in the course of jointly reminiscing, both realize that they shared a class, but it need not mean that, at the time, they both were aware of each other jointly attending to the professor. Surprisingly, though, we engage in joint reminiscing all the time. What does it take for us to engage in joint reminiscing? Specifically, how can two or more people jointly refer to an object that is long gone—or, at least, that is not present in their surrounding? In this chapter, I offer a threepart answer to this question. First, I suggest that our capacity to remember intentional objects during memory retrieval depends on our capacity to direct our attention inwardly toward the relevant component of a memorial content—a mental act I call, in the spirit of Prinz (2007), mental ostension. Second, I argue that, in order for us to refer to remembered intentional objects, we must possess the ability to refer to them indirectly or “deferredly” by way of mentally ostending toward a present mental content; in short, we must be capable of deferred mental ostension. Third, I claim that in order to jointly reminisce, we must have the capacity to guide someone else’s attention inwardly toward the relevant aspect of the mental content we want them to focus on so that they become aware of the past object we are deferredly ostending. I call this concerted deferred mental ostension. I explain each element of this account in turn.

2  Memory and Mental Ostension Imagine this event. I am grocery shopping, strolling down the aisles, when all of the sudden I hear a female voice, behind my back, calling my name. Think of what happens as a result of my hearing this brief sequence of phonemes. First, as I was silently focused on a particular visual scene, the noise made my attention shift from the shelves onto the stimulus behind my back. Given the

202  Felipe De Brigard silence around me, any auditory stimulus would have done that, of course. But this noise was a particularly relevant sequence of phonemes: it was my name. Had it been any other sound against a noisy background, I may not have heard it. My brain is attuned to certain noises that are socially relevant for me, like my name, and as result it makes me conscious of them even if I had been sensitized to background noises of equivalent pitch and volume (Wood & Cowan, 1995). This shift of attention to an exogenous stimulus, which my brain had already recognized as socially relevant, in turn shifts the cognitive mode I am in to what Tulving (1983) called “retrieval mode”: a mental state in which I am poised to retrieve information from memory. This shift occurs as I turn my back toward the source of the stimulus, no more than 500 milliseconds after its onset. The face, the tone of voice, and the mannerisms of the woman behind me constitute the perceptual cues with which I now try to recognize her—but to no avail. Perhaps noticing my facial expression of confusion, she appends her call with a new utterance: “We met last week at your party.” This new string of auditory information, added to already in-process perceptual cues, reactivated further sensory contents toward which I now turn my attention inwardly. Now I am covertly surveying different sensory mental contents. These contents are presented to me as blurry snapshots, maybe even quick footages, of scenes featuring my house and my friends in situations I recognize as having happened last week during my party. All of the sudden, there is a match between the perceptual cues and the sensory contents I’m aware of—a phenomenon Semon (1904/1921) called “ecphory.” My attention has been focused upon a particular region of a scene in which I see a person that highly resembles my interlocutor. She’s dancing, wearing a funny-looking hat. (1) I remember that hat! I say now to the woman in front of me, and she smiles approvingly. No more than a few seconds elapsed since she said my name. To make sense of this example, we need first to understand how we become aware of retrieved memorial contents, which in turn requires us to understand how we manage to retrieve memorial contents. Let us start with retrieval. Most philosophical accounts of memory retrieval have been mere speculations based on the commonsensical idea that experiences are somehow saved in a metaphorical storehouse, where they lose vivacity over time as though they were accumulating dust, awaiting their eventual retrieval during recollection. Recent developments in cognitive psychology and neuroscience have shown that this view is mistaken. For one, memory consolidation—i.e., the physical process by means of which the brain changes so as to encode experienced information in a memory trace—is a highly selective process. Not all the information that was initially perceived is encoded, and not all of the information that is encoded is available for retrieval. Much of our sensory information is lost due to inattention

Memory, Attention, and Joint Reminiscing  203 and working memory limits, as well as normal decay caused by lack of rehearsal and selective consolidation during sleep (Paller & Voss, 2004). In addition, the encoded information does not remain stable over time. Almost four decades of research in the cognitive psychology of false and distorted memories have shown that, during retrieval, memories become malleable and prone to being contaminated by extraneous information (Roediger, 1996). Finally, evidence also suggests that events that were only sketchily encoded can nonetheless be remembered with detail via pattern-completion processes that fill in the missing information in surprisingly reliable ways (McClelland et al., 1995). As a result, there is now wide consensus among neuroscientist regarding the reconstructive character of our memories (Schacter et al., 1998; Schacter & Addis, 2007). Remembering does not consist of the exact reproduction of previous experiences, but rather of the reconstruction of previously entertained mental contents by way of reactivating the brain regions that processed them during encoding (Rugg et al., 2008; De Brigard, 2012). Reactivation, of course, is not all there is to remembering, as we need to tell apart memories of previous events from experiences of current events. The brain manages to do that by way of incorporating, during retrieval, brain regions that were not involved during encoding and by redeploying some of the same regions for different purposes. In particular, whereas encoding recruits the sensory cortices and the medial temporal lobes, retrieval additionally recruits pre-frontal and parietal cortices (Buckner & Wheeler, 2001), which likely engage meta-cognitive processes such as source monitoring. To better understand this process, recall the previous example and consider how my memory of the woman’s hat gets first encoded and then retrieved. Suppose that, at my party, I did in fact attend to the woman and her hat. My sensory cortices first processed this fleeting perception in a distributed manner (i.e., visual information in occipital cortex, auditory information in auditory cortex, etc.). Since I did pay attention to her and to her hat, (most) of this perceptual information was processed by my working memory, and some of it was, in turn, bound together—presumably by the hippocampus—as a single, unified event. Neurophysiological evidence suggests that the area CA3 of the hippocampus carries out this binding and stores a sort of index of the episode (McClelland, McNaughton, & O’Reilly, 1995). However, this index does not include any sensory information per se. Instead, it records the manner in which the pattern of sensory activation during my perceptual experience occurred in order to re-enact it at retrieval (De Brigard, 2017). Thus, when presented with a cue—in this case, the utterance of my name—the brain gets into retrieval mode, which apparently is subserved by the fronto-polar cortex (Rugg & Wilding, 2000). Using every piece of sensory data as a potential cue for retrieval (e.g., the woman’s voice, her physique, etc.) my brain tries to get the hippocampal index to reactivate a perceptual pattern. Yet it is only upon the pronunciation of the right

204  Felipe De Brigard cue—in the example, a contextual-semantic piece of information—that ecphory is achieved, and the right index gets to reactivate, more or less, the pattern of neural activity in which it was when I first perceived the woman at my party. Incidentally, the fact that every time a memory trace is reactivated it occurs in a different neuronal and experiential context (e.g., the mental state one is in at the time of recollecting) means that each reactivation of a memory is also an instance of reconsolidation (Moscovitch et al., 2005), which helps to explain why retrieval makes memories vulnerable to distortion (Hardt, Einarsson, & Nader, 2010). In sum, the content of my memory is the result of a complex process of sensory reactivation in which subpersonal level representations are bound together to reconstruct the perceived content during retrieval. Now, how did I become aware of this content? More specifically, how is it that the retrieved content presents to me as being about this woman’s hat at my party? My suggestion is that it becomes conscious when I covertly focus my attention on the region of my retrieved representation depicting the woman’s hat. Only then was I able to (subpersonally) match the retrieved content with my present perception, and only then was I able to recognize her as the woman I am talking to right now. Additionally, it was only when the attended content of my representation became the focus of my conscious experience that I was able to say that I remember her wearing a hat at my party. In other words, it was by way of mentally delineating a particular region of my intentional content that this aspect of the scene was mentally highlighted to me, and it was this highlighting that made it available to my conscious reporting. This process is basically the memorial equivalent of what Campbell (2002) calls the Causal Hypothesis for visual perception: “When, on the basis of vision, you answer the question, ‘Is that thing F?’ what causes the selection of the relevant information to control your verbal response is your conscious attention to the thing referred to” (p. 13). My claim is that the same mechanisms by means of which you consciously attend to a region of space are responsible for the experiential highlighting in a memory experience. I call this experiential highlighting “mental ostension” (tantamount to what Prinz (2007) calls “mental pointing”). To mentally ostend an aspect of an intentional content is to focus one’s attention inwardly toward such aspect. And mental ostension is the mechanism by means of which the mental content—or the region of the mental content—we attend to becomes available to consciousness (De Brigard & Prinz, 2010). This hypothesis finds strong support in results coming from cognitive psychology and neuroscience. In the early years of cognitive science, it was thought that attention was necessary for encoding but not for retrieval. Specifically, it was argued both on the basis of attentional deficits (Critchley, 1953) and experimental manipulations (Baddeley et al., 1984) that impaired attention negatively impacts recollection only at encoding, not at retrieval. However, Fernandes and Moscovitch (2000) showed that this conclusion is not warranted when the secondary task is material-congruent.

Memory, Attention, and Joint Reminiscing  205 For example, if the retrieval task is verbal, a word-based secondary task would be more detrimental to successful recollection than a digit-based or a picture-based task (Fernandes et al., 2005). More recent studies have further demonstrated that under deep versus shallow encoding conditions (Hicks & Marsh, 2000), as well as strategic versus nonstrategic encoding (Lozito & Mulligan, 2006), divided attention tasks are detrimental during retrieval and can negatively affect not only hit rates but also meta-memory judgments (Skinner, Fernandes, & Grady, 2009). The neuropsychological evidence has also been challenged. Although damage in posterior parietal cortex usually causes attentional deficits (e.g., hemispatial neglect), it is typically thought that retrieval processes remain intact. But a study of two patients with damage in ventral parietal cortex showed severely diminished free-recall of autobiographical memories relative to controls (Berryhill et al., 2007). Indeed, when considered as a recall test, the classic study of Bisiach and Luzzatti (1978) on hemineglect further supports the claim that attention may be required at retrieval. In this study, Bisiach and Luzzatti (1978) asked a patient with severe hemispatial neglect to remember the main square in Milan, the city he lived in all his life. Although his language capacities where impeccable, his report omitted all the buildings to the left of the square when he remembered it facing one direction. Then he was asked to imagine crossing the square and turning back so that now he’d be facing the opposite side. Again, he failed to report the left-hand buildings—even though those were the buildings he had just reported! A final piece of evidence supporting this hypothesis comes from neuroimaging. The involvement of attentional areas of the parietal cortex during recollection is a frequent finding in PET and fMRI studies of episodic retrieval (Rugg & Henson, 2002). As a result, some theorists suggest that the parietal cortex may be playing a similar role during recollection as it does during perception. According to their “Attention to Memory” (AtoM) hypothesis (Cabeza et al., 2008), for instance, the dorsal parietal cortex, which is usually associated with top-down attention, is involved in voluntary, goal-directed retrieval, whereas the ventral parietal cortex, which is usually involved in bottom-up attention, appears to be associated with involuntary recollection. A related hypothesis suggests that the parietal cortex may play a role analogous to the working-memory buffer, gating stored information for decision-making and action (Wagner et al., 2005). Finally, the Cortical Binding of Relational Activity (CoBRA) suggests that the parietal cortex may selectively modulate the reactivation of disaggregated sensory components during retrieval in order to bind them together—reconstruct them—as a unified recollective experience (Shimamura, 2011). Despite the subtle differences among these views, what matters is that there is general agreement in that retrieval related activity in partietal cortex signals its critical role in the selection (either voluntary or involuntary), modulation (either top-down or bottom-up), and maintenance of internally generated information. And if we accept a functional description of attention as the

206  Felipe De Brigard filtering, selection, and modulation of information (De Brigard & Prinz, 2010), then the filtering, selection, and modulation of internally generated information (Chun, Golomb and Turk-Browne, 2011), of which memorial contents are a sub-class, is tantamount to accepting that internal attention is needed for retrieving memorial contents (De Brigard, 2012). So far, the evidence reviewed supports the claim that internal attention is needed for retrieving the contents of our memories, but is it needed to render such contents conscious? The question as to what are the precise mechanisms that allow us to become aware of certain mental contents is hotly debated. Nevertheless, attention plays a pivotal role in the most prominent neural theories of consciousness, including the influential Global Neuronal Workspace (GNW) hypothesis (Dehaene & Changeux, 2000; 2011). Roughly, the GNW postulates two computational spaces in the brain: local, informationally encapsulated, and specialized processing networks, and a distributed, nonencapsulated associative GNW. Contents that are processed by local networks can become conscious when they are broadcasted onto the GNW. Critically, attention is suggested as the mechanisms that allows this informational amplification to take place. Do we have any evidence to the effect that internal attention at retrieval modulates the informational amplification needed for broadcasting contents to the GNW? I think we do. Consider neurophysiological evidence. We know that attention acts upon local networks by modulating their synchronized firing (Steinmetz et al., 2000). These neuronal changes are correlated with increases in the gamma frequency, which predicts not only successful encoding (Sederberg et al., 2003; Paller, Voss, & Westerberg, 2009) but also retrieval of old, and correct rejection of new, items (Gruber et al., 2004; Osipova et al., 2006; Jensen, Kaiser, & Lachaux, 2007). Moreover, in a study involving intracranial electroencephalographic recordings, Sederberg and colleagues (2007) discovered that the same pattern of gamma-frequency activity that predicts successful encoding reappears at retrieval. This oscillatory activity emerges in the prefrontal cortex and the hippocampus, and then spreads onto the sensory cortex (Osipova et al., 2006), following the same pattern of activity observed in the prefrontal/medial—temporal/parietal cortex network underlying conscious retrieval. Neuropsychological results give further support to the claim that internal attention gates memories into consciousness. If, as hypothesized, parietal regions modulate the availability of local sensory representations to the GNW, one should expect a diminished sense of “re-experiencing” in patients whose parietal lesions hinder such broadcasting. This is precisely what Berryhill et al. (2007) report: patients with bilateral parietal lesions report fewer episodic details and lower levels of vividness during freerecall of autobiographical memories, suggesting that a reduced number of sensory contents were actually made available to their conscious experience. Relatedly, Davidson et al. (2008) reported that patients with parietal lesions showed a reduced number of “remember” responses, which are associated

Memory, Attention, and Joint Reminiscing  207 with increased subjective experience of recollection, relative to both “know” responses and controls (Drowos, Berryhill, & Olson, 2010). Finally, Simons et al. (2010) found that patients with bilateral parietal damage showed lower confidence levels for source recollection tasks, a result they interpret as suggesting that parietal lobe lesions impair subjective experience of episodic recollection. The view that internal attention is required for conscious recollection is consistent with their interpretation. Let’s take stock. In this section I argued that retrieving episodic information involves reconstructing subpersonal level representations by binding reactivated sensory information in a process that involves interactions among the parietal, the medial-temporal, and the pre-frontal cortices (De Brigard, 2014; 2017). Next, I suggested that the content that becomes available for consciousness is that toward which we direct our internal attention. Finally, I suggested that attended contents become conscious when they are rendered available to the GNW, which in turn poise them for use in working memory, and thus linguistic production. When the resultant verbal production is vocal it constitutes an utterance that—modulo communicative intentions and good faith—aims at reporting the mental state of which it is an effect. If this account of what happens when I utter (1) in a situation like the one exemplified previously is roughly correct, then we have the components that we need to clarify the first part of the three-part answer to the question of joint reminiscing with which I started this chapter: to mentally ostend an aspect of a mental content is to internally attend such a content. We need now to see how we can go from mental ostension to reference and, then, to full-fledge joint reminiscing.

3  Remembering as Deferred Mental Ostension If the previous account is on the right track, remembering a past object involves, in the first instance, our capacity to mentally ostend toward an experiential content, and it is by way of mentally ostending toward such content that we can make it available for conscious reporting. As it stands, however, this view poses a difficult question. In the case of perception, the object that is mentally ostended—or ‘experientially highlighted,’ in Campbell’s verbiage (2002)—is in direct contact with the perceiver. Indeed, in the relational (realist) view that Campbell puts forth, the object becomes a constituent of the experiential content. Thus, for Campbell, there is no need to separate the intentional content and its object when it comes to making them available for consciousness. However, in the case of memory, the object of one’s recollection isn’t in direct contact with the rememberer. In fact, the object of one’s memory not only is not present when we remember it; it usually no longer exists. How can we be aware of an object or event with which we are no longer in direct contact? One possibility is to go the direct realist route (Reid, 1785/1849). According to this view, remembering is tantamount to direct perception in

208  Felipe De Brigard that the intentional objects are directly apprehended. Intentional contents, particularly representational contents, are thus disposed of. For direct realism has it that remembering is just like perceiving, except that its objects—i.e., that which is remembered—do not exist in the present: they exist in the past. Although relatively popular in the past (Laird, 1920), direct realism for memory fell in disrepute. It faces, after all, difficult obstacles. For one, direct realism suggests an analogy between memory and perception, but it does not specify the extent to which they are similar, or how to accommodate their obvious differences. Memories, for instance, are phenomenologically different from perceptions, and they are usually coarser. Memory and perception also differ in the capacity to provide us with discriminatory information. For instance, while we can visually discriminate similar shades of red when perceived simultaneously, we can’t when relying solely on memory (Nemes, Parry, & McKeefry, 2010). Also, memories decay and are often blurry and lifeless. Nonsalient events tend to be more easily forgotten than salient ones, even if the salient ones occurred much before in time. It is hard to see what the equivalent of this kind of saliency effect would be for perception. Finally, there is the problem of false memories. Many of our veridical memories are actually the result of the same mechanisms that give us non-veridical memories (Schacter, 1995, De Brigard, 2014). But nonveridical memories are about events that never occurred. Thus, the direct realist would have to explain not only how can memory be in direct contact with an event that no longer exists—or that exists in the past—but also with events that never existed. Some metaphysical maneuvering could potentially solve these issues, but I doubt we want to pay such price when the alternative is to accept the existence of representational contents (Furlong, 1948). Aware of these problems, Campbell suggests a different non-representational alternative, based upon McCormack and Hoerl’s notion of temporal decentring: “The ability to temporally decenter is the ability to consider alternative temporal perspectives on events and to understand the relationship of these perspectives to one’s current perspective” (McCormack & Hoerl, 1999; Evans, 1982). Accordingly, Campbell suggests that our capacity to refer to remembered objects or events depends upon our capacity for temporally decentering. It is only when we acquire the capacity to temporally decenter that we can grasp the truth-conditions of judgments tensed at times different from when they are uttered. Accordingly, in order to understand the sentence: (2) I see that you are wearing a hat at my party when uttered in the presence of the object of attention (i.e., the person wearing a hat at my party), we only need to be able to grasp the truth-conditions of the judgment as it applies to the current situation. But in order to understand (1) we need to be able to move away from the current temporal situation, and grasp the truth-conditions of the judgment as if it had been made

Memory, Attention, and Joint Reminiscing  209 at a different time, namely the relevant moment in the past. Therefore, there is no need for direct contact with the past object, nor a reference to any intermediary mental representations. All that is required is the acquisition of a particular skill—i.e., temporal decentering—so that we can refer to the object of our conscious recollection as if we have been talking about it at a different time (Campbell, 2002: 181) Although I am not completely unsympathetic to this view, I find it unsatisfactory for two reasons. First, according to McCormack and Hoerl (1999), the development of episodic memory depends upon our acquisition of temporal decentering, which in turn depends upon the acquisition of the concept of personal/perspectival time. Although this hypothesis seems to fit some data in the developmental literature, it has a hard time accommodating data coming from neuropsychology. After all, individuals with amnesia are perfectly capable of using personal/perspectival concepts, thus they are perfectly capable of temporal decentering, even though their episodic memory is damaged (Craver et al., 2014; De Brigard & Gessell, 2016). This suggests that temporal decentering is independent of episodic memory. Second, defining something as a cognitive skill or capacity does not preclude it from requiring representations, whether conscious or unconscious. Motor skills, for instance, may require representations in the form of programs to be executed or motor sequences (Pavese, 2015). The good news is that we can keep the intuition that mental ostension is the mechanism by means of which we can refer to the object of our memories without having to accept temporal decentering. I suggest that what allows us to refer to past objects when we are consciously attending to a particular mental content that presents itself as being about a previous experience, is the covert equivalent of our overt capacity to demonstrate deferredly. Notice that the root of the problem we are facing is that what we mentally point at when we remember is not identical to what we refer to. Consider (1) again. When I uttered (1) I am not talking about my mental experience but about the past event involving this woman’s hat at my party. It is an event that no longer exists. But what I am inwardly attending to—what I am mentally ostending—is a region of the intentional content I am being aware of, right now, as I am having the mental experience of remembering the woman’s hat. Schematically, if ‘p’ stands for the intentional object of my memory, and ‘r’ stands for the intentional content of my memory, according to my proposed account, when I remember that p I am talking about p while ostending r. Linguistically, the phenomenon of ostending at a certain thing ‘r’ in order to refer to a different thing ‘p’ is known as deferred ostension (Quine, 1968: 194). Consider the classical example due to Evans (1981: 199). We are walking down the street and I point toward a parked car covered with parking tickets. Pointing at it I say, “That man is going to be sorry.” The intuition here is that even though I am pointing at the car—that is, even though my demonstration (Kaplan, 1989) is directed toward the car—the object demonstrated or referred to is not the car, but the owner of the car.

210  Felipe De Brigard Or consider the situation in which I point to a set of footprints and say, “He must be giant!” or the case in which I am holding a copy of The Confederacy of Dunces and say, “He’s my favorite author” (Borg, 2002). These too are cases in which I am pointing at something (e.g., footprints, a book) while referring to something else (e.g., whatever animal left the footprints, John Kennedy Toole). My suggestion is that the same sort of phenomenon occurs when we remember episodic memories. To understand what the objects of our memories are, and consequently to be able to talk about the objects of our memories, we first learn how to mentally refer to something that is not perceptually present in one’s environment but that nonetheless is present in our conscious experience. Developmental psychologists have debated for decades whether preverbal children have episodic memory.1 Nonetheless, probably all developmental psychologists agree that deferred imitation of action sequences does in fact demonstrate the emergence of episodic memory (Barr, RoveeCollier, & Campanella, 2005). Moreover, older adults with medial temporal lobe damage as well as individuals with developmental amnesia have trouble with this task, further suggesting its intimate relation with episodic memory (Adlam et al., 2005). In this paradigm, infants are shown a relatively unusual sequence of actions with a particular object. For instance, the experimenter may show the child that in order to get the key out of the box she needs to first hit the box three times with the tip of the magic wand and then once with the bottom. Then the child is either left alone (and recorded) or the experimenter leaves for a few minutes and comes back with the wand, asking the infant whether she can get the key out of the box. Prior to 6 months of age, infants are completely incapable of reproducing previously learned action sequences. There is some evidence that they can perform deferred imitations of brief sequences after 6 or 7 months of age, as long as the retention interval—i.e., the elapsed time between study and test—is kept short (Barr & Hayne, 1996). Gradually, children learn how to perform action sequences that are increasingly more complex, that have longer retention intervals, and that are retrieved with less specific cues (Hayne, Boniface, & Barr, 2000). By the second year of age, deferred action sequences are pretty much established. Notice that, prior to 6 months of age, infants are capable of pointing. If one shows a 4-month-old the magic wand, she can point at it. Nonetheless, she does not see it as related to anything that happened before. It is just another object in the visual field, however interesting it may be. After 6 or 7 months of age, though, the infant appears to be able to see the magic wand as something more than a mere present object. She sees the magic wand as related to a previous event. The mental content elicited by the perception of the wand is now experienced as being about something other than the want. The wand becomes a cue. Now, the experimenter is able to ostend at the wand, while the infant perceives it, and ask for the right sequence of actions: “Can you get the key out of the box?” The fact that the infant can indeed come up with the right sequence of actions strongly suggest that she

Memory, Attention, and Joint Reminiscing  211 knows that one can talk about a previous ‘p’—a sequence of actions—while pointing at a present ‘r’—the magic wand. As time goes by, the perceptual cues can become less and less concrete, that is, less and less similar to the perception of the original event. Eventually, a pretend wand can elicit the memory, then just the waging of a finger, the uttering of a word. Suddenly, the demonstration of the cue, and perhaps the cue itself, becomes irrelevant. All that matters is that the cue can elicit the retrieval of the right sort of mental content and that it can experientially highlight the relevant property of the resultant conscious experience. Neither my hearing the woman’s voice nor my seeing her face succeeded in triggering the right memorial content. It was only when she gave me the contextual information that ecphory occurred, and the right intentional content was then retrieved.2 Now, the sensory information I have been presented with mentally highlights certain aspect of that content, which is experienced by me as a reinstatement of the perceptual event of seeing this woman dancing and wearing a hat. Mental ostension is, thus, an acquired skill, and deferred mental ostension is a way we learn to use mental pointing to refer to something else—usually that which caused the retrieved intentional content to begin with.3 My suggestion, therefore, boils down to no more than this: we can talk about the intentional objects of our memories because we can refer deferredly to them by mentally ostending toward the intentional contents we experience when retrieved by the right cue. Remembering a past object is a case of deferred mental ostension.4

4 Joint Reminiscing as Concerted Deferred Mental Ostension In the second section, I argued that memorial contents are reconstructed out of subpersonal level representations via a process of pattern-completion that reactivates, more or less,5 the sensory cortices that were engaged during the perception of the remembered event. Then I argued that we become conscious of these mental contents when we direct our attention inwardly to them. I called that process mental ostension. I suggested that the memorial contents we mentally ostend at are thereby reportable, as they have been poised for verbal control in working memory. Then, in the third section, I claimed that mental pointing was not enough to explain how we get to talk about objects that aren’t in the surroundings of the rememberer. I argued that the capacity to talk about something not present while ostending something present was required for the rememberer to be able to talk about remembered objects. Following the convention in linguistics, I called that capacity deferred mental ostension. Now, in this last section, I suggest that to acquire the capacity to jointly reminisce we need to learn how to orient our attention inwardly alongside other coreminiscers in order to mentally ostend memorial representations with the same contents, which in turn allows us to speak about the objects that those contents represent.

212  Felipe De Brigard Consider, once again, the situation in which I utter (1). Imagine that, after my brief encounter at the supermarket, I ran into a friend whom I know was at my party, and the following dialogue takes place: “I just ran into the woman who spilled wine in my carpet at the party.” “Which one?” my friend asks. “I don’t remember her name,” I reply. “You mean the woman who was wearing a hat?” “That one!” I say. What just happened? Think of what occurred during this brief exchange. I ran into my friend and by mentioning my recent encounter at the supermarket, I shift his cognitive mood toward that of reminiscing. When he asks, “Which one?” I assume he’s trying to single out a particular individual from his own memory experience of the party. In other words: my opening sentence serves as a verbal trigger for his own memory trace of the party. Now he’s surveying, via his own topdown attentional mechanisms, his own intentional content. But, of course, he does not know whom am I talking about yet. There were many women at my party. He thus asks for a distinctive feature that may help him single her out: her name. Since I don’t know her name he tries a new one: the hat. He is now mentally pointing toward the region of his intentional content depicting the woman in a hat, so he asks for confirmation. My saying “That one!” confirms that we are talking about the same remembered object and signals the moment we realize we are jointly reminiscing. Let me stress this point. Most discussions of perceptual joint attention appear to make the act of pointing—what Kaplan called “demonstration” (Kaplan, 1989)—essential for the process to effectively take place. But one can engage in joint attention without any demonstration overtly taking place by any of the attendees. An object can demonstrate itself, as it were, by making itself salient in one’s experiential field. Suppose you are watching a soccer game when, all of a sudden, an enthusiastic fan runs from one side of the court to the other wearing no clothes. The event did not disturb the development of the game, but it was enough to grab the attention of many people in the audience, including yours and your friend’s. “Do you see that?” your friend asks. There is no need for him to overtly point toward the enthusiastic fan. Your attention, just like your friend’s, has been disengaged from what it was focused on before—the player with the ball, presumably—and it has moved onto a new target: the zealous fan. The demonstrated object is its own demonstration. Likewise, one can guide someone else’s attention toward a particular target without having to use overt pointing. One can help the other person navigate the perceptual field using intermediate salient targets as reference points. Suppose you fail to notice the naked fanatic because it failed to disengage your attention from the soccer ball. Thus, when your friend asks you whether you’ve seen that, you rightly ask, “what?” Given the distance between the naked fan and your seats, pointing is useless. And given the fact that he’s holding a hot dog with one hand, and a beer with the other, hand-waving is out of the question. So, he finds a landmark, a salient reference point, and orients your attention from there. “See the side referee? Draw an imaginary line from

Memory, Attention, and Joint Reminiscing  213 him to the goal, and you’ll see what I’m talking about.” Your attention has been reoriented, and now you are both jointly attending at the same target. I believe that an equivalent process goes on in the case of joint reminiscing. I can expect my friend, whom I know was at the party, to have encoded much of the same information I encoded then. The information we both encoded isn’t identical, of course. Even if we were looking at the woman from the same side of the room at the exact same time, we both occupy different spatial locations, so our perspectives are going to differ. But these differences need not matter. Memorial contents represent their objects with varying degrees of correctness, and just as there may be differences within a subject between the way an event was originally perceived and the way it presents itself during recollection, there may be also subtle differences between subjects that still allow us to talk about the same memorial content being entertained by two or more joint reminiscers. Just as in the case of the fan in the soccer game, I can guide my friend’s attention to highlight a particular aspect of his intentional content—the woman with a hat—so it becomes the target of his mental ostension. I can use—as in my imaginary example—a reference to a salient feature of the object itself: the hat. But I could have also oriented my friend’s inward attention using other reference points: e.g., she was dancing by the window. The process can also go the other way around. Upon remembering this woman, my friend may be able to reorient my attention toward a different aspect I did not remember at the time—her perfume, say, or the fact that she had brought a delicious bottle of wine. The capacity to mutually coordinate each other’s attention so as to consciously highlight (approximately) the same mental contents I call “concerting.” Consequently, our capacity to talk about the very objects represented by the intentional contents we are conscious of during joint reminiscing would be concerted mental deferred ostension. Memory allows a temporal dimension of concerted mental deferred ostension that isn’t present in perception: we can direct each other’s attention along a temporal line. In other words, we can mutually direct each other attention toward memorial contents depicting events that occurred before or after a certain target event. For instance, when jointly reminiscing about the woman at the party, my friend can reorient my attention toward the beginning of the party, and mentally highlight to me the fact that she brought a bottle of wine. In fact, he could even guide my attention backwards in time, reminding me that the party was not the first time I met that woman. “Remember, about a month ago, we had that picnic at the park?” This kind of concerted mental ostension along temporal dimensions is unique to memorial contents, and empirical evidence strongly suggests that concerted mental ostension plays a fundamental role in children’s learning how to talk about their memories. Consider, first, some linguistic data. As Clark (1978) observed, demonstrative terms are usually among the first ten words uttered by English-speaking children, and always among the first 50. Moreover, there is a positive linear

214  Felipe De Brigard correlation between the transition from dyadic to tryadic attention and the use of demonstratives (Iverson and Goldin-Meadow, 2005), which has led some theorists to suggest that a critical linguistic function of demonstratives is to help to coordinate the focus of attention among interlocutors (Diessel, 2006). Consistently, results from developmental psychology suggest that events and/or objects that were jointly attended by children and caregivers during encoding had higher retrieval rates than those that were individually attended or not attended at all (Haden et al., 2001). Moreover, the effect was additive if the event/object was both jointly attended and jointly talked about. Experiments employing collaborative retrieval, which occurs when the caregiver guides the retrieval of the child with the use of wh-questions, shows improvements in several memory measures, including number of episodic details, narrative coherence, and subsequent recall, among others. To illustrate, consider an excerpt from an example with a 2-year-old (Fivush et al., 1994): Mother: Remember when Mommy and Daddy and Sam went in the car for a long time and went to Grandma’s house? Child: (Shakes head yes) Mother: And what did we see when we were in the car? Remember Daddy showing you outside the car? What was it? Child: I don’t know. Mother: Do you remember we saw some mountains and we went to that old house and what did we do? We took off our shoes and walked on the rocks. What else did we do? Who was there? This is an example of collaborative retrieval with plenty of scaffolding, that is, plenty of retrieval support offered by the mother to the child during retrieval. Now compare this example with another one, reported by Hoerl and McCormack (2004), that involves much less scaffolding during a joint reminiscing session with a 3-year-old: Mother: Child: Mother: Child: Mother: Child:

What happened to your finger. I pinched it. You pinched it. Oh boy, I bet that made you feel really sad. Yeah . . . it hurt. Yeah, it did hurt. A pinched finger is no fun . . . But who came and made you feel better? Daddy!

These narratives, Hoerl and McCormack observe, exploit causal links between experienced events in order to guide the children backward or forward in time. And this, I believe, is a clear example of concerted mental deferred ostension. The mother starts off highlighting a particular mental content for the child and invites her to explore certain aspects of that

Memory, Attention, and Joint Reminiscing  215 content like the emotion she felt when it happened. Then there is a temporal exploration via focusing her attention in a particular causal link: the transition from being in pain to getting better. Contents that are jointly recovered become reference points from when one can call attention to other aspects of the retrieved intentional content, as in the case of the striker in the soccer game mentioned previously. Mother and child are, thus, jointly reminiscing an event that occurred later in time via consciously attending to a different memorial content representing the effect of the event depicted by the previously attended memorial content. Therefore, when jointly reminiscing, attended contents can become not only spatial but also temporal reference points.

5 Conclusion To conclude, let me recap the main points of this chapter. First, I suggested that remembering an object involves inward attention toward an aspect of a retrieved memorial content. Following the terminology introduced by Campell (2002), I suggested that in doing so a particular mental content is highlighted or ostended. In turn, I argued that the attentional mechanisms behind mental ostension make the intentional content available for verbal reporting.6 In addition, I argued that for the speaker to successfully refer to the object of her memory, the capacity to mentally point to a present conscious content while referring to a nonpresent intentional object is required. In analogy with the linguistic phenomenon, I called this capacity deferred mental ostension. Finally, I claimed that for two or more people to engage in joint reminiscing, and thus to be able to successfully refer to the same past object, they are required to mutually coordinate their attention toward relevantly similar regions of their memorial contents. I called this processed concerted mental ostension. Only when there is concerted mental ostension it is possible for two or more remembers to refer to the same past object. Joint reminiscing is, therefore, concerted deferred mental ostension.7

Notes 1 Some developmental psychologists suggest that visual pair comparison tasks, whereby babies are presented with novel versus familiar objects and their kicking rates are measured, are good indications of the origins of episodic memory. However, many others disagree, as it is always possible to interpret this paradigm as tapping at implicit rather than explicit memory. 2 Presumably, a working hippocampus is required for ecphory to take place. Absent the right sort of hippocampal index, the process of pattern completion required for sensory reactivation is hindered, so no mental content upon which to turn one’s inward attention is retrieved. This would explain why individuals with medial temporal lobe damage fail at the deferred action sequences paradigm and, incidentally, partly explains why they fail to retrieve unconsolidated memories. 3 So far, I have only talked about mental deferred ostension, in analogy with linguistic deferred ostension. I wonder, however, if what I have said here may have

216  Felipe De Brigard also some application to the linguistic phenomenon as well. In a comprehensive study on deferred ostension, Emma Borg (2002) shows that the strategy of treating deferred uses of demonstratives as a different semantic kind of indexicals, is wrong-headed. She presents persuasive arguments to the effect that the differences ought to be accommodated at the pragmatic level. Indeed, she suggest that the same pragmatic rule that works for perceptual uses of demonstratives also works for deferred uses, as long as the child learns that there is more than one way to demonstrate an object. Her proposal, then, “is simply that there are lots of ways to draw an object to attention to facilitate the use of a referring expression, and pointing directly to the object is just one way amongst others—other ways which include pointing at a related object” (Borg, 2002, p. 509). The development of episodic memory may provide the psychological basis for one of these forms, and it is likely that other forms are similarly developed (see, for instance, Hoerl & McCormack, 2004, where learning to refer to distal causes via pointing at current perceptual events is explored). 4 My claim here is one of necessity but not sufficiency. Deferred mental ostension is necessary for memorial reference but likely not sufficient. It is possible that, in addition, we may need to believe that the retrieved mentally ostended content represents an object that is no longer present. (Thanks to Jordi Fernandez for this suggestion.) 5 I say “more or less” because neurological evidence suggests that even though there is reactivation at the systems level, there are changes that occur at the local neural level. The precise relationship between the sensory reactivation at the systems level and the neural changes at the local level is unknown. 6 In my view, uttering a sentence such as (1) to express one’s intentional content at the time of recollection is tantamount to describing one’s content of experience. As such, the utterance used to express the intentional content of a memory experience has to be understood as a description of that content, and it need not reflect the structure of the content at all (Crane, 2009). 7 Previous versions of this paper were presented at the Department of Philosophy at the University of Utah, the V Colombian Congress of Philosophy in Medellín, and the Department of Philosophy at the Universidad Autónoma Metropolitana, in Mexico City. Many thanks to all of those audiences. Thanks also to Paul Henne, Jordi Fernandez, and an anonymous reviewer for their comments.

References Adlam, A., Vargha-Khadem, F., Mishkin, M., & de Haan, M. (2005). Deferred imitation of action sequences in developmental amnesia. Journal of Cognitive Neuroscience, 17(2), 240–248. Baddeley, A. D., Lewis, V., Eldridge, M., & Thomson, N. (1984). Attention and retrieval from long-term memory. Journal of Experimental Psychology: General, 13, 518–540. Barr, R., & Hayne, H. (1996). The effect of event structure on imitation in infancy: Practice makes perfect? Infant Behavior and Development, 19, 253–257. Barr, R., Rovee-Collier, C. K., & Campanella, J. (2005). Retrieval protracts deferred imitation by 6-month-olds. Infancy, 7, 263–284. Berryhill, M. E., Phuong, L., Picasso, L., Cabeza, R., & Olson, I. R. (2007). Parietal lobe and episodic memory: Bilateral damage causes impaired free recall of autobiographical memory. Journal of Neuroscience, 27, 14415–14423. Bisiach, E., & Luzzatti, C. (1978). Unilateral neglect of representational space. Cortex, 14, 129–133.

Memory, Attention, and Joint Reminiscing  217 Borg, E. (2002). Pointing at Jack, Talking about Jill: Understanding deferred uses of demonstrative pronouns. Mind and Language, 17(5), 489–512. Buckner, R. L., & Wheeler, M. E. (2001). The cognitive neuroscience of remembering. Nature Reviews Neuroscience, 2, 624–634. Cabeza, R., Ciaramelli, E., Olson, I. R., & Moscovitch, M. (2008). The parietal cortex and episodic memory: An attentional account. Nature Reviews Neuroscience, 9, 613–625. Campbell, J. (2002). Consciousness and reference. Oxford: Oxford University Press. Chun, M., Golomb, J., & Turk- Browne, N. B. (2011). A taxonomy of external and internal attention. Annual Review of Psychology, 62, 73–101. Clark, E.V. (1978). From gesture to word: On the natural history of deixis in language acquisition. In J. S. Bruner & A. Garton (Eds.), Human growth and development (pp. 85–120). Oxford: Oxford University Press. Crane, T. (2009). Is perception a propositional attitude? Philosophical Quarterly, 59(236), 452–469. Craver, C., Kwan, D., Steindam, C., & Rosenbaum, R. S. (2014). Individuals with episodic amnesia are not stuck in time. Neuropsychologia, 57, 191–195. Critchley, M. (1953). The parietal lobes. London: Edward Arnold. Davidson, P. S. R., Anaki, D., Ciaramelli, E., Cohn, M., Kim, A., Murphy, K. J., Troyer, A. K., Moscovitch, M., & Levine, B. (2008). Does lateral parietal cortex support episodic memory? Evidence from focal lesion patients. Neuropsychologia, 46, 1743–1755. De Brigard, F. (2012). The role of attention in conscious recollection. Frontiers in Psychology, 3, 29. De Brigard, F. (2014). Is memory for remembering? Recollection as a form episodic hypothetical thinking. Synthese, 191(2), 155–185. De Brigard, F. (2017). Memory and the intentional stance. In B. Huebner. (Ed.), The philosophy of Daniel Dennett. Oxford & New York: Oxford University Press. De Brigard, F., & Gessell, B. S.* (2016). Time is not of the essence: Understanding the neural correlates of mental time travel. In S. B. Klein, K. Michaelian & K. K. Szpunar (Eds.), Seeing the future: Theoretical perspectives on future-oriented mental time travel (pp. 153–180). Oxford & New York: Oxford University Press. De Brigard, F., & Prinz, J. (2010). Attention and consciousness. Wires Interdisciplinary Reviews, 1(1), 51–59. Dehaene, S., & Changeux, J. (2000). Reward-dependent learning in neuronal networks for planning and decision making. Progress in Brain Research, 126, 217–229. Dehaene, S., & Changeux, J. (2011). Experimental and theoretical approaches to conscious processing. Neuron, 70, 200–227. Diessel, H. (2006). Demonstratives, joint attention, and the emergence of grammar. Cognitive Linguistics, 17, 463–489. Drowos, D. B., Berryhill, M. E., Andre, J., & Olson, I. R. (2010). True memory, false memory, and subjective memory after parietal lobe damage. Neuropsychology, 24, 465–475. Evans, G. (1981). The varieties of reference. Oxford: Oxford University Press. Fernandes, M. A., & Moscovitch, M. (2000). Divided attention and memory: Evidence of substantial interference effects at retrieval and encoding. Journal of Experimental Psychology: General, 129, 155–176.

218  Felipe De Brigard Fernandes, M. A., Moscovitch, M., Ziegler, M., & Grady, C. (2005). Brain regions associated with successful and unsuccessful retrieval of verbal episodic memory under divided attention. Neuropsychologia, 43, 1115–1127. Fivush, R. (1994). Constructing narrative, emotion and gender in parent-child conversations about the past. In U. Neisser, & R. Fivush (Eds.) The Remembering Self: Construction and accuracy of the life narrative (pp. 136–157). New York: Cambridge University Press. Furlong, E. J. (1948). Memory. Mind, 57, 16–44. Gruber, T., Tsivilis, D., Montaldi, D., & Müller, M. M. (2004). Induced gamma band responses: An early marker of memory encoding and retrieval. Neuroreport, 15, 1837–1841. Haden, C. A., Ornstein, P. A., Eckerman, C. O., & Didow, S. M. (2001). Motherchild conversational interactions as events unfold: Linkages to subsequent remembering. Child Development, 72, 1016–1031. Hardt, O., Einarsson, E. Ö., & Nader, K. (2010). A Bridge over troubled water: Reconsolidation as a link between cognitive and neurotraditions. Annual Review of Psychology, 61, 141–167 Hayne, H., Boniface, J., & Barr, R. (2000). The development of declarative memory in human infants: Age-related changes in deferred imitation. Behavioral Neuroscience, 114, 77–83. Hicks, J. L., & Marsh, R. L. (2000). Toward specifying the attentional demands of recognition memory. Journal of Experimental Psychology: Learning, Memory and Cognition, 26, 1483–1498. Hoerl, C., & McCormack, T. (2004). Joint reminiscing as joint attention to the past. In N. Eilan, C. Hoerl, T. McCormack, & J. Roessler (Eds.), Joint attention: communication and other minds. Oxford: Oxford University Press. Iverson, J. M., & Goldin-Meadow, S. (2005). Gesture paves the way for language development Psychological Science, 16, 367–371. Jensen, O., Kaiser, J., & Lachaux, J. (2007). Human gamma-frequency oscillations associated with attention and memory. Trends in Cognitive Sciences (Regul. Ed.), 30, 317–324. Kaplan, D. (1989). Demonstratives. In Themes from Kaplan. Oxford: Oxford University Press. Laird, J. (1920). A study in realism. Cambridge: Cambridge University Press. Lozito, J. P., & Mulligan, N. W. (2006). Exploring the role of attention during memory retrieval: Effects of semantic encoding and divided attention. Memory & Cognition, 34, 986–998. McClelland, J. L., McNaughton, B. L., & O’Reilly, R. C. (1995). Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory. Psychological Review, 102, 419–457. McCormack, T., & Hoerl, C. (1999). Memory and temporal perspective: The role of temporal frameworks in memory development. Developmental Review, 19, 154–182. Moore, C., & Dunham, P. J. (1995). Joint attention: Its origins and role in development. Hillsdale, NJ: Lawrence Erlbaum Associates. Moscovitch, M., Rosenbaum, R. S., Gilboa, A., Addis, D. R., Westmacott, R., Grady, C., McAndrews, M. P., Levine, B., Black, S. E., Winocur, G., & Nadel, L. (2005). Functional neuroanatomy of remote episodic, semantic and spatial

Memory, Attention, and Joint Reminiscing  219 memory: A unified account based on multiple trace theory. Journal of Anatomy, 207, 35–66. Nemes, V. A., Parry, N. R. A., McKeefry, D. J. (2010). A behavioural investigation of human visual short-term memory for colour. Ophthalmic & Physiological Optics, 30, 594–601. Osipova, D., Takashima, A., Oostenveld, R., Fernandez, G., Maris, E., & Jensen, O. (2006). Theta and gamma oscillations predict encoding and retrieval of declarative memory. Journal of Neuroscience, 26, 7523–7531. Paller, K. A., Voss, J. L., & Westerberg, C. E. (2009). Investigating the awareness of remembering. Perspectives on Psychological Science, 4, 185–199. Paller, K., & Voss, J. (2004). Memory reactivation and consolidation during sleep. Learning & Memory, 11, 664–670. Pavese, C. (2015). Practical senses. Philosophers’ Imprint, 15(29), 1–25. Prinz, J. (2007). Mental pointing: Phenomenal knowledge without concepts. Journal of Consciousness Studies, 14(9–10), 184–211. Quine, W. V. O. (1968). Ontological relativity and other essays. New York: Columbia University Press. Reid, T. (1785/1849). Essays on the intellectual powers of man. Edinburgh: McLachlan, Stewart, & Co. Roediger, H. L. (1996). Memory illusions. Journal of Memory and Language, 35, 76–100. Rugg, M. D., & Henson, R. N. A. (2002). Episodic memory retrieval: An (event-related) functional neuroimaging perspective. In A. Parker, E. Wilding, & T. Bussey (Eds.), The cognitive neuroscience of memory: Encoding and retrieval (pp. 3–37). Hove: Psychology Press. Rugg, M. D., Johnson, J. D., Park, H., & Uncapher, M. R. (2008). Encoding-retrieval overlap in human episodic memory: A functional neuroimaging perspective. Progress in Brain Research, 169, 339–352. Rugg, M. D., & Wilding, E. L. (2000). Retrieval processing and episodic memory. Trends in Cognitive Sciences, 4(3), 108–115. Schacter, D. L. (1995). Memory distortion: History and current status. In D. L. Schacter (Ed.), Memory distortion: How minds, brains, and societies reconstruct the past (pp. 1–43). Cambridge, MA: Harvard University Press. Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering the past and imagining the future. Philosophical Transactions of the Royal Society B, 362, 773–786. Schacter, D. L., Norman, K. A., & Koutstaal, W. (1998). The cognitive neuroscience of constructive memory. Annual Review of Psychology, 49, 289–318. Sederberg, P. B., Schulze-Bonhage, A., Madsen, J. R., Bromfield, E. B., Litt, B., Brandt, A., & Kahana, M. J. (2007). Gamma oscillations distinguish true from false memories. Psychological Science, 18, 927–932. Sederberg, P. B., Kahana, M. J., Howard, M. W., Donner, E. J., & Madsen, J. R. (2003). Theta and gamma oscillations during encoding predict subsequent recall. Journal of Neuroscience, 23, 10809–10814. Semon. (1904/1921). The mneme. London: Allen & Unwin. Shimamura, A. P. (2011). Episodic retrieval and the cortical binding of relational activity. Cognitive, Affective, & Behavioral Neuroscience, 11, 277–291. Simons, J. S., Peers, P. V., Mazus, Y. S., Berryhill, M. E., & Olson, I. R. (2010). Dissociation between memory accurary and memory confidence following bilateral parietal lesions. Cerebral Cortex, 20, 479–485.

220  Felipe De Brigard Skinner, E. I., Fernandes, M. A., & Grady, C. L. (2009). Memory networks supporting retrieval effort and retrieval success under conditions of full and divided attention. Experimental psychology, 56, 386–396. Steinmetz, P. N., Roy, A., Fitzgerald, P. J., Hsiao, S. S., Johnson, K. O., & Niebur, E. (2000). Attention modulates synchronized neuronal firing in primate somatosensory cortex. Nature, 404, 187–190. Tulving, E. (1983). Elements of episodic memory. Oxford: Clarendon Press. Wagner, A. D., Shannon, B. J., Kahn, I., & Buckner, R. L. (2005). Parietal lobe contributions to episodic memory retrieval. Trends in Cognitive Sciences (Regul. Ed.), 9, 445–453. Wood, N., & Cowan, N. (1995). The cocktail party phenomenon revisited: How frequent are attention shifts to one’s name in an irrelevant auditory channel? Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 255–260.

Part V

Memory Failures Concepts and Ethical Implications

11 Forgetting

Matthew FriseForgetting

Matthew Frise

1 Introduction This chapter is about how you will forget this chapter. It’s about what forgetting anything is. We forget often, and psychologists research why. But neither they nor philosophers have tried much to unearth the nature of forgetting. The little shoveling in the area has turned topsoil only. This is odd, since forgetting is philosophically important. It looks essentially connected to, and as important as, remembering—a popular topic in the philosophy of memory. Fifty years ago, the Philosophical Review published C.B. Martin and Max Deutscher’s landmark article “Remembering,” which proposed an analysis of remembering. The literature on their paper and topic is now sizable. Commonsense puts remembering on one pole of a spectrum, and forgetting on the opposite pole. If commonsense is correct here, then a handle on forgetting looks crucial to a handle on remembering. We fully grip the poles together or not at all. Yet “forget” appears only three times in Martin and Deutscher’s paper. Forgetting also matters in epistemology. The fact that ordinary humans forget introduces an evaluative challenge. We forget evidence, as well as counterevidence or “defeaters.” This forgetting can be a kind of losing. But losing evidence or counterevidence can affect what it is reasonable to believe. Since epistemology is concerned with theories of reasonable belief, understanding what counts as forgetting will be crucial to evaluating these theories.1 Without this understanding, it will be hard to tell whether an apparent case involving forgetting is a counterexample to a theory. What’s more, the pattern of ordinary human forgetting looks epistemically significant. Kourken Michaelian (2011: 400) argues that, contrary to a standard view in epistemology, normal human forgetting “approximates a virtue” rather than a vice. It is a mean between too much forgetting and too much remembering. If Michaelian or the standard view is right, then forgetting underlies a normative epistemic status. Forgetting has no small significance in metaphysics either. Consider personal identity, that is, how a person persists or is identical to something at another time. One take on this is that personal identity consists in having

224  Matthew Frise certain memory connections.2 And forgetting looks like a wearing down of memory connections. Too much forgetting might destroy personal identity, on some theories. Testing these theories requires clarity on when subjects do and don’t forget. There are other areas in philosophy where forgetting matters, but turn now to psychology. Most often, when psychologists talk of forgetting, they talk about its causes, not its nature. Research focuses on the trauma, disorders, and commonplace memory processing that makes forgetting more likely, not on what forgetting is. When psychologists do say what forgetting is, their accounts are often incomplete and superficially incompatible with one another. Further reflection on forgetting will help reveal whether they really disagree with each other, whether they have ultimately distinct phenomena in mind, and whether some accounts are simply mistaken. For reasons like these, I develop and defend a theory of forgetting in this chapter. First, I survey some viable choice-points for theorizing about forgetting, points where forgetting is somewhat murky. Then I state the theories of forgetting that flit in the literature. I introduce two tests for a theory of forgetting, and show that the best theories from the literature fail both tests. Finally, I introduce a new theory of forgetting, the learning, access failure, dispositional (LEAD) theory. I argue that the LEAD Theory of Forgetting is the lead theory of forgetting. The causes and normative status of forgetting don’t get further airtime.

2  Dimensions of Forgetting The dimensions of forgetting I will cover are notable, but not exhaustive. Keeping them in mind helps us articulate a more exact theory of forgetting. It turns out that what we say about some dimensions of forgetting limits what we can sensibly say about others. As we’ll see, that introduces problems for some theories of forgetting. Some dimensions to consider are: Ontological Category At a most general level, what kind of thing is forgetting? Maybe forgetting is a mental state—a state of mind. It is something that a subject is in at a time, like the state of being in pain, or feeling bliss, or desiring the truth. Another option is that forgetting is a mental process. It is something mental that unfolds over time, like the process of calculating a sum, or forgiving someone, or coming up with a joke. Some ordinary forgetting-attributions mask these options. When I tell you merely that “Maria is forgetting her password,” I may be saying something about her current mindset, or instead about how she is changing. It could be that these options aren’t exclusive. Some forgetting may be a state, and some may be a process. Within any of its dimensions, forgetting may show some diversity.

Forgetting  225 Content Set aside what forgetting itself is. What kind of thing is it that we forget? The content of forgetting could be propositional. To forget is to forget that p. Your friend forgets that Plato taught Aristotle, and you are forgetting that Tuesday is your mother’s birthday. Or the content of forgetting could be objectual. To forget is to forget o, where o is an event, experience, or other object; I have forgotten my first birthday party, and I am forgetting my childhood neighbor. Or, the content of forgetting could be procedural. To forget is roughly to forget how to φ, where φ-ing is the performing of an action. The unfortunate magician forgets how to perform his magic trick, and the hungry fisher is forgetting how to fillet a fish. Different memory systems would naturally correspond to forgetting with different content types. Propositional forgetting would likely have a special relation to semantic memory, that is, memory for propositions; objectual forgetting would likely have a special relation to episodic memory, our memory for experienced events; and procedural forgetting would have a special relation to procedural memory, our memory for how to do things and for skills. If some forgetting has one type of content and other forgetting has another type, more than one memory system is likely responsible for human forgetting.3 Relation to Content Forgetting appears to be a mental relation between a subject and a content. But a subject can bear many mental relations to a content. Suppose the content is propositional. A subject who forgets that p is related to p in a certain way. What distinguishes that relation from others the subject bears to p? Instead of forgetting that p a subject may, say, desire that p or imagine that p. Something characterizes forgetting, setting it apart from these other relations. Perhaps loss characterizes it. To forget is, in some sense, to have something no longer. What exactly is lost is a further question. Michaelian (2011: 403) says a “record” or mental representation is lost, and others (e.g., McGrath, 2007, p. 1; Naylor, 2015, p. 377) suggest a belief is lost. Another characterization of the forgetting relation is that of failing to retrieve (Arango-Muñoz, 2013). Forgetting is not a matter of losing something, but rather of failing to access it. Some relation characterizes forgetting. Whatever it is, another question concerns its minimal duration: how long must the relation endure to count as forgetting rather than as something else? The minimum may be short, such that even temporary loss or retrieval failure could count as forgetting. Alternatively, the relation may have to be effectively permanent. If something is forgotten it is locked away in, or eliminated from, the mind. It would have to be relearned altogether in order to be mentally accessed. Some philosophers (e.g., Pappas, 1987, p. 153) think there is more than one kind of forgetting, and they individuate the kinds by minimal relation duration.

226  Matthew Frise When the relation is permanent, it is one circumstance in which we hear the subject described as “completely” forgetting. Scale The last dimension I’ll cover is the magnitude-range of forgetting. Can forgetting vary in intensity? The simplest answer is that it cannot. Forgetting is binary. You either forget or you do not. There’s no middle ground, no partial forgetting, no milder or sharper forgetting. Then again, it could be that forgetting is gradable. It admits of degrees or occurs to varying extents. As Daniel Schacter (2001: 33) puts it, there is “incomplete rather than total forgetting that leaves in its wake scattered shards of experience. Vague impressions of familiarity, general knowledge of what happened, or fragmentary details of experience.” When forgetting intensity is high enough, it is another circumstance in which we hear the subject described as “completely” forgetting.

3  Theories of Forgetting A theory of forgetting fills in some dimensions of forgetting. A more complete theory completes more dimensions. The peeps in the literature about forgetting’s nature express stances on some dimensions I’ve covered. Some philosophers and psychologists state in passing simple views about forgetting. It may be most charitable to understand these views as incomplete theories, stating just a necessary or sufficient condition for forgetting. Even so, I will show that the simple views face immediate problems. Then I will go over more developed theories of forgetting. One simple view is that you are forgetting whatever it is you aren’t recalling. That is: Simple Access Failure Theory. If S does not recall x, then S forgets x.4 The Simple Access Failure Theory classifies forgetting as a state. You are in the state of forgetting anything that you aren’t recalling. But this is not quite right. At any given time, we aren’t recalling most things that we remember quite well. This is in part because we are making no attempt to recall any of this. But it would be incorrect to count us at any given time as forgetting most of what we remember! A moment ago, for example, you weren’t recalling what you did last night. You weren’t trying recall it. Still, we shouldn’t conclude that you were forgetting it. There is a simple fix. You aren’t forgetting just anything you aren’t recalling. Rather, you are forgetting only what you aren’t recalling in a certain situation. What’s the situation? It could be free recall, a period where you try to recall select items (in no particular order, using any strategy). Given this fix, we have: Less Simple Access Failure Theory. If S does not recall x during free recall targeting x, then S forgets x.5

Forgetting  227 This theory makes an improvement. It doesn’t imply that at any given time you are forgetting most of what you remember. This is because you aren’t in free recall at just any given time, and because when you are in free recall you aren’t trying to recall most of what you remember. Unfortunately, a problem remains. Suppose that right now, during free recall, I am trying to recall what it is you did last night. If I am unsuccessful, we should not conclude that I am forgetting what you did last night. This is because at no point have I had any idea what you did last night. I never learned what you did. The Less Simple Access Failure Theory, however, oddly implies that I am forgetting what you did last night. What’s more, the Less Simple Access Failure Theory is far from complete, giving just a narrow sufficient condition for forgetting. It doesn’t tell us anything about the forgetting that occurs outside of free recall. And, as we’ll see, forgetting outside of free recall is common. If, during free recall, you do not recollect something that you learned and that you’re trying to recall, your memory in some sense fails. The final simple view is that forgetting is just any sort of memory failure: Simple Memory Failure Theory. If S’s memory fails (with respect to x), then S forgets (x).6 On the Simple Memory Failure Theory, forgetting looks once again like a state. It at least sometimes is a state of having memory failure with respect to something. Compared to the previous theory, this theory makes a step forward. On this theory, memory failures other than those occurring during free recall count as forgetting. However, this theory also makes a step backward. It avoids being too narrow about forgetting simply by being too broad. Memory can fail in a number of ways that don’t involve forgetting. It can fail by leaving us open to confabulating, adding incorrect details into what is recollected. It can fail by leading to various biases, or by continually activating unwelcome memories.7 These memory failures are not automatically instances of forgetting. Of course, we could refine the idea behind the Simple Memory Failure Theory. Not just any memory failure is forgetting. Rather, forgetting is a special kind of memory failure. Just what kind is it? As we noted with the Less Simple Access Failure Theory, our focus should not be limited to failures during free recall. Also, the relevant memory failure should exclude failures to access information one never learned. I don’t think there is a good, simple way to meet all the criteria here, so I won’t say more about the Simple Memory Failure Theory. The simple theories proved to be too simple. I will introduce a pair of more sophisticated theories, and then test them together. The simple theories all identify forgetting as something that can be a state. The first sophisticated theory departs notably from this, however, focusing on forgetting as a process. Timothy Williamson (2000: 34) takes this to an extreme, saying, “Not all factive attitudes constitute states; forgetting is a process.” A factive

228  Matthew Frise attitude is an attitude that guarantees the truth of its content. Seeing that p is often thought to be factive. If you see that Brian has a mullet, Brian indeed has a mullet. Williamson thinks forgetting is factive—you can forget only truths, not falsehoods. For Williamson, it is critical that all forgetting is a process, never a state. He (2000: 34) thinks “knowing is the most general factive stative attitude,” that is, anyone in a factive attitude state that p knows that p. (This, he thinks, is evidence that the concept of knowledge plays a central role in our thinking.) But at first glance, it doesn’t look like your forgetting that p guarantees your knowing that p. On the contrary, forgetting seems to explain a loss of knowledge. So, if forgetting is factive, it had better turn out that forgetting isn’t a state. For if it is a factive state, knowledge isn’t the most general factive stative attitude, and we lose evidence that the concept of knowledge is so special. Williamson suggests, then: Process Theory. S forgets that p only if and because S has a factive process attitude toward p.8 As formulated, the Process Theory says just this: forgetting guarantees there is some process involving an attitude toward a true proposition, and forgetting is explained by this process. It doesn’t say that forgetting is the process attitude, nor that forgetting cannot be a state. So, the Process Theory is weaker than Williamson’s view. Also, friends of the Process Theory may offer a different account for nonpropositional (objectual and procedural) forgetting. None of this will matter for my evaluations of the Process Theory in the next section. At any rate, if the Process Theory is incorrect then Williamson’s view is incorrect, since Williamson’s view entails the Process Theory. And if the Process Theory is incorrect, there is less evidence that knowledge plays the special role that Williamson says it plays. I’ll return to this point in Section 6. The other sophisticated, and most promising, theory of forgetting currently in the literature, understands propositional and objectual forgetting as a loss of information. The information is lost from long-term memory, which includes short-term memory but not working memory. This gives us: Information Loss Theory. S forgets x only if and because S loses a record of x from S’s long-term memory.9 Understand loses here broadly. Sometimes loss is the elimination of a record from long-term memory—the information fades or gets deleted— so the loss is all else being equal permanent. But more often it is an inaccessibility of information despite appropriate stimuli. That is, even though the subject is presented with relevant retrieval cues, the record remains at least temporarily inaccessible.10 On the Information Loss Theory, forgetting looks like a state, though it could also be a process involving this

Forgetting  229 state. The state is that of having dropped something from long-term memory, or of having something in long-term memory that cannot be retrieved. The Information Loss Theory skirts the snares that tripped the Less Simple Access Failure Theory—it theorizes about all propositional and objectual forgetting, and it makes forgetting require learning (because anything lost from long term memory was learned). And the Information Loss Theory does not conflate just any memory failure with forgetting. So it beats, or perhaps simply develops, the Simple Memory Failure Theory. In fact it’s looking shipshape.

4  The Metacognition Test and Prospection Test The Process Theory and Information Loss Theory fashion up forgetting in quite different ways. Still, they have similar flaws. This section offers two tests for an adequate theory of forgetting. Each checks whether a theory can accommodate certain intuitive data about forgetting. It is bad to fail either test. Failure generates an argument against the theory. I claim that the Process Theory and Information Loss Theory fail each test, and so they are in trouble. They may face separate challenges as well, but for simplicity I consider just those they share. First, there is the metacognition test. Not only do we cognitively process information, but we also monitor and control this processing. This is metacognition. One thing we monitor, typically unconsciously, is our own production of information (from, for example, memory). This results in a metacognitive feeling, which gives phenomenological feedback about our cognitive processing. Any of a broad range of metacognitive feelings may arise, depending on whether information is indeed produced, on the details of any produced information, and on how it is produced (e.g., quickly, slowly). One common metacognitive feeling is the feeling of forgetting. It can be a frustratingly unspecific feeling. As you are about to leave your home for the airport, you feel you are forgetting . . . something. You can’t tell what, despite trying. You are failing to retrieve some target information. This metacognitive feeling is not only common, but it is also generally accurate evidence of forgetting.11 The feeling may sometimes arise when there is no relevant forgetting, but that is the exception. A theory of forgetting should not conflict with any of this. What’s more, it should make sense of cases in which, intuitively, the feeling of forgetting is accurate; the theory should imply that they are indeed cases of forgetting. These are the standards of the metacognition test. The Process Theory does not meet these standards. It could be that processes altogether lack phenomenology, but that is not why the Process Theory fails the metacognition test. For even if no process is phenomenological, it could be that the process of forgetting involves a succession of states, and that one of these can be a phenomenological state like the feeling of forgetting. The Process Theory could then explain any accurate feeling of forgetting as occurring in connection with a state in the process of forgetting.

230  Matthew Frise No, the Process Theory fails the metacognition test elsewhere. Suppose forgetting can indeed be a process. The process is not in each case ongoing. It ends. And sometimes, afterward, a relevant feeling of forgetting arises. And sometimes, this feeling is accurate—there is in fact forgetting after the process of forgetting has finished. Take a simple example. I study a list of obscure words having to do with astrophysics. Much later, you ask me to reproduce the list. I experience the feeling of forgetting. I am in fact forgetting what was on the list. So, my feeling is accurate. But the Process Theory does not make sense of this. The process of forgetting what was on the list began and ended some while ago. Once the process ends, that forgetting ends. The feeling of forgetting is a feeling having to do at least partly with how things currently stand; it is a feeling of current forgetting, not purely past. So on the Process Theory, any feeling of forgetting after the process of forgetting ends is inaccurate. According to it, I am not forgetting when having a feeling of forgetting upon trying and failing to recollect the astrophysics terms that I had learned. But that is wrong. Despite its merits the Information Loss Theory stumbles on the metacognition test too. The Information Loss Theory explains forgetting in terms of lost target information. A feeling of forgetting will be accurate only if the target information either has been eliminated or is inaccessible when the feeling occurs. In many cases the target information really is lost, but not so in many others. These other cases are those where the target information has not been eliminated and remains accessible in the face of relevant cues, but the information simply is not accessed. Suppose you supply helpful retrieval cues, reminding me that the words on the studied list were on astrophysics, that some words start with this or that letter, and that others rhyme with these or those words. On this occasion I am unsuccessful in recollecting what was on the list, but I could succeed on another occasion. Yet I am indeed forgetting, and there is no inaccessibility or elimination of information here that explains this. The problem is not just in the eyes of pedants; the difference between a failure to access and inaccessibility is not trivial.12 I fail to access my destination if something halts me en route to it. But that is not to say my destination is inaccessible. If it were inaccessible, all my routes to it would be blocked, or I would not have the power to access it. Inaccessibility is a strong dispositional relation. Forgetting can result from a one-off disconnect, a nondispositional and weaker relation. Sometimes this weaker relation explains why a feeling of forgetting is accurate, even when the target information is accessible. That is why the Information Loss Theory doesn’t pass the metacognition test. The second test, the prospection test, focuses on certain cases of prospective memory failure. Roughly, prospective memory is memory for intended future action. Some prospective memory failure counts as forgetting.13 Suppose I decide I will call you at noon, but when noon comes I am so preoccupied that it never occurs to me to pick up my phone. I needn’t have a

Forgetting  231 feeling of forgetting at noon, nor any other relevant metacognitive feeling. I simply fail to perform my intended act when I am in the appropriate circumstance. The failure to recollect may result from a shortage of suitable retrieval cues—I didn’t set an alarm as a reminder—or from suitable retrieval cues being available yet ineffective—my alarm went off, but I didn’t recall what it was for. Regardless, it seems I am forgetting, as a direct result of some failure of prospective memory. Call this prospective forgetting. The Process Theory poorly explains prospective forgetting. Prospective forgetting can happen in particular circumstances or at specific times. At noon I forget to call you. I am not forgetting before or after noon. But processes take time. They occur over time, not at times. So the Process Theory requires that forgetting is temporally extended. It can describe me, at noon, as being somewhere in the process of forgetting, but it cannot describe my forgetting as entirely localized to noon. But it is natural to describe some prospective memory failure as occurring just at one time. This would be the time of the intended action. Since processes are temporally bloated, the Process Theory can’t explain all prospective memory failure. The Information Loss Theory struggles on the prospection test for the same reason it struggled on the metacognition test. The theory explains forgetting in terms of eliminated or inaccessible information. Now, in prospective forgetting, an intended act is not performed at the chosen time or circumstance. Something or other is not accessed then. Suppose what is not accessed is the intention to act. This is not to say that anything is inaccessible or has been eliminated at the time of forgetting. The intention can remain accessible, but just not accessed. Better retrieval cues, or better luck with retrieval cues, could have led to access. I could have accessed my intention to call you, and would have accessed it if I had set my alarm, or if I recollected what it was for when it went off. The Information Loss Theory does not account for all prospective forgetting. On a different theory of what loss consists in, the Information Loss Theory could do better on the metacognition and prospection tests. The alternative theory counts not just elimination and inaccessibility as loss, but also access failure. Advocates of the Information Loss Theory do not yet count access failure as loss, but perhaps the metacognition and prospection tests will be their gadfly. However, this alternative theory does not seem promising— it seems to imply that there are losses where, intuitively, there aren’t any. Suppose you are playing a trivia game, and an answer to a trivia question is stuck “on the tip of your tongue.” You are failing to access it. This new theory apparently counts this answer as lost. This tip of the tongue state does seem like forgetting of a sort, but it needn’t involve loss. It is precisely because you have not lost the answer that it feels like it is on the tip of your tongue. Because you have not lost the answer, you can reasonably rule out many possible answers (“Let’s see, the fifth U.S. president wasn’t James Madison . . .”), and why you can in some cases detect linguistic features (“But his name is very similar to ‘James Madison’ . . .”). The revised theory

232  Matthew Frise of loss looks incorrect. Advocates of the Information Loss Theory gain little if they adopt it in order to pass the metacognition and prospection tests.

5  The LEAD Theory of Forgetting No theory of forgetting in the literature is fully adequate. Each fails to account for some forgetting. I offer a new theory. It aims to be maximally general, leaving open as many dimensions of forgetting as possible. So, it aims to be a maximally unified account of forgetting, showing what the various types of forgetting fundamentally have in common. Insofar as it meets this aim, it has an asset. (We’ll see it has other assets too.) Perhaps nothing unites the various forgetting phenomena beyond a family resemblance or even coincidence—English-speakers just happen to call some unrelated things “forgetting.” But if there is a sturdier connection, we should look for it. I will build up the necessary and sufficient conditions for forgetting one by one. To start: S forgets x to extent e at t iff . . . Here, x can be propositional, objectual, or procedural. So the theory is general with respect to content type. It is not general with respect to ontological category, however. It indexes forgetting to a time t, and so it is not about forgetting over time—that is, not about the process of forgetting. For now I will say what just the state of forgetting is, and later will explain the process in terms of the state. Finally, forgetting can be gradable on this theory. One forgets to an extent, e. If forgetting can be gradable, it is a further and unanswered question just how many grades there are. The grades could be few (slight, moderate, strong) or many (extremely slight, very slight, slight. . .). But the theory is also compatible with some and even all forgetting being binary. When the magnitude range of forgetting is binary, then one forgets to the greatest extent if at all. Since the theory accounts for both binary and gradable forgetting, it is general in scale. The first necessary condition for forgetting is: (1) S has learned x by t, and. . . You cannot forget what you never learned. According to (1), forgetting requires learning by the time of forgetting. Normal human learning results from experience and from memory processing information that originates in experience. Learning can also have unusual causes, such as brain-tampering or trauma.14 At any rate, not all learning will be conscious. It is an open question whether we count as learning information that enters only working memory and not long-term memory, or information that is a part of experience but never attended to.15 In short, (1) ensures that we forget only what we have gotten.

Forgetting  233 The next necessary condition for forgetting is: (2) S fails to extent e to internally access x at t, and. . . At the time of forgetting, the subject fails to access internally whatever is being forgotten. If you are internally accessing something—that Plato taught Aristotle, your experience of your breakfast—you aren’t altogether forgetting it. Whatever you internally access, you have, in a way that forgetting precludes. The internally qualification here is imprecise but important. Suppose I ask you what you were doing exactly one year ago. You try to recall, without success. I then pull up some footage of what you were doing. Merely by watching the footage, you are in a sense accessing what you were doing.16 Still, you forget what you were doing, even as you watch. The kind of access forgetting cannot abide is strictly internal. For simplicity I leave ‘internally’ implicit from here on. Like forgetting, access failure appears to occur to greater and lesser extents. I can bring to mind just a few details of a past experience, or a host of details, or a smattering in between. According to (2), a subject forgets something only to the extent that she fails to access it. If she completely forgets, it must be that she completely fails to access. Accessibility is another matter, however. For all (2) states, something can be completely forgotten and yet accessible, even highly accessible. So we diverge here from the Information Loss Theory. Conditions (1) and (2) say forgetting requires learning and access failure. Forgetting requires more, though. At any given time, I am failing to access most of what I have learned—even most of what I still know—but I am not forgetting all of this bulk. The final element for forgetting is a dispositional relation to what is forgotten. But it turns out that either of two dispositional relations will do. The last necessary condition for forgetting, then, is disjunctive: (3) (a) at t, S intends to internally access x under some description by t, or (b) x is internally inaccessible to S to extent e at t. Forgetting involves either intending to internally access, or internal inaccessibility. I will explain these dispositional relations in turn. According to (a), one last condition for forgetting can be intending to access. Together with the first two conditions, this tells us that forgetting can be failing to access something that was learned and is intended to be accessed. There is more than one way that a subject can be, at the time of forgetting, intending to access something by then. One obvious option: the subject is at the time of forgetting in fact attempting to access what is forgotten. If, for example, I am trying to recollect your last name right now, I am thereby intending to access it right now. If my efforts are unsuccessful, and I have indeed learned your last name, then right now I am forgetting

234  Matthew Frise it. Now, if I am forgetting your last name, I am failing to access it. So the content of my current attempt to recollect your last name cannot be directly identifying it (for example, as Nieves)—direct identification looks like access. Rather, the content of my recollective attempt identifies it indirectly, by appropriately describing it (for example, as Maria’s last name). That is why (a) says the subject who forgets x intends to access x under some description. We now have a good explanation of why the feeling of forgetting is accurate, when accurate. While a subject has the feeling of forgetting and is indeed forgetting, she is attempting (perhaps involuntarily) to access something she had learned, satisfying (1). In virtue of attempting, she is intending to access what she had learned, satisfying (a). But she is unsuccessful, satisfying (2). These three conditions jointly suffice for forgetting. Metacognition test: passed. There is a second option for how a subject can be intending to access what she is at the time forgetting: the subject formed this intention in the past and has kept it. That is, the subject has a suitable intention stored or standing when forgetting. A closer look at prospective memory will help us make sense of this. Prospective memory, again, is memory for intended future action. This morning I decide I will call you at noon. Something sustains this plan, allowing me to follow through on it. That is prospective memory. Although we commonly talk as though a mere act is the content of prospective remembering and forgetting, the content really is an intention. I don’t remember to call you, I remember my intention to call you. Now, since I don’t intend to call you just yet, for now I file away my intention to call you. I don’t want it filed away forever. I want to retrieve it at the right time, noon. In fact, normally, I intend to retrieve it at the right time—I have an intention to access my intention to act. To see this, note how it is possible yet unusual for me to intend a future act but not intend to access the intention to act. This might occur if, for example, I suspend or lose my intention to access my intention to call you, even while keeping the intention to call. To my relief, my wife tells me she’ll remind me to call you; I cease trying to remember to call you, though I still intend to call. Prospective memory commonly involves an intention to retrieve an intention for future action. I suggest that in prospective forgetting, the subject still intends to retrieve, but at the desired time fails to retrieve.17 Suppose I haven’t outsourced the reminder task to my wife. I intend to access at noon my intention to call you. Noon comes, and I do not access the intention to call, but I still have the intention to access it. I am forgetting to call you. More exactly, I am forgetting my intention to call you. If, however, I have totally discharged the reminder task, I cannot be at noon forgetting my intention to call you. I no longer intend to access my intention to call, so I am not forgetting that intention. I can of course not remember at noon to call you, but that is not the same as forgetting.

Forgetting  235 Conditions (1), (2), and (a) are satisfied in any prospective forgetting case. At noon I fail to access my intention to call—fulfilling (2)—yet at noon I intend to access the intention to call—fulfilling (a). I formed my intention to call you by the time of forgetting it, and this acquisition of an intention is learning of a sort, a taking in. This fulfills (1). Prospection test: passed. Now for (b), the second disjunct in condition (3). It places an internal inaccessibility condition on forgetting. Given (b) and conditions (1) and (2), we see forgetting can be something different from what accounts for the accurate feeling of forgetting and for prospective forgetting. Forgetting can be failing to access something that was learned and is inaccessible. This explains cases where what is forgotten is something that has been eliminated from memory or rendered at least temporarily irretrievable. The forgetter needn’t have any metacognitive feeling or standing intention to access what is forgotten. With respect to your old locker combination you have no feeling of forgetting, nor any plan to retrieve it. Yet you forget it, and will continue to forget it, and have been forgetting it for some while. It’s inaccessible to you. Inaccessibility can be partial. This is vague, but intuitive enough. If many details of a previously experienced event are now inaccessible, yet other details remain accessible, then the event itself is partially inaccessible. Condition (b) links the extent of forgetting with the extent of the inaccessibility of what is forgotten. Forgetting is greater or lesser as inaccessibility is greater or lesser. Of course, condition (2) links the extent of forgetting with something else— the extent of the failure to access what is forgotten. This raises a question: if both (2) and (b) are satisfied, and the extents of the access failure and inaccessibility are unequal, to what extent does the subject forget? Is it to the extent of the access failure, or to the extent of the inaccessibility? Answer: the lower of the two. The access failure and inaccessibility are each at least that high. Now, the inaccessibility will always be equal to or less than the access failure. That is, something will be at least as accessible as it is accessed. Sometimes what is forgotten is highly accessible while simply not accessed. What’s forgotten, though, is never highly accessed while inaccessible. So, since the extent of inaccessibility is the lower of the two when there is inequality, it determines the extent of forgetting. This way, you don’t count as forgetting your current and well-remembered locker combination when not trying to retrieve it. When not trying to retrieve it, you are failing to a maximal extent to access it. Still, it is highly accessible to you—it is inaccessible to no extent—so you aren’t forgetting it to any extent. There are no other conditions on forgetting. I propose, then, the learning, access failure, and dispositional—for short, LEAD—theory of forgetting: LEAD Theory. S forgets x to extent e at t iff (1) S has learned x by t, and

236  Matthew Frise (2) S fails to extent e to internally access x at t, and (3) (a) at t, S intends to internally access x under some description by t, or (b) x is internally inaccessible to S to extent e at t. The LEAD Theory says that a subject who forgets is failing to access something both learned and either inaccessible or intended to be accessed. I have been all too brief about what learning, internal access, and inaccessibility consist in. Those elements of the LEAD Theory deserve exclusive analysis on another day. It is enough work for now to establish just the structure of forgetting. Even by its structure we can see the LEAD Theory has a number of virtues. Jointly, they are strong evidence for the theory. I have laid out two virtues already: the LEAD Theory passes both the metacognition test and the prospection test. What’s more, the LEAD Theory accounts for all forgetting that the Information Loss theory accounts for, where the target information is irretrievable. It is a further virtue of the LEAD Theory that it unites various kinds of forgetting. It shows for example what prospective forgetting has in common with the forgetting that comes with a metacognitive feeling. It is general with respect to content type, connecting propositional, objectual, and procedural forgetting. And so on. There are many species of forgetting, and the LEAD Theory picks out their genus. Another virtue of the LEAD Theory is that it is not in jeopardy if human memory turns out to be generative. Empirical research suggests our memory is not a simple warehouse that shelves the items we deposit and later withdraw. Instead, it looks like memory disassembles, discriminatingly discards, redesigns, and reassembles, even in cases where we end up withdrawing a fairly faithful representation of the past.18 Memory constructs, not just a little, and not just when malfunctioning. This presents no problems for the LEAD Theory. If memory meddles with what we give it, it doesn’t follow on the LEAD Theory that we are forgetting, since the meddling needn’t affect accessibility. Accessing could be a generative process; accessibility could be a matter of generative power, and inaccessibility could result from generation troubles and not just storages troubles. On the LEAD Theory, forgetting is compatible with a few things that might surprise us. This is a virtue of the theory because the compatibility is correct but hard to achieve. Forgetting is compatible with knowing, for example. Some ways of forgetting leave room for knowledge. Only certain ways preclude it.19 Suppose in a trivia game, you are asked who the fifth U.S. president was. You know it was James Monroe. But this knowledge remains dispositional, even after you are asked—it is standing, not before your mind. You have a feeling of forgetting, and are trying but failing to access this information you learned. So, you are forgetting something you simultaneously know, namely, that James Monroe was the fifth U.S. president. While having this feeling of forgetting, you might even have a

Forgetting  237 second metacognitive feeling about the same information—the feeling of knowing! Sometimes both feelings are accurate. Similarly, on the LEAD Theory, forgetting is compatible with remembering. When a subject is forgetting x because she is trying but failing to access x, she can on many theories still be remembering x.20 To be remembering and forgetting the same thing at once is not rare. I am forgetting your last name. Nonetheless I am confident I still remember it, and so I continue to try to retrieve it. Eventually I successfully retrieve it, and this is because I remembered it all along. The remembering had just been dispositional or standing. In light of its virtues, I conclude that the LEAD Theory is the one to beat.

6  Process Forgetting The LEAD Theory does not, however, tell the full story of forgetting. It accounts for forgetting with a variety of content types, relations to content, and scales. But it is just a theory of stative forgetting, forgetting as a state. It is not, in other words, general with respect to ontological category. A final chapter in the full story of forgetting says what the process of forgetting is, and says what relation the process bears to the state. I’ll sketch a reductionist theory of the process of forgetting (or process forgetting). It takes a simple view of what a process in general is: a sequence of states. If a process of any kind is just a sequence of states, then the process of forgetting is just a sequence of states. This suggests a formula: build up a view of process forgetting by identifying its constitutive states and their relation. I propose that the constitutive states are states of forgetting, and that their relation is one of increasing strength. That is, the process of forgetting is that of going from a state of lesser forgetting to a state of greater forgetting. More formally: Reductionist Process Theory. S is in the process of forgetting x from t to t+n iff for every pair of sequential times tm and tm+1 in the interval from t to t+n, (1) S forgets x to extent e1 at tm, (2) S forgets x to extent e2 at tm+1, and (3) e1 < e2. Since a process is a sequence of states on this theory, a process is extended through time. So, the left-hand side indexes the process of forgetting to spans of time. Whatever the potential extents of forgetting are, they increase over every time during the process of forgetting. This respects our intuitive judgments about process forgetting. When it seems someone is in the process of forgetting, it seems the subject ends up in a greater state of forgetting than before. If the extent of forgetting seems fixed over a stretch of time, the

238  Matthew Frise subject does not seem to be in the process of forgetting during it. The subject could, however, be in a persisting state of forgetting over those times. This points the spotlight at a potential ambiguity in our forgettingattributions. That someone is forgetting over time underdetermines whether the person is in a persisting state of forgetting, or is instead in the process of forgetting. On the LEAD Theory, someone could count as being in the state of forgetting over a long stretch of time—the subject could continue to forget something long forgotten. This seems odd if taken to suggest that as time passes the subject is repeating a process of forgetting, or is forgetting more severely something already completely forgotten. But a continuing state of forgetting does not suffice for a continuing process of forgetting. Continuously forgetting can be like continuously knowing, rather than continuously coming to know. There are two ways to keep forgetting. The Reductionist Process Theory does not follow from the LEAD Theory. It is compatible with any gradable theory of stative forgetting. It is an alternative to Williamson’s strong process theory, which denies that there is any stative forgetting. If the Reductionist Process Theory is correct, then Williamson’s strong process theory is incorrect, and several of his other claims from Section 3 teeter. On the Reductionist Process Theory, since there are processes of forgetting, there are states of forgetting. But if we grant Williamson that forgetting is factive, then the state of forgetting is a factive stative attitude that does not require knowledge. So, contra Williamson, knowledge is not the most general factive stative attitude. As a result, there is less evidence for Williamson’s claim that the concept of knowledge is central in our thinking.

7 Conclusion I have surveyed many dimensions of forgetting and challenged the few theories about it on offer. I have put forth a unified theory of the state of forgetting, and a theory of the process of forgetting, making the relation between the process and state clear. Together, my two main proposals remember all forgetting.21

Notes 1 See especially Conee and Feldman (2011, p. 304). For discussion, see Frise (2015; 2017a). 2 A view commonly attributed to John Locke. Behan (1979) reads him differently. 3 Cf. Michaelian (2011, p. 402) and Tulving (1983, p. 47). 4 Halamish et al. (2011, p. 632) and Tulving and Pearlstone (1966, p. 389). 5 Friedman and Castel (2011). 6 Bernecker (2010, p. 198). 7 See Schacter (2001). 8 Cf. Koriat et al. (2004, p. 651).

Forgetting  239 9 Harris et al. (2010, p. 255) and Michaelian (2011, pp. 402–4). Cf. Roediger et al. (2010, p. 2) and Tulving (1974, p. 74). 10 See Tulving and Pearlstone (1966) and Michaelian (2011, pp. 403–7). I note that elimination guarantees inaccessibility—what you no longer have, you cannot access. So, the loss relations appear to be just varieties of inaccessibility. 11 Cf. Arango-Muñoz (2013) and Halamish et al. (2011). See also Arango-Muñoz and Michaelian (2014) and Michaelian (2016). 12 Mere access failure explains the mere retrieval failure of information, which seems to have special epistemological significance. See Frise (2017b, section 4). 13 Cf. Annis (1980, p. 330) and Schacter (2001, pp. 51–60). 14 Cf. Conee and Feldman (2011, p. 305). 15 Schacter (2001, p. 27) thinks we forget even information that was only in working memory. Michaelian (2011, p. 402) disagrees. Presumably Schacter would say that information was learned, and Michaelian would not. 16 Cf. Martin and Deutscher (1966, pp. 182–3) on prompting. 17 McDaniel and Einstein (2007, p. 239) argue that the “retrieval of the prospective memory intention can occur even when the intention to retrieve has been suspended.” That is, prospective remembering can be spontaneous, not requiring an intention to retrieve the intention to act. This is orthogonal to my claim that prospective forgetting requires an intention to retrieve. 18 See Frise (2018) and Michaelian (2016). 19 Cf. Moon (2012, p. 356). 20 This is so on the simulation theory and various causal theories of remembering (Michaelian, 2016, Chs. 5–6). Incidentally, Bernecker (2008, p. 27) says “the notion of forgetting can be adequately defined only by way of appeal to the notion of remembering.” I see it as a further bonus that the LEAD Theory makes no such appeal. 21 I thank David James Barnett, Earl Conee, Dorothea Debus, Christoph Hoerl, Corey Maley, Kevin McCain, Kourken Michaelian, Andrew Moon, Denis Perrin, Sarah Robins, two anonymous referees, and an audience at the 2016 Memory and Subjectivity Conference for helpful discussion of a draft of this paper. I wrote this while supported by a grant from the Templeton Religion Trust. The views expressed here may not reflect those of the Templeton Religion Trust.

References Annis, D. B. (1980). Memory and justification. Philosophy and Phenomenological Research, 40(3), 324–333. Arango-Muñoz, S. (2013). Scaffolded memory and metacognitive feelings. Review of Philosophy and Psychology, 4(1), 135–152. Arango-Muñoz, S., & Michaelian, K. (2014). Epistemic feelings, epistemic emotions: Review and introduction to the focus section. Philosophical Inquiries, 2(1), 97–122. Behan, D. P. (1979). Locke on persons and personal identity. Canadian Journal of Philosophy, 9(1), 53–75. Bernecker, S. (2008). The metaphysics of memory. New York: Springer. Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press. Conee, E., & Feldman, R. (2011). Replies. In T. Dougherty (Ed.), Evidentialism and its discontents. Oxford: Oxford University Press.

240  Matthew Frise Friedman, M. C., & Castel, A. D. (2011). Are we aware of our ability to forget? Metacognitive predictions of directed forgetting. Memory & Cognition, 39(8), 1448–1456. Frise, M. (2015). Epistemology of memory. In J. Fieser & B. Dowden (Eds.), Internet encyclopedia of philosophy. Retrieved from www.iep.utm.edu/epis-mem/ Frise, M. (2017a). Internalism and the problem of stored beliefs. Erkenntnis, 82(2), 285–304. Frise, M. (2017b). Preservationism in the epistemology of memory. The Philosophical Quarterly, 67(268), 486–507. Frise, M. (2018). Eliminating the problem of stored beliefs. American Philosophical Quarterly, 55(1), 63–79. Halamish, V., McGillivray, S., & Castel, A. D. (2011). Monitoring one’s own forgetting in younger and older adults. Psychology and Aging, 26(3), 631–635. Harris, C. B., J. Sutton, & A. Barnier. (2010). Autobiographical forgetting, social forgetting and situated forgetting. In S. D. Sala (Ed.), Forgetting (pp. 253–284). London: Psychology Press. Koriat, A., Bjork, R. A., Sheffer, L., & Bar, S. K. (2004). Predicting one’s own forgetting: The role of experience-based and theory-based processes. Journal of Experimental Psychology: General, 133(4), 643–656. Martin, C. B., & Deutscher, M. (1966, April). Remembering. Philosophical Review, 75, 161–196. McDaniel, M. A., & Einstein, G. O. (2007). Spontaneous retrieval in prospective memory. In J. S. Nairne (Ed.), The foundations of remembering: Essays in honor of Henry L. Roediger, III. New York: Psychology Press. McGrath, M. (2007). Memory and epistemic conservatism. Synthese, 157(1), 1–24. Michaelian, K. (2011). The epistemology of forgetting. Erkenntnis, 74(3), 399–424. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Moon, A. (2012). Three forms of internalism and the new evil demon problem. Episteme, 9(4), 345–360. Naylor, A. (2015). Justification and forgetting. Pacific Philosophical Quarterly, 96(3), 372–391. Pappas, G. S. (1987). Suddenly he knows. In S. Luper (Ed.), The possibility of knowledge: Nozick and his critics. Totowa, NJ: Rowman & Littlefield. Roediger III, Henry L., Weinstein, Y., & Agarwal, P. K. (2010). Forgetting: preliminary considerations. In S. D. Sala (Ed.), Forgetting (pp. 1–22). New York: Psychology Press. Schacter, D. L. (2001). The seven sins of memory: How the mind forgets and remembers. Boston: Mariner Books. Tulving, E. (1974). Cue-dependent forgetting. American Scientist, 62(1), 74–82. Tulving, E. (1983). Elements of episodic memory. Oxford: Oxford University Press. Tulving, E., & Pearlstone, Z. (1966). Availability versus accessibility of information in memory for words. Journal of Verbal Learning and Verbal Behavior, 5(4), 381–391. Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University Press.

12 On the Blameworthiness of Forgetting Sven BerneckerOn the Blameworthiness of Forgetting

Sven Bernecker

Broadly speaking there are two kinds of phenomena the ethics of forgetting is concerned with. First, there are the moral and legal ramifications of the fact that we sometimes forget about our past wrongdoings. Consider two agents accused of the same kind of wrongdoing. One agent has suffered memory loss, albeit through no fault of his own, and cannot remember anything about the action he is accused of. In this case, the forgetting may be used as evidence of incompetence to stand trial or in mitigation of criminal and moral responsibility. The other agent intentionally brought about the memory loss (say, by taking a forgetting pill) before committing the wrongdoing. The voluntary nature of his forgetting has the consequence that it does not mitigate his responsibility (Birch, 2000). This suggests that whether the fact that an agent has forgotten about his past wrongdoing can be used as an excuse for a wrongdoing crucially depends on whether the forgetting was brought about intentionally and voluntary. Besides the moral assessment of past wrongdoings the agent has forgotten about, there is the moral assessment of the harm an agent brings about as a consequence of forgetting something. This is the issue the chapter is concerned with. There are at least two ways in which forgetting might be harmful. Sometimes it is the forgetting itself that constitutes the harm. For instance, it is common for your partner to feel hurt if you forget her birthday and for a close friend to be upset if you forget his name. People expect to be remembered by those who are important to them. The fact that you forget your partner’s birthday is taken as evidence that the partner is not sufficiently important to you. Besides cases where the forgetting itself constitutes the harm there are cases where forgetting gives rise to another event that, in turn, constitutes the harm. An example of the latter kind is a baby suffering from a heat stroke because it was forgotten inside a parked car on a hot day. The moral assessment of harm caused by forgetting crucially depends on whether the forgetting is brought about intentionally or unintentionally. If the memory loss is brought about unintentionally, the harmed person is usually upset, but when the memory loss is intentional and voluntary the harmed person often feels anger and resentment directed toward the

242  Sven Bernecker forgetting subject (Driver, 2009, pp. 84–85). This is nicely illustrated in the movie Eternal Sunshine of the Spotless Mind where the protagonist Joel (Jim Carrey) realizes that his ex-girlfriend Clementine (Kate Winslet) chose to undergo a memory erasure procedure offered by Lacuna, Inc., to overcome the pain of their breakup. The technicians at Lacuna wiped out all memories of Joel and of their prior relationship. Joel feels betrayed by Clementine’s intentional memory zap. But since he wants to give their relationship another chance he too undergoes the memory erasure process. Joel asks Lacuna, Inc., to not only erase all of his memories of Clementine and of his relationship with her but also his memories of the feeling of betrayal upon learning that she decided to erase her memories of him. In this chapter, I argue that forgetting has ethical impact in three ways. First, psychological studies show that we sometimes have control over forgetting. Hence the claim that we cannot be morally responsible for forgetting because forgetting is outside of our control is based on a misunderstanding of the psychology of forgetting. Once this psychology is understood it becomes apparent that there are instances in which we do have control over our forgetting and are therefore morally responsible for harm brought about as a consequence of forgetting something. The second claim concerns instances in which we have control over our ability to remember something at a later point in time. We can learn, for instance, that as an individual we are bad at remembering people’s names and birthdays. Our awareness of this fact should prompt us to take extra steps to improve our memories regarding names and birthdays. When we fail to take these extra steps and then forget, we are morally responsible. This type of moral responsibility is similar to the moral responsibility someone who intentionally got drunk bears for doing regrettable acts while drunk. The third claim concerns instances where we have neither control over our forgetting something nor control over our ability to remember something. Intuitively a person can be morally responsible for forgetting something even if it is beyond her control that she forgets that thing and even if she did everything that can be reasonably expected of her to remember that thing. The literature on culpable ignorance contains two accounts of the blameworthiness of cases of forgetting over which the agent has no control: attributionism and the liberalized awareness condition of moral responsibility. Rather than playing one account off against the other, I will give a neutral assessment of their strengths and weaknesses. Section 1 distinguishes between two kinds of forgetting: trace decay and interference. Section 2 reviews the two standard conditions for moral responsibility: the control condition and the awareness condition. Section 3 draws on work in the psychology of forgetting to argue that there are things an agent can do to actively forget something. The picture whereupon, in principle, it is beyond a person’s control whether she forgets something does not hold up to scrutiny. Section 4 discusses the tracing account of wrongdoing due to forgetting. On the tracing account, the blameworthiness

On the Blameworthiness of Forgetting  243 of the forgetting can be traced back to an intentional action the agent performed earlier. The point of Section 4 is to argue that not all cases of blameworthy forgetting can be captured by the tracing account. Section 5 is a critical discussion of an attempt to explain the blameworthiness of nontracing cases of forgetting in terms of a liberalized awareness condition of moral responsibility. Section 6 is a critical discussion of the attributionist explanation of the blameworthiness of nontracing cases of forgetting. Section 7 contains some concluding remarks.

1  Theories of Forgetting Forgetting in long-term memory indicates that an item of information that was once learned is either permanently lost or temporarily inaccessible.1 According to the decay theory, forgetting occurs as a result of traces in longterm memory declining in strength with time. The competing interference theory states that what causes forgetting is the disruption and obscuration of memory traces by preceding or succeeding learning. Interference manifest itself in two ways: when previous learning interferes with later learning and retention it is called proactive interference; when later learning disrupts memory for earlier learning it is called retroactive interference (cf. Baddeley, 2014, ch. 6). It has proven difficult to experimentally verify the automatic degeneration of memory traces postulated by the decay theory. One of the problems is that in any real-life situation the time between learning something and recalling will be filled with all kinds of different events. It is therefore impossible to rule out that any forgetting that takes place between leaning and recall is the result of trace decay rather than interference effects. On the other hand, there is solid experimental support for the interference theory by means of paired-associate learning.2 This may explain why, for many years, the interference theory was the majority view. But even though the interference theory has been supported in numerous studies it is not clear to what extend the findings from paired-associate learning in a laboratory setting can be applied to forgetting in everyday life. So even though decay (indexed by time) and interference (indexed by amount of distracting information) have historically been viewed as competing accounts of forgetting there is a recent tendency to view them as complementary processes. Hardt, Nader, and Nadel (2013), for instance, argue that in certain brain areas, notably the hippocampus, interference-driven forgetting is minimal and decay causes most forgetting; but in other brain areas it is the other way round (cf. Altmann & Gray, 2002). When a piece of information is forgotten due to the relevant trace having disappeared, it is permanently gone. When forgetting occurs because of interference, it is sometimes possible that the memory comes back if the agent is provided with the appropriate retrieval cue. So while forgetting due to trace decay is permanent, forgetting due to interference may be only

244  Sven Bernecker temporal. From a first-person point of view, it is usually indistinguishable whether a memory item is merely inaccessible or whether it has been deleted. In this chapter, I focus on cases of forgetting where the relevant information is still stored in the memory system and could be retrieved if the appropriate retrieval cues were provided. My reasons are two-fold. First, interference-driven forgetting happens much more frequently than decaybased forgetting. Even a forgotten item of information can typically be recognized at a rate that greatly exceeds chance levels, can be relearned at an accelerated rate, and can often be recalled in special circumstances that reinstate certain cues from the past—all of which constitute evidence that such items have not been lost from memory in any absolute sense. (Bjork & Vanhuele, 1992: 156, as cited in Michaelian, 2011, pp. 403–404) Second, the ethical implications of forgetting due to inaccessibility (and interference) seem to be more interesting than those of forgetting that takes the form of trace elimination. In other words, it seems more promising to try to prevent interference from rendering certain memory items inaccessible than to try to stop the decay of a particular memory trace.

2  Two Components of Moral Responsibility Intuitively there are two kinds of conditions that can be used to excuse oneself from moral responsibility: ignorance of the nature of one’s action and lack of control over one’s action. Consequently, the standard account of moral responsibility lists two necessary conditions: a control condition and an awareness condition. The control condition specifies the type and degree of control the agent needs to have over her action for her to be morally responsible for it. A crude form of the control condition states: Control Condition: S is morally responsible for performing (not performing) action A only if it is within S’s power to perform (not to perform) A. A lot of ink has been spilled over the question of whether the kind of control required for moral responsibility is compatible with causal determinism. For the purpose of this chapter we can put the issue of determinism to one side. The awareness condition specifies the type and degree of the agent’s awareness of the likely consequences of his action and its moral status for the attributability of moral responsibility. The idea is that an agent cannot be blamed for the harm she is causing if she is unaware, and has no reason to be aware, that her action risks or leads to harm. Given that blame responds

On the Blameworthiness of Forgetting  245 to contempt or the absence of concern for others, there is nothing in the mental life of an ignorant agent that could be the proper object of blame; or so the standard conception of moral responsibility claims.3 The awareness condition of moral responsibility is said to have two components. The first component states that an agent can be held responsible for doing something only when she is aware that she is doing it. And since an action may be intentional under some description and unintentional under another (Anscombe, 1957, pp. 11–12) the agent needs to be aware of what she is doing under the appropriate description. The second component of the awareness condition states that the agent can be held responsible for doing something only when she is aware of why she is doing it. In other words, it is a condition on moral responsibility that the psychological state that moves the agent to perform (or omit) some action matches the consideration that counts in favor of performing (or omitting) some action (Alfano, 2016). When both components are combined, we get a condition that looks something like this: Awareness Condition: S is morally responsible for action A only if (i) S is aware of doing A (under the appropriate description) and (ii) S does A while believing that A is morally wrong (in the case of a blameworthy action) or permissible/obligatory (in the case of a praiseworthy action).4 Obviously, one cannot be aware of why one is doing what one is doing unless one is aware of what one is doing. I focus on the first component of the awareness condition. The reason is that the main worry about blaming someone for forgetting something has to do with the agent not being aware of what she is doing, rather than with the agent not being aware of her motives.5 The fact that we blame people who cause harm due to having forgotten something seems to fly in the face of the standard conception of moral responsibility. It is commonly thought that forgetting is not under our control. But if forgetting is not under our control and if it is wrong to blame someone for something that they have no control over, then we ought not to blame someone for forgetting something, not even if the forgetting is harmful. Furthermore, forgetting does not seem to meet the awareness condition. The element of awareness commonly required for blameworthiness is absent in forgetting. If an agent has forgotten at t that p, then she is not aware at t of the likely consequences of not acting on the basis of p. For if the agent were aware of the likely consequences of not acting on the basis of p, then she would most likely act on the basis of p. So how can an agent be responsible for an action she is not aware of performing?6 The point of the chapter is that, contrary to what many people think, we can indeed be morally responsible for our forgetting. There are instances of forgetting where the control condition is satisfied, either because we have control over our forgetting (Section 3) or because we have control over our

246  Sven Bernecker ability to remember at a later point in time (Section 4). Moreover, it seems that a person can be morally responsible for forgetting something even if it is beyond her control that she forgot that thing. The literature contains two accounts of the blameworthiness of forgetting over which the agent has no control. One of these accounts modifies the awareness condition so that it is satisfied in some cases of forgetting (Section 5). The other account argues that neither awareness nor control are necessary for moral responsibility (Section 6).

3  Directed Forgetting It is common to think that there are no techniques for forgetting specific items that are non-drastic in the sense of not significantly diminishing one’s general epistemic competencies or abilities.7 This rules out, among other things, forgetting as a result of inflicting a serious brain injury on oneself. This section discusses the possibility of non-drastic techniques for forgetting specific items in light of recent work in psychology. There are a number of pharmacological tools available that can make us forget. Medications such as scopolamine, benzodiazepines, and kinase inhibitors have the proven effect of inducing amnesia.8 While these medications, when they work properly, block memory consolidation altogether there are other medications, namely beta-blockers that merely dampen the strength of a memory by reducing both its emotional intensity as well as its factual richness. Beta-blockers such as propranolol can reduce the risk of post-traumatic stress disorder (PTSD) without wiping out the memory in question. PTSD develops as a result of extensive amounts of stress hormones being released at the time of a traumatic event. When propranolol is taken shortly after a traumatic event this can disturb the consolidation of the traumatic memories and thus prevent the disorder from developing.9 An advantage of propranolol over other amnesia-inducing medications is that it can be used to target specific items of information encoded in the memory system. Arguably the most interesting method of forgetting is known as directed forgetting. Directed forgetting is impaired memory arising from an instruction to forget a learned piece of information. In a typical directed forgetting experiment, subjects are presented with a list of words (list 1). After the presentation of list 1, one group of subjects is instructed to forget the words on the list. The other group of subjects is instructed to continue remembering the list.10 Then both groups of subjects are presented a second list of words (list 2) and are asked to study the words. Next the subjects are asked to recall all of the previously presented items, including both to-be-forgotten and to-be-remembered words. These experiments show two things: first, a forget-instruction leads to poorer memory for the targeted material (list 1) than a remember-instruction, and second a forget-instruction sometimes leads to better memory for materials learned following that instruction (list 2).

On the Blameworthiness of Forgetting  247 These two effects are taken to be evidence for retrieval inhibition initiated by the forget-instruction (Bjork, Bjork, & Anderson, 1998). The instruction to forget triggers inhibitory mechanisms that temporarily reduce the accessibility of the to-be-forgotten information in memory, which explains why subjects who received the forget-instruction have a harder time to recall the words of list 1. Moreover, this inhibition reduces proactive interference from list 1 and thus facilitates retrieval of list 2 items. This is why subjects who have forgotten some of the words on list 1 remember more of the words of list 2 than the subjects who did not receive a forget-instruction and thus still remember more of the words of list 1.11 Forget-instructions have been shown to work not only for lists of random words but also for autobiographical material. Regardless of the kind of material studied and regardless of its emotional valence, directed forgetting instructions are surprisingly effective. In an experiment conducted by Joslyn and Oakes (2005) participants in the group that received the forgetinstruction recalled approximately 24% fewer personal events from the previous week than participants in the group that received the rememberinstruction. In an experiment conducted by Barnier et al. (2007) the forgetgroup recalled 11% fewer words from a list than the group that was asked to remember the words. It is also worth noting that this effect occurred after only a single effort on the part of the participants to voluntarily inhibit the retrieval of the said items. Barnier et al. therefore speculate that the repeated use of retrieval-inhibition techniques will make it impossible for a subject to recall a learned item not only in the short term but also in the long run: Our success in reliably inducing inhibition of recently recalled autobiographical memories in the laboratory with a simple, unrepeated procedure suggests that in everyday cognition, much more powerful effects might be present. Spontaneous and repeated use of a directed forgetting procedure on the same knowledge may induce much stronger and enduring inhibitory effects. In everyday life, there may be many opportunities for repeated directed forgetting, and inhibitory control of autobiographical memories may be both common and effective. (Barnier et al., 2007, 319, as cited in Matheson, 2013, 201–202) As was mentioned at the beginning of the section, the standard view has it that forgetting cannot be under our control in the sense that there are no non-drastic techniques for forgetting specific items. When this view is combined with the view that blame requires control on the part of the agent, then it follows that it is wrong to blame someone for forgetting something. This argument is problematic for at least two reasons. First, the studies on directed forgetting show that it is sometimes possible to make oneself forget something. That said, most forgetting takes place unintentionally and involuntary. Second, just because an agent does not have complete control

248  Sven Bernecker over something she does need not mean that she cannot be blamed for that which she does. Moral responsibility may not require control on the part of the agent. There will be more on this issue in Section 6.

4  Tracing Blame After having discussed techniques that allow us to control our forgetting I will now discuss the control we have over our ability to remember at a later point in time. There are things we can do to improve the likelihood that we will remember them later. Among these techniques is writing oneself reminder notes, keeping a calendar, getting enough sleep, and not overcrowding one’s memory with useless information. Alternatively one can simply think about the thing one wishes to remember later. For the less frequently a stored piece of information is retrieved from memory the more likely it is that it will be forgotten. As Michaelian notes, “the memory system renders records inaccessible (in part) according to their retrieval history” (2011: 420). As we saw before, the moral assessment of harmful forgetting crucially depends on what led the agent to forget. If the agent did not follow the steps we ordinarily follow when we want to make sure that we remember a certain piece of information later, then we hold him accountable to a greater extent than if the agent did everything in his power to try to remember and still forgot. Prima facie, this is baffling. How could it be appropriate to blame someone for forgetting something if she does meets neither the control nor the awareness condition at the time she forgets? According to the tracing account an agent can be morally responsible for a wrongdoing even if she does not meet the control condition or awareness condition at the time of action—provided the action can be causally traced back to an earlier choice or action and provided the agent met the control and awareness condition at the earlier time. Following H.M. Smith’s (1983: 547) classical terminology, we can say that the tracing interpretation of ignorant wrongdoing involves a sequence of two actions: a benighting act and a subsequent unwitting wrongful act. The benighting act is one “in which the agent fails to improve (or positively impairs) his cognitive position” and in virtue of which he subsequently performs the unwitting wrongful act. Talbert (2016: 141) summarizes the tracing account as follows: Tracing principles help us explain how a person can be responsible for an action when, at the time of action, she lacked some element of control or knowledge that seems crucial for moral responsibility. We substantiate the claim of responsibility in such a case by tracing the agent’s lack of control or knowledge back to decisions that were made when the agent’s abilities (and/or knowledge) were not compromised. Thus, the drunk driver is morally responsible because the actions he takes while impaired can be traced back to unimpaired choices.12

On the Blameworthiness of Forgetting  249 The upshot is that we are blameworthy for our ignorant and unintentional conduct if we are blameworthy for our ignorance or unintentionality. To see the tracing theory in action, consider two agents, A and B, who have both forgotten the birthdays of their respective partners. But while it is the first time that agent A forgot his partner’s birthday, agent B has often in the past forgotten the birthday of his partner. Intuitively agent B is more blameworthy than agent A because he is, or at least could be, aware of his tendency to forget birthdays and is therefore expected to take extra precautions to prevent these memory failures from reoccurring. The upshot is that if an agent has epistemic access to the fact that he easily forgets birthdays, then it is reasonable to expect him to take measures that will remind him and to actively search for momentarily inaccessible memory items. And when he fails to take these extra steps to remember the birthdays of close friends and family and then forgets them, he is blameworthy for forgetting. The blame can be traced back to an earlier time when the agent knew the birthday and had the choice of setting up reminders that would have prevented the forgetting. Needless to say, the degree of blameworthiness for harmful forgetting has to do, among other things, with the relationship between the agent and the harmed individual(s). To see this reconsider the forgotten birthday case. When the person in question is your partner, you are expected to work harder to remember her birthday than if she is, say, a distant relative. The idea here is that one has special obligations to those to whom one stands in some special relationship, e.g., one’s friends, family members, colleagues, and those to whom one has made promises or commitments of some sort. The tracing view is a powerful strategy for explaining why it is reasonable to hold a wrongdoer morally responsible even if she is not aware of doing anything wrong and when she has no control over what she is doing. However, it is questionable that the tracing view captures all cases of blameworthy forgetting. A number of authors maintain that there are cases of wrongdoing where the agent is intuitively responsible even though the wrongdoing cannot be traced back to a past act or omission that grounds the agent’s present moral responsibility. Matthew Talbert, for instance, holds that “it is at least possible that some forgetful or inattentive agents are blameworthy without ever having committed a past culpable action” (2016: 147). To get a sense for the limits of the tracing account consider the following example:13 Forgotten Baby. Jill is a single mother of a 7-year-old boy and a 10-month-old girl. She has to pick up her son from elementary school. Usually the baby is at home while she does the school run. However, today the babysitter cancelled as Jill is walking out the door, and so she takes the baby along. Although it is very hot, the pick-up has never taken more than a few minutes, so Jill leaves the sleeping baby in the car while she goes to gather her son. This time, however, Jill

250  Sven Bernecker is greeted by a tangled tale of misbehavior, ill-considered punishment, and administrative bungling which requires several hours of indignant sorting out. During that time, the baby languishes, forgotten, in the locked car. When Jill and her son finally make it to the parking lot, they find the baby unconscious from heat prostration.14 The first thing to notice about this case is that there is a lot more at stake than in the forgotten birthday case discussed earlier. Though it can be quite upsetting when a partner forgets one’s birthday, the degree of harm inflicted on a person by not remembering her birthday does not compare with the degree of harm inflicted on a baby by actively endangering its life. Everything else being equal, when there is a lot at stake, we expect the agent to take more precautions to not forget the memory item in question than when the stakes are low. In a situation where the consequence of forgetting something could result in someone’s death, the agent must double and triple check to make sure that he has not forgotten. What qualifies as an adequate procedure for remembering the birthdays of friends and family may not count as an adequate procedure for remembering things that are more important. The importance of a memory item is in part a function of the degree of harm caused if the said item is forgotten. At first sight, it might seem that the forgotten baby case is perfectly compatible with the tracing account. Jill’s culpability for the baby’s heat stroke can be traced back, say, to her culpability for not setting an alarm on her cell phone while she is away from the car. Alternatively, we can trace Jill’s culpability back to her decision to take the baby on the school run in the first place. She should have asked a neighbor to watch her baby while she picks up her son from school. It is important to realize, however, that while these are perfectly good explanations for why Jill is culpable for endangering the life of the baby they do not explain the blameworthiness of Jill’s forgetting. To drive this point home, suppose that we trace Jill’s culpability for the baby’s heat stroke back to her decision to not set an alarm on her cell phone. In this case, we are blaming Jill for intentionally bringing about a situation that required her to remember that the baby is in the car. But to blame Jill for deciding to rely on her unaided memory is one thing; to blame her for failing to remember is quite another. For the trace account to explain the blameworthiness of Jill’s forgetting we have to stipulate that Jill knew, or should have known, on the basis of past experience, that she easily forgets important things. If she knew this about herself and if she nevertheless decided to rely on her unaided memory in a life-and-death situation, then and only then can we trace her forgetting to what H.M. Smith calls an ‘benighting act,’ i.e., an act the subject has awareness of, and control over, and that gives rise to the wrongful act over which the subject has no control because she is unaware of committing the act. However, if, prior to the wrongful act, Jill’s memory for important things had never failed her, then she does not have compelling reasons to

On the Blameworthiness of Forgetting  251 mistrust her ability to remember these things. And if she doesn’t have good reasons to mistrust her ability to remember important things, the tracing account cannot explain the blameworthiness of her forgetting. In other words, the tracing account can explain the blameworthiness of harmful forgetting only if it assumed that the subject is, or has good reasons to be, aware of the reliability of her memory to be below average. It seems undeniable that there are cases of harmful forgetting where, prior to the subject unwittingly committing the wrongful act, she has no good reasons to question the reliability of her memory. The Forgotten Baby case can be spelled out to fit this mold. But this raises the question of how we can explain the blameworthiness of cases of harmful forgetting where the wrongful act is not traceable to a benighting act? The following two sections are concerned with explanations of moral responsibility in nontracing cases of forgetting.

5  The ‘Rational Man’ vs. You and Me The literature on culpable ignorance contains two accounts of the blameworthiness of nontraceable cases of forgetting: the liberalized awareness condition of moral responsibility and attributionism. This section is a discussion of the liberalized awareness condition. Attributionism is the topic of the following section. We can account for the intuition that Jill is blameworthy for forgetting the baby by modifying the awareness condition on moral responsibility. According to the version of the awareness condition stated in Section 2, an agent is blameworthy for an action only if she is aware of what she is doing. Jill doesn’t meet this condition at the time she commits the wrongful act because she is oblivious of having left the baby in the parked car. For the awareness condition to be applicable to the forgotten baby case the condition needs to be strengthened in something like the following way: Liberalized Awareness Condition: S is morally responsible for action A only if (i) S is aware of doing A (under the appropriate description) or a rational person in S’s situation would be aware of doing A (under the appropriate description) and (ii) S does A at least partly on the basis of the belief that A is morally wrong (in the case of a blameworthy action) or permissible/obligatory (in the case of a praiseworthy action).15 Applied to the forgotten baby case, the idea is that Jill is blameworthy for the baby’s heat stroke because a rational person in her situation would have remembered the baby in time. The motivation behind the liberalized awareness condition is expressed in the following quotation: Ignorance, whether circumstantial or normative, is culpable if the agent could reasonably have been expected to take measures that would have

252  Sven Bernecker corrected or avoided it, given his or her capabilities and the opportunities provided by the social context, but failed to do so either due to akrasia or due to the culpable, nonakratic exercise of such vices as overconfidence, arrogance, dismissiveness, laziness, dogmatism, incuriosity, self-indulgence, contempt, and so on. (FitzPatrick, 2008, p. 609) The liberalized awareness condition raises two kinds of issues. First, there is the issue of what, if anything, a rational person may forget. Second, there is the issue of what characterizes a rational person in the agent’s situation and how much the rational person would remember in a specific situation. I will discuss these issues in turn. Forgetting has negative consequences but it also serves valuable functions. Forgetting promotes subjective well-being by limiting access to negative memories and by reducing unpleasant affect; it orients information processing for the present and the future by ensuring that beliefs are current and updated; and it provides a basis for obtaining semantic and procedural beliefs by allowing for abstraction and automatization (Nørby, 2015). Yet despite the fact that forgetting has adaptive value, it might still be irrational. The claim that forgetting is irrational is due to John Broome (2013: 178), who maintains that rationality demands persistence of belief as well as persistence of intention. The former requirement states that it is irrational to abandon a belief unless one has a sufficient reason to let go of it. And the persistence of intention requirement states that, if you have an intention, then it is irrational to stop having that intention without a sufficient reason. Broome further holds that forgetting does not provide a sufficient reason for ceasing to have a certain belief or a certain intention and that the persistence conditions therefore require that we do not forget. Broome’s claim that forgetting is irrational rests on the idea that any change in an agent’s confidence with respect to p must be the result of some change in the evidence the agent has for the belief that p. Given this view, it is rational for you to abandon the belief in p only if you have acquired new evidence that suggests that not-p. This position is plausible only in so far as we focus on intentional changes in a person’s confidence with respect to p. But a change in confidence with respect to p that is the result of having forgotten p usually takes place unintentionally (cf. Section 3). And to the extent that forgetting is a passive mental event, it is not irrational. As Williamson (2000: 219) states: “[unintentional] forgetting is not irrational; it is just unfortunate.” A rational agent may not intentionally forget anything, but he may unintentionally forget something. Rationality does not require perfect memory. Granted that unintentional forgetting is not irrational, the question arises of how much a rational person would remember in a specific situation. According to the liberalized awareness condition, an agent is responsible for the harm that arises from his having forgotten something if a rational

On the Blameworthiness of Forgetting  253 person would have remembered the thing in question. Another way of putting this point is to say that an agent is blameworthy for the harm caused by his forgetting p if, had he actively searched for p in his memory, he would have been able to recall p. The problem, however, is to give a general specification of the kinds of activities an agent has to undergo to qualify as having actively searched for a memory item. In Section 4, we saw that the kinds of activities that we expect an agent to engage in so as to recall something depend, among other things, on the importance of the memory item (where the importance is a function of the degree of harm caused if the said item is forgotten). The point is that I don’t see how we could say with any reasonable degree of precision what kinds of memory-jogging activities a rational person (as opposed to you and me) would engage in so as to remember something and how much the rational person would be able to remember. Yet the fact that the proposal under consideration can’t give a definite verdict even in a simplified hypothetical case like Forgotten Baby seems to be a good reason to rule it out.

6  Attributability vs. Control Let us step back and remind ourselves of where we are in the dialectical structure. We started out explaining the blameworthiness of harmful forgetting by tracing it to an earlier act the subject had awareness of and control over. The earlier act gives rise to the unwitting forgetting over which the subject has no control. We saw that the problem with the tracing account is that it can only explain the blameworthiness of forgetting if it is assumed that the agent is, or has good reasons to be, aware of the reliability of her memory to be below average. This gave rise to the question of how to account for the blameworthiness of forgetting in cases where the tracing account does not get a foothold because the subject has no good reasons to question the reliability of her memory. One strategy is to weaken the awareness condition so that it is met by the subject while she forgets the item in question and brings about harm. Another strategy, to be discussed in this section, is to develop a notion of moral responsibility that is completely independently of the awareness and control conditions. If it can be shown that moral responsibility requires neither awareness nor control, then it would seem that the moral assessment of harmful forgetting poses no special problem. Attributionism is the view that agents need not meet the control and awareness conditions to be properly blamed or praised for their behavior. According to attributionism, for an agent to be blameworthy for his wrongdoing all that is needed is that the behavior is attributable to the agent and to his orientation toward other people. An action is attributable to the agent if it is expressive of his emotional reactions, spontaneous attitudes, and values. Even though the emotional reactions and attitudes are nonvoluntary they give a good insight into the agent’s moral personality. Sometimes the

254  Sven Bernecker picture provided by these reactions and attitudes is even better than that provided by the agent’s voluntary actions, which may be performed for strategic reasons so as to disguise his real feelings.16 To see attributionism at work, consider once again the forgotten birthday case. Forgetting one’s partner’s birthday is usually not the product of an act that is intentionally aimed at bringing about the memory lapse. Moreover, if the subject has never before forgotten about his partner’s birthday then, prior to the incident, he has no good reason to doubt his ability to remember the birthday without relying on external tools such as calendar entries. But if he has no good reasons to question his ability to remember his partner’s birthday, then the tracing account doesn’t get a foothold. To the extent that we think that the memory lapse is nevertheless blameworthy we have to resort to an account of blameworthiness that differs from the tracing account. According to attributionism, the memory lapse is blameworthy if it is expressive of the agent’s orientation toward other people, in this case his partner. It would, for example, be expressive of the agent’s orientation toward his partner if he generally didn’t care much about her and if he neither acknowledged his fault nor apologized upon learning about the memory lapse. If the agent didn’t feel the need to apologize, he would manifest disregard toward the way his actions affect his partner’s feelings. Sher (2009: 24) illustrates attributionism by means of an example of a dog (as opposed to a baby) being forgotten in a parked car by its owner called Alessandra. Sher labels this case Hot Dog. Alessandra, Sher argues, is blameworthy for the dog’s heat stroke but the reason she is blameworthy is not (as the tracing view would have it) that she “negligently failed to do something that would have prevented her from forgetting [the dog]” (2009: 36). Instead Alessandra is blameworthy for harming the dog because her behavior has fallen below the standard of care a dog owner owes to her pet and because the behavior is reflective of an underlying blameworthy evaluative attitude. Sher writes: “if . . . Alessandra were less solicitous of her children, or was made less anxious by conflict—then she would not have forgotten about the dog” (2009: 92). Thus for the attributionist to be able to ascribe blame in a case of harmful forgetting he has to argue that the harmful behavior is representative of the agent’s moral character in general and her orientation toward the harmed individual in particular. There are no doubt cases where the attributionist strategy is appropriate; but there are also instances of harmful forgetting where the agent’s orientation toward other people leaves nothing to be desired. To see this consider a version of Hot Dog where Alessandra loves her dog dearly and where she has, up to the incidence, been a model dog owner. The attributionist strategy seems ad hoc in cases where an agent has no history of wrongdoings of the specific kind and where the harmful behavior does not fit with the agent’s overall character. Why should we have to accuse Alessandra of being too

On the Blameworthiness of Forgetting  255 solicitous of her children or of not genuinely loving her dog just so that we can blame her for the one time she has forgets the dog in the car? Why do we have to suppose that the wrongdoing is always reflective of the agent’s character? The worry is that the attributionist account of the blameworthiness of forgetting may tempt us to misinterpret the agent’s moral character for the sake of being able to attribute blame for the wrongdoing she committed.

7 Conclusion It is a mistake to think that we cannot be morally responsible for forgetting because, as a matter of principle, forgetting is outside of our control. Sometimes we do have control over our forgetting. When forgetting is under our control there is no question that it is the proper object of praise and blame. But we can also be morally responsible for forgetting something when it is beyond our control that we forget that thing. The literature contains three accounts of the blameworthiness of forgetting over which the agent has no control—the tracing account, the liberalized awareness condition, and attributionism. The point of the chapter was to examine each of these accounts and spell out the pros and cons. Even though these are competing accounts of the blameworthiness of harmful forgetting they are compatible with one another. In particular, it is possible to come up with a position that endorses the tracing account for certain kinds of harmful forgetting and attributionism for other kinds of harmful forgetting. But this combination of the tracing account and attributionism is the topic of another chapter.17

Notes 1 For a conceptual analysis of forgetting, see Frise (2018). 2 Paired-associate learning involves the pairing of two items (usually words)—a stimulus and a response. For example, words such as ‘candle’ (stimulus) and ‘table’ (response) may be paired, and when the learner is prompted with the stimulus (‘candle’), he responds with the appropriate word (‘table’). 3 In Section 5, we will see that proponents of attributionism challenge the idea that moral responsibility requires awareness. 4 This condition is modeled after Haji (2008: 90). 5 The control and the awareness conditions of moral responsibility are, of course, connected in the sense that if an agent does not know what she is doing, then she does not have (complete) control over her behavior (Talbert, 2016, 154 n10). 6 Harm brought about by forgetting is, of course, not the only putative counterexample to the awareness condition of moral responsibility. Forgetting is a kind of ignorance. Besides forgetting ignorance can be brought about by intoxication, distraction, fatigue, and intense emotions such as rage. 7 The notion of an epistemically non-drastic technique for forgetting is borrowed from Matheson (2013, pp. 197–8). 8 Cf. Caine et al. (1981), King (1992), Pastalkova et al. (2006), and Shema et al. (2007), all cited in Liao (2017, p. 374).

256  Sven Bernecker 9 See Pitman et al. (2002) and Vaiva et al. (2003). For a discussion of the ethical issues connected with the use of beta-blockers for curing PTSD see Evers (2007) and Kober (2006). 10 See Bjork (1989, Bjork and Bjork (1996), and MacLeod (1998). The description of the experiment is taken from El Haj et al., 2011, p. 994. 11 MacLeod (1998) claims that sometimes list 1 forgetting occurs without an enhancement of the memory for items of list 2. 12 Talbert (2016, pp. 130–141) distinguishes between different versions of the tracing view. For a discussion of tracing views see: Peels (2011), H.M. Smith (2011), Timpe (2011), and Vargas (2005). 13 This example is modeled after an example by Sher (2009: 24). In Sher’s example, it is a dog instead of a baby that is forgotten in the car. 14 Since 1998, 376 children have died in the U.S. of heatstroke because they were forgotten by caregivers in cars (see http://noheatstroke.org/). The psychologist David Diamond talks about the forgotten baby syndrome. According to Diamond, when people perform habitual actions, they operate on autopilot. When in fact things are different, the brain creates an alternative reality that matches how things usually are when the person engages in a habitual action. The tragic memory lapse dubbed ‘forgotten baby syndrome’ occurs when an agent is guided by that alternative reality (see http://psychology.usf.edu/faculty/data/ddiamond/ baby-sy.pdf). 15 The concept of a reasonably intelligent and impartial person goes back to antiquity (cf. Lucas, 1963). Component (i) of the liberalized awareness condition is similar to Vargas’ formulation of the awareness condition: “For an agent to be responsible for some outcome (whether an action or consequence) the outcome must be reasonably foreseeable for that agent at some suitable prior time” (Vargas, 2005, p. 274). 16 H.M. Smith, 2011, pp. 118–9; Talbert, 2016, pp. 52–61. Among the proponents of attributionism are Frankfurt (1988), Scanlon (1998), Sher (2009), and A. Smith (2005). The negation of attributionism is called volitionism. Among the proponents of volitionism are Sidgwick (1907, p. 59–61), Taylor (1970, pp. 241–52), and Wallace (1994, pp. 131–2). 17 For comments on a previous draft and for conversations related to the topic of the paper I am deeply grateful to Maura Priest and two anonymous reviewers. I would also like to acknowledge the financial support from an Alexander-vonHumboldt Professorship Award.

References Alfano, M. (2016). Moral psychology: An introduction. London: Polity. Altmann, E. M., & Gray, W. D. (2002). Forgetting to remember: The functional relationship of decay and interference. Psychological Science, 13, 27–33. Anscombe, G. E. M. (1957). Intention. Oxford: Blackwell. Baddeley, A. (2014). Essentials of human memory. New York: Psychology Press. Barnier, A. J., Conway, M. A., Mayoh, L., Speyer, J., Avizmil, O., & Harris, C. B. (2007). Directed forgetting of recently recalled autobiographical memories. Journal of Experimental Psychology: General, 136, 301–322. Birch, C. (2000). Memory and punishment. Criminal Justice Ethics, 19, 17–31. Bjork, E. L., & Bjork, R. A. (1996). Continuing influences of to-be-forgotten information. Consciousness and Cognition, 5, 176–196. Bjork, E. L., Bjork, R. A., & Anderson, M. C. (1998). Varieties of goal-directed forgetting. In J. M. Golding & C. M. MacLeod (Eds.), Intentional forgetting: Interdisciplinary approaches (pp. 103–137). Mahwah, NJ: Lawrence Erlbaum.

On the Blameworthiness of Forgetting  257 Bjork, R. A. (1989). Retrieval Inhibition as an adaptive mechanism in human memory. In H. L. Roediger & F. I. M. Craik (Eds.), Varieties of memory and consciousness: Essays in honor of Endel Tulving (pp. 309–330). Hillsdale: Lawrence Erlbaum. Bjork, R. A., & Vanhuele, M. (1992). Retrieval inhibition and related adaptive peculiarities of human memory. In J. F. Sherry & B. Sternthal (Eds.). Advances in consumer research (Vol. 19, pp. 155–160). Provo: Association for Consumer Research. Broome, J. (2013). Rationality through reasoning. London: Wiley-Blackwell. Caine, E. D., Weingartner, H., Ludlow, C. L., Cudahy, E. A., & Wehry, S. (1981). Qualitative analysis of scopolamine-induced amnesia. Psychopharmacology (Berl.), 74, 74–80. Driver, J. (2009). Memory, desire, and value. In C. Grau (Ed.), Eternal sunshine of the spotless mind (pp. 80–93). London: Routledge. El Haj, M., Postal, V., Le Gall, D., & Allain, P. (2011). Directed forgetting of autobiographical memory in mild Alzheimer’s disease. Memory, 19, 993–1003. Evers, K. (2007). Perspectives on memory manipulation: Using beta-blockers to cure post-traumatic stress disorder. Cambridge Quarterly of Healthcare Ethics, 16, 138–146. FitzPatrick, W. J. (2008). Moral responsibility and normative ignorance: Answering a new skeptical challenge. Ethics, 118, 589–613. Frankfurt, H. (1988). The importance of what we care about. In H. Frankfurt (Ed.), The importance of what we care about (pp. 8–94). Cambridge: Cambridge University Press. Frise, M. (2018). Forgetting. In D. Debus, D. Perrin, & K. Michaelian (Eds.), New directions in the philosophy of memory (ch. 11). London: Routledge. Haji, I. (2008). Incompatibilism’s allure: Principle arguments for incompatibilism. Ontario: Broadview Press. Hardt, O., Nader, K., & Nadel, L. (2013). Decay happens: The role of active forgetting in memory. Trends in Cognitive Sciences, 17, 111–120. Joslyn, S. L., & Oakes, M. A. (2005). Directed forgetting of autobiographical events. Memory and Cognition, 33, 577–587. King, D. J. (1992). Benzodiazepines, amnesia and sedation: Theoretical and clinical issues and controversies. Human Psychopharmacology, 1, 79–87. Kober, A. J. (2006). Therapeutic forgetting: The legal and ethical implications of memory dampening. Vanderbilt Law Review, 59, 1561–1626. Liao, S. M. (2017). The ethics of memory modification. In S. Bernecker & K. Michaelian (Eds.), The Routledge handbook of the philosophy of memory (pp. 373–382). London. Routledge. Lucas, J. R., (1963). The philosophy of the reasonable man. Philosophical Quarterly, 13, 96–106. MacLeod, C. M. (1998). Directed forgetting. In J. M. Golding & C. M. MacLeod (Eds.), Intentional forgetting: Interdisciplinary approaches (pp. 1–57). Mahwah, NJ: Lawrence Erlbaum. Matheson, D. (2013). A duty of ignorance. Episteme, 10, 193–205. Michaelian, K. (2011). The epistemology of forgetting. Erkenntnis, 74, 399–424. Nørby, S. (2015). Why forget? On the adaptive value of memory loss. Perspectives on Psychological Science, 10, 551–578. Pastalkova, E. Serrano, P., Pinkhasova, D., Wallace, E., Fenton, A. A., & Sacktor, T. C. (2006). Storage of spatial information by the maintenance mechanism of LTP. Science, 313, 1141–1144.

258  Sven Bernecker Peels, R. (2011). Tracing culpable ignorance. Logos and Episteme, 2, 575–582. Pitman, R., Sanders, K. M., Zusman, R. M., Healy, A. R., Cheema, F., Lasko, N. B., Cahill, L., & Orr, S. P. (2002). Pilot study of secondary prevention of posttraumatic stress disorder with propranolol. Biological Psychiatry, 51, 189–192. Scanlon, T. M. (1998). What we owe to each other. Cambridge, MA: Harvard University Press. Shema, R., Sacktor, T. C., & Dudai, Y. (2007). Rapid erasure of long-term memory associations in the cortex by an inhibitor of PKM zeta. Science, 317, 951–953. Sher, G. (2009). Who knew? Responsibility without awareness. New York: Oxford University Press. Sidgwick, H. (1907). The methods of ethics. Chicago: University of Chicago Press. Smith, A. (2005). Responsibility for attitudes: Activity and passivity in mental life. Ethics, 115, 236–271. Smith, H. M. (1983). Culpable ignorance. Philosophical Review, 92, 543–571. Smith, H. M. (2011). Non-tracing cases of culpable ignorance. Criminal Law and Philosophy, 5, 115–146. Talbert, M. (2016). Moral responsibility. Cambridge: Polity. Taylor, R. (1970). Good and evil. New York: Palgrave Macmillan. Timpe, K. (2011). Tracing and the epistemic condition on moral responsibility. The Modern Schoolman, 88, 5–28. Vaiva, G., Ducrocq, F., Jezequel, K., Averland, B., Lestavel, P., Brunet, A., & Marmar, C. R. (2003). Immediate treatment with propranolol decreases posttraumatic stress disorder two months after trauma. Biological Psychiatry, 54, 947–949. Vargas, M. (2005). The trouble with tracing. Midwest Studies in Philosophy, 29, 269–291. Wallace, J. R. (1994). Responsibility and the moral sentiments. Cambridge, MA: Harvard University Press. Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University Press.

13 Consent Without Memory1

Carl F. Craver and R. Shayna RosenbaumConsent Without Memory

Carl F. Craver and R. Shayna Rosenbaum

1 Introduction Can someone with episodic amnesia consent to participate in a scientific experiment? Episodic memory is widely regarded as the capacity to remember knowingly experiences of specific events from one’s personal past (Tulving, 1985). Individuals with severe episodic amnesia sometimes cannot episodically remember even a single such event and fail to lay down new episodic memories going forward. In some special cases, such individuals have maximally severe deficits in episodic memory while their intellectual ability and other cognitive functions (e.g., semantic recall, procedural learning, working memory) remain in control range (Rosenbaum et al., 2005). Such individuals afford the unique opportunity to investigate how episodic memory specifically contributes to other cognitive capacities and more generally, to the distinctive lives of persons. We can thus invert the opening question: What, if anything, does episodic memory contribute to the fact that persons can, and so deserve the right to, consent? Episodic memory can be thought of as having two components: a constructive component (Mullaly & Maguire, 2014; Rosenbaum et al., 2009; Rubin & Umanath, 2015; Schacter et al., 2012) for generating specific experiences out of stored representations across and among different modalities (visual, tactile, olfactory, etc.), and an autonoetic component, giving one the sense that one is re-experiencing something one did in the past (Tulving, 1983; Mahr & Csibra, forthcoming). Individuals with textbook episodic amnesia can no longer reconstruct or re-experience details of their personal pasts in memory (Tulving, 1985; Rosenbaum et al., 2005). In both science and popular culture, the ability to re-live our personal pasts in memory is accorded a central position in our lives as persons. Among cognitive scientists, episodic memory is increasingly associated or identified with other capacities that are similarly, intuitively central: the capacity to imagine personal futures (e.g., Atance & O’Neill, 2001; Klein, Loftus, & Khilstrom, 2002; Michaelian, 2016), to imagine counterfactual pasts (Schacter et al., 2012; de Brigard, 2014), to engage in “mental time travel” (Tulving, 1985, 1993; Suddendorf & Corballis, 2007; Boyer, 2008),

260  Carl F. Craver and R. Shayna Rosenbaum and to have a subjective sense of temporality (Dalla Barba and La Corte, 2013). Individuals with episodic amnesia are frequently described as trapped in a “permanent present tense” (Corkin, 2013) with no sense of their pasts or futures. Oliver Sacks (1985), for example, described the life of “Jimmie G.,” a person with episodic amnesia resulting from Korsakoff’s syndrome, as a “Humean froth,” a “fluttering on the surface.” This popular view about the centrality of episodic memory to our lives as persons has yet to be tested. Neuropsychologists do not yet have a moral profile of individuals with different varieties of amnesia: a complex characterization of if, and if so, how the absence of episodic memory impacts on the capacities required for one to deserve, e.g., the right to consent. This is partly because we do not have a clear sense of which capacities are and are not required to deserve that right. But it is also partly because we are only beginning to understand what episodic memory allows us to do that could not be done with other memory or cognitive systems, and so what episodic memory contributes, in the here and now, to our capacities as agents or persons in the fullest senses of those terms. The case of consent provides a manageable domain in which to address this immense and important topic. The abstract, philosophical question of consent confronts the hard reality of practice every time a researcher visits an individual with cognitive deficits to perform even the most innocuous of experimental tests. A decision must be made whether the individual is capable of consenting, whether consent should be given by proxy, or, as in our group, whether consent should be obtained from both the participant and the proxy (the “dual consent” approach). Recent discussions of Suzanne Corkin’s approach to the case of Henry Molaissen (H.M.) have drawn public attention to the fact that our ordinary idea of “consent” is put under extreme pressure when working with cognitively impaired individuals (Dittrich, 2017, pp. 327–329; Begley, 2016; Levine, 2016). If individuals with episodic memory lack imagination, cannot simulate possible personal futures, cannot re-live their mistakes, cannot consider or evaluate counterfactuals, are trapped in an eternal present, as our best current theories of the function of episodic memory suggest, then they plausibly lack key skills required to make decisions for themselves. It would perhaps be better for a guardian to make the decisions on their behalf. Our group has relied on a dual consent approach precisely because it leaves open the possibility that the amnesic individual might be able to consent on their own. We begin (Section 2) by explaining why this approach does not, in fact, provide an ethically superior alternative to consent by proxy. The remainder of the chapter addresses whether episodic memory is required for consent. Section 3 argues from abstract considerations that consent plausibly does not require episodic memory or perhaps even memory more generally. In Section 4 we argue from empirical findings for the conclusion that individuals with amnesia might have the psychological capacities required to decide whether to consent for themselves to an experiment.

Consent Without Memory  261 We focus here on the case of Kent Cochrane (1951–2014; known in the scientific literature as K.C.), the best known and most studied example of someone with specific, episodic memory deficits (Rosenbaum, 2005; see also Branswell, 2014). Kent’s episodic amnesia gradually became apparent after a 1981 motorcycle accident in which he sustained significant brain trauma, including (but not limited to) key areas of the medial temporal lobes known to be involved in episodic memory. Kent’s episodic amnesia, which extended backward over his entire life and prevented him from forming new episodic memories going forward, was singular both in its severity (Kent could not remember a single personal experience) and in its apparent specificity. Yet what we learned from Kent has important implications for other individuals whose episodic amnesia may be less surgical and thorough.

2  Dual Consent? The dual consent approach involves two consentees in the consent process: e.g., Kent and his mother, Ruth. This approach is designed to solve a practical dilemma that arises in working with individuals with cognitive impairments. Namely, we do not know, before testing him, whether Kent is able to consent. If we give him that right when he does not deserve it, we subject him to the very risks consent statutes were designed to avoid. However, if we deny him that right when he deserves it, we rob him of the very status consent statues were designed to protect. Neither choice is obviously correct, and either could do him harm. The dual consent approach promises a third way: If Kent can consent (and has been properly informed and is acting voluntarily), then he consents when he signs the document. Ruth’s approval is ethically superfluous (even if it is practically or legally beneficial). If Kent cannot consent, Ruth’s consent takes center stage. Kent’s would-be “consent” is transformed into “mere assent,” i.e., mere expressed willingness to go along with the plan. If Kent doesn’t assent, or wants to stop the testing at some later time, that overrides anything he or Ruth said before. The dual consent approach addresses the practical dilemma by covering the bases, allowing Kent to consent if he can, and allowing a proxy to do so if he cannot. Does the dual consent approach uphold the values consent statutes were designed to protect? The Nuremberg Code was adopted in the face of Nazi atrocities to place standards on the use of human subjects in scientific experiments. Two main values stand out in this document: the prevention of abuse and the protection of autonomy (i.e., the ability to make decisions for oneself). It is also clearly designed to safeguard public trust in science. The dual consent approach arguably protects Kent from abuse better than a policy of allowing him to consent for himself. It must do so at least as well as allowing Ruth to consent on his behalf. And either of these ­options is clearly preferable to the illegal option of leaving the decision to the experimenter. Dual consent interposes two layers of consent between

262  Carl F. Craver and R. Shayna Rosenbaum the experimenter and the participant, making it less likely that a harmful experiment will be allowed to proceed and less likely that any abuses will occur than if the participant makes the decision alone. For this reason, the dual consent approach also safeguards public trust; it encourages the sense that researchers and clinicians care about protecting potentially incapacitated participants from harm. Because the policy is flexible and easy to apply, it makes for a convenient way to handle the contingencies that might arise in any particular individual, a particular individual over time, or across very different individuals with overlapping deficits. Finally, the dual consent model has the virtue of not delivering potentially harmful self-knowledge or, worse, inaccurate and injurious moral psychological diagnoses. Compared to proxy consent, the dual consent approach does not imply, in the very process of soliciting consent, that Kent is incapable of giving it. By remaining agnostic, the dual consent approach avoids introducing the perhaps unwarranted stigma of incompetence under the guise of protecting Kent from harm. In fact, however, besides this keeping up of appearances, there is no moral difference in Kent’s status between the dual consent approach and consent by proxy. This equivalence would be especially apparent if Kent and Ruth disagreed with one another about continuing an experiment (a scenario that never happened, to our knowledge). Under those conditions, the redundant control mechanisms so useful in achieving the utilitarian objectives discussed previously are pitted against one another. If Kent consents, and Ruth does not (perhaps because Kent enjoys the testing, and Ruth has better things to do), we are forced to address the very premise the dual consent approach is designed to sidestep: We have to make a decision about whether Kent can consent on his own. If he cannot, the dual consent model delivers the right verdict. But if he can, Ruth’s opinion is no more relevant than the opinion of any other mother about what her 60-year-old son does. He should speak for himself. This problem reveals something intellectually unsatisfying about the dual consent approach even when Kent and Ruth agree. If Kent is competent to make decisions for himself, the necessity of Ruth’s consent enshrined in the dual consent model is not merely gratuitous; it renders Kent’s retained decision-making power illusory. In fact, if Kent had thought it through, he might have recognized the dual consent model as cold comfort or worse. So the dual consent approach does not really sidestep the ethical question. We have to ask directly whether someone robbed of their episodic memory capacity, and all that goes with it, might nonetheless deserve the right to consent.

3  Consent Without Memory Intuitively and prereflectively, memory seems to play a crucial role in keeping track of what one has consented to and what one has not. If Kent

Consent Without Memory  263 cannot possibly remember giving his consent, one might reasonably surmise, he cannot keep track of what he has consented to and what he has not. The relevant thread between the moment of consent and its moment of fruition is severed, and “consent” loses all meaning (Dittrich, 2016, pp. 329–331). This intuition, however, does not fit with things we already know about consent. For example: It is typically given in writing. We have this practice precisely because we know our memories often fail. The document provides evidence of a binding agreement undertaken at the moment of consent. Should doubts arise as to whether consent was granted or about what ­exactly one consented to, the document legally trumps the frail constructions of memory. This mundane fact should lead us to question whether the thread binding the moment of consent and its moment of fruition must be wound from episodic memory specifically. If the bond between these times might be constituted by a piece of paper, there is little reason to think consent requires that one be able to reactivate a vivid, experiential memory of the moment consent was given. We do not require such things for monthly bank account deductions to fund public radio, for marriages, or for sex. We do not videotape or photograph such moments to preserve the consent in their fine detail. If episodic detail is so unnecessary in our ordinary dealings with consent, why should it take center stage when we think about consenting to a scientific experiment? This way of thinking will no doubt be congenial to externalists in the philosophy of mind (Clark & Chalmers, 1998). The persistence conditions for consent need not be satisfied “inside the skull” but might just as easily be borne by a piece of external scaffolding. From this perspective, the thread between the moment of consent and its fruition might equally be written on paper, stored in digital traces, or tattooed across one’s chest. Perhaps Kent, left to his own devices, could not consent; but perhaps with Ruth’s help, a pencil and paper, or a mobile phone, he could (see Fluharty & Priddy, 1993; Sohlberg & Mateer, 1989). A second mundane fact, however, undermines even the idea that consent requires a thread between the moment of consent and the moment of fruition: We routinely require and give consent for actions we cannot possibly remember, think about, or understand at the moment of fruition. We consent to surgery during which we will be unconscious. We consent to having our organs harvested when we will be dead. In neither case is it even possible to remember consenting when the moment of fruition arrives. Nor is there any knowledge the consentee must entertain (or, arguably, possess). Indeed, the consenter might no longer even exist. Consent, it would seem, can outlast our thoughts, our mental lives, and even our bodies (cf. Nagel, 1970). If so, the key causal players in consent occur at the time consent is granted. In giving one’s consent, one thereby (immediately) permits someone to do something that they were not previously permitted to do. That change in one’s ethical relations to others is not, in the first instance, a change in the

264  Carl F. Craver and R. Shayna Rosenbaum causal structure of the world. It is rather a change in its ethical structure: in what is permissible and what is not. To understand how persons are related to the causal mechanisms studied in the psychological and brain sciences, we must cross a divide between two very different ways of “keeping track of” human beings and their relations. We must relate a world of norms to a world of causes (Haugeland, 1993). From the normative perspective, persons are nodes in a network of commitments, obligations, permissions, deservings, and the like, that make some future possibilities permissible and others not. We should not be looking for the material basis of consent’s persistence but rather the material conditions for the existence of consenters in the first place. We should rather ask: What kinds of causal mechanisms must nature assemble to create creatures that properly can be said to, and so to deserve the right to, consent for themselves?

4 Consent in Amnesia: Informed, Voluntary, and Fully Capacitated Even if consent requires no memorial or material bridge to persist, it is possible that Kent was cognitively impaired in ways that prevented him from being fully informed, from consenting voluntarily, or from being fully capacitated to make decisions for himself. As psychologists and philosophers have identified episodic memory with other, far-reaching cognitive capacities, it has become easier to suppose that a person with episodic amnesia must be critically debilitated in capacities required for giving autonomous consent. The available evidence, however, does not support this supposition. 4.1  Informed Consent Did Kent’s episodic amnesia prevent him from being informed or from understanding what he was being asked to do? Anecdotally, nothing in our experience raised any doubts on this front. Kent never displayed any difficulty understanding the consent process or any of the tests he was being asked to perform. He was regularly able to describe the test in his own words and appeared to grasp the meagre risks and benefits in each case. On one occasion, Kent was able to explain what consent is and how consent works. Without prompting, he repeated his right to withdraw. This is precisely what one would expect given that Kent’s semantic and procedural knowledge from before his accident were largely preserved (Rosenbaum et al., 2005; Tulving, 1985). The same symptom profile has now been observed in many individuals; indeed, some individuals with episodic memory deficits resulting from perinatal trauma have graduated high school and held jobs (e.g., Vargha-Khadem, Gadian, & Mishkin, 2001). There is, in short, no reason to suspect that Kent should fail to understand what he was being asked to do or what he was being asked to give up in the consent encounter.

Consent Without Memory  265 One might wonder, however, whether a conceptual grasp of the consent situation suffices for a complete understanding of what consent involves. If, for example, amnesic individuals are trapped in a permanent present tense (Corkin, 2013) or lack a subjective temporality (Dalla Barba and La Corte, 2013), they arguably cannot fully appreciate that they are being asked now to permit a future act. Or perhaps they lack the requisite understanding of time (Hoerl, 1999). Individuals with episodic amnesia indeed have difficulties generating details of future personal events in direct proportion to the extent of their amnesia (Kwan et al., 2010; Rosenbaum, 2005). Kent, for example, could not generate a single future scenario involving himself (Tulving, 1985). Asked what it felt like to try to answer the question, he described it as “like being asleep,” “blank,” or like being in an empty room and, “having a guy tell you to go find a chair.” Despite these very severe deficits, a test of Kent’s temporal knowledge revealed no significant deficits. He could list significant life events (e.g., his birth, his graduation, a car accident, a train wreck) and order them on a timeline in roughly the correct sequence. That is, he understood what J.M.E. McTaggart (1908) called the B-series: time construed as a series of events arranged earlier or later than one another. Kent’s semantic grasp of the B-series also extends into the future. Individuals with amnesia can predict future events (e.g., their deaths, global resource conflicts, climate change) despite being unable to construct vivid personal scenarios involving those events. This dissociation is also seen in neurotypical individuals (Klein, Loftus, & Kihlstrom, 2002). Kent also clearly understood McTaggart’s A-series: time ordered as past, present, or future relative to an indexical “now” (contra Hoerl, 1999; but see McCormack & Hoerl, 1999). On a battery of questions designed to probe his grasp of the A-series, Kent correctly understood the concepts of past, present, and future and understood, for example, that the past is irrevocable and that the present influences the future and not vice versa (Craver et al., 2014). He was, in this respect, like other individuals with episodic amnesia (Klein, 2002). Semantic knowledge of the past, present, and future, in short, allows individuals with episodic amnesia subjectively to escape a permanent present tense. In fact, this finding has now been replicated in individuals with developmental amnesia (Vargha-Khadem, unpublished data), indicating that one need never have had a fully functioning episodic memory system to acquire a conceptual grasp of the A-series or to understand the idea that one is consenting to a future event. Kent also had definite opinions about how he should orient his behavior with respect to time. We tested him and three other individuals with episodic amnesia on the Zimbardo Time Perspective Inventory (Zimbardo and Boyd 2015/1999), a battery of questions designed to assess the extent to which a person is positive or negative about the past, oriented toward the future, fatalistic about the present, and hedonistic about the present. Kent had no difficulty understanding the questions or generating meaningful answers.

266  Carl F. Craver and R. Shayna Rosenbaum Perhaps surprisingly, Kent scored in the first percentile on measures of present hedonism and only slightly higher for measures of fatalism (Kwan et al., 2013). That is, he positively disavowed temporal perspectives bound to the here and now. In fact, his highest scores showed positivity toward the past and a preference for valuing the future. None of the other individuals with amnesia had hedonism or fatalism as their dominant orientations; in fact, none rose above the fiftieth percentile in these categories. It would appear, then, that one can maintain healthy and typical attitudes toward the past, present, and future in the absence of episodic temporal self-projection. Yet perhaps one could have this conceptual knowledge but be unable to use it for the purposes of making complex moral decisions, as would be required in a consent encounter. Perhaps episodic construction is required to imagine or assess moral actions (Casebeer & Churchland, 2003; Thagard, 2007; Darwin, 1871). To address this question, we tested Kent and eleven other individuals with episodic amnesia on a battery of moral scenarios designed to study moral decision-making in healthy adults (Greene et al., 2001). None of them had any difficulty understanding the questions. They each seemed to consider the advantages and disadvantages of the possible actions in each scenario (Craver et al., 2016). Consistent with Greene’s motivating hypothesis, Kent did show a significant bias toward utilitarian answers: he explained in each case that the good of the many outweighs the good of the few. While unusual outside academic philosophy, this response profile nonetheless showed that Kent could understand complex, highstakes, and high-conflict moral scenarios and reason his way to an answer comparable to the answers that, e.g., John Stuart Mill and Greene would give. Importantly, however, no other individual with amnesia displayed this utilitarian bias; in fact, one of them showed the opposite bias, i.e., toward deontological answers. The others fell between these extremes, like controls. The main point is that even individuals with episodic amnesia can decide what to do in complex moral situations, such as the situation one faces when giving consent.2 In sum, there is no evidentially grounded reason to suppose that an individual with episodic amnesia must lack the conceptual resources required to be informed in the consent proceedings. 4.2  Voluntary Consent Let us turn, then, to whether Kent was capable of consenting voluntarily. One might come to doubt his ability to do so by considering the implications of Kent’s amnesia for his identity over time, for his self-knowledge, or for his unique vulnerability to forms of coercion. Starting with Locke (1690), many philosophers have held that a person’s identity over time is constituted by threads of episodic memory. If those threads are severed, Kent1, at the moment of consent, is not the same person as Kent2, who engages in the experimental task. If so, Kent1’s

Consent Without Memory  267 voluntary consent is irrelevant to Kent2, who is forced to undergo the experimental test without having agreed to do so (see Dittrich, 2016). As intuitive as this idea is, the simple Lockean view of personal identity is viciously circular (see Butler, 1975/1736; Schechtman, 1990). For a memory to ground one’s identity over time, the memory has to be a veridical memory, not merely an apparent memory. And for the memory of an event to be veridical, the person has to have experienced the earlier event. So to remember an event veridically presupposes, and cannot ground, personal identity over time. A more serviceable view holds that diachronic identity is reducible to a complex web of interlocking causal connections between states of consciousness at different times (Parfit, 1984). These causal connections are not limited to episodic memories, but also include other kinds of memories (e.g., semantic or procedural memory), preserved beliefs and attitudes, emotions and moods, prolonged perceptual experiences and imaginings, etc. If so, then episodic memory does not contribute uniquely to the preservation of the self over time, and Kent’s identity might be maintained by causal threads independent of episodic systems. This intuitive idea that loss of memory severs one’s right to speak for one’s future self loses its justification (see Craver, 2012). But even if one cannot shake the Lockean intuition that Kent1 is not metaphysically identical to Kent2, that alone would not imply that Kent1 should be preventing from making decisions for Kent2. After all, Kent2 and Kent1 share all the same semantic knowledge, the same preferences, the same quirks of habit, and, not trivially, the same body. Even if they are distinct persons, in some technical, metaphysical sense, Kent1 would seem to have a better claim to speaking for Kent2 than anyone else. So given the choice between Kent1 and Ruth, Kent1 is arguably the better judge of what Kent2 would want. Still, one might wonder whether individuals with episodic amnesia really have sufficient self-knowledge to act voluntarily. To act voluntarily, one might think, is a matter of acting in accordance with your own beliefs and values. If episodic memories ground our knowledge of who we are, what we think, what we like, etc., then an individual with episodic amnesia might not know themselves well enough to act voluntarily, in the fullest sense of the word. In considering this possibility, one should note that there is no clear threshold for how much self-knowledge one has to have to act voluntarily. Unimpaired people vary considerably in the extent to which they form an explicit sense of self and in the accuracy of the self-conceptions they form (see Vazire & Carlson, 2011). Furthermore, people often act out of character voluntarily. Yet without settling these matters, the important question for us is whether episodic memory contributes something to the sense of self that cannot just as easily be contributed by one’s semantic self-knowledge. Selfknowledge can be stored and accessed episodically or semantically, and

268  Carl F. Craver and R. Shayna Rosenbaum these two forms of self-knowledge, as Kent helped to demonstrate, can be dissociated from one another (Rosenbaum et al., 2005; Tulving, 1993a). When we think of ourselves as ambitious, creative, kind, mischievous, or selfish, we subsume ourselves under abstract, semantic categories. Such categorical knowledge is plausibly involved in consent situations: the consentee must know themselves well enough to know what they could and could not tolerate being done to them, which values they would like to maximize, etc. Tulving repeatedly tested both Kent and Ruth on their knowledge of Kent’s categorical traits and desires, having them describe in separate tests Kent’s personality before and after the accident. When Tulving asked them to describe Kent’s personality in general and his personality before the accident, their answers matched 73%. Assuming Ruth was a good judge of Kent’s personality, these findings suggest that Kent has a fairly accurate conception of himself. It suggests further that his self-conception evolved to match his post-accident personality. His updating clearly did not rely on an episodic store of exemplars: Kent had no episodic memories. This feature of Kent’s amnesic syndrome accords well with independent evidence from other individuals with amnesia (e.g., Klein, Loftus, and Kihlstrom, 1996) and from unimpaired individuals (e.g., Klein, Sherman, & Loftus, 1996). Having made these points, it should be acknowledged that individuals with episodic memory deficits are clearly more dependent for their daily needs on caretakers than are neurotypical individuals. Their mobility is limited, and they may be more in need of social interaction outside the home (Davidson, 2012). They are likely accustomed to treating doctors with authority. They are perhaps likely to expect treatment benefits from the experiment even if they have been told to expect none. And in these ways, they are susceptible to influence by external factors that might impinge upon the voluntariness of their decisions. Yet it is not clear that the possible impingements in this case are any greater than those imposed on the undergraduate in need of spending money, a cancer patient out of standard treatment options, or a person consenting to a liver transplant. That is, Kent’s vulnerability along this dimension is not a consequence of the memory deficit per se but rather of his being impaired in a way that might leave him with expectations and dependencies healthy and wellhealed individuals do not have. Kent’s special vulnerabilities in this respect do not disqualify him as a consentee. They should rather lead researchers and administrators to design consent procedures that scaffold the intact capacities for voluntary choice that cognitively impaired individuals might retain. 4.3  Capacitated Consent Much of the evidence discussed in the previous two sections concerns what Kent believes: about the world, about himself, about time, and about the appropriateness and inappropriateness of different courses of action. But it does not address the key question of whether Kent’s actual decision-making suffered because of his amnesia.

Consent Without Memory  269 It is easy to imagine that individuals with episodic amnesia must be erratic, akratic, and impulsive, acting without regard to future consequences. If Kent was trapped in a permanent present with no sense of the past or the future, one might expect this to have impacted how he made decisions. One might expect Kent to have been impulsive and risk-prone, taking immediate gains without regard for future consequences. Or one might expect Kent to have adopted a more fatalistic orientation toward the future, as if his future is dead or “blank.” However, as detailed next, the results of the few studies of decision-making in which Kent and other individuals with episodic amnesia have participated did not support these intuitive expectations. To assess Kent’s impulsivity and risk-taking independently of one another, we used the Toronto Gambling Task, which was specifically designed for use in cognitively impaired individuals (Floden et al., 2008). As operationalized in this task, an impulsive person tends to act at the first opportunity, without regard to future consequences. A risk-taker, in contrast, shows a decided preference for low-probability/high payoff options. These tendencies are detected in ascending and descending presentations of a card selection task. Participants sit in front of a computer screen. In the ascending condition, cards appear on the screen one at a time, with some delay, up to a maximum of five cards. In the descending condition, five cards appear on the screen, and one is removed at a time, with some delay, until only one card remains. Participants can stop the dealer at any time. If the “winning card” is in their hand when they stop the dealer, the participant wins a payout inversely proportional to the number of cards in their hand. A risk-taker will tend to stop the presentation with very few cards in both the ascending and the descending condition. An impulsive person will tend to choose the first available option in each condition (i.e., with few cards in the ascending condition and many cards in the descending condition). Kent’s patterns of behavior were neither risk-prone nor impulsive (Rosenbaum et al., 2016). In fact, he displayed a general conservatism, preferring to wait until three or more cards had been dealt in both the ascending and descending conditions. These experiments all involved testing Kent’s choices about immediate rewards. This is unlike the consent situation, which requires one to consider future benefits and costs. Does the inability to episodically imagine future events impair one’s ability to make decisions about future rewards? Perhaps vividly imagining the future is essential for investing it with any value at all. This appears not to be the case. On multiple occasions, Kent took a standard test that asks subjects repeatedly to choose between a fixed reward in the present and a reward of a greater value in the future. The choices are arranged so as to converge on an indifference point, where the decisional forces in either direction are at equilibrium. We found that Kent (and three other individuals with amnesia) discounted the value of future rewards and did so well within the range of controls (Kwan et al., 2012; 2013; 2015). This means that Kent could invest the future with

270  Carl F. Craver and R. Shayna Rosenbaum value despite being unable to imagine concrete, personal, future scenarios; otherwise, one would expect him to discount far more rapidly than controls. Building on these results, there is also some evidence that Kent was able to anticipate the possibility of regretting his choices and to factor that anticipated regret into present decisions. Given the choice between a surething payoff of one million dollars and an 89% chance of one million, a 10% chance of five million, and a 1% chance of nothing, most people will take the sure thing. A natural and intuitive explanation of this effect is that people imagine themselves in the unfortunate 1%; i.e., in the future with nothing but regret. This anticipated regret then tilts the decision in favor of the first option. If Kent were unable to anticipate his regret in such choice situations, he should follow the expected utilities (which are greater in the second option). But Kent’s behavior was statistically indistinguishable from controls. He was thus arguably able to anticipate future regrets and to consider them in making present decisions (Craver et al., 2014; in contrast, see Hoerl and McCormick, 2016). In short, the available evidence demonstrates that Kent retained many of the capacities plausibly involved in the kinds of decisions persons make. Furthermore, his exercise of those capacities left him well within the range of neurotypical subjects and so within the range one might think is required to make decisions for one’s self. These studies indicate that Kent’s explicit decision-making capacities were largely intact and that his decisions on these tasks were indistinguishable from those of healthy controls. However, one might reasonably worry that episodic amnesia impairs one’s ability to update one’s knowledge (and so desires and preferences) based on experience. Many individuals with episodic amnesia, including Kent, fail the Iowa Gambling task (Gutbrod et al., 2006; Gupta et al., 2009; Rosenbaum et al., 2016). This task is designed to assess one’s ability to update one’s preferred actions in response to repeated experience of probabilistic outcomes of different values (Bechara et al., 1994). In this task, the participant is instructed to make a number of sequential draws from four piles of cards. The piles are assembled such that two deliver large payoffs but, on balance, lose more than they gain. The other two piles deliver more modest gains and losses and, on balance, gain more than they lose. The point of the task is to determine whether the participant gradually learns to prefer the advantageous decks over the disadvantageous decks. Kent’s performance was in the second percentile. He selected from the disadvantageous decks twice as often as the advantageous decks; at no point did his preferences begin to deflect in the direction of the advantageous deck (Rosenbaum et al., 2016). Kent’s performance on the Toronto Gambling task indicate that his responses likely do not reflect an impulsive character or bias resulting from his brain damage. Instead, these findings appear to indicate that Kent does not update his knowledge of the world with the same facility as do neurotypical controls. These findings raise the question of whether individuals with episodic amnesia can effectively exercise their right to withdraw from an experiment.

Consent Without Memory  271 Though they know their preferences and can make reasonable decisions in light of them, the fact that they might not update their preferences with experience raises the possibility that they are impaired in learning that the experiment is no longer beneficial or fun or, perhaps, that it is costing them more in the long-run than they are gaining. These findings raise the burden on experimenters and caretakers to be especially careful to guard against this occurrence: to allow for frequent breaks, to monitor participants for signs of discomfort, and to regularly re-administer the consent process. It is important to bear in mind, however, that the IGT results concern relatively subtle assessments of anticipated reward. When the reward and punishment signals are stronger (i.e., when something is truly unpleasant) individuals with episodic amnesia learn to avoid unpleasant and noxious stimuli (as demonstrated in the Claparède task).

5 Conclusion: Relating Episodic Memory to Our Moral Lives The question of Kent’s competence to consent is unlikely to be settled by the conceptual and empirical considerations described in this chapter. On the contrary, we are only beginning to farm this rich and fertile valley at the confluence of science and philosophy. The data to date concern too few individuals and approach these questions with only limited experimental tools. A deeper appreciation of the relationship between consent and amnesia requires a more systematic investigation of the agential and person-level competencies of individuals with episodic amnesia than we have completed and a deeper understanding of the role that episodic memory plays in our lives as persons than philosophy has achieved. Yet perhaps for now, the question is more interesting than the answer. Psychologists do not typically ask how episodic memory might figure in the person-level game of keeping track of our entitlements and commitments in relation to one another, or wonder (in practice) how those commitments are grounded in experiential, episodic knowledge of the world (see Mahr & Csibra, 2017 for a notable exception). They do not ask what role episodic memory plays in our ability to alter our positions in ethical space by undertaking obligations, granting permissions, forgiving transgressions, and taking responsibility. The very real question of how to obtain consent from individuals with episodic amnesia is in fact a convenient and tractable entry point into the question of how persons are possible in a world of causes. The point of this chapter has not been to answer this question with finality but to sketch an approach whereby philosophy and neuropsychology might inter-animate one another in the search for answers. Yet we hope to have established one important conclusion: Our armchair intuitions about the role of memory in our lives are a very poor guide to the psychological reality of how episodic memory mechanisms figure in our informed, voluntary choices.

272  Carl F. Craver and R. Shayna Rosenbaum Kent and his family remained committed for over thirty years to participate in scientific experiments. They granted researchers from various disciplines (psychology, neurology, philosophy) access to their homes, their lives, and their minds, resulting in over 40 scientific publications (and counting) and a research program dedicated to the study of episodic memory. And this prolonged arrangement was sustained by the careful and courteous stewardship of a few key scientists, beginning with Endel Tulving. Should the scientists that follow in their footsteps manage to build and sustain such working relationships with the individuals and families that is required to work on these important topics, and should they approach these questions with philosophical humility, grounded in sensible ideas about what personhood, agency, consent, promising, forgiveness, etc., are, we might open a path to learning more about how nature has managed to create a kind of creature that can consent and that deserves the distinctive respect we a­ ccord to persons.

Notes 1 Thanks to Andrè Ariew, Pascal Boyer, Brian Carpenter, Julia Driver, Len Green, Charlie Kurth, Joel Myerson, Pamela Speh, Amy Ravin, Roddy Roediger, Richard Rubin, Marya Schechtman, and Eric Wiland for useful comments discussions. Thanks especially to Endel Tulving and Donna Kwan for inspiration and experimental work. And finally, thanks to Kent, Ruth, and the Cochrane family for allowing the kind of access to their lives that makes work like this possible. 2 McCormick et al. (2016) report heightened deontological tendencies in five individuals with hippocampal damage. Given that we typically do not know moral response profiles prior to hippocampal damage, it is difficult to assess causation in these cases. With such low sample sizes, differences in response profiles might arise simply by chance. Speculatively, KC’s utilitarian performance might be explained by an additional lesion to the vmPFC discovered only post-mortem.

References Atance, C. M., & O’Neill, D. K. (2001). Episodic future thinking. Trends in Cognitive Sciences, 5(12), 533–539. Bechara, A., Damasio, A.R., Damasio, H., & Anderson, S.W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition 50: 7–15. Begley, S. 2016. MIT challenges The New York Times over book on famous brain patient—Scientific American. Scientific American. Retrieved June 1, from www.scientificamerican.com/article/mit-challenges-the-new-york-times-overbook-on-famous-brain-patient/ Boyer, P. (2008). Evolutionary economics of mental time travel? Trends in Cognitive Sciences, 12(6), 219–224. Branswell, H. (2014). Toronto amnesiac whose case helped rewrite chapters of the book on Memory Dies, Toronto Star. April 1. Retrieved from Thestar.com June 1, 2017. www.thestar.com/news/gta/2014/04/01/toronto_amnesiac_whose_case_helped_ rewrite_chapters_of_the_book_on_memory_dies.html

Consent Without Memory  273 Butler, J. (1975/1736). “Personal Identity” from The analogy of religion, natural and revealed, to the constitution and course of nature. In John Perry (Ed.), Personal Identity (pp. 99–106). University of California Press. Casebeer, W. D., & Churchland, P. S. (2003). The neural mechanisms of moral cognition: A multiple-aspect approach to moral judgment and decision-making. Biology and Philosophy, 18, 169–194. Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 7–19. Corkin, S. (2013). Permanent present tense: The unforgettable life of the amnesic patient, HM (Vol. 1000). New York: Basic Books. Craver, C. F. (2012). A preliminary case for amnesic selves: Toward a clinical moral psychology. Social Cognition, 30(4), 449–473. Craver, C. F., Keven, N., Kwan, D., Kurczek, J., Duff, M. C., & Rosenbaum, R. S. (2016). Moral judgment in episodic amnesia. Hippocampus, 26(8), 975–979. Craver, C. F., Cova, F., Green, L., Myerson, J., Rosenbaumm, R. S., Kwan, D., & Bourgeois-Gironde, S. (2014). An allais paradox without mental time travel. Hippocampus, 24(11), 1375–1380. Craver, C. F., Kwan, D., Steindam, C., & Rosenbaum, R. S. (2014). Individuals with episodic amnesia are not stuck in time. Neuropsychologia, 57, 191–195. Dalla Barba, G., & La Corte, V. (2013). The hippocampus, a time machine that makes errors. Trends in Cognitive Sciences, 17(3), 102–104. Darwin, C. (1871). The descent of man, 2 Vols. London, 81, 130–131. Davidson, P. S. R., Drouin, H., Kwan, D., Moscovitch, M., & Rosenbaum, R. S. 2012. Memory as social glue: Close interpersonal relationships in amnesic patients. Frontiers in Psychology, 3, 531. De Brigard, F. (2014). Is memory for remembering? Recollection as a form of episodic hypothetical thinking. Synthese, 191(2), 155–185. Dittrich, L. (2016). Patient HM: A story of memory, madness, and family secrets. New York: Random House Trade Paperbacks. Floden, D., Alexander, M. P., Kubu, C. S., Katz, D., & Stuss, D. T. (2008). Impulsivity and risk-taking behavior in focal frontal lobe lesions. Neuropsychologia, 46(1), 213–223. Fluharty, G., & Priddy, D. (1993). Methods of increasing client acceptance of a memory book. Brain Injury, 7(1), 85–88.Gupta, R., Duff, M. C., Denburg, N. L., Cohen, N. J., Bechara, A., & Tranel, D. (2009). Declarative memory is critical for sustained advantageous complex decision-making. Neuropsychologia, 47, 1686–1693. Gutbrod, K., Kroužel, C., Hofer, H., Müri, R., Perrig, W., & Ptak, R. (2006). Decision-making in amnesia: Do advantageous decisions require conscious knowledge of previous behavioural choices? Neuropsychologia, 44, 1315–1324. Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., & Jonathan, D. C. (2001). An fMRI investigation of emotional engagement in moral judgment. Science, 293(5537), 2105–2108. Haugeland, J. (1993). Mind embodied and embedded. In J. Haugeland (Ed.), Having thought: Essays in the metaphysics of mind. Cambridge, MA: Harvard University Press. Hoerl, C. (1999). Memory, amnesia and the past. Mind & Language, 14(2), 227–251. Hoerl, C., & McCormack, T. (2016). Making decisions about the future. In K. Michaelian, S.B. Klein, K.Szpunar, eds Seeing the future: Theoretical perspectives on future-oriented mental time travel. Oxford, U.K.: Oxford University Press.

274  Carl F. Craver and R. Shayna Rosenbaum Klein, S. B. (2002). Memory and temporal experience: The effects of episodic memory loss on an amnesic patient’s ability to remember the past and imagine the future. Social Cognition, 20(5), 353–379. Klein, S. B., Chan, R. L., & Loftus, J. (1999). Independence of episodic and semantic self-knowledge: The case from autism. Social Cognition, 17(4), 413–436. Klein, S. B., Cosmides, L., Tooby, J., & Chance, S. (2002). Decisions and the evolution of memory: Multiple systems, multiple functions. Psychological Review, 109(2), 306. Klein, S. B., Loftus, J., & Kihlstrom, J. F. (1996). Self-knowledge of an amnesic patient: Toward a neuropsychology of personality and social psychology. Journal of Experimental Psychology: General, 125(3), 250. Klein, S.B., Loftus, J., & Khilstrom, J.F. (2002). Memory and temporal experience: The effects of episodic memory loss on an amnesic patient’s ability to remember the past and imagine the future. Social Cognition, 20: 353–379. Klein, S. B., Sherman, J. W., & Loftus, J. (1996). The role of episodic and semantic memory in the development of trait self-knowledge. Social Cognition, 14(4), 277–291. Kwan, D., Carson, N., Addis, D. R., & Rosenbaum, R. S. (2010). Deficits in past remembering extend to future imagining in a case of developmental amnesia. Neuropsychologia, 48(11), 3179–3186. Kwan, D., Craver, C. F., Green, L., Myerson, J., Boyer, P., & Rosenbaum, R. S. (2012). Future decision-making without episodic mental time travel. Hippocampus, 22(6), 1215–1219. Kwan, D., Craver, C. F., Green, L., Myerson, J., Gao, F., Black, S. E., & Rosenbaum, R. S. (2015). Cueing the personal future to reduce discounting in intertemporal choice: Is episodic prospection necessary? Hippocampus, 25(4), 432–443. Kwan, D., Craver, C. F., Green, L., Myerson, J., & Rosenbaum, R. S. (2013). Dissociations in future thinking following hippocampal damage: Evidence from discounting and time perspective in episodic amnesia. Journal of Experimental Psychology: General, 142(4), 1355. Levine, B. (2016). When patients can’t give informed consent to be research subjects. Scientific American Blog Network. Accessed June 1. https://blogs. scientificamerican.com/mind-guest-blog/when-patients-can-t-give-informed-consentto-be-research-subjects/ Locke, J. (1690). An essay concerning human understanding. London: Thomas Bassett. Mahr, J., & Csibra, G. (2017). Why do we remember? The communicative function of episodic memory. Behavioral and Brain Sciences, 1–93. McCormick, C., Rosenthal, C. R., Miller, T. D., & Maguire, E. A. (2016). Hippocampal damage increases deontological responses during moral decision making. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 36(48): 12157–12167. https://doi.org/10.1523/JNEUROSCI.0707-16.2016 McCormack, T., & Hoerl, C. (1999). Memory and temporal perspective: The role of temporal frameworks in memory development. Developmental Review, 19(1), 154–182. McTaggart, J. M. E. (1993). The unreality of time, Mind XVII (1908). Reprinted in Time, Ed. Jonathan Westphal and Carl Levenson. Hackett. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press.

Consent Without Memory  275 Michaelian, K., Klein, S. B., & Szpunar, K. K. (2016). Seeing the future: Theoretical perspectives on future-oriented mental time travel. Oxford: Oxford University Press. Mullally, S. L., & Maguire, E. A. (2014). Memory, imagination, and predicting the future: A common brain mechanism?” The Neuroscientist, 20(3), 220–234. Nagel, T. (1970). Death. Noûs, 4(1), 73–80. Parfit, D. (1984). Reasons and persons. Oxford: Oxford University Press. Rosenbaum, R. S., Kwan, D., Floden, D., Levine, B., Stuss, D. T., & Craver, C. F. (2016). No evidence of risk-taking or impulsive behaviour in a person with episodic amnesia: Implications for the role of the hippocampus in future-regarding decision-making. The Quarterly Journal of Experimental Psychology, 69(8), 1606–1618. Rosenbaum, R. S., Köhler, S., Schacter, D. L., Moscovitch, M., Westmacott, R., Black, S. E., Gao, F., & Tulving, E. (2005). The case of KC: Contributions of a memory-impaired person to memory theory. Neuropsychologia, 43(7), 989–1021. Rubin, D. C., & Umanath, S. (2015). Event memory: A theory of memory for laboratory, autobiographical, and fictional events. Psychological Review, 122(1), 1. Sacks, O. (1985). The man who mistook his wife for a hat and other clinical tales. New York: Harper & Row. Schacter, D. L., Addis, D. R., Hassabis, D., Martin, V. C., Spreng, R. N., & Szpunar, K. K. (2012). The future of memory: Remembering, imagining, and the brain. Neuron, 76(4). doi:10.1016/j.neuron.2012.11.001. Schechtman, M. (1990). Personhood and personal identity. The Journal of Philosophy, 87(2), 71–92. Sohlberg, M. M., & Mateer, C. A. (1989). Training use of compensatory memory books: A three stage behavioral approach. Journal of Clinical and Experimental Neuropsychology, 11(6), 871–91. Suddendorf, T., & Corballis, M. C. (2007). The evolution of foresight: What is mental time travel, and is it unique to humans? Behavioral and Brain Sciences, 30(3), 299–313. Thagard, P. (2007). The moral psychology of conflicts of interest: Insights from affective neuroscience. Journal of Applied Philosophy, 24, 367–380. Tulving, E. (1983). Elements of episodic memory. Oxford: Oxford University Press. Tulving, E. (1985). Memory and consciousness. Canadian Psychology/Psychologie Canadienne, 26(1), 1. Tulving, E. (1993a). Self-knowledge of an amnesic individual is represented abstractly. The Mental Representation of Trait and Autobiographical Knowledge About the Self. Advances in Social Cognition, 5, 147–156. Tulving, E. (1993b). What is episodic memory? Current Directions in Psychological Science, 2(3), 67–70. Vargha-Khadem, F., Gadian, D. G., & Mishkin, M. (2001). Dissociations in cognitive memory: The syndrome of developmental amnesia. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 356(1413), 1435–1440. Vazire, S., & Carlson, E. N. (2011). Others sometimes know us better than we know ourselves. Current Directions in Psychological Science, 20(2), 104–108. Zimbardo, P. G., & Boyd, J. N. (2015/1999). Putting time in perspective: A valid, reliable individual-differences metric. In Time perspective theory; Review, research and application (pp. 17–55). New York: Springer.

Part VI

The Content and Phenomenology of Episodic and Semantic Memory

14 The Remembered

Mark RowlandsThe Remembered

Understanding the Content of Episodic Memory Mark Rowlands

1  Propositional Content Philosophical thinking about mental content, to a considerable extent, been dominated by a model supplied by the propositional attitudes. The expression ‘propositional attitude’ was originally coined by Bertrand Russell (1912) to denote any mental state that is attributed by way of a that-clause. Jones believes that snow is white, that grass is green, or that the cat is on the mat. What follows the ‘that’ is a complete sentence (“Snow is white,” “Grass is green, “or “The cat is on the mat”). A complete sentence has a meaning or as philosophers sometimes put it, expresses a proposition. Hence, we arrive at the expression, ‘propositional attitude.’ To believe something (or to think it, hope it, fear it, expect it, and so on) is to stand in a certain relation (believing, thinking, etc.) to a proposition—where this is understood as the meaning of the sentence that follows the occurrence of ‘that.’ Thus, in the grip of this sort of picture, philosophers came to think of mental content as propositional or at least, proposition-like. The content of mental states has the same sort of structure as a proposition. In the content of the belief that the cat is on the mat there is a component that stands for the (subject) cat, a component that stands for the (object) mat, and a component that stands for the (predicate) relation of being on. At the same time, it is difficult to overlook the fact that many mental states are not attributed by way of a that-clause. I intend to make a snowman, cut the grass, and feed the cat. Any attempt to render attributions of intention in that-clause terms—I intend that I make a snowman, that I cut the grass, and that I feed the cat—seem contrived at best. “To make a snowman,” “to cut the grass,” “to feed the cat,” are not complete sentences and, as such do not express propositions. How, then, are we to understand the content of intentions? Perception occupies a curious middle ground: attributions of perception being seemingly expressible in both with and without a thatclause. I can see the cat on the mat, and I can also see that the cat is on the mat. Are these two different perceptual states? They might be. Or they might not. Episodic memory, however, is a particularly interesting case because here there seems to be a well-known vocabulary in situ that precludes attributing

280  Mark Rowlands such memory by way of a that-clause. I remember, let us suppose, falling out of a tree on my tenth birthday. This is an episodic memory. Suppose we try to render this in that-clause terms: I remember that I fell out of a tree on my tenth birthday. This we cannot do—without turning the episodic memory into a semantic memory. We have turned it into a memory of a fact, and that seems to be the hallmark of semantic memory. In the case of perception, it seems possible to attribute perception both with and without a that-clause. In the case of intention, there was at least a contrived—perhaps desperate— way of attribution via a that-clause. But in the case of episodic memory, this is what we cannot do. This makes the case of episodic memory, I shall argue, especially interesting. When I attribute an episodic memory, I must do so by way of a dependent clause rather than a complete sentence. A dependent clause does not express a proposition. How, then, are we to understand the content of episodic memory? That is the subject of this chapter. It may be that episodic memory is merely a curiosity. However, I suspect that this is not, in fact, the case. I suspect that no mental content is really propositional, and its appearance as such derives from the distorting influence of the language we use to both attribute and describe such content. Episodic memory, for this reason, provides us with a crucial template for understanding mental content more generally. However, to argue for this broader claim here would take us beyond the scope of this chapter. Here, I shall focus on understanding exactly what the content of episodic memory is.

2  The Content of Episodic Memory 1: Episodes Semantic memory fits the propositional template nicely: semantically, I remember that something is the case—that Paris is the capital of France, that a platypus is monotreme, and so on. This is no surprise: the category of semantic memory seems to be nothing more than a sub-category of belief (Rowlands, 2009; Klein, 2014). That is, while not all beliefs are semantic memories, all semantic memories are beliefs. I find persuasive Klein’s (2014) argument, that this casts doubt on the utility of thinking of semantic memory as a form of memory at all—but this issue is tangential to my current concerns. Whether or not it really merits the appellation ‘memory,’ we can understand the content of a semantic memory as having propositional structure: the structure of the sentence that follows the ‘that.’ But episodic memory is not attributed in this way. How do we understand the content of episodic memory? One obvious option is to take our cue from the name and understand episodic memory as memory of an episode. An episode is a state-ofaffairs, broadly construed. And a state-of-affairs is an arrangement of objects and properties, including relational properties. When the content is propositional, the corresponding state-of-affairs must, it seems, have propositional structure. For example, one constituent of the state of affairs

The Remembered  281 corresponds to the subject of the proposition, another to the predicate etc. There is, however, no reason for supposing the content of episodic memory is propositional. Nevertheless, it must have some structure: it cannot simply be a collection of objects and properties: these must be arranged in a specific way. To episodically remember the cat being on the mat is not to episodically remember the mat being on the cat. How do we get the right structure? Since episodes are, in effect, events, we might borrow from a well-known theory of events, associated with Jaegwon Kim, among others: the propertyexemplification theory. According to this, any event should be understood as the instantiation of a property by an object at a time. That is, any event has the structure [x, P, t], where x is the constitutive object, P is the constitutive property, and t is the constitutive time of the event. Thus, I might episodically remember me (a constitutive object) instantiating a constitutive property (falling out of a tree) at a constitutive time (on my tenth birthday). When I episodically remember, I remember episodes understood in this way. The problem with any attempt to understand episodic memory as memory of episodes is that it threatens to collapse the distinction between episodic and semantic memory. This is for the simple reason that many semantic memories are also memories of episodes. The semantic memory that Paris is the capital of France is not a memory of an episode, but the semantic memory that Vesuvius erupted in 79 CE certainly seems to be. In semantically remembering this, I remember that a certain object (Vesuvius) instantiated a certain property (erupting) at a certain time (79 CE). If we do not want to lose the distinction between episodic and semantic memory, we should not understand the content of episodic memory simply as an episode. I shall argue later that the notion of episode does have a significant role to play in the content of episodic memory, but to properly play this role it needs supplementation with something else.

3  The Content of Episodic Memory 2: Experiences What is missing from the bare idea of remembering an episode is what often goes by the name of mental time travel. I do not remember the eruption of Vesuvius as an episode I formerly experienced. This might suggest we should understand episodic memory as memory of experiences. This is a common, and historically respectable way of thinking about episodic memory. John Locke, for example, took memory to be a power of the mind “to revive perceptions, which it has once had, with this additional perception annexed to them, that it has had them before” (Locke, 1690/1975, p. 150). William Brewer defines episodic memory in similar terms, as a ‘reliving’ of the individual’s phenomenal experience from a specific moment in their past, accompanied by a belief that the remembered episode was personally experienced by the individual in their past (Brewer, 1996, pp. 60–61).1 On this construal, then, experiences are the immediate objects of experiential memory. To the extent that episodic memory involves the recall of episodes,

282  Mark Rowlands this recall proceeds via the recall of the experiences that accompanied them. Episodes may be the distal objects of episodic recall; but its proximal objects are always experiences. The well-documented (see Locke, 1971; Nigro & Neisser, 1983) Debus, 2007; Sutton, 2010; Eich et al., 2011; Goldie, 2012; McCarroll, 2015) phenomenon of perspective switching provides a serious problem for this understanding of episodic memory. Often episodes are remembered from a perspective one could not have had when that episode occurred. Suppose, to purloin a memory I anticipate my children one day having, you remember sitting in the back of the family car singing a song called “The Man,” by a band called The Killers, with your brother when you were 7 years old. Let us suppose this event actually did happen. However, you remember this from a perspective only others could have had of that event: from a third-person perspective—as it might have been witnessed by your parents, turning around to watch you rather than as experienced by you. This phenomenon is often expressed in the distinction between ‘field’ and ‘observer’ memories. As Goldie puts it: “In field memory, one remembers ‘from the inside,’ the events as they took place. In observer memory, one remembers ‘from the outside’ so that one is oneself part of the content of what one remembers.”2 In perspective switching one can shift from remembering an episode in which one was involved to remembering how you might have appeared to others as that episode unfolded. Perspective switching militates against the understanding of episodic memories as memory of experiences. In this sort of case—“The Man” memories—the experiences that you seem to remember are ones that you never had. In other words, the claims that (1) episodic remembering is remembering of experiences, and (2) perspective switching in episodic remembering is relatively common, yields (3) many of our episodic memories are false. It is, of course, well known that many of our episodic memories are false or inaccurate (see Neisser & Harsch, 1992). But the problem with understanding episodic memory as memory of experiences is that it blurs the distinction between truth and falsity in an unacceptable way. To see why, consider two scenarios. In the first, it is true that the 7-yearold you once sat in the back of the family car singing “The Man” with your brother. However, your perspective has switched: you remember this from a perspective, and so remember experiences, you never actually had. If episodic memory is memory of experiences, this memory is false. In the second scenario, the event of your singing “The Man” with your brother, in the back of the family car, never actually happened at all: your apparent memory is, let us suppose, the result of confabulation. In this scenario, your episodic memory is also false. But it is false in a very different way. We do not want to lose the distinction between these two different ways in which episodic memories can be false. Therefore, we should abandon the idea that episodic memories are simply memories of experiences. Whether or not the episode in question actually happened is important to the status of the

The Remembered  283 memory in a way that is lost if we think of episodic memory simply as the remembering of experiences. One might object: we can capture the difference between the two kinds of falsity simply on the basis of experiences. The field and observer memories of the same episode are sufficiently similar for us to be able to deny that the perspective-switched memories are false.3 However, is this claim really true? Consider, for example, the visual component of the two sorts of memory. In one memory, the person who remembers pictures himself as he imagines he must have looked to someone else: that is, he sees his face, as he imagines it must have looked to another. This is an observer perspective. But in the field perspective, he sees the fields and mountains that pass by as he sings, and the faces of his parents in front of him. On what basis can we say that these memories are similar? Admittedly, there is a song involved in both memories: the auditory components are the same. But that is not enough to justify an overall claim of similarity. The perspective-switched version of the same memory could easily, in other circumstances, be two distinct memories, one true and the other false. Therefore, the claim that there is sufficient similarity of perspective-switched memories for us, on this basis alone, deny the falsity of perspective-switched memories should be treated with considerable suspicion. Please note that I am not claiming perspective-memories are false. That is what I wish to deny. Rather, my claim is that if we regard the content of episodic memory as simply experiences, this yields the conclusion that perspective-switched memories are false. This conclusion needs to be resisted. And to do this, we need to reject the idea that episodic memories are simply memories of experiences. I suspect the only real temptation for judging the perspective-switched memories to be similar stems from the fact that they are both memories of the same episode. If this is correct, then it is the episode, rather than the experiences, that is doing the work required to ground the distinction between the two sorts of falsity. If so, this is once again indicative of the indispensability of the episode to our understanding of the content of episodic memory. In understanding this content, we need to incorporate both experiences (otherwise we will lose the distinction between episode and semantic memory) and episode (otherwise we will lose the distinction between falsity and perspective-switching).

4  The Content of Episodic Memory 3: The Fregean Model There is a well-known account of intentional states that will allow us to insert both episode and experience into the content of memory: we think of the content as an episode-subsumed-under-a mode-of-presentation. We might label this account the Fregean model, after its principal inspiration, Gottlob Frege. According to this model, any episodic memory should be analyzed into (1) the act of remembering, (2) the episode remembered, and (3) the mode of presentation of that episode.

284  Mark Rowlands You remember, let us suppose, singing “The Man,” with your brother, in the back seat of the family car when you were 7 years old. Let us suppose this episode actually did take place. Your memory is, therefore, of an episode. This, by itself, is not sufficient for the memory to qualify as episodic. To thus qualify, the episode needs to be subsumed under a specific mode of presentation: an experiential mode of presentation of a quite specific sort. Minimally, you must remember the episode as one that you formerly experienced (that is, in this case, orchestrated or performed). What makes a memory episodic is not that it is a memory of an episode but, rather, that it is a memory of an episode that is subsumed under this specific mode of presentation. There is quite a bit built into this notion of remembering as. Consider a semantic autobiographical memory: you remember that when you were three your parents took you on a trip to Europe. You now, however, have no episodic recall of the trip. This would not count, in the sense in which I am employing the term, as remembering: you do not remember the European trip as one that happened to you.4 You believe it happened to you. You have a semantic autobiographical memory of the episode happening to you. But episodically remembering an episode as an episode one formerly encountered involves an experiential component that is lacking in the case of semantic autobiographical memory: the episode must be experienced as an episode one formerly encountered rather than merely thought of as an episode that one formerly encountered.5 This general Fregean apparatus for thinking about the content of mental states in general is familiar. It is a relatively straightforward application of the general model to the specific case of episodic memory—to which, I think, it fits both naturally and nicely. It is the next question that is the most important—and also the most overlooked. What sort of thing must the content of episodic memory be if it essentially presents an episode as one formerly encountered by the subject—that is, as falling under modes of presentation such as I have seen this before or I have done this before?

5  Memories vs. Photographs In answering this question, it is instructive to contrast episodic memories with photographs. There are many problems with a photographic model of memory—and it is not possible to overcome many of these problems merely by replacing a static photograph with a moving film. For example, photographs (or films) are, for example, notoriously indiscriminate at capturing detail, quite independently of the photographer’s intentions or attention— that is what photo bombing is all about. Memories are not like this. But there is one, less familiar, problem that is important for our purposes. There is nothing in the scene presented by a photograph that is essentially de se. That is, there is nothing in a photograph that guarantees that the scene depicted is presented as one the viewer formerly witnessed, orchestrated,

The Remembered  285 or otherwise encountered. This is so even if the viewer is actually depicted in the scene: the viewer may not recognize him- or herself as the person in the scene. Photographs are, in this respect, akin to nonindexical descriptions. Consider Perry’s famous argument for what he calls the ‘essential indexical’ (Perry, 1979). To be aware that I am currently typing this page it is not sufficient to be aware that a philosopher of Welsh descent currently employed by the University of Miami is typing this page—for I may not know that I am a philosopher of Welsh descent currently employed by the University of Miami. No matter what description one employs, or how many of them one employs, it will only guarantee awareness of self if we add the extra premise: and I satisfy this description or descriptions. Similarly, your awareness of an episode depicted in a photograph or film clip provides no guarantee that the episode will be presented as one you have formerly encountered. This is true even if you are present in the scene. For you might not recognize yourself as you. A photograph may portray an episode that once happened to you, and you may be in the photograph with the episode happening to you. But there is no guarantee that you will see the photograph as depicting an episode that once happened to you. The case of photographs, however, diverges from that of the essential indexical in at least one respect. Recognition of the person depicted in the photograph as you is a necessary but not sufficient condition of recognizing the episode portrayed as one you have formerly encountered. You might recognize yourself, but still not remember the scene depicted in the photograph: “Yes, that’s me all right. But I can’t remember where this was or what I was doing there.” This is a perfectly possible reaction, indeed far from unlikely once you reach a certain age. Once again, this reaction has no echo in the case of episodic memory. If you episodically remember, then what you remember must be presented to you as something you formerly encountered. If it is not, then it is not an episodic memory. The moral seems to be this: The mode of presentation I have encountered this before is an essential feature of episodic memories, but only a contingent feature, at best, of photographs. The relevant question, now, is this: is: why is there this difference between memories and photographs? The answer, I shall argue, can be found in a certain theme identifiable in the work of the later Wittgenstein.

6  The Two-Factor Model A prominent theme of the Philosophical Investigations (Wittgenstein, 1953) was that meaning something by a sign does not consist in an inner state or process. Wittgenstein adopted a particular conception of an inner state or process: it consists in an item “coming before one’s mind.” Meaning something by a sign, therefore, does not consist in an item coming before one’s mind. In particular, an item’s coming before one’s mind—equivalently, one’s

286  Mark Rowlands being presented with a particular object of awareness—is neither necessary nor sufficient for meaning something by a sign. Against the idea that it is sufficient Wittgenstein advised to conduct a phenomenological investigation: look and see! If we do so, we will realize that much of the time we use signs in perfectly meaningful ways without being presented with any particular object of awareness. For our purposes, however, the argument against sufficiency is more important. Wittgenstein’s argument against sufficiency is based on the idea that any object of awareness has the logical status of a symbol. A symbol is an item that has both syntax and semantics and, crucially, the semantics is something that must be supplied by an act of interpretation. Written and spoken words are the most obvious examples of symbols: a written word—a sequence of shapes against a contrastive background—can mean anything. In order to mean something, it must be interpreted. The same, Wittgenstein argued, is true of anything that comes before the mind—whether the item is external (words, pictures, etc.) or internal (for example, a mental picture). Consider, for example, a picture of a dog. In principle, this picture might mean or signify any number of things. It might signify a particular dog, or dogs in general. It might be used to signify a mammal, or a furry thing, or a thing with four legs, or a tail, or floppy ears. It might be used, if the picture is of a certain sort, to signify happiness, or sadness, or laziness, boredom or comfort. In order to mean one thing, rather than all possible alternatives, the picture must be interpreted. The same is true, logically speaking, if we move the picture inwards—make it a mental picture rather than a physical one. Understood as something that ‘comes before the mind’—as an object of awareness—the mental picture could mean any number of things. To mean something in particular, it must be interpreted. At this juncture, there are two ways of interpreting Wittgenstein. The first is constructive. If, to mean something, an item that comes before our mind must be interpreted, then the task is to find what supplies the relevant interpretation. Wittgenstein developed a line of argument—it later became known as the rule-following paradox—to show that there are no facts about an individual that supply the required interpretation. Then, according to the constructive interpretation, Wittgenstein’s appeal to custom or practice can be enlisted to show that there are other facts—facts about the customs or practices in force in a given linguistic community that do provide the required interpretation. This constructive interpretation of Wittgenstein’s argument raises at least two difficult questions. First, is this the correct interpretation of Wittgenstein? Second, independently of whether it is Wittgenstein’s view, does it work? I think negative answers to both questions can be defended (See Rowlands, 2006, 2010 and 2016 for further discussion), but I shall not argue for these answers here. Instead, my focus is on another interpretation of Wittgenstein that is more pertinent to my current concerns. The interpretation I have in mind is based on the observation that providing constructive solutions to philosophical problems contradicts many facets of Wittgenstein’s avowed philosophical method. On this

The Remembered  287 interpretation, Wittgenstein’s concern is not with providing a positive account of how meaning is possible, but to elucidate the assumptions that led us to think there is a problem in the first place (see McDowell, 1992). On this deflationary interpretation, the primary blame should be assigned to the assumption that meaning involves an item ‘coming before the mind’ of a subject. This assumption tempts one into adopting a twofactor model of how mental items relate to the world. On the one hand, there is the item that “comes before the mind,” and on the other is the interpretation of that item that links it to the world. In themselves, the items that come before the mind are semantically inert: in principle, they could be about many things, even anything. Only by being interpreted by the subject can these items be about anything else. On the proposed, deflationary account of Wittgenstein’s arguments, they target this two-­ factor model of meaning. Assuming such a model is, as Wittgenstein would put it, is the decisive movement in the conjuring trick, the very one we thought most innocent. There is no reason to suppose that meaning something by a sign can be factored into (i) an item that is, intrinsically, semantically inert, and (ii) an act of interpretation that supplies the semantics. That is, we should reject the idea that meaning something by a sign consists in encountering an object of awareness that is, in itself, semantically inert, combined with a mental act that supplies an interpretation to this object. We reject, that is, the twofactor model of meaning. If items that were, in themselves, semantically inert did ‘come before the mind’ then we would need an act of interpretation to supply them with a semantics. And we may have a very hard time understanding what could supply that. But if there are no such items, then no such interpretational act is needed. Whether this view can be correctly attributed to Wittgenstein, it is the view I wish to defend regarding the content of episodic memory. This is what the reflections on the inadequacies of the photographic model are intended to highlight. Remembering an episode is not like being acquainted with a photograph. The photograph is, in itself, semantically inert—it could represent an indefinite number of things—and therefore requires interpretation. What is remembered is not like this. I do not, for example, remember a face, and certain transformations it undergoes, and then subsequently identify it as that of my father—not if this memory is accurately characterized as a memory of the face of my father. I do not even do this unconsciously. To remember the face of my father is to remember the face-as-the-face-of-my-father. Of course, in other circumstances, I might remember my father’s face without remembering it as the face of my father. Suppose for example, that my father had deserted his family when I was very young. I have a dim recollection of a man’s face, a recollection that manifest itself from time-to-time, but no idea who it is. Nevertheless, I remember this face as something: for example, as a face from my past whose owner I cannot identify. To remember a face is to remember it as the face of someone, or, least, as a face seen in some circumstances. (I remember

288  Mark Rowlands the face I saw last summer, even though I do not know the identity of the person to whom it belongs. The interpretation here is this: I remember it as the face I saw last summer.) We might put the point in this way. Let us call what I episodically remember, in any given case, the remembered. Then, we can say that some or other as is essentially built into any token of the remembered. Of course, I always see a photograph in some way or other too. But this is because I bring to the photograph an act of interpretation that is distinct from the photograph itself: the photograph exists independently of the act of interpretation. Thus, that the photograph, at any given time, represents this or that, or something else, is a contingent feature of the photograph. This is not how it is with the content of episodic memory. That an episodic memory is a memory of this, that or something else, is not a contingent but essential feature of the memory. That is, when I episodically remember something, I remember this something precisely as something. There is no remembered to which the ‘as’ must be subsequently attached, by an act of interpretation, as a logically posterior addendum. The remembered is always, essentially, remembered as something.6 The crucial question, then, is: why this is not true of the photograph? What is the difference between the remembered and the photograph that allows the ‘as’ to be essentially built into the former but not the latter. To see what this difference is, we should remember that the two-factor model is entirely appropriate to the photograph. On the one hand there is the photograph, and on the other, there is the act of interpretation that allows us to form beliefs as to what it is about. The two-factor model is appropriate precisely because the existence of the photograph is logically independent of the act of seeing, and interpreting, it. It exists if I do not see it, and even if no one sees it. This independence of photograph and act of interpretation, therefore, makes the two-factor model unavoidable. I might be presented with a photograph of the face of my father and not recognize it as him. But this cannot happen when I episodically remember the face of my father: if I, in fact, episodically remember the face of my father, then I must remember it as the face of my father.7 The photograph contains no inherent ‘as’ because it exists independently of the act of seeing. This is why a two-factor model is appropriate in explaining the semantics of the photograph. But if such a model is not appropriate to the remembered, then this supports (inductively) the claim that the content of episodic memory does not exist independently of the act of remembering. The implications of this lack of independence of act of remembering and content remembered are significant. In the rest of this chapter, I shall briefly outline some of these.

7  The Face of My Father, April 26, 1965 Much recent debate in work on memory has concerned the extent to which memory is reconstructive. Very roughly, memory is constructive to

The Remembered  289 the extent it involves nonperceptual operations performed on content at the time of encoding. Memory is reconstructive to the extent it involves operations performed on memory content at the time of retrieval. The arguments for reconstruction in memory have typically been empirical. There is, for example, the groundbreaking work of Karim Nader (2003) and colleagues on reconsolidation in memory, which strongly supports the idea that memory is reconstructive. Here, I shall supplement these empirical arguments with an a priori argument. Given the nature of the content of such memory, episodic memory must be reconstructive. There was really no other way for it to be. One of my earliest memories is of a nosebleed I suffered one night, on the small settee in the small house in which I spent the first six years of my life. The memory in question is that of the face of my father—in particular, of certain transformations this face underwent—in the early hours of April 26, 1965 (GMT), which I know because that was the date of the second Cassius Clay (as he then was)-Sonny Liston fight. The order of events apparently transpired thusly: A few seconds before the opening bell, my nose begins to bleed. My father dutifully runs off to get some tissues. When he returns, mere moments later, the fight is over. Clay has knocked Liston out with a punch so fast that many had trouble seeing it. But that is not important. What is important, for my purposes at least, is my father’s face. His face is a study in confusion, flitting between the TV screen and me—as if I were somehow responsible for the events unfolding in black and white. What is going on? Is this a rerun of the end of the first? (I assume, no doubt retrospectively, that thoughts such as these were running through his head.) Then I remember his face slowly transforming from confusion, to suspicion (was this it, had he missed it?), and from suspicion to resigned acceptance (he had, indeed, missed it) to joy (my father was an admirer of Clay’s intelligence and skill). That memory was about my father, not me. But I am still in there: these scenes, the transformations undergone by my poor father’s face, are ones that are presented, precisely, as ones that I once experienced: as events that I once saw. One should not get hung up on the veracity of this memory, whether it is a real memory or confabulation. At the very least, while I have good reasons to suppose this episode actually happened, I have equally good reasons to believe some of its elements are inaccurate. (I remember a nose bleed, but it was really my desire for warm milk that was the reason for my father temporary vacation of the TV room.) The memory, if that is what it is, functions, here, as the central element in a thought experiment. In this context, there is a striking feature of this memory, and one particularly pertinent to the issue of reconstruction. When this fight took place, my father would have been a relatively young man, in his early-to-mid thirties. However, the face I remember, slowly transforming from confusion to suspicion to acceptance to joy is the face of an old man. It is the face he wore in his final years. I think I know why this is: it can be traced to the paucity of photographs at that time—a paucity that will be difficult to grasp for

290  Mark Rowlands someone who has grown up in the selfie-age. Making a photograph was a long, drawn-out affair, which involved a camera that wouldn’t work at least 50 percent of the time, a trip to a pharmacist (a chemist, as they were known back then), and a wait of approximately two weeks while the photos were developed. As a result, people couldn’t, in general, be bothered. Photographs were something reserved for special occasions: holidays, birthdays, and the like. My parents were not averse to capturing their children in celluloid, but they certainly weren’t going through this rigmarole for photographs of themselves. As a result, photographs of my father were few and far between. With this historical context in mind, consider what might have happened if, in my memory, the face slowly transforming, in stages, from suspicion to joy were the face of my father as a young man. Then, there would be a significant chance I would not recognize him. Episodic memory, as we have seen, is not like this. If I do episodically remember the face of my father, transforming in these ways, then I remember it precisely as something I formerly encountered. If it is true that, because of the dearth of photographs, I cannot remember the face of my father as a young-ish man then, in order to experience these transformations of the face as something I have previously encountered, the face has to be changed to make it recognizable to me. If the face hadn’t been changed in this way, then there is no necessary reason why this episode would be presented to me as one that I formerly experienced. This episode can only be presented as one that I formerly experienced if the face that it presents is presented, precisely, as the face of my father. Thus, given the paucity of photographs from that time in my life, the face has to be updated to make it recognizable, precisely, as the face of my father. The only frame of reference I have that could be used to supply an interpretation is the face of my father I actually remember—the face my father of his later years. One should be clear what this example claims and what it does not. I do not assume, of course, that our ability to recognize previously encountered scenes and people is dependent on the availability of photographs of ­relevant scenes and people.8 Of course, some people might be able to remember the face of their younger father without photographs. I am, in this case, apparently not one of those people. But my abilities or lack thereof are not the issue. As I mentioned earlier, one is free to treat this example as a thought experiment. The essential elements of this thought experiment are (i) ‘I’ (or some random person if you prefer) episodically remember the face of my father during a specific episode from a long time ago, and (ii) ‘I,’ for whatever reason, am unable to remember what my father looks like at that time. In such circumstances, the face of my father must be transformed from the way it actually was at that time into a form that allows me to recognize it as the face of my father. In short: if, in episodic memory, an episode is presented, necessarily, as one that I formerly encountered, then some memories must be reconstructive. This is not to say how much reconstruction is involved. If someone is effortlessly

The Remembered  291 able to remember the younger face of their father without photographic help, then in having this sort of memory, little or no reconstruction might be required. But in other circumstances, such as the ones outlined previously, significant reconstruction might be required. We can see, in outline, the circumstances that require reconstruction: I remember an episode as one that happened to me, but I don’t remember specific elements of the episode. Put in terms of the apparatus introduced earlier, we might say that my father’s face is the constitutive object of the episode. I remember my father’s face (instantiating the property of) transforming, but I do not remember the face and so I have to reconstruct this constitutive object to make it necessarily true that I experience this episode as one I formerly encountered. Any particular token of episodic remembering must involve sufficient reconstruction to underwrite the sense that the episode presented is one formerly encountered. In some cases, this may entail little or no reconstruction. In other cases, the necessary reconstruction may be substantial.

8  Memory and the Self It is not unreasonable to suppose—and has indeed been supposed by many—that our episodic memories play a significant role in making us the people we are. After all, what could make us the people we are if not the experiences we have had on these tracks through space and time that are our lives? And how could these experiences be retained—and so play a role in shaping who we are in the present—if not through episodic memories? This intuition has received several, quite different, theoretical articulations, but it is the underlying intuition that is of interest here. The intuition seems to fall foul of two inconvenient truths. The first is that we forget—a lot! For each one of us, forgotten experiences far outnumber the remembered. Second, even when we do remember, memories often bear little relation to the experiences of which they purport to be records (See Neisser & Harsch, 1982 for a classic study). It is as if you were writing an autobiography—something that you hoped to be an accurate and honest recounting of your life. However, when you open the book, you find vast swathes of redactions, which greatly outnumber the printed words. Worse, although you are not in a position to know this, the sentences that remain are not particularly accurate records of the episodes you encountered. Some of them bear virtually no relation to the episodes you encountered. In such circumstances, how can this book—this purported autobiography—be the book of you? On the one hand, one might hope there is enough salvageable content to do the trick. Much content may be lost, and much of the remaining content may be inaccurate. But, if there is just enough content left to tell a coherent story of a life, perhaps that is all we need? Perhaps. It is this sort of idea that underlies the memory theory of personal identity. But the view of episodic memory content outlined here suggests a very different picture. The person

292  Mark Rowlands is not something constructed out of episodic memory contents that are, in themselves, neutral with regard to who has them. Rather, the person is already in the content of episodic memory. It is not as if the person emerges subsequent to our putting the contents together, arranging them, in so far as we can, into a coherent whole. Rather, the person who remembers is already in the memory content she has retained. The person is in her memories in the way, very roughly, that a sculptor is in her sculpture. The person has shaped—transformed—the episode encountered into a content remembered. The person has to do this, for there is nothing in an episode encountered that guarantees it will be recalled as an episode one formerly encountered. For this to occur, the episode must be transformed—reconstructed—into content that is, necessarily, de se. The person who episodically remembers transforms the episode encountered into content recalled. And she does this by subsuming the episode under the mode of presentation, encountered (witnessed, orchestrated etc.) before. For the episode to be thus subsumed— as we saw in the case of the memory of the face of my father—the episode might need to be reconstructed into something that can be subsumed under this mode of presentation. Sometimes the reconstruction will be minimal, and sometimes it will be far from this. But the possibility of reconstruction must always there, for this is required to transform episode encountered into content remembered. The episode must be reconstructed for it to be capable of fitting into a person’s life, and it is that same person performs that reconstruction. The indelible stamp of the person, therefore, lies on the content of many of her episodic memories.

Notes 1 Brewer calls this ‘recollective’ rather than ‘episodic’ memory. For purposes of exposition, I shall treat this as one form of what I am calling episodic memory. 2 Goldie (2012, p. 49). 3 I would like to thank Dorothea Debus for this objection. 4 I would like to thank an anonymous referee for allowing me to clarify this point. 5 In earlier work (1999, 2009), I have argued that the distinction between semantic and episodic memory is one of degree rather than kind. The present point reinforces this idea. Thought can be more or less rich in phenomenology. Given that certain veridicality conditions are met, the richer the phenomenology of a thought that an episode happened the more this thought approximates an experience of the episode as happening to you. 6 I’d like to thank another anonymous referee for allowing me to clarify this point. If I were a philosopher of a certain bent, I might be tempted to put this point in terms of distinction in the scope of the model operator: wide scope in the case of photographs, narrow scope in the case of episodic memory content. Necessarily a photograph is seen in some or other way—but precisely which way is a contingent matter. But in the content of an episodic memory, an episode is necessarily presented in some way or other, Helpful to some, but perhaps baffling to others. I have, therefore, relegated this way of stating matters to a footnote. 7 This is not the same, as the earlier discussion indicated, as episodically remembering something that is the face of my father. I can episodically remember something

The Remembered  293 that is the face of my father without episodically remembering the face of my father. 8 I would like to thank an anonymous referee for allowing me to clarify this point.

References Brewer, W. F. (1996). What is recollective memory? In D. C. Rubin (Ed.), Remembering our past. Cambridge: Cambridge University Press, pp. 19–66. Debus, D. (2007). Perspectives on the past: A study of the spatial perspectival characteristics of recollective memories. Mind and Language, 22, 173–206. Eich, E., Handy, T., Holmes, E., Lerner, J., & McIsaac, H. (2011). Field and observer perspectives in autobiographical memory, 14th Sydney Symposium on Social Psychology, University of New South Wales, March 15–17. Goldie, P. (2012). The mess inside: Narrative, emotion and the mind. Oxford: Oxford University Press. Klein, S. (2014) What memory is? WIREs Cognitive Science. doi:10.1002/wcs.1333 Locke, D. (1971). Memory. London: Palgrave Macmillan. Locke, J. (1690/1975). An essay concerning human understanding (P. H. Nidditch, Ed.). Oxford: Oxford University Press. McCarroll, C. (2015). Point of view in personal memory: A philosophical investigation, Ph.D. Thesis, Macquarie University. McDowell, J. (1992). Meaning and intentionality in Wittgenstein’s later philosophy. Midwest Studies in Philosophy, 17, 30–42. Nader, K. (2003). Memory traces unbound. Trends in Neurosciences 26(2), 65–72. Nigro, G., & Neisser, U. (1983). Point of view in personal memories. Cognitive Psychology, 15, 467–482. Perry, J. (1979). The problem of the essential indexical. Nous, 13(1), 3–21. Rowlands, M. (1999). The body in mind: Understanding cognitive processes. Cambridge: Cambridge University Press. Rowlands, M. (2006). Body language: Representation in action. Cambridge, MA: MIT Press. Rowlands, M. (2009). Memory. In P. Calvo & J. Symons (Eds.), Routledge companion to philosophy of psychology (pp. 336–345). London: Routledge. Rowlands, M. (2010). The new science of the mind: From extended cognition to embodied phenomenology. Cambridge, MA: MIT Press. Rowlands, M. (2016). Memory and the self: Phenomenology, science, autobiography. New York: Oxford University Press. Russell, B. (1912). The problems of philosophy. London: Henry Holt. Sutton, J. (2010). Observer perspectives and acentered memory: Some puzzles about point of view in personal memory. Philosophical Studies, 148, 27–37. Wittgenstein, L. (1953). Philosophical investigations (E. Anscombe, R. Rhees, & G. H. von Wright, trans. & E. Anscombe, Ed.). Oxford: Blackwell.

15 The Past Made Present

Matthew SoteriouThe Past Made Present

Mental Time Travel in Episodic Recollection Matthew Soteriou

Some philosophers and psychologists suggest that the episodic recollection of some past event or action that you have witnessed or performed is analogous to traveling back in time. It is a form of ‘mental time travel.’ Some also suggest that just as episodic recollection can be regarded as a form of mental time travel, so too can episodes of imagining future events. For example, Thomas Suddendorf and Michael Corballis suggest we have a “general faculty of mental time travel that allows us not only to go back in time, but also to foresee, plan, and shape virtually any specific future event” (2007: 299).1 However, it is not entirely clear how we are to understand and unpack this notion of mentally traveling backward and forward in time. On first reflection there is something puzzling to the idea. One might think that all conscious mental acts fall within the subject’s experienced present and that the experienced present associated with any such mental act is tethered to the moment at which that conscious mental act occurs. Such conscious mental acts may be tensed thoughts about the past and future, but it is not at all clear that such tensed thoughts can themselves amount to anything like a form of mental time travel. Indeed I take it that part of the appeal of invoking the notion of mental time travel is to signal that certain ways of relating to the past and future (through episodic memory and imagination) are importantly different from merely thinking about the past and future by entertaining tensed thoughts about earlier and later times. So what sense can be made of the proposal that episodic recollection makes possible a form of mental time travel? An answer to that question, I suggest, will be intimately bound up with the question of how we are to understand the role of imagery in episodic recollection. Moreover, one’s stance on the role of imagery in episodic recollection will in turn have a significant bearing on the position one adopts in debates about the similarities between, and differences between, episodic recollection and imagination.2 So the question of how we should understand the proposal that episodic recollection provides for a capacity to travel mentally back in time is not without significance. In this chapter I shall start by introducing and drawing out the consequences of one way of accommodating the proposal. This “re-enactment” view, as I shall label

The Past Made Present  295 it, can be motivated by working through various puzzles associated with the suggestion that episodic recollection amounts to a form of mental time travel. However, while it is a view that is not obviously untenable, I think that there is a preferable alternative, and in the remainder of the chapter I shall work toward elaborating that alternative. That elaboration will explain and draw on connections between the notion that absence can be made present in perceptual imagination, and the notion that the past can be made present in episodic recollection.

1  The Re-enactment View of Mental Time Travel Sometimes the gloss that is put on the idea that episodic recollection amounts to a form of mental time travel invokes a notion of being “mentally transported” to an episode or experience in one’s past. That idea is captured in John Sutton’s remark that “in remembering episodes or experiences in my personal past . . . I am mentally transported away from the social and physical setting in which I am currently embedded.”3 Sometimes the gloss that is put on the appeal to mental time travel invokes the idea that episodic recollection allows one to “re-experience” or “re-live” episodes or experiences in one’s past. For example, Endel Tulving famously claimed that episodic memory “makes possible mental time travel through subjective time, from the present to the past, thus allowing one to re-experience . . . one’s own previous experiences” (2002: 5). One way of connecting those two ideas (the idea of being mental transported across time, and the idea of re-experiencing past events) is as follows. The notion of time travel brings with it the idea of a form of transportation to another temporal location that can make that temporal destination temporally present to the time traveler. If episodic recollection can in some sense enable one to re-live or re-experience some past experience, then that past episode can thereby be made present to one in a way that is akin to being mentally transported to its temporal location. I take it that something like that idea lies behind a suggestion that Hopkins makes as to how we might unpack this talk of mental time travel. He writes, [J]ust as travelling back in time would allow us to experience the events then occurring, so this form of memory gives us access to episodes from our past in a way that is like experiencing them. This distinguishes episodic memory from memory in factual (or ‘semantic’) form. Either might concern an event from one’s past, but, while entertaining factual memories of the event is to call to mind what one believes about it, summoning an episodic memory is something like living it anew. (2014: 313–14) What sense can be made of this idea that episodic recollection makes possible the “re-experiencing” of past events? One might think that it is the imagistic

296  Matthew Soteriou elements of episodic recollection that provide for that distinctive way of “re-experiencing” past events. At its simplest the thought would be that our acts of perceptually imagining objects and events are phenomenologically similar to our conscious perceptual experiences of objects and events; and since imagining an event is somewhat like actually experiencing the event, the fact that episodic recollection involves imagery makes apt the suggestion that this is a form of memory that allows one to “re-experience” events from one’s past. In slightly more developed form, the proposal might be that when one recollects a past event, one reconstructs that past event with imagery, and one thereby simulates a past experience of the event. As the current act of recollection is something like the present simulation of a previous experience, it serves to make present the event recollected in a way that is akin to experiencing it now, and hence living through it, once again.4 A concern that might be raised for this proposal is the following. If an episode of recollection serves to make present a past event by simulating a current experience of that past event, then that conscious act of simulation will thereby represent that past event as concurrent with the episode of recollection. But this would be to misrepresent the temporal location of the recollected event, for of course the event recollected occurs earlier than the episode of recollecting it. The most straightforward reply to this worry is to say that when one simulates an experience in recollecting a past event, the past event is not itself represented as concurrent with the act of recollection. Rather, what happens is this. One imagines an event that is of the same kind as the event previously witnessed, and in so imagining, an event of that kind is thereby made present to one, and it is this imaginative act that simulates one’s past experience of the previously witnessed event. A subsequent concern that might attend this way of defending the proposal is that it may end up making the imagery associated with episodic recollection a mere accompaniment of the act of remembering. The concern I have in mind here is connected with remarks that Russell made when outlining his early account of memory in The Problems of Philosophy and in his 1913 manuscript Theory of Knowledge.5 In The Problems of Philosophy he writes, There is some danger of confusion as to the nature of memory, owing to the fact that memory of an object is apt to be accompanied by an image of the object, and yet the image cannot be what constitutes memory. This is easily seen by merely noticing that the image is in the present, whereas what is remembered is known to be in the past. (1912: 114–5) In the same vein, in Theory of Knowledge he writes, In the first place, we must not confound true memory with present images of past things. I may call up now before my mind an image of a

The Past Made Present  297 man I saw yesterday; the image is not in the past, and I certainly experience it now, but the image itself is not memory. The remembering refers to something known to be in the past, to what I saw yesterday, not to the image which I call up now. (1992: 9–10) In response to this concern it can be said that the generation of imagery in episodic recollection can itself be regarded as an achievement of memory, and not merely an accompaniment of memory, if the generation of the imagery depends, in the right way, on the relevant past perception and is controlled by information that is retained as a result of that past perception. Compare the following example that Martin and Deutscher discuss in their seminal 1966 paper ‘Remembering’: Suppose that someone asks a painter to paint an imaginary scene. The painter agrees to do this and, taking himself to be painting some purely imaginary scene, paints a detailed picture of a farmyard . . . His parents then recognize the picture as a very accurate representation of a scene which the painter saw just once in his childhood . . . Although the painter sincerely believes that his work is purely imaginary, and represents no real scene, the amazed observers have all the evidence needed to establish that in fact he is remembering a scene from childhood. (1966: 167–8) Here we are encouraged to acknowledge that this act of painting is an achievement of memory precisely because the act depends in the right way on the painter’s past perception, even though the painter hasn’t intentionally drawn on knowledge that he has retained as a result of that past perception. So likewise, we might have similar reasons for thinking that the generation of imagery in episodic recollection can itself be regarded as an achievement of memory, and not simply some mental activity that is the mere accompaniment of some other cognitive act that is the genuine act of remembering. However, the comparison with Martin and Deutscher’s well-known example of the painter might give one pause to wonder whether the appeal to imagery in episodic recollection can itself be enough to secure the idea that this form of memory is akin to a form of mental time travel. It is not at all compelling to think that the painter is mentally transported to any past temporal location when he is painting the scene he previously witnessed. So why should we think that in a case in which the generation of imagery depends, in the right way, on some past perception, the effect is to mentally transport the subject to some earlier temporal location? In response it could be conceded that the generation of imagery will not itself suffice to mentally transport one to an earlier time. For that would require a further act of imagining. In particular it would require the

298  Matthew Soteriou cognitive act of imagining or supposing that one’s current act of perceptual imagination is a past experience of an event.6 We might call the account that incorporates that idea a “re-enactment” view of the mental time travel involved in episodic recollection, for it offers a model of mental time travel that is comparable to activity engaged in by those who re-enact historical battles. Those who enjoy that hobby re-create past events by doing things now, and the things they now do are presented to them as temporally present. But they also pretend, or imagine or treat what they now do as some past event. So likewise, according to this “re-enactment” view of the mental time travel involved in episodic recollection, when you recollect some past event you generate some mental imagery, and that act is presented to you as present, but you also pretend or imagine or suppose that this present act is a past experience of an event. That proposal suggests that there is a direct correspondence between mental time travel in episodic recollection and mental time travel in the imagining of future events (or past events). In all cases one mentally transports oneself to another temporal location through imaginatively projecting oneself to that location, and one imaginatively projects oneself across time by imagining that one occupies the relevant temporal location as one engages in acts of perceptual imagination.7 A defender of this sort of account of episodic recollection might say the following in its favor. The proposal accommodates the idea that in acts of episodic recollection an event-kind is made present in perceptual imagination in a way that is akin to simulating a perceptual experience of that eventkind; and this accommodates the idea that episodic recollection is somewhat like experiencing, or living through, that event-kind once again. It also accommodates the idea that this is an achievement of memory. Moreover, it accommodates the idea that one is mentally transported to another temporal location; and since the mechanism of mental transportation is a cognitive act of imagining or supposing that the current act of perceptual imagination is a past experience of an event, the proposal avoids committing to the idea that episodic recollection invariably misrepresents the temporal location of the event recollected. However, detractors of this sort of proposal may complain that the account falters in taking the sensory aspects of episodic recollection to concern the present, and in assuming that it takes a cognitive act of imagining and/or a belief in order to connect the episode of recollection with some past event. A related complaint might target the account’s proposal that the sensory aspects of episodic recollection concern an event-kind, rather than a particular past event. Some argue that it is through relating one to a particular past event that the sensory aspects of episodic recollection afford one a means of re-apprehending that past event—a way of re-apprehending that past event that can account for one’s ability to retain cognitive contact with that individual event, and demonstratively refer to it in a distinctive way.8 But that proposal seems difficult to accommodate if one holds that (a) the sensory aspects of episodic recollection concern an event-kind rather

The Past Made Present  299 than a particular past event and (b) it is only one’s beliefs about particular events in one’s past that makes it possible for one to cognitively relate those sensory aspects of recollection to particular past events. These considerations may not tell decisively against what I have called the “re-enactment” account of episodic recollection. But if one is sympathetic to such complaints, one might think that the source of the account’s error lies in the way that it takes too seriously this talk of mental time travel. In its attempt to accommodate that metaphor, it proposes that in episodic recollection an event is represented as temporally present. That proposal in turn leads to the suggestion that the sensory aspects of episodic recollection concern the present and the suggestion that the sensory aspects of episodic recollection concern an event-kind, rather than a particular past event. The most obvious way to avoid such results is to be skeptical of the metaphor. And indeed some have expressed such skepticism. For example, Alex Byrne writes, [the] striking metaphor of “mental time travel” is misleading. Time travellers have no special experience of the past—when Dr. Who steps out of his TARDIS in the Silurean Period he experiences the events going on then as present. But in episodic recollection events appear as past. (2010: 25) And in a similar vein, Mohan Matthen writes, It is phenomenologically inaccurate to claim, as Tulving (perhaps inadvertently) does, that the memory . . . presents itself as about the present . . . and must therefore be referred to the past by an accompanying belief. Yet that is the implication of the supposition that episodic memory presents itself in just the same way as the original experience. In an important respect, episodic memory is nothing like “time travel.” If I literally travelled back in time to yesterday’s lunch, I would not only have an experience that felt that it is about the present: in fact, it would be about the present. (2010: 8) In what follows I shall be arguing that these concerns with the re-enactment account of episodic recollection need not lead one to reject as inapt the metaphor of mental time travel. One need not retreat to a view on which episodic recollection is no more like mental time travel than is tensed thought about the past, for there is an alternative way of understanding the idea that there is a distinctive respect in which a past event is made present in episodic recollection. This alternative can accommodate the proposal that episodic recollection amounts to a form of mental time travel, but it can do so without committing to the re-enactment account and its problematic features.

300  Matthew Soteriou I shall be suggesting that to make sense of the distinctive way in which the past is made present in episodic recollection we first need to address a related issue about perceptual imagination: we need to make sense of the distinctive way in which absent objects and events can be made present in perceptual imagination. Sartre’s work, The Imaginary, offers important insights on that very issue. In following the trail of Sartre’s insights, I shall develop a proposal about the representation of time in perceptual imagination. This will then be applied in an account of the representation of time in episodic recollection. As we shall see, that account offers a way of accommodating the proposal that episodic recollection amounts to a form a mental time travel, but without committing to a “re-enactment” view.

2  Absence Made Present in Perceptual Imagination In The Imaginary, Sartre suggests that the act of perceptually imagining an object can have much in common with the perception of an image of an object—e.g., the experience of looking at a photo or caricature of someone you know. According to Sartre, “These various cases all act to ‘make present’ an object” (2005: 19). Looking at a photo of Pierre, and also imagining Pierre in sensory imagination can serve to ‘make present’ Pierre. However, in each case Pierre is not there and we know he is not there. That latter point is connected with a characteristic of imagination that Sartre discusses in a section of his book entitled The imaging consciousness posits its object as a nothingness. There Sartre suggests “The characteristic of the intentional object of the imaging consciousness is that the object is not there and is posited as such” (2005: 13). He writes that imagination “gives its object as not being”—as “absent to intuition” (2005: 14). A question to which this naturally gives rise is the following. How are we to understand the respect in which the object of imagination is “made present” to one when one imagines it, granting that the object imagined is not literally made present, and is given as ‘not there’? Consider first the experience of looking at a photograph of your friend. You see the absent friend in something that is present but not identical to your friend—i.e., the photograph. There may be a respect in which the absent object (your friend) is ‘made present’ by your looking at the photograph and seeing your friend in the photograph. But in this case, there are two objects you are aware of—the photo as well as your friend. One of them is present and one of them is absent. If we were to pursue too closely the similarity between the act of perceptually imagining your absent friend and the experience of looking at a photo of your absent friend, we run the risk of reifying mental images as objects of awareness in imagination, and that is something that Sartre warns against. He writes, Whether I perceive or imagine that chair, the object of my perception and that of my [imagining] are identical: it is that straw-bottomed chair

The Past Made Present  301 on which I sit. It is simply that consciousness is related to this same chair in two different ways. (2005: 7) When discussing the notion of a mental image of Pierre as one imagines Pierre, he writes, To tell the truth, the expression ‘mental image’ gives rise to confusion. It would be better to say ‘consciousness of Pierre-as-imaged” or “imaging consciousness of Pierre.” (2005: 7) The imaging consciousness that I have of Pierre is not a consciousness of an image of Pierre: Pierre is directly reached, my attention is not directed at an image, but at an object. (2005: 7) If we follow Sartre in this rejection of the reification of mental images, we can qualify our question as follows. In the case of perceptual imagination, how might we accommodate the respect in which the imagined object is both ‘made present’ and ‘given as absent’ without reifying a mental image and positing two objects—e.g., as well as Pierre, “a certain portrait of Pierre in consciousness”? (2005: 6) The solution to that puzzle, I think, lies in explaining what is correct in a further suggestion that Sartre makes. This is the suggestion that “the time of the object as imaged is an irreality” (2005: 129). Sartre suggests that in perceptual imagination, “There is an absenteeism of time as of space” (2005: 131). This involves denying that “the time of the flowing of the image consciousness is the same as the time of the imaged object” (2005: 129). The time of imagining and the imagined time are, Sartre says, “radically separated” (2005: 129). I think Sartre is on to something important here, and what I think is correct in Sartre’s proposal can be approached by considering two further characteristics of the phenomenology of perceptual imagination that Sartre points to. First, his suggestion that imagining consciousness is given to one as an act of ‘spontaneity,’9 and second, his suggestion that there is something perspectival in perceptual imagination that is connected with what is perspectival in perception.10 I shall now explore each of those suggestions in some detail, starting with the claim about the ‘spontaneity’ of imagination. 2.1  The ‘Spontaneity’ of Imagination How should one go about characterizing the distinction between receptivity and spontaneity, if one thinks that perception is an act of receptivity, whereas imagination is an act of spontaneity? One shouldn’t simply say

302  Matthew Soteriou this: perception is receptive in so far as it is passive, where ‘passive’ means the experience ‘occurs unbidden,’ and/or is ‘not subject to the will.’ That doesn’t suffice to capture the kind of passivity that is distinctive of a ‘receptive’ faculty, for acts of perceptual imagination can occur unbidden and be nonagential, and moreover, agency can be exercised over one’s receptive faculty. However, there is, nonetheless, a difference in the way we are able to exercise agency over each (perception and imagination) that does have to do with a distinctive respect in which receptive occurrence is passive. This in turn, I suggest, is connected with the distinctive temporal phenomenology of receptive occurrence in the stream of consciousness. In summary my proposal is this. (1) There is a respect in which the receptivity of perception is reflected in its phenomenology—and in particular in aspects of its temporal phenomenology. (2) This is relevant to a respect in which, from the subject’s point of view, distinctive constraints are operative on the way in which she can exercise agency over this ‘receptive’ aspect of mind. (3) The relevant constraints aren’t operative when it comes to the agency we are able to exercise over nonperceptual elements of the stream of consciousness, such as imagination. I shall now explain each of these claims in turn, starting with claim (1)—the suggestion that the receptivity of perception is reflected in aspects of its temporal phenomenology. In the case of perception, it seems to one as though the temporal location of one’s perceptual experience depends on the temporal location of whatever it is that one’s experience is an experience of. This is connected with the following point: in the normal case, in explaining why the perception occurs when it occurs, one cites the temporal location of the object of experience. Contrast here the exercise of one’s ability to recollect facts or past events, imagine or think about things. In explaining why these mental acts occur when they occur one doesn’t cite the temporal locations of the intentional objects of the acts, unless one is explaining the occurrence of a particular kind of perceptually based thinking.11 In general, when asking why a mental episode occurs when it occurs, in the case of receptive occurrence we cite the temporal location of the intentional object, and in the case of a nonreceptive occurrence we do not. There is a respect in which the temporal location of one’s perception seems to one to be passive with respect to the temporal location of its object. So it is not just that in the case of perceptual experience a mental episode occurs unbidden, for this can equally be true of other mental acts, including conscious cognitive acts, such as conscious thoughts, as well as acts of perceptual imagination. In the case of perceptual experience, what is distinctive is this: from the subject’s point of view, the course taken by her experience depends on, and is determined by, how things are now with the object of experience. There is then a distinctive sense in which such perceptual occurrences are to be thought of as passive effects on us. In the normal case, perceptual experience of an object is causally sustained by its object. We don’t find an analogue of this in the case of conscious activity

The Past Made Present  303 that isn’t perceptual, or perceptually based, such as conscious imagination, conscious recollection, or conscious calculation and deliberation.12 Let us now move to claim (2). This is the suggestion that the way the receptivity of perception is reflected in aspects of its temporal phenomenology is relevant to a respect in which, from the subject’s point of view, distinctive constraints are operative on the way in which she can exercise agency over this ‘receptive’ aspect of mind. From the subject’s point of view, the only way in which she can exercise agency over the course taken by her perceptual experience is by exercising agency over the obtaining of a relation to the object of perception—by initiating and/or sustaining a relation to the object of experience—e.g., looking at it, watching it, attending to it. In so doing, the subject exercises agency over the course of her perceptual experience by determining which objects, features etc. now affect her—i.e., by determining which objects now determine the course taken by her perceptual experience. And from the subject’s point of view it seems as if the following constraint operates on the way in which she can exercise agency over the obtaining of this relation: the temporal location of the obtaining of the relation depends on the temporal location of the object to which she is so related. From the subject’s point of view, she can only be so related to what is now present (in the temporal sense). This is a distinctive constraint on the way in which agency can be exercised with respect to the receptive faculty, and doesn’t apply to other aspects of mind. Moving on the claim (3), we can see that this constraint isn’t operative when it comes to the agency we are able to exercise over nonperceptual elements of the stream of consciousness, such as conscious thinking, imagining, and recollecting. In the case of conscious thinking, imagining, and recollecting, the temporal locations of the mental acts involved are not determined by the temporal locations of their intentional objects.13 For example, the course taken by one’s conscious thinking is not determined by the temporal location of what one thinks about. One’s thinking can range over the past, the future, and the atemporal, as well as the present. In summary, the following phenomenological claim is being proposed about perception, or sensibility, as receptive faculty: from the subject’s point of view, the domain of sensibility is restricted to items that fall within the bounds of the ‘temporal present.’ This constrains the way in which one can exercise agency over this receptive faculty. Spontaneity is not so restricted. In that respect, spontaneity is ‘free and spontaneous’ in a way that our receptive faculty of sensibility is not. I now want to move on to Sartre’s suggestion that there is something perspectival in perceptual imagination that is connected with what is perspectival in perception. Discussions of this idea usually focus on aspects of spatial representation, but I think we also need to consider the nature of temporal perspective in perception and its analogue in perceptual imagination. However, that will first require clarifying the respect in which perceptual experience is associated with a temporal point of view.

304  Matthew Soteriou 2.2  Temporal Perspective in Perceptual Awareness There is a respect in which perceptual experience doesn’t seem to be temporally perspectival in quite the way that some perceptual experience seems spatially perspectival—for example the way vision seems to be spatially perspectival. In the case of vision, the spatial point of view afforded by one’s experience does phenomenologically seem to be perspectival in the following respect: it seems to involve the perception of X from Y, where X and Y are spatial locations, and where X is not identical to Y. Whereas, the temporal point of view that is afforded by perception does not phenomenologically seem to be perspectival in that way. That is to say, it doesn’t seem to involve the perception of X from Y, where X and Y are temporal locations, and where X is not identical to Y. Perceptual experience does seem to feature objects of awareness that are non-instantaneous. So there is a respect in which perceptual experience does seem to afford conscious awareness of an interval of time. However, phenomenologically speaking, it doesn’t seem as though it affords one a perceptual point of view on that interval of time from a temporal location that is distinct from that interval of time (e.g., from a temporal location that falls within that interval of time). How then should we understand the respect in which perceptual awareness of the relevant interval of time nonetheless brings with it something like a temporal perspective? I suggest that the following two negative phenomenological points can be made about the interval of time one seems to be afforded conscious awareness of in perception. (i) The boundaries of the temporal interval do not seem to mark out for their subject the boundaries of time. This is connected with the Kantian claim that in vision, the region of space that you seem to be aware of is presented as a sub-region of a region of space; and likewise, the temporal interval you are consciously aware of is presented as a sub-interval of an interval of time. (ii) The boundaries of the temporal interval do not mark out for their subject the temporal boundaries of their experience—i.e., boundaries of the temporal interval do not mark out for their subject the beginning and endpoints of their perceptual experience. For example, if I am continuously watching a continuously moving object over some extended period of time, just as it doesn’t seem to me as though the object stops moving at sub-intervals of that extended period of time, it doesn’t seem to me as though my experience of the object stops occurring at sub-intervals of that extended period of time. To these negative remarks about the phenomenology we can add the following more positive remarks about the interval of time one seems to be afforded conscious awareness of in perception. There is an asymmetry in one’s psychological orientation to what falls on either side of the boundaries of the relevant interval of time. This asymmetry in one’s psychological orientation to what falls on either side the boundaries of the relevant interval of time amounts to a tensed temporal orientation to the immediate past and

The Past Made Present  305 immediate future. That is to say, the boundaries of the relevant interval of time mark out for the subject of experience the boundaries between past and future. In consequence, everything that falls within the temporal interval is given as falling within an interval of time that intervenes between past and future. So everything that falls within the interval is thereby experienced by one as temporally present, insofar as it is experienced as falling within an interval of time that intervenes between what is given to one as past and future. It is this asymmetry in one’s psychological orientation to what falls on either side of the interval—an asymmetry that constitutes a tensed temporal orientation to the past and future—provides for a sense in which perceptual awareness of an interval of time brings with it something like a temporal perspective.14 Now let us return to the idea that there is something perspectival in perceptual imagination that is connected with what is perspectival in perception—starting with some of the more familiar claims about spatial representation in perception and imagination. 2.3  Spatial and Temporal Perspectives in Perceptual Imagination When a subject sensorily imagines a scene, she typically imagines a spatial point of view on objects within the imagined scene. For example, in visualizing an array of objects, some objects may be imagined as being to left, and others to the right, from an imagined point of view. In saying that the spatial point of view, and not just the array of objects, is itself imagined, I mean the following. The center of origin of the spatial point of view from which objects are visualized to the left and right is not determined by the actual spatial location and orientation of the subject who is visualizing. For example, suppose you are lying in bed on your back with your head facing toward the ceiling, and suppose that, so situated, you close your eyes and visualize a mountain range. You do not thereby imagine the mountain range as occupying a spatial location relative to your actual location—i.e., somewhere above the spatial location that is actually occupied by your bed. If you happen to move your head as you visualize that scene, you do not thereby imagine a change in the spatial location of the scene you visualize. As you visualize, any change in your actual spatial location is consistent with no change in the imagined spatial location of the scene you imagine, and consistent with no change in the spatial location of the origin of the imagined point of view from which aspects of the scene are imagined as being to the left and right. Suppose that you stop visualizing a mountain range and instead start visualizing a beach. In such a case there may be nothing to determine the represented spatial relations between these imagined scenes—i.e., the mountain range and the beach—at least not if that question is not settled by your intentions in so imagining. These aspects of spatial representation in imagination can be explained by the following proposal: in imagining an object or event, one imagines (and

306  Matthew Soteriou thereby represents) a perceptual perspective on that object or event. Suppose we agree with Sartre that an act of imagining is an act of spontaneity. What is now being suggested is this. When one imagines an object, that act of imagining is an act of spontaneity that represents a perceptual perspective on that object. So it is an act of spontaneity that represents a perspective on the intentional object that is afforded by an act of receptivity. As the act of imagining is an act of spontaneity, its temporal location is not determined by (and doesn’t seem to its subject to be determined by) the temporal location of its intentional object. But by representing a perspective on the object that is afforded by an act of receptivity, it represents a temporal perspective on the object that is afforded by an act of receptivity. Which is to say, it represents a temporal perspective that presents that object as temporally present. As the temporal location of the act of imagining is not determined by the temporal location of its intentional object, the temporal location of the represented perceptual perspective is also not determined by the temporal location of the act of imagining (and hence needn’t be presented as coincident with the temporal location of the act of imagining). So in the case of imagination, the represented temporal location of the represented temporal present is not determined by (and hence needn’t be presented as coincident with) the temporal location of the act of imagining. Putting all this together delivers the following proposal: The act of imagining is the representation of a temporal present, and the represented temporal location of that represented temporal present is not determined by the temporal location of the act of imagining. For instance, when you visualize a friend walking toward you, there is a respect in which the successive temporal parts of her approach are each imagined as being temporally present—e.g., now she is walking toward the traffic light, now she stops at the traffic lights, now she is crossing the road, and so on. However, the temporal location of your act of imagining does not determine a represented temporal location of the event you imagine. That is to say, when you imagine your friend walking toward you, you needn’t thereby be imagining that her approach occurs at the actual time of your act of imagining. You could be imagining a future encounter, or you could be imagining a past encounter you wished for, and indeed the question of the time of the imagined event could be left entirely open. Note that the thesis being proposed is not that the represented temporal location of an intentional object of one’s imagining cannot be determined by anything. Clearly this can be determined by one’s intentions in so imagining. One can intend to imagine a future event, intend to imagine a past event, or intend to imagine a present event. But the key point for our purposes is this: the represented temporal location of the intentional object of imagination is not determined by anything that’s independent of one’s intentions in so imagining. In particular, it isn’t determined by the temporal location of the act of imagining. This is what I take to be correct in Sartre’s suggestion that

The Past Made Present  307 the “the time of the object as imaged is an irreality,” and what is correct in his proposal that we should deny that “the time of the flowing of the image consciousness is the same as the time of the imaged object.”15 We are now in a position to provide an answer to the Sartrean question I posed earlier about the way in which absence can be made present in perceptual imagination: in the case of perceptual imagination, how do we accommodate both ‘presence’ and ‘absence’—the respect in which the imagined object is both ‘made present’ and yet ‘given as absent,’ as ‘not there’—without reifying the mental image and positing two objects? First, consider what we are now in a position to say about the distinctive respect in which in perceptual imagination the intentional object is ‘made present.’ Imagination represents a temporal perspective on its intentional object. It represents the temporal perspective afforded by an act of receptivity. That represented temporal perspective is one that presents the intentional object of imagination as ‘temporally present.’ And it is in that respect that imagination provides one with a distinctive way of ‘making present’ its intentional object. Now consider what we can also say about the respect in which the intentional object that is ‘made present’ in imagination is nonetheless given as ‘not there.’ The temporal present that is represented in imagination isn’t presented as coincident with one’s actual present—i.e., the actual time of one’s act of imagining. This answer to the Sartrean question about the way in which absence can be made present in perceptual imagination depends on a clarification of the way in which perceptual imagination can serve to represent a ‘here and now’ that is not presented to one as one’s actual here and now—i.e., that is not presented to one as the actual here and now that is determined by the place and time of the act of imagining. It depends on making sense of the idea that perceptual imagination offers a way of representing a ‘here and now’ that isn’t shackled to the spatiotemporal location of the mental act of representing. Let us now consider how we might apply that proposal in making sense of the notion that the past is ‘made present’ in episodic recollection, and in a way that can capture the idea that episodic recollection offers a form of mental time travel.

3  Mental Time Travel in Episodic Recollection The account of perceptual imagination that I have sketched opens up the prospect of an account of the representation of time in episodic recollection that can accommodate the idea that such mental acts involve the representation of a past event as temporally present, and in a way that avoids the re-enactment view of episodic recollection. The proposal is this. When one episodically recollects a past event, one’s act of recollection is an act of spontaneity that represents a perceptual perspective on that past event—so it is an act of spontaneity that represents a perspective on that past event that is afforded by an act of receptivity.16 As the act of recollection is an act

308  Matthew Soteriou of spontaneity, its temporal location is not determined by (and doesn’t seem to its subject to be determined by) the temporal location of its intentional object. But by representing a perspective on the intentional object that is afforded by an act of receptivity, it represents a temporal perspective on the intentional object that is afforded by an act of receptivity. Which is to say, it represents a temporal perspective that presents that past event as temporally present. So the act of recollection involves the representation of a temporal present, and the represented temporal location of that represented temporal present is not determined by the temporal location of the act of recollection. To say that much isn’t yet to distinguish episodic recollection from an act of imagining a past event. So, assuming we want to respect that difference, what more can be added? I think we can find a way to respect a significant difference between episodic recollection and an act of imagining a past event by addressing the following question: In the case of episodic recollection, what determines the temporal location of the represented temporal present? In the case of perceptual imagination, the temporal location of the represented temporal present is determined, if at all, by the subject’s intentions in so imagining. Nothing about the act of imagining determines that the represented event takes place in the subject’s present, past, or future, independently of the subject’s intentions in so imagining. Whereas, by contrast, in the case of episodic recollection, we can make the following set of claims. The temporal location of the represented temporal present is determined by the temporal location of the past event that is represented, and that which determines the temporal location of the past event that is represented is whatever it is that determines which particular past event is represented. This is because particular events, unlike particular objects, cannot continue to exist at different temporal locations. So that which determines the temporal location of the represented temporal present is whatever it is that determines which particular past event is represented. And in the case of episodic recollection (in contrast with imagination) the question of which particular past event is represented is not determined by the subject’s imaginative intentions in so representing. It is determined by the causal ancestry of the memory. At the outset I suggested that part of the appeal of invoking the notion of mental time travel is to signal that the way of relating to one’s past and future that is afforded by episodic recollection and imagination is significantly different from merely entertaining tensed thoughts about earlier and later times. The account that I have offered of the representation of time in perceptual imagination and episodic recollection respects that difference. According to it, acts of perceptual imagination and episodic recollection allow one to slip the knot of one’s actual present in a distinctive way, for they provide one with a distinctive way of representing entities as temporally present. In such acts the temporal location of the represented temporal present isn’t shackled to the actual time of representing. The way in which this account accommodates the notion of mental time travel in episodic recollection differs in some significant respects from the

The Past Made Present  309 “re-enactment” view that I sketched in Section 1. Recall that on the reenactment view, sensory aspects of episodic recollection concern the subject’s actual present—which is to say they concern the time of recollecting. It is this aspect of the account that leads to its proposal that sensory aspects of episodic recollection concern event-kinds, rather than a particular past event. For proponents of the view will likely want to avoid committing to the claim that episodic recollection invariably misrepresents the temporal location of the event recollected, and by saying that sensory aspects of episodic recollection concern event-kinds, rather than a particular past event, that commitment is avoided. On the re-enactment view, the generation of imagery may offer a way of simulating the kind of perceptual experience one previously underwent, but the generation of imagery and consequent simulation of that kind of perceptual experience will not suffice to mentally transport one to another temporal location. So on the re-enactment view, in all cases, in order to mentally transport oneself to another temporal location, one needs to imaginatively project oneself across time by imagining or supposing that one occupies the relevant temporal location as one engages in acts of perceptual imagination. By contrast, according the alternative account I have just been outlining, sensory aspects of episodic recollection are not tied to the subject’s actual present. The represented temporal present associated with such imagistic episodes does not coincide with the time of recollecting. The temporal location of the represented temporal present is, rather, determined by the temporal location of the particular past event that is recollected. So this alternative account of episodic recollection can accommodate the idea that sensory aspects of episodic recollection concern past temporal locations, rather than the time of recollecting, and it can accommodate the idea that sensory aspects of episodic recollection concern particular past events, rather merely concerning event-kinds. It can accommodate the idea that those past events and their temporal locations are ‘made temporally present’ to you in episodic recollection in a way that transports you to their temporal location, rather than transporting those past events to your current temporal location as you recollect. Moreover, the means by which this is achieved doesn’t require any cognitive act of imagining or supposing on your part. Indeed the account I have proposed allows that one might episodically recollect a past event even if one is unsure whether one is genuinely remembering anything. Even in this instance, there is a respect in which one is mentally transported to a past location whether one realizes it or not. For in that act of episodic recollection an actual past event is represented as temporally present, and the temporal present that is thereby represented is the past time at which the event took place.17

Notes 1 See also Schacter et al., 2007. 2 For examples of discussions of these debates, see Debus, 2014, and Hopkins, 2014 and Hopkins (forthcoming).

310  Matthew Soteriou 3 Sutton, 2009, p. 217 4 It should be said that there are a variety of different simulation approaches to episodic recollection that I am not here distinguishing. See for example, Schacter, Addis, & Buckner, 2008; Mullally & Maguire, 2014; Shanton & Goldman, 2010; and Michaelian, 2016. 5 Some have suggested that Russell’s account of memory changed by the time he began work on Theory of Knowledge. For discussion of that debate, see Martin, 2015. 6 Arguably, a ‘feeling of familiarity’ accompanying the imagery would not suffice for transporting one to another temporal location, for it would only suffice for the sense that an event of the same kind had been previously experienced. An appeal to a feeling of familiarity is made in Russell’s later (1921) account of memory, and also Broad’s (1925) account. Matthen (2010) presents an account in which a feeling of familiarity plays a rather different role. 7 According to the re-enactment view I have in mind, in the case of episodic recollection, the act of mentally transporting oneself to another temporal location will likely be grounded in knowledge and beliefs one has about particular events in one’s past. Imagination, or supposition, nonetheless has a crucial role to play in the account, because the account attempts to accommodate the notion that something present (i.e., the simulated experience, which occurs at the time of recollection) is represented as occupying an earlier temporal location; and imagining or supposing that what is now present occupies an earlier temporal location is the obvious way to avoid the charge that this involves a problematic form of misrepresentation of the temporal location of that which is now present. 8 Different proposals about the way in which episodic recollection can provide for a distinctive form of cognitive contact with a particular past events can be found in McDowell, 1978, Campbell, 2001, Hoerl, 2001, Martin, 2001, and Debus, 2008. For a discussion of puzzles associated with certain ways of accommodating that proposal by way of the claim that the past event is a constituent of the episode of recollection, see Martin, 2015. 9 Sartre, 2005, p. 14. 10 This is a characteristic of imaging that Sartre refers to as ‘The Phenomenon of Quasi-Observation’ (2005, pp. 8–11). 11 In the case of episodic recollection, the event recollected will of course be earlier than the act of recollection. However, that fact does not itself explain why an act of recollection occurs when it occurs. For example, that I just now recollected a particular event in my childhood (rather than, say, fifteen minutes earlier) cannot be explained by appeal to the fact that the past event I recollected is earlier than my act of recollection. 12 Compare again the case of episodic recollection. When one tries to recollect some past event and one mentally reaches for some patch of the past, that past event cannot initiate and causally sustain, and hence determine the course of, some current episode of recollecting what happened. This is connected with the fact that if one so chooses one can recollect the different temporal stages of an earlier sequence of events in an order that differs from the order in which they occurred. 13 It might be suggested that in the case of episodic recollection, when one recollects a past event one initiates an epistemic relation of ‘acquaintance’ with that past event; and so in recollecting, one exercises agency over the obtaining of that relation. However, even if one accepts that proposal, the important point of contrast with perception is this: in the case of episodic recollection the temporal locations of the mental acts that are involved in the exercise of that agency are not determined by the temporal locations of the intentional objects of those acts. See footnotes 11 and 12.

The Past Made Present  311 14 The proposal I have just sketched has affinities with aspects of Husserl’s account of time-consciousness. Husserl (1905) appeals to asymmetrical psychological orientations that are temporal—both retention and protention—in his account of the way in which the ‘now’ in perception is the hub of orientations to ‘temporal fringes’ of the ‘now.’ If consciousness did not transcend the ‘now’ there would no awareness of the present as such. One finds similar ideas in Brian O’Shaughnessy’s discussion of temporal experience, where he suggests that there is an irreducible ‘co-presence’ of past and future in the experienced present. He says at one point, “Close up the past, wall off the future, and you cover over the present too” (O’Shaughnessy, 2000, p. 62). 15 Sartre, 2005, p. 129. 16 Note that to accept this much is not to commit to the claim that episodic recollection necessarily involves the recollection of a particular past perceptual experience one had. I do not have the space here to consider whether there are adequate grounds for accepting that further claim. 17 I am very grateful to the editors of this volume, to two anonymous referees, and to Bill Brewer for helpful comments on an earlier draft of this paper.

References Broad, C. D. (1925). The mind and its place in nature. London: Routledge & Kegan Paul. Byrne, A. (2010). Recollection, perception, imagination. Philosophical Studies, 148(1), 15–26. Campbell, J. (2001). Memory demonstratives. In C. Hoerl & T. McCormack (Eds.), Time and Memory (pp. 169–186). Oxford: Oxford University Press. Debus, D. (2008). Experiencing the past: A relational account of recollective memory. Dialectica, 62(4), 405–432. Debus, D. (2014). “Mental Time Travel”: Remembering the past, imagining the future, and the particularity of events. Review of Philosophy and Psychology, 5(3), 333–350. Hoerl, C. (2001). The phenomenology of episodic recall. In C. Hoerl & T. McCormack (Eds.), Time and memory: Issues in philosophy and psychology (pp. 315– 338). Oxford: Oxford University Press. Hopkins, R. (2014). Episodic memory as representing the past to oneself. Review of Philosophy and Psychology, 5(3), 313–331. Hopkins, R. (Forthcoming). Imagining the past: On the nature of episodic memory. In F. Dorsch & F. Macpherson (Eds.), Perceptual memory and perceptual imagination. Oxford: Oxford University Press. Husserl, E. (1905/1964). The phenomenology of internal time-consciousness (M. Heidegger, Ed. & J. S. Churchill, Trans.). Bloomington, IN: Indiana University Press. Martin, C. B., & Deutscher, M. (1966). Remembering. Philosophical Review, 75, 161–196. Martin, M. G. F. (2001). Out of the past: Episodic recall as retained acquaintance. In C. Hoerl & T. McCormack (Eds.), Time and memory. Oxford: Oxford University Press. Martin, M. G. F. (2015). Old acquaintance: Russell, memory, and problems with acquaintance. Analytic Philosophy, 56(1), March, 1–44.

312  Matthew Soteriou Matthen, M. (2010). Is memory preservation? Philosophical Studies, 148(1), 3–14. McDowell, J. (1978). On the reality of the past. In A. Hookway & A. Pettit (Eds.), Action and interpretation. Cambridge & New York: Cambridge University Press. Michaelian, K. (2016). Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press. Mullally, S. L., & Maguire, E. A. (2014). “Memory, imagination, and predicting the future: A common brain mechanism? The Neuroscientist, 20(3), 220–234. O’Shaughnessy, B. (2000). Consciousness and the world. Oxford: Oxford University Press. Russell, B. (1912). The problems of philosophy (G.N. Clark, Gilbert Murray and G. De Beer, Eds.). Home University Library. London: Williams & Norgate. Russell, B. (1921). The analysis of mind. London: Allen and Unwin. Russell, B. (1992). Theory of knowledge the 1913 manuscript. London: Routledge. Sartre, J. P. (2005). The imaginary: A phenomenological psychology of the imagination (J. Webber, Trans.). London: Routledge. (originally L’Imaginaire 1940) Schacter, D. L., Addis, D. R., & Buckner, R. L. (2007). Remembering the past to imagine the future: Theprospective brain. Nature Reviews Neuroscience, 8, 657–661. Schacter, D. L., Addis, D. R., & Buckner, R. L. (2008). Episodic simulation of future events: Concepts, data, and applications. Annals of the New York Academy of Sciences, 1124(1), 39–60. Shanton, K., & Goldman, A. (2010). Simulation theory. WIREs Cognitive Science, 1(4), 527–538. Suddendorf, T., & Corballis, M. C. (2007). The evolution of foresight: What is mental time travel, and is it unique to humans? Behavioral and Brain Sciences, 30(3), 299–313. Sutton, J. (2009). Remembering. In P. Robbins & M. Aydede (Eds.), The Cambridge handbook of situated cognition (pp. 217–235). Cambridge: Cambridge University Press. Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53(1), 1–25.

16 Remembering Past Experiences

Christoph HoerlRemembering Past Experiences

Episodic Memory, Semantic Memory, and the Epistemic Asymmetry Christoph Hoerl There seems to be a distinctive way in which we can remember events we have experienced ourselves, which differs from the capacity to retain information about events that we can also have when we have not experienced the relevant events ourselves but just learned about them in some other way. Psychologists and increasingly also philosophers have tried to capture this difference in terms of the idea of two different types of memory: episodic memory and semantic memory. Yet, the demarcation between episodic memory and semantic memory remains a contested topic in both disciplines, to the point of there being researchers in each of them who question the usefulness of the distinction between the two concepts.1 In this chapter, I outline a new characterization of the difference between episodic memory and semantic memory, which connects that difference to what is sometimes called the ‘epistemic asymmetry’ between the past and the future, or the ‘epistemic arrow’ of time. My proposal will be that episodic memory and semantic memory exemplify the epistemic asymmetry in two different ways, and for somewhat different reasons, and that the way in which episodic memory exemplifies the epistemic asymmetry is manifest to the remembering subject in a way in which this is not the case for semantic memory. I will start with a brief sketch of some of the existing work on the epistemic asymmetry, before turning to the question as to how exactly the idea of the epistemic asymmetry applies in the case of memory, and whether it might apply in a somewhat different form to different kinds of memory. The latter questions are, I believe, questions of independent interest that have to date been largely ignored in the literature on the epistemic asymmetry.2 However, as I will try to show, focusing on them can also help shed new light on some already existing characterizations of the contrast between episodic and semantic memory, for instance by providing the materials for fleshing out a sense in which episodic memory involves the preservation of a distinctive form of cognitive contact with events, as well as being past-directed in a way semantic memory isn’t. My primary aim is to set out an agenda for work in this neglected area of research, which is why some of my claims remain fairly programmatic at this stage.

314  Christoph Hoerl

1  The Epistemic Asymmetry Philosophers working on the metaphysics of time often speak of a number of different asymmetries or ‘arrows’ that time seems to exhibit: that events become successively present in the direction of the future, rather than the past (the ‘arrow of time’), that entropy increases over time in the same direction (the ‘thermodynamic arrow’), or that causes always precede their effects (the ‘causal arrow’). The general question philosophers are typically interested in, in this context, is how these arrows might be related to one another, and in particular whether some of them might ground others in a way that reveals as illusory certain asymmetries that time appears to exhibit, in the sense that they are due not actually to the metaphysical nature of time itself, but rather to some contingent features of how things are arranged in time.3 One of the asymmetries that philosophers have also discussed in this context concerns our knowledge of the past versus our knowledge of the future, or what is sometimes called the ‘epistemic arrow.’ There is clearly some sense in which our knowledge of the past is different from our knowledge of the future. Yet, as the following quote from David Albert brings out, it can be difficult to get a more precise fix on what exactly the difference at issue comes to: The sort of epistemic access we have to the past is different from the sort of epistemic access we have to the future. This (to put it mildly) nobody doubts. And nonetheless . . . there is a vast . . . literature nowadays about the alleged difficulty of specifying exactly what that difference is. It’s often pointed out, for example, that the difference certainly does not consist in our having knowledge of the past but none of the future. We do, after all, have knowledge of the future. We know (for example, and not less certainly than we know much of what we know of the past) that the sun will rise tomorrow. And if it’s said that we know more of the past than we do of the future, this seems (according to the usual way of talking) true enough, but (as it stands) not particularly informative—it seems to give us nothing at all that we can reason any further with, nothing that (as it were) we can sink our teeth into. Sometimes the focus is shifted to differences between the methods by which we come to know things about the past and the future. It’s said (more particularly) that there can be such things as records only of the past; but this is almost always immediately followed up with whining about the perennial elusiveness of exactly what it means to be a ‘record,’ and cluelessness follows again. (Albert, 2000, p. 113) I will discuss some of the issues alluded to in this passage in more detail in what follows. For the moment, I just want to note that what there has been to date

Remembering Past Experiences  315 by way of a philosophical discussion regarding the nature and source of the epistemic asymmetry has actually been conducted almost exclusively within the context of the philosophy of science and the metaphysics of time, rather than epistemology itself. Perhaps connected with this, existing discussions have also not primarily been concerned with the way in which knowledge or memory themselves may exhibit a temporal asymmetry. Rather, as indicated in the quote from Albert, the main focus has ultimately been on the general idea that we can have records of the past but not of the future, where such records have typically been conceived of as things that we can acquire knowledge from. Within the context of the present chapter, there is only space for a very brief and crude sketch of some of the work that has been going on in this area. One useful starting point is the observation that there are ways we have of finding out about things in the world that do not display a past/ future asymmetry. Laplace’s demon provides a vivid demonstration of how, under the assumption of determinism, both past and future states of the world could, in principle, be inferred from the present state of the world using dynamical laws. We may not have the demon’s comprehensive knowledge of the present state of the world or his powers of deduction, and our world may also not be governed by strict determinism; yet, as long as we stick to the method of making inferences across time using dynamical laws, it is not obvious why these factors should differentially affect our ability to do so in the future and the past direction, respectively. The only thing that really breaks the parity between the past and the future and introduces the epistemic asymmetry, then, is that there is also a separate way in which we can have or obtain knowledge of nonpresent events or states of affairs, other than by applying dynamical laws, but it is only past events and states of affairs that this separate way can give us epistemic access to. This is what the idea that there can be records of the past, but not of the future, is trying to capture. To use a paradigmatic example discussed by Reichenbach (1956), a footprint on a beach can tell us of a person walking there in the past, but there is nothing that could tell us in an analogous way about future events.4 Attempts to explain the existence of the epistemic asymmetry, thus understood, typically link it to another asymmetry in time: the thermodynamic asymmetry. This consists in the fact that the world, as we experience it, is marked by an earlier/later entropy-gradient. As Jill North (2011, p. 313) explains: Our everyday experience is largely of physical processes that occur in only one direction in time. A warm cup of coffee, left on its own in a cooler room, will cool down during the day, not grow gradually warmer. A box of gas, opened up in one corner of a room, will expand to the volume of the room; an initially spread-out gas won’t contract to one tiny corner. A popsicle stick left out on the table melts into a hopeless mess; the hopeless mess sadly won’t congeal back into the original popsicle.

316  Christoph Hoerl Each of these processes is one of a transition from a lower-entropy state to a higher-entropy state, and we only find these transitions occurring in one direction in time, and never in the reverse direction.5 Crudely, what is meant by an entropy-increase here is that, as it is sometimes put, a system in a state that exhibits a particular form of order (all the warmth concentrated in the coffee cup, all the gas condensed in the box in the corner, the popsicle occupying a small, well-defined place on the table) evolves into one that exhibits less order of this kind. How might the thermodynamic asymmetry be connected to the epistemic asymmetry? Consider again the example of a footprint on a beach. The section of the beach containing the footprint is a physical system that, in virtue of containing the footprint, exhibits a (relatively) low state of entropy: the way the grains of sand are distributed involves a relatively sharp closed boundary between two regions: the region inside the footprint and the region outside it. Given what we have just said about the way such systems evolve over time, this is a type of state that we would not expect to arise spontaneously just from the movements of the grains of sand on the beach. Left on its own, we would expect the footprint to disappear after a while, as the sand gets blown around in the wind; we would never expect to see the reverse of this process. This is what makes the footprint, considered as a particular kind of physical arrangement, capable of acting as a record.6 What it records is an interaction between that section of the beach and some other physical system itself exhibiting a low state of entropy: a human walking along the beach. This is what the footprint is a record of, but, given the direction of the thermodynamic asymmetry, in so far as this is what is recorded by the footprint, it must be something that lies in the footprint’s past, rather than in its future. Something broadly like this story is accepted in much of the current literature, although there have also been a number of criticisms that suggest that, at the very least, it needs to be refined if it is supposed to provide an accurate account, e.g., of the level of detailed knowledge we take ourselves to have about the past, and of the variety of ways in which we can derive inferences about the past from present evidence.7 I will not rehearse these existing debates in what follows. Rather, my focus will be on a different aspect in which this story, even if true, remains rather incomplete. As I want to argue, it gives us an account of only one type of phenomenon connected with the epistemic asymmetry, not least leaving out the particular aspect of the epistemic asymmetry that we are arguably most intimately acquainted with.

2  The Epistemic Asymmetry and Semantic Memory Part of my aim in relating the distinction between episodic memory and semantic memory to the general issue of the epistemic asymmetry of time is to provide a taxonomy of several different types of ‘records,’ all of which

Remembering Past Experiences  317 exhibit a type of past/future asymmetry, but with each of them doing so in a somewhat different form.8 Or so I will argue. A first question to consider in this context is in what respects memories, in general, differ from phenomena such as the footprint on the beach. I will assume that, at some level of abstraction, memories, too, may be described as records, and there may also ultimately be a common underlying explanation, perhaps of the kind sketched in the previous section, as to why both memories and other types of records exhibit a past/future asymmetry (where the nature of this asymmetry is to be spelled out further for the case of memories in what follows). Yet, there is clearly also a sense in which we are dealing with phenomena of two somewhat different kinds. One obvious way in which memories are different from phenomena such as the footprint on the beach is that the former are themselves epistemic states, they embody pieces of knowledge. The latter, by contrast, are items from which pieces of knowledge about the past can be obtained, but doing so first requires the subject drawing an inference (e.g., from the existence of the footprint to the past presence of another person on the beach). One way of explaining this difference is in terms of the idea that memories, whilst falling under the general category of records, are records specifically of epistemic activity; even more specifically, they are records of the formation of the very epistemic state that the subject’s subsequent remembering consists in, her memory consisting precisely in the fact that it has been preserved.9 In so far as items such as the footprint on the beach also count as records, by contrast, they are typically records of nonepistemic activity. And even if some of them may, in some sense, also be described as records of epistemic activity—the knot I have put in my handkerchief to remind myself of something may perhaps qualify as some such—they are not themselves epistemic states formed by this activity.10 The fact that memories themselves embody knowledge, whereas phenomena such as footprints on the beach are merely items from which knowledge may be obtained by inference, constitutes an important difference between the two.11 In the case of semantic memory, however, there is also a further important dimension of difference from the case of footprints. I have said that, like a footprint, semantic memories are records, though they differ from a footprint in being records specifically of epistemic activity. I have also said that semantic memories embody knowledge (rather than just serving as items from which knowledge can be derived). Yet, crucially, the knowledge that semantic memories embody is not knowledge of the epistemic activity they are records of. The knowledge they embody derives from that activity, but it is not knowledge about that activity. This becomes particularly clear when we consider that there are no restrictions regarding the tense of beliefs that can be stored in semantic memory. The astronomer who has figured out that there will be a lunar eclipse next week can retain this knowledge in semantic memory, but her memory is clearly not a record of that lunar eclipse, which is still to occur

318  Christoph Hoerl in the future. Rather, it is a record of her activity of having calculated the date of the eclipse. Similarly for memories of tenseless facts: After years of study, the mathematician may remember the Cauchy-Peano theorem, but her memory is clearly not a record of the theorem in the sense of that term associated with the epistemic past/future asymmetry. This means that in the case of semantic memories with a past-tensed content, too, it is important to distinguish between the question as to what is being remembered, i.e., what the content of the memory is, and what the memory is a record of. There is also an important consequence of this when it comes to specifying the precise sense in which semantic memory exhibits an epistemic asymmetry. Crucially, there is no built-in past/future asymmetry in the knowledge semantic memories can embody. Any such asymmetry is at best a quantitative one—i.e., one concerning the amount of knowledge retained that concerns the past and the future, respectively—and, where such a quantitative asymmetry exists, it is (with one exception that I will get to) inherited from an asymmetry connected to the particular epistemic activity the relevant semantic memories are records of, rather than being explained by the nature of semantic memory itself. Thus, it would come as no surprise, for instance, that an archaeologist’s semantic memory contains a great deal of knowledge about the past, since the main method of inquiry they pursue in their professional life consists in interpreting archaeological records, which are evidence only of past, and not future, events. In an exactly analogous way, though, the semantic memory of a scientist programming the future flight path of an interplanetary space probe could be expected to contain a great deal of knowledge of the future, because of the many calculations she has carried out to determine the future orbits of the planets to be encountered by the probe on its mission. The one exception that is probably worth mentioning separately here are semantic memories about events that are in fact based on our own past experiences of these events. Amnesic patients often retain knowledge about events that have happened to them, even though they are unable to recollect them in episodic memory,12 and whilst some of that knowledge may have been acquired, say, through later testimony, there is no reason to rule out that some of it might also trace back to their own experiences of the relevant events, even though the only form in which it is now still available is that of a semantic memory. In this sort of case, the time of the event the memory is concerned with in fact coincides with that of the epistemic activity it is a record of—a feature that, as I will suggest, semantic memories with this specific sort of history share with episodic memories. And of course, semantic memories with this specific sort of history are, in virtue of this and given the general temporal asymmetry of records, restricted to carrying information about past events. Yet, crucially, because the knowledge regarding the relevant events has only been retained in the form of semantic memories, there is an important sense in which the epistemic asymmetry is no more manifest in these memories

Remembering Past Experiences  319 than it is in other semantic memories concerning past events. Specifically, the fact that the time of the event the memory is concerned with coincides with that of the epistemic activity it is a record of is not manifest as part of the memory itself. It is in just this respect, I now want to argue, that there is a crucial difference between semantic and episodic memory.

3  Episodic Memory and the Epistemic Asymmetry Just as blanket statements to the effect that we know more about the past than we do about the future are of little help in trying to explicate the epistemic asymmetry (if not outright false), so it is with another sort of blanket statement sometimes made in this context, viz. that we can remember the past but not the future. As should be clear from what I said in the previous section, when it comes to semantic memories, their contents can as easily range over the future as over the past, or they can even comprise tenseless truths such as those of mathematics. Semantic memories do exhibit the epistemic asymmetry, in so far as they fall under the general category of being records, but their doing so does not show up in the knowledge they can embody; it is not manifest in the knowledge possession of which having the relevant memories consist in. Yet there is arguably one specific form of memory to which the statement that we can remember the past but not the future does apply, and this is what is commonly referred to as episodic memory. To a first approximation, episodic memory is the type of memory that allows us to remember particular events themselves, as they happened, in a way that is different from simply retrieving information about them of the kind that can also be retained in semantic memory. And in contrast to semantic memory, it seems to be the knowledge that is retained in epistemic memory itself that exhibits a temporal asymmetry. We can remember events in this specific way only if they have already happened. Moreover, it is not just de facto the case that we can have episodic memories only of past events, and not of future ones; this is also something that is obvious to the remembering subject. Episodic recollection itself seems to involve a distinctive awareness, on the part of the subject, of the remembered events as lying in the past, in virtue of the particular type of mental state it is.13 As should be clear from what has been said so far, this specific way in which episodic memory exhibits a past/future asymmetry cannot simply follow from the type of epistemic asymmetry also exhibited by semantic memory. Neither, it seems, can it simply derive from the one at issue in our capacity to make inferences about past events based on phenomena such as footprints on a beach. So we need to ask what grounds this more specific type of asymmetry. Earlier, I suggested a general characterization of the distinctive sense in which memories are records, according to which what is distinctive about memories is that they are records specifically of epistemic activity; and even

320  Christoph Hoerl more specifically, they are records of the formation of the very epistemic state that the subject’s subsequent remembering consists in, her memory consisting precisely in the fact that it has been preserved. In the case of semantic memories, however, I also suggested that, whilst they are records of epistemic activity, the knowledge they embody is not knowledge of that epistemic activity; what is retained in memory is rather just the epistemic state that was formed in this activity, which can have a variety of different contents, depending on what the relevant epistemic activity was. What I now want to suggest is that, whilst what is retained in episodic memory, too, is an epistemic state that was formed through the epistemic activity the memory is a record of, in episodic memory that epistemic state, at the same time, constitutes knowledge of that epistemic activity. Thus, having an episodic memory is both a matter of having a record of past epistemic activity and having knowledge concerning that past epistemic activity; these two things coincide in episodic memory in a way in which this is not the case, I have suggested, for semantic memory. Moreover, this means that, in episodic memory, the epistemic asymmetry is manifest to the remembering subject in a way in which this is not the case for semantic memory. At first, this characterization of episodic memory might appear in tension with the initial one that I gave at the beginning of this section, where I described it as the capacity to remember particular past events themselves, as they happened. But I think the two characterizations can be made compatible with each other, and indeed we can draw an explanatory link between them, if we think of episodic memory as preserving knowledge of events by preserving knowledge of our experiences of those events. That is to say, episodic memories embody knowledge of the epistemic activity through which they themselves were formed in the more specific sense that what is remembered in episodic memory is one’s experience of an event, and it is in virtue of remembering that experience that the subject also remembers the event itself in the distinctive way involved in episodic recollection. Similar sorts of considerations are also sometimes framed in terms of the idea that episodic memories are subject to a ‘previous awareness condition’—i.e., that we can only recall events in episodic memory that we have experienced ourselves.14 Yet it is not obvious that existing accounts of episodic memory can give a satisfactory explanation as to why this condition holds in the distinctive way in which it holds for episodic memory. For instance, an explanation in terms of the causal processes underlying episodic memory seems unable to do justice to the fact that it seems introspectively obvious to us that the condition holds. Similarly, simply writing such an introspective awareness into the definition of what episodic memory is (as, e.g., Owens, 1996, suggests) seems ad hoc. What we need is an explanation of how the fact that episodic memory manifestly involves retaining knowledge about past experiences makes possible the distinctive awareness

Remembering Past Experiences  321 of events we can enjoy in episodic memory, rather than just being an addon to it. I think a more promising approach to this issue is to consider exactly what kind of knowledge about experience episodic memory, distinctively, might be said to preserve. In particular, I want to consider a suggestion made by Matthew Soteriou (2008), according to which episodic memory involves retention of knowledge of our own past experiences more specifically in the sense that what is preserved in episodic memory is knowledge of ‘what it was like’ to experience the remembered event. What is retained in episodic memory, in other words, is knowledge specifically of the conscious experiential character of the relevant past experience.15 How might thinking about episodic memory as preserving knowledge specifically about what it was like to experience certain past events help make intelligible also the idea that it involves a distinctive way of retaining knowledge of past events, as well as exhibiting the epistemic asymmetry in a distinctive way that is manifest to the remembering subject herself? Here I think it might help to draw on some ideas in the existing literature on consciousness on the special status of knowledge regarding the conscious experiential character of experience. One issue frequently noted in this literature is that there seems to be an essential connection between such knowledge and first-hand experience: The only epistemic means by which we can come by knowledge of the conscious experiential character of an experience is by having the experience ourselves.16 The point here is also sometimes made by saying that there is a particular way in which experience is epistemically transformative (Paul, 2014): It furnishes us with knowledge of a kind that we have no other epistemic means of obtaining—knowledge that itself is sometimes described in the vocabulary of ‘experience,’ as in the notion of experiences it is possible to have and accumulate a stock of. The existing literature in this area is mostly concerned with knowledge of the conscious experiential character of types of experience, and the role experience itself plays in such knowledge. But what I want to suggest is that the points made in that literature also have important implications for our knowledge of the conscious experiential character of token experiences, in a way that lets us understand the special status of episodic memory, if it is construed as the retention specifically of such knowledge.17 Consider a scenario of a type described by A. J. Ayer, in which someone describes an incident in your past of which they were a witness. In this sort of case, there is a possibility that you may believe the other person, and that you may even be able to visualize the event based on your general knowledge, but without remembering it. As Ayer points out, though, in this situation it can also happen that you suddenly start remembering again. The transformation may be uncertain. One says ‘I do dimly recollect it,’ being still not quite sure whether one does or not. . . . But it may also be

322  Christoph Hoerl that all of a sudden the event comes back to one quite clearly. One has no doubt that one remembers it. (Ayer, 1956, p. 146. See also Evans, 1982, p. 308; Campbell, 2001, p. 173) What I am suggesting is that we can understand this situation as one in which the subject recovers the retained knowledge of a particular experiential episode and its conscious experiential character as that episode.18 Why think that the capacity to retain knowledge of this type constitutes a separate, distinctive category of memory? Note that, because of the indispensable role that experience itself plays in equipping us with that knowledge in the first place, in so far as the subject retrieves that knowledge, its source is, at the same time, obvious to the subject herself—it is not knowledge that, like the knowledge retrieved from semantic memory, can leave its source open; it is knowledge that the subject could only obtain through going through the experience itself. And it is in virtue of this that the memory can be said to involve a distinctive form of knowledge of the subject’s particular past experience itself, and by extension of the event it was an experience of, as it happened. As we might also put it, the knowledge about a past event that is retained in episodic memory is the knowledge of just how our encounter with that event has added to our stock of experiences, and it is in virtue of this that it constitutes knowledge of that particular event itself and manifestly locates it in our past. Thus, there is a specific way in which episodic memory exhibits an epistemic asymmetry, which is grounded not just in the fact that episodic memories fall under the general characterization of a record, but in something specifically to do with the knowledge that is retained in episodic memory, i.e., the fact that episodic memory is the retention of knowledge of what it was like to experience the remembered event, the only epistemic means of acquiring which is through the experience itself.19 As we might also put it, the distinctive way in which episodic memories put us in touch with particular events and locates them in the past is rooted in the fact that it is memory for the particular epistemic transformation we underwent when we experienced the event and learned what it was like to experience it—a type of knowledge we could only obtain on that occasion. This is why, in episodic memory, the time of the event the memory embodies knowledge of and the time it is a record of coincide in a way that is transparent to the remembering subject herself.

4  Coda: Episodic Memory and Time I have sketched a characterization of the difference between episodic and semantic memory based on the idea that, whilst they both exemplify the epistemic asymmetry of time, each of them does so in a somewhat different way. Furthermore, I have also suggested that both forms of memory, in turn,

Remembering Past Experiences  323 exemplify the epistemic asymmetry in a way that differs from the way in which this is the case for records such as a footprint on a beach. As I already hinted at toward the beginning of this chapter, interest in the epistemic asymmetry in recent metaphysics and philosophy of science is mainly driven by the ambition to give an account of the origins of this asymmetry (along perhaps with further asymmetries in our thinking about time that may be grounded in it) that is compatible with the idea that the fundamental dynamical laws that govern our universe are time-symmetric, as current physics suggests.20 More specifically, the core agenda in this context is typically that of unmasking a pervasive tendency we seem to have toward thinking of temporal asymmetries such as the epistemic arrow as being due to the metaphysical nature of time itself. If what I have said in this chapter is along the right lines, might it also perhaps be able to contribute to this unmasking project in a new way? I will conclude with some brief remarks on this issue, which are admittedly even more speculative than what I have said so far. Contrary to the four-dimensionalist view of the universe suggested by modern physics, our everyday understanding of time conceives of time as completely different from space. However, there are several, potentially separable, dimensions of this perceived difference, which is part of the reason why metaphysical accounts of time that diverge from the fourdimensionalist picture come in several different flavors, depending on which of those dimensions of difference they foreground, such as presentism, the ‘growing block’ view, or the ‘moving spotlight’ view.21 Put briefly, we may identify three such dimensions of difference as follows: that the present moment in time is somehow special, that there is a fundamental difference between the past and the future, and that there is an irreversible ‘passage’ or ‘flow’ of time. What I now wish to suggest, as against the background of the preceding discussion, is that the epistemic asymmetry might in fact contribute to our everyday picture of time in two somewhat different ways, connected to two different such ingredients. The epistemic asymmetry exhibited by records generally, i.e., the fact that we can have records only of the past and not of the future, is arguably one source of the idea that the past is fundamentally different from the future. Some work may be required to spell out in detail how exactly it gives rise to this idea, but there seems to be some intuitive sense in which we tend to think of the reason why we have records of the past but not of the future in terms of ideas such as the idea that the past is ‘fixed’ whereas the future is ‘open,’ or that the past is real but the future isn’t. To get right the particular way in which we tend to think of the future as ‘open’ and the past as ‘fixed,’ however, it seems we also need to appeal to the somewhat separate, if connected, idea of an irreversible ‘passage’ or ‘flow’ of time that turns what first belongs to the open future into something that then belongs to the fixed past. And here we can perhaps see a separate role specifically for episodic memory to play in our everyday understanding of

324  Christoph Hoerl time.22 In this chapter, I sought to identify a distinctive way in which episodic memory exhibits an epistemic asymmetry, in addition to the asymmetry that comes with memories belonging to the general category of a record. I traced this distinctive way in which episodic memory exhibits an epistemic asymmetry back to the particular kind of knowledge that is retained in episodic memory, namely knowledge of what it was like to experience the remembered event—or, as I have also put it, knowledge of the particular epistemic transformation that experiencing the event consisted in. What is special about this kind of knowledge, I argued, is that it is only the experience itself that can equip us with it, which can explain why, in episodic memory, it is transparent to the remembering subject that the time of the event the memory embodies knowledge of and the time it is a record of coincide. Implicit in these considerations, I believe, is the thought of a form of irreversibility time has for us, from our epistemic point of view, that is specifically associated with the fact that we can engage in episodic recollection. To gain the knowledge of what it is like to experience a particular event, we must first wait until we undergo that experience, after which the knowledge of what it was like is retained in episodic memory. That time is, in this sense, irreversible from our epistemic point of view is not just a thought that we apply in our thinking about the past. It can also play a crucial role in our thinking about the future. Experiences stay with us, and deciding about the future is in part a matter of deciding on what experiences we want to look back on.23, 24

Notes 1 The episodic/semantic distinction originates with Tulving (1972), and has been refined by Tulving in a number of other works (Tulving, 1985, 2002; Wheeler, Stuss, & Tulving, 1997). Rival ways of carving up the domain of memories are suggested, for instance, in Bernecker (2010) and Rubin and Umanath (2015). 2 An exception is Huggett (forthcoming). He approaches questions about the connection between memory and the epistemic asymmetry largely from within the philosophy of physics. My main focus will be on epistemology. 3 On the importance of this distinction, see Price (1996, ch. 1). Different ways of fleshing out this general project are also suggested, e.g., in Horwich (1987), Ismael (2016), Callender (2017), and Fernandes (2017). 4 The example is originally from Schlick (1925). 5 What, in turn, explains this fact is still the subject of considerable debate. There is a fairly straightforward statistical story to be told about why we should expect entropy to increase over time, i.e., why we should expect low-entropy states to evolve into higher-entropy states. The problem is that the relevant statistics are temporally neutral: the same considerations should lead us to think that entropy increases in the past direction too. There is therefore a second explanation needed as to what introduces the past/future asymmetry. See, e.g., Callender (2016) for discussion. 6 Throughout this paper, I will use phrases such as ‘is a record of’ and also ‘remembers’ as factive, i.e., as implying veridicality (see also the focus on knowledge in what follows). My question is: In so far as there are such states as being a

Remembering Past Experiences  325 (veridical) record of something, or (veridically) remembering something, what constitutes an object’s or person’s being in those states? I will set aside the existence of other, non-veridical states that we might, on occasion, be unable to distinguish from those veridical states, and the separate set of philosophical issues they might raise. 7 For some challenges to this story, see, e.g., Earman (1974) and Horwich (1987). For a revision, according to which there is a more fundamental asymmetry that grounds both the epistemic asymmetry and the thermodynamic arrow, see Albert (2000). 8 As indicated by the scare quotes, I am not entirely happy with this use of the word ‘record.’ I need a word that generalizes over all the different particular phenomena that result in a past/future asymmetry in our epistemic standing, and I have chosen to go with the word ‘records’ as it is already being used in the existing literature on the epistemic asymmetry. Yet, it is also an important part of my argument that there are crucial differences between some of these phenomena, which I also take to be indicated by the fact that, e.g., a footprint on a beach is perhaps better described as a trace, and that memories, even if they can be described as records in some sense, are not necessarily memories of the thing they are a record of (as I will explain). 9 For this reason, speaking of memories as a type of record is compatible with Dummett’s (1993, pp. 420f.) claim that “[m]emory is not a source, still less a ground of knowledge: it is the maintenance of knowledge formerly acquired by whatever means.” It is plausible to think that things that can be a record of something can also, at the same time, be a record of a number of other things. In the case of memory, though, its being a record of the relevant event of knowledgeacquisition is what makes a memory a memory. 10 This is not to rule out that states of objects outside a person’s body could, under suitable circumstances, form part of that person’s epistemic state. I take it whether they can, and, if so, under what circumstances, is precisely one of the things under discussion in debates about the ‘extended mind thesis’ (Clark & Chalmers, 1998). 11 I have focused specifically on the contrast between these two categories of phenomena. A further category of records that I will leave to one side for present purposes are records involving representational media such as written words or images. Like the footprint, they are not by themselves epistemic states (though see the preceding footnote), but in other respects they might be thought to share more features with memories. For instance, it is not just knowledge about the past that we can obtain from written texts or images—a diary entry may inform me of an appointment next week, a map may show the course of next month’s marathon. As I will go on to discuss, semantic memory is similarly not restricted to retaining knowledge just about the past. 12 Patient KC, for instance, whilst being described as having lost all capacity for episodic recollection, can describe a number of aspects of his life before the accident that brought about his amnesia. See Craver et al. (2014). 13 There are certain approaches to episodic memory that allow for the possibility of cases in which subjects are in fact recollecting a past event, even though it doesn’t even seem to them that they are representing a past event—the de facto obtaining of a causal connection between the subject’s present mental state and a specific past experience of theirs being supposedly sufficient to make it a case of episodic recollection (see, e.g., Martin and Deutscher, 1966). I argue against this way of conceptualizing episodic memory elsewhere (see Hoerl, 2014a). Another hypothetical situation in which one might claim that somebody who recollects an event in episodic memory need not necessarily think of the event as lying in

326  Christoph Hoerl the past is that of a person travelling back into the past but remembering events from the period in the future before they stepped into the time machine. As David Lewis (1976) has shown, though, cases of time travel force us to make a distinction between what he calls ‘personal time’ and what he calls ‘external time,’ and whilst it is true that the time traveler might think of some events she remembers as being located in the future with respect to external time, this is compatible with thinking that she will nevertheless also be aware of them lying in the past with respect to her personal time (i.e., they belong to a period of her life when she was biologically younger). I am grateful to a referee for prompting me to clarify these points. 14 The term ‘previous awareness condition’ was coined by Shoemaker (1970), though the idea goes back at least as far as Locke (1690) and Reid (1785). For discussion, see also Martin (2001). 15 This claim should be distinguished from the claim that the remembering subject’s current mental state must faithfully resemble her past experience in all respects before she can be said to have such knowledge. For instance, it is at least not obvious that one cannot be said to remember what it was like to experience a certain event if one remembers it from an ‘observer’ rather than a ‘field’ perspective. See also McCarroll and Sutton (2017) on related issues. 16 Discussion of this point is particularly prominent in the context of work on the so-called ‘ability hypothesis,’ according to which phenomenal knowledge consists in a particular form of practical knowledge (Lewis, 1990; Nemirow, 1990). But I think the latter should be seen as a further claim, intended as an explanation of why the point holds. 17 For further discussion of related ideas, see also Hoerl (forthcoming). 18 Compare here also Martin & Deutscher’s (1966) discussion of the phenomenon of ‘prompting.’ 19 Saying that this constitutes a distinct sense in which episodic memory exhibits an epistemic asymmetry is compatible with it too having a deeper grounding in whatever asymmetry grounds the general asymmetry of records, e.g., the thermodynamic asymmetry. 20 At least by and large. There seems to be a breakdown of complete time-reversal invariance in the interaction between certain sub-atomic particles. Even if this is so, however, it is far from clear how this might explain the existence of the relevant asymmetries on the level of our everyday experience. See also Wallace (2013). 21 For an overview and discussion of these accounts see, e.g., Miller (2013). 22 For a related argument, see also Hoerl (2014b). 23 This is an important theme in Paul (2014, 2015). See also Hoerl and McCormack (2016). 24 Work on this chapter was supported by AHRC grant AH/P00217X/1. For comments on earlier versions, I am grateful to Patrick Burns, Alison Fernandes, Teresa McCormack, and two anonymous referees, as well as to audiences at the 2017 Annual Meeting of the European Society for Philosophy and Psychology, and a meeting of the Warwick Mind and Action Research Seminar.

References Albert, D. Z. (2000). Time and chance. Cambridge, MA: Harvard University Press. Ayer, A. J. (1956). The problem of knowledge. Harmondsworth: Penguin. Bernecker, S. (2010). Memory: A philosophical study. Oxford: Oxford University Press.

Remembering Past Experiences  327 Callender, C. (2016). Thermodynamic asymmetry in time. In E. N. Zalta (Ed.), Stanford encyclopedia of philosophy (Winter 2016 ed.). Retrieved from https://plato. stanford.edu/archives/win2016/entries/time-thermo/ Callender, C. (2017). What makes time special. Oxford: Oxford University Press. Campbell, J. (2001). Memory demonstratives. In C. Hoerl & T. McCormack (Eds.), Time and memory: Issues in philosophy and psychology (pp. 169–186). Oxford: Oxford University Press. Clark, A., & Chalmers, D. J. (1998). The extended mind. Analysis, 58(1), 7–19. Craver, C. F., Kwan, D., Steindam, C., & Rosenbaum, R. S. (2014). Individuals with episodic amnesia are not stuck in time. Neuropsychologia, 57, 191–195. Dummett, M. (1993). Testimony and memory. In The seas of language (pp. 411– 428). Oxford: Oxford University Press. Earman, J. (1974). An attempt to add a little direction to “the problem of the direction of time.” Philosophy of Science, 41(1), 15–47. Evans, G. (1982). The varieties of reference (J. McDowell, Ed.). Oxford: Clarendon Press. Fernandes, A. (2017). A deliberative approach to causation. Philosophy and Phenomenological Research 95(3), 686–708. Hoerl, C. (2014a). Remembering events and remembering looks. Review of Philosophy and Psychology, 5(3), 351–372. Hoerl, C. (2014b). Time and the domain of consciousness. Annals of the New York Academy of Sciences, 1326(Flow of Time), 90–96. Hoerl, C. (Forthcoming). Episodic memory and theory of mind: A connection reconsidered. Mind & Language. Hoerl, C., & McCormack, T. (2016). Making decisions about the future: Regret and the cognitive function of episodic memory. In K. Michaelian, S. Klein, & K. Szpunar (Eds.), Seeing the future: Theoretical perspectives on future-oriented mental time travel (pp. 241–266). Oxford: Oxford University Press. Horwich, P. (1987). Asymmetries in time: Problems in the philosophy of science. Cambridge, MA: MIT press. Huggett, N. (Forthcoming). Reading the past in the present. In B. Loewer, B. Weslake, & E. Winsberg (Eds.), Time’s arrow and the probability structure of the world. Cambridge, MA: Harvard University Press. Ismael, J. (2016). How physics makes us free. Oxford: Oxford University Press. Lewis, D. (1976). The paradoxes of time travel. American Philosophical Quarterly, 13(2), 145–152. Lewis, D. (1990). What experience teaches. In W. G. Lycan (Ed.), Mind and cognition (pp. 29–57). Oxford: Blackwell. Locke, J. (1975/1690). An essay concerning human understanding (P. H. Nidditch, Ed.). Oxford: Clarendon Press. Martin, C. B., & Deutscher, M. (1966). Remembering. Philosophical Review, 75, 161–196. Martin, M. G. F. (2001). Out of the past: Episodic recall as retained acquaintance. In C. Hoerl & T. McCormack (Eds.), Time and memory: Issues in philosophy and psychology (pp. 257–284). Oxford: Oxford University Press. McCarroll, C. J., & Sutton, J. (2017). Memory and perspective. In S. Bernecker & K. Michaelian (Eds.), The Routledge handbook of philosophy of memory (pp. 113– 126). Abingdon: Routledge.

328  Christoph Hoerl Miller, K. (2013). Presentism, eternalism, and the growing block. In H. Dyke & A. Bardon (Eds.), A companion to the philosophy of time (pp. 345–364). Chichester, West Sussex: John Wiley & Sons, Ltd. Nemirow, L. (1990). Physicalism and the cognitive role of acquaintance. In W. G. Lycan (Ed.), Mind and cognition (pp. 490-499). Oxford: Blackwell. North, J. (2011). Time in thermodynamics. In C. Callender (Ed.), The Oxford handbook of philosophy of time (pp. 312–350). Oxford: Oxford University Press. Owens, D. J. (1996). A Lockean theory of memory experience. Philosophy and Phenomenological Research, 56(2), 319–332. Paul, L. A. (2014). Transformative experience. Oxford: Oxford University Press. Paul, L. A. (2015). What you can’t expect when you’re expecting. Res Philosophica, 92(2), 1–23. Price, H. (1996). Time’s arrow and Archimedes’ point: New directions for the physics of time. Oxford: Oxford University Press. Reichenbach, H. (1956). The direction of time (M. Reichenbach, Ed.). Berkeley: University of California Press. Reid, T. (2002/1785). Essays on the intellectual powers of man (D. Brookes & K. Haakonssen, Eds., The Edinburgh edition of Thomas Reid). Edinburgh: Edinburgh University Press. Rubin, D. C., & Umanath, S. (2015). Event memory: A theory of memory for laboratory, autobiographical, and fictional events. Psychological Review, 122(1), 1–23. Schlick, M. (1925). Naturphilosophie. In M. Dessoir (Ed.), Die Philosophie in ihren Einzelgebieten: Lehrbuch der Philosophie (pp. 397–492). Berlin: Ullstein. Shoemaker, S. (1970). Persons and their pasts. American Philosophical Quarterly, 7(4), 269–285. Soteriou, M. (2008). The epistemological role of episodic recollection. Philosophy and Phenomenological Research, 77(2), 472–492. Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), Organization of memory. New York: Academic Press. Tulving, E. (1985). Elements of episodic memory. Oxford: Oxford University Press. Tulving, E. (2002). Episodic memory: From mind to brain. Annual Reviews of Psychology, 53, 1–25. Wallace, D. (2013). The arrow of time in physics. In H. Dyke & A. Bardon (Eds.), A companion to the philosophy of time (pp. 262–281). Chichester, West Sussex: John Wiley & Sons, Ltd. Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic memory: The frontal lobes and autonoetic consciousness. Psychological Bulletin, 121(3), 331–354.

17 On Seeming to Remember*

Fabrice TeroniOn Seeming to Remember

Fabrice Teroni

Philosophers and psychologists often distinguish episodic or personal memory from propositional or semantic memory. A vexed issue concerns the role, if any, of memory “impressions” or “seemings” within the latter. According to an important family of approaches, seemings play a fundamental epistemological role vis-à-vis propositional memory judgments: it is one’s memory seeming that Caesar was murdered, say, that justifies one’s judgment that he was murdered.1 Yet, it has been convincingly argued that these approaches lead to insurmountable problems and that memory seemings are not wellsuited to play this justifying role. As a result, many contemporary accounts of propositional memory dispense with these seemings altogether. Is the idea that memory seemings play a key role in propositional memory really the result of bad theorizing? My aim is to shed light on this issue, which I will approach as follows. In Section 1, I contrast episodic memory with propositional memory so as to clarify the nature of the latter. According to the account I put forward, episodic memory consists in the preservation of acquaintance with objects and events, whereas propositional memory consists in the preservation of thought contents. In Section 2, I turn my attention to the contrast between propositional memory contents and propositional memory as an attitude. I argue that they play different roles. Memory contents satisfy a past awareness constraint and a causal constraint; the attitude of remembering explains why we are inclined to endorse these contents. This distinction leads me to explore the attitude of remembering, and I argue, in Section 3, that the most appealing account of this attitude is in terms of feelings of familiarity. In Section 4, I turn my attention to the epistemology of propositional memory and revisit the claim that propositional memory judgments are justified by memory seemings. In so doing, I contend that the attitude of remembering plays an exclusively explanatory role and does not contribute to the epistemology of propositional memory judgments. I conclude by drawing a more general lesson regarding the respective roles of attitudes and contents.

330  Fabrice Teroni

1  Memory: Propositional vs. Episodic Let me start by introducing the sort of phenomena that fall under the label of “propositional memory.” I shall do so by contrasting these phenomena with what happens when we remember episodically. We often attribute memories to people with the help of the verb “to remember” followed by nominal constructions, as when we say “Mary remembers her first encounter with her mother-in-law,” or “John remembers the first movements of the symphony.” We thereby imply (among other things) that Mary and John have been acquainted with the events and objects they remember. This is sometimes called “personal” and sometimes “episodic” memory.2 What happens when we remember episodically? First, we must have been acquainted with the objects or events that are remembered, where the past acquaintance is typically perceptual—for instance, Mary saw her motherin-law. Second, this past acquaintance informs our memory of these objects or events. Episodic memory is phenomenologically rich: it is, for Mary, “as if” she was seeing her mother-in-law and, for John, “as if” he was hearing the overture of the symphony. That is to say, memory here consists in a preserved acquaintance or cognitive contact (Byrne, 2010; Martin, 2001). What I shall call memory images make one aware of the relevant events or objects.3 By contrast, propositional memory is typically (but, of course, not only) manifested in one’s knowledge of historical and mathematical facts.4 When Sam remembers that Caesar was murdered, or Mary remembers that the square root of 625 is 25, we face cases of propositional memory. In a nutshell, propositional memory consists in the preservation of propositional contents. Thanks to this kind of memory, propositional contents that we have judged or merely entertained remain available for thought and are typically accepted or endorsed at later times.5 If this is what propositional memory amounts to, then it no doubt differs from episodic memory. In particular, propositional memory does not involve memory images. Consider Sam, who has preserved the content that Caesar was murdered and is disposed to endorse it. First, Sam is plainly not in a position to enjoy a memory image that would make him aware of Caesar’s murder. After all, he has not been acquainted with this event—perceptually or otherwise—and so cannot have a preserved acquaintance with it in the form of a memory image (McGrath, 2007; Teroni, 2017).6 Second, while propositional memory arguably always traces back to a past learning event, it does not presuppose episodic memory of this event. Actually, it does not even presuppose that one has preserved any information about it. For instance, Sam may remember that Caesar was murdered while having no inkling as to how he acquired this information, let alone be in a position to re-live, say, the relevant classroom experience. Third, the idea that propositional memory involves memory images would be hard to make sense of in many cases, as in the case of Mary who remembers that the square root of 625 is 25.

On Seeming to Remember  331 I hope the foregoing is enough to capture the very familiar phenomena that I describe as propositional memory. In what follows, I shall be exclusively concerned with this kind of memory.

2  Propositional Memory: Content vs. Attitude I will start exploring propositional memory with the help of the important, yet surprisingly often overlooked, distinction between memory contents and memory as a mode or an attitude (Locke, 1971, chap. 1; Matthen, 2010).7 To explain what is at stake, let me start by noting that two distinct features of a psychological state may lead one to describe it as a memory. One may describe a psychological state as a memory first because its content satisfies some constraints. Recent discussions of propositional memory have almost exclusively focused on the specification of these constraints. Given my aim here, I suggest that we adopt the following rough characterization of propositional memory contents: a propositional content p is a memory content for S if and only if (i) it was the content of a past representation of S that p, and (ii) its occurrence in S is caused by this past representation.8 The content that Caesar was murdered is a memory content for Sam in virtue of (i) Sam having previously judged that Caesar was murdered, and (ii) this content being actually available to Sam because he judged so.9 A content may thus fail to qualify as a memory content because the subject was not previously aware of it, or because, despite having been aware of it, his past awareness does not explain the actual availability of that content.10 Observe now that the contrast between memory contents and other sorts of contents holds independently of the subject’s inclination to judge memory contents to be true. On the one hand, someone may entertain a memory content without being inclined to endorse it—the fact that a content satisfies the previous constraints does not imply that there is an inclination to judge it true. In retrospect, we typically characterize the situation in which a memory content does not go together with such an inclination by saying that we did not realize, at the time, that we were remembering. On the other hand, we may also be inclined to endorse a content that fails to qualify as a memory content, either because we erroneously believe that it does so qualify, or because we have other reasons for endorsing it. Let me now turn to the second feature of a psychological state that may lead one to describe it as a memory. This second feature is the specific attitude that the subject has toward a content. The fact that there is such a distinction between memory contents and the attitude of remembering is plain given that they can independently vary. The attitude of remembering sometimes targets contents that are not memory contents—in such cases, we often say that what we seem to remember is a figment of our imagination. Moreover, we have already acknowledged that this attitude may be absent when we entertain memory contents. Memory content is one thing, the attitude of remembering another.

332  Fabrice Teroni How should we go about characterizing the attitude of remembering? Contrary to what I have just said regarding memory contents, we cannot dissociate the attitude of remembering from the subject’s inclination to endorse the content that it targets. As opposed to someone who entertains a content or supposes that it is true, someone who remembers the content is inclined to endorse it, and he will endorse it if he has no reason to doubt that it is true or that he has preserved it. As Burge (1993: 465) rightly observes, propositional memory (“purely preservative memory” in his terminology) preserves contents with judgmental force, i.e., we are inclined to endorse these contents when we remember. For that reason, an account of the attitude of remembering is an account of a tendency to endorse some contents, a tendency that explains the crucial role propositional memory plays in our cognition. Now, as this tendency is obviously not exclusive to the attitude of remembering, is it possible to say a bit more in order to clarify what we are after? I shall attempt to do so by focusing first on the correctness conditions of mental states in general and then by applying this idea to propositional memory. Consider the following psychological states: judging that p, conjecturing that p, and remembering that p. These attitudes take the same content, which is why they share an important part of their correctness conditions. Yet, the fact that these are different attitudes has itself a distinct impact on their respective correctness conditions.11 Roughly, it is because one judges that the psychological state of judging that p is correct if and only if the content is true; it is because one conjectures that the state of conjecturing that p is correct if and only if the content is probable; and it is because one remembers that the state of remembering that p is correct if and only if that content was the content of a past representation that is causally responsible for its being now available. These differences in the correctness conditions trace back to the contribution of the different attitudes. In light of these observations, we can conclude that the attitude of remembering plays an important explanatory role: it explains why we are inclined to endorse some contents out of a sensitivity to the fact that these contents are available to us because they have been previously represented. Our task in Section 3 is to examine what this attitude and sensitivity amount to. But let me first wrap up this discussion of the contrast between memory contents and memory as an attitude. An account of memory contents should proceed in terms of a previous awareness constraint and a causal constraint. As opposed to this, an account of the attitude of remembering is an account of a distinctive tendency to endorse some contents.

3  Explanation: The Nature of the Attitude This section revolves around the following issue. In propositional memory, we typically endorse whatever we remember. If Sam remembers that Caesar was murdered, then in all likelihood (and if he has no conflicting evidence),

On Seeming to Remember  333 he will endorse this as a fact. We have just concluded that this is a characteristic of the attitude of remembering. Given that a central aspect of this attitude is a tendency to endorse the content it targets, how should we understand it? In what follows, I do not explore all the available options—I rather focus on two unpersuasive accounts before introducing my favorite one.12 The first account consists in regarding our inclination to endorse memory contents as a primitive and not further explainable trait of propositional memory, at least insofar as the first-person level is concerned (Goldman, 1993a).13 This amounts to saying that, from the subject’s perspective, it is simply that she finds herself so inclined. No memory attitude that would help her to make sense of this inclination is available to her. Sensitivity to the fact that contents are available because they have been previously represented is a characteristic of subpersonal processes and is not made manifest at the first-person level. Now, this flat denial of a first-person level explanation as to why we are typically inclined to endorse memory contents should come as a surprise. This is because there appears to be a clear difference between two distinct cases. You can be aware that you are simply inclined to endorse a content. Alternatively, you can be aware that you are so inclined because, as we often say, you seem to remember (e.g., Cullison, 2010). In the latter case, reference to remembering appears to be explanatory of the inclination to endorse the content, but the first account fails to acknowledge this.14 I do not mean to suggest that we should always look for a first-person level explanation of the inclination to endorse memory contents. The suggestion is more modest: typically, memory makes itself manifest at the first-person level as the source of the inclination to endorse a content.15 This is why the first account is not persuasive. We often explain why we are inclined to endorse memory contents by saying that we seem to remember. Which explanation do we offer when we do so? To answer this question, let me say a few words about “look” constructions, which function in very similar ways as “seem to remember.” Look constructions, it is frequently observed, lend themselves to different readings (Chisholm, 1957, chap. 4; Maund, 2003, chap. 7). We can read “It looks as if it will rain” as expressing the belief that one has good reasons to think it will rain. This is the epistemological reading. “I seem to remember that p” is read epistemologically when it is read as expressing my belief that I have good reasons to think that a content is a memory content. In light of the rough characterization we use here, this is the belief that I did previously represent a content and that this past event explains why the content is now available to me.16 The second account of the attitude of remembering recruits the epistemological reading of “seem to remember.” It claims that beliefs about contents explain our inclination to endorse memory contents. More precisely, the attitude of remembering is understood in terms of the

334  Fabrice Teroni functional role that individuates memory contents: this attitude consists in the subject’s belief that a content he has in mind actually plays this role.17 Sensitivity to the fact that contents have been previously represented is now a characteristic of the first-person level, since it takes the shape of this belief. That being said, this account of the inclination to endorse memory contents is not very appealing for at least three reasons. First, it rests on unconvincing psychological assumptions. We are surely prone to surmise that contents play a functional role when there is room for doubt and when alternative explanations for their availability are salient to us. To illustrate, suppose that Michelle tells her friend Sam that he sometimes makes up historical facts. He may then think about it, conclude that the content that Caesar was murdered is available to him because he learned it somewhere, and as a result endorse it. However, the claim that we should explain all memory judgments in this way is difficult to reconcile with their typical immediacy. Second, and in direct connection, the explanation is far too intellectualistic (Goldman, 1993a, 1993b). Children make memory judgments well before they understand the nature of memory contents—grasping relations between thought contents across time is, after all, a substantial cognitive achievement. This provides the opportunity to come back to the claim put forward in Section 2, according to which pastrelated correctness conditions in memory are the result of the attitude of remembering. One reason for this claim is that attitudes are at play well before subjects have the capacity to understand what they are. The account under discussion raises a worry precisely because it rejects this claim in identifying the attitude of remembering with complex beliefs. Third and finally, the explanation is circular. How can Sam believe that he previously judged that p, if not by endorsing other propositional memory contents? For these reasons, beliefs about the functional role played by memory contents cannot explain the inclination to endorse them that is characteristic of remembering. To understand this attitude, we should look elsewhere. To home in on the attitude of remembering, we have to come back to the “seem to remember” locution. Similar to “this looks red to me,” “I seem to remember” lends itself to a phenomenological reading, as opposed to an epistemological one. Read this way, these expressions refer to specific experiences. “Looking red” refers to the distinctive visual experience that red surfaces elicit, and “seeming to remember” refers to an experience that is characteristic of propositional memory. Now, it is hardly deniable that experiences often accompany memory judgments—this is why so many philosophers have referred to impressions of remembering, memory impressions, or seemings (e.g., Audi, 1995; Pollock, 1974; Pollock & Cruz, 1999; Russell, 1921). The real issue is not the existence of these seemings, but their nature and their role(s) within propositional memory. As regards their nature, I shall rest content with a few observations (see Teroni, 2017). First, it is important to emphasize that memory seemings differ from the memory images distinctive of episodic memory. Memory

On Seeming to Remember  335 seemings do not constitute a retained acquaintance with the relevant objects or events, and they are phenomenologically much poorer than memory images. Seeming to remember that Caesar was murdered is not being in a state akin to seeing his murder. More specifically, these seemings do not vary as a function of what is retrieved. They do no more than tag the content as being a memory, and they differ from one another only insofar as they are more or less intense.18 Second, among the many approaches to these seemings, a quite appealing one claims that they consist in feelings of familiarity. In the phenomenological sense, to “seem to remember” a content is for it to feel familiar.19 In the present context, this means that propositional memory contents typically feel familiar, which is why we are inclined to endorse them. This is what the subject’s sensitivity to the fact that contents have been previously represented amounts to, and it is squarely located at the first-person level. But we now explain the inclination to accept memory contents in a way that is not too intellectualistic and is faithful to the fact that we typically endorse memory contents immediately, not as a result of reasoning. An additional virtue of an account in terms of feelings of familiarity is that it explains typical mistakes of self-attribution, which are due to illusions of familiarity. This makes for an attractive account of the attitude of remembering.20 Now that we have a clearer picture of the attitude of remembering, let me emphasize an important consequence of our having carefully kept content and attitude apart. The attitude of remembering need not accompany memory contents, which may in addition be endorsed for a variety of reasons. But observe that we are in a position to say that, when the inclination to endorse these contents is explained by a feeling of familiarity, the attitude of remembering plays a distinctive first-person level explanatory role. It is only when feelings of familiarity come about that we endorse these contents because we (seem to) remember.21 Let me conclude this section. The best explanation of why we are typically inclined to endorse memory contents proceeds in terms of feelings of familiarity that characterize the attitude of remembering. The next issue regards the epistemological consequences of the explanatory role characteristic of this attitude.

4  Justification: Content vs. Attitude In the previous section, we recruited memory seemings to explain why we are typically inclined to endorse memory contents. I now turn my attention to a key epistemological issue regarding propositional memory. Do memory seemings justify the judgments we make when we remember? I will explore this issue by examining a widespread approach about propositional memory that gives a positive answer to this question. According to this approach, there is a distinctive justification in propositional memory: the memory seemings that typically explain why

336  Fabrice Teroni we judge also justify our endorsing the contents we happen to retrieve.22 If you seem to remember that Caesar was murdered, i.e., if that content feels familiar, then you are justified in judging that he was murdered (e.g., Audi, 1995; Pollock, 1974; Pollock & Cruz, 1999). At least, you are justified if you are aware of no defeaters, which may have to do with information suggesting that the judgment is false or that you could not have learned the relevant fact. In this sense, the justification provided by feelings of familiarity is defeasible or prima facie. Why is this approach popular? The following line of thought constitutes an important source of motivation. In many, if not most, cases of propositional memory, we have lost track of our reason(s) for having judged that p in the first place. Sam is likely to have forgotten why he came to judge that Caesar was murdered, for instance. Claiming that these memory judgments are unjustified would therefore generate a substantial form of skepticism. Now, according to a widespread approach, whatever contributes to the justification of a judgment must be accessible by the subject who makes that judgment—this is internalism about justification (Pappas, 2014). To the extent that one sides with internalism, a sensible move in the epistemology of propositional memory consists in avoiding skepticism by claiming that memory seemings—experiential states to which we have access when we retrieve contents—justify our endorsing these contents.23 Accounting for the justification of propositional memory judgments in terms of memory seemings should look quite convincing—indeed, it has convinced many philosophers. Yet, this account faces a serious worry insofar as memory seemings turn out to be insufficient for the justification of propositional memory judgments (Annis, 1980; Naylor, 1982). Suppose that Michael came to judge that Caesar died in his bed on the basis of fanciful reasons. Years later, he entertains the thought that Caesar died in his bed, and it so happens that he seems to remember that this was the case and is aware of no defeater for that judgment. The account under discussion would assess Michael’s memory judgment that Caesar died in his bed as justified. This is a claim we have every reason to avoid: memory cannot function as a generative epistemological source in this way.24 After all, it is because Michael judged on the basis of fanciful reasons that his judgment was unjustified when he made it for the first time. The fact that the content now feels familiar surely cannot turn this unjustified judgment into a justified one. Yet, the account as it stands is committed to saying that the justification of propositional memory judgments is completely divorced from past reasons to judge. This is inacceptable. In a slogan, “garbage in, garbage out” (Jackson, 2011). This parallels the more general conclusion reached by some philosophers regarding the justificatory role of seemings: it is not seemings in general, but only some of them that justify.25 In particular, if a seeming has a problematic aetiology, then it cannot justify a subject to endorse the content that seems true to him (Bergman, 2013; Jackson, 2011; Markie, 2013). In propositional

On Seeming to Remember  337 memory, aetiology of course takes center stage, since the feeling of familiarity originates in a situation in which the subject has already judged for good or bad reasons.26 The dependence of the seeming on a past situation in which the belief was acquired is what makes it unappealing to claim that endorsing a content because it feels familiar is sufficient for justification. Memory is not an independent source of justification; it can only transmit it (McGrath, 2007; Naylor, 1982; Teroni, 2014).27 Memory seemings do not generate a new type of justification, an observation that chimes well with the fact that retaining and manifesting a belief is not a matter of basing it on new evidence. This conclusion can be strengthened if we remind ourselves of the nature of these seemings. Feelings of familiarity toward contents are reliably caused by the fact that these contents have been entertained. So, they may justify the judgment that they have been entertained. But they cannot justify our endorsing these contents: the fact that a content has already been entertained is hardly a reason to think it true. Let me highlight three aspects of the conclusion we have reached so far, namely the conclusion that feelings of familiarity cannot justify our endorsing of the contents they target. First, this conclusion goes against something we may call “present-tense internalism,” i.e., the claim that all the factors relevant to the justification of a judgment are accessible to the subject at the time he makes that judgment (Teroni, 2014). The conclusion is nevertheless compatible with more relaxed forms of internalism, for instance those according to which these factors must have been accessible at some point in the past (at the time the belief is acquired, for instance).28 Second, the claim is limited to feelings of familiarity and is not meant to carry over to other experiences we may have when we remember. In Section 1, we distinguished episodic from propositional memory by saying that memory images are characteristic of the former but not of the latter. When we episodically remember, it is as if we were perceiving the relevant events or objects again. Nothing I have said here militates against the idea that memory images constitute an original form of justification—we may have to adopt distinct accounts of propositional and episodic memory justification (Teroni, 2014). More specifically, there is an important difference between propositional and episodic memory. When, and only when, we remember episodically, the truth-maker of the judgments we make is manifest to us. For instance, we judge that a friend wore a yellow dress at a party because we remember her wearing the dress. This may well have epistemological consequences (Conee, 2013; Hoerl, 2001; Teroni, 2014). Third, and most importantly, the conclusion that feelings of familiarity are not sufficient for justification leaves their exact epistemological role underspecified. Here is how I think we should go about specifying that role. If we insist on aetiological issues in the epistemology of propositional memory, then the net result is that we are justified in endorsing a memory content because it feels familiar only if we had good reasons to make the judgment in the first place.29 This is the extent of memory’s epistemological

338  Fabrice Teroni dependence. Now, there are two options at this stage. One option is to say that the judgment we make when we remember is justified by the feeling of familiarity provided it has the right aetiology. This is tantamount to saying that the reasons we had to pass the judgment in the first place—our past reasons, we may say—play no epistemological role when we remember. Alternatively, we may say that the judgment is justified by these past reasons. I think we should favor the second option. Here is why. According to the first option, feelings of familiarity play an epistemological role only when they are filtered by considerations that are typically outside of the subject’s ken when he remembers. If we concede this much, however, it is difficult to see why we should deny that past reasons play an epistemological role at the time of memory. Once we accept that endorsing a content is justified (or not) as a function of the subject’s past reasons, why not wholeheartedly embrace the idea that preserving a justified belief is not a matter of basing it on new evidence? Let me emphasize that the option under discussion should receive no support from the fact that feelings of familiarity play an explanatory role. In Section 3, we granted that these feelings explain our inclination to endorse the relevant contents. Memory seemings do make a psychological difference; they contribute to making the judgment intelligible from the subject’s perspective. This is why I insisted that we should distinguish a simple inclination to endorse a content from the inclination to endorse it because we seem to remember. The present issue is whether memory seemings additionally play an epistemological role.30 Once explanatory and epistemological issues are distinguished, the claim that feelings of familiarity justify endorsing the contents they target is difficult to support. Suppose that Sam acquired the belief that Caesar was murdered because he attended a lecture on the Roman Empire with you ten years ago. Suppose, moreover, that he now makes this judgment because he did attend this lecture. I submit that, intuitively, his judgment is justified—provided, of course, there are no defeaters. However, the absence of a feeling of familiarity is not one of these defeaters. If you come to learn that no feeling of familiarity accompanies Sam’s retrieving this content, you will not revise your assessment of his judgment.31 So, judgments based on past reasons— judgments made because one had these reasons—are justified in the absence of feelings of familiarity.32 This suggests that ascribing a justificatory role to these feelings betrays a confusion between an explanation as to why we sometimes judge out of a sensitivity to what we have previously represented (which consists in a content’s feeling familiar) and what justifies this judgment (the past reasons). In other words, to advocate the option under discussion is to confuse the attitude of remembering—which only explains why we make memory judgments—and constraints on contents that make them (justified) memory contents—constraints that specify what happened at the time the belief was acquired, as well as the relation between what happened then and what is happening at the time of memory.

On Seeming to Remember  339 These considerations motivate the second option distinguished previously, which is often described as the past reasons theory (Annis, 1980; Bernecker, 2010; Naylor, 1982). According to this theory, insofar as past reasons have not been defeated in the meantime, they justify endorsing the same content at the time of memory. If Sam’s reason for judging that Caesar was murdered is that he hears a teacher say so, this reason—provided that, for example, Sam did not come across information suggesting that the teacher is a fraud—justifies Sam’s memory judgment that Caesar was murdered.33 The past reasons theory avoids the worries attached to the claim that seeming to remember a content justifies endorsing that content. Of course, many refinements would be required in order to develop a complete account of propositional memory justification along the lines recommended by the past reasons theory.34 These refinements shall not concern me here. What I want to emphasize is that the past reasons theory does not, in itself, explain why we make memory judgments. Past reasons for judging that p may justify us in making the same judgment at a later time, but they do not explain why we are inclined to make that judgment when we remember. In light of the distinction between explanatory and justificatory issues related to propositional memory, we can conclude that the past reasons theory should remain open to a variety of explanations as to why we make propositional memory judgments, and in particular to first-person level explanations in terms of feelings of familiarity. We have reached the following conclusion: given that feelings of familiarity do not justify, the most appealing approach to the justification of propositional memory judgments is a version of the past reasons theory. With the help of this theory, we have dissociated the first-person level explanation of why we make memory judgments (viz., because of a phenomenologically manifest attitude of remembering) from what justifies these judgments (which relates to the aetiology of memory contents).

5  Conclusion: Generalizing the Lesson I shall now bring together and generalize the conclusions of Section 4 (feelings of familiarity explain why we endorse memory contents) and Section 5 (these feelings do not justify). These conclusions contain a more general lesson: having an attitude toward a given content is seldom a reason to endorse it. It is, I suggest, because feelings of familiarity characterize memory as an attitude that they are unfit to justify the contents they target. Why is that so? In general, the occurrence of an attitude does not carry any implication regarding whether we are justified in endorsing the content of this attitude. Consider these cases. “Why do you think that p is true?—Because I believe that p.” “Why do you think that o is dangerous?—Because I am afraid of it.” “Why do you think that p is true?—Because I seem to remember that p.”

340  Fabrice Teroni It is fair to say that reference to any of these attitudes is not a good answer to these epistemological questions. I have illustrated the fact that attitudes need not justify the judgments to which they incline us with belief, fear, and memory. These attitudes cannot justify the relevant judgments for the same reason. The reason is that believing, having an emotion and remembering are all based, albeit in different ways, on other psychological states. Believing may be based on other beliefs, on perception, or on introspection. To be afraid of a dog, one must be aware of it independently of one’s emotion, for instance by perceiving it (Deonna & Teroni, 2012). Fear would in that case be based on a perceptual experience. Similarly, someone who remembers a content must have previously thought it. Remembering is in this (different) sense based on past mental states. This basing relation has the following consequence. In order to assess whether the judgments we make because we have these attitudes are justified, we should turn our attention to the states on which they are based. Believing is obviously an inclination to endorse a content. But to assess whether endorsing that content is justified, we should turn our attention to whatever the belief is based on. The judgment that the dog is dangerous, which one is inclined to make because one is afraid of it, is justified only if the perceptual experience on which fear is based meets some constraints—for instance, if it represents the approach of a growling dog. Drawing attention to the fact that one’s fear is based on a perceptual experience with this content explains why one is justified in judging that the dog is dangerous. In a parallel fashion, we should assess the justification of the judgment we are inclined to make when we seem to remember as a function of the past mental state on which the memory is based. As this is typically a past judgment, the justification of the memory judgment depends on the justification of a past judgment, and so on one’s reasons to make it. The attitudes of believing, fearing, and remembering do not justify the relevant judgments. Yet, if my argument has been successful, attitudes are still needed to explain our inclination to endorse the relevant contents. In the same way as one might not have been inclined to judge that the dog is dangerous if one had not been afraid of it, one might not have been inclined to judge that Caesar was murdered if one did not seem to remember that so was the case. These explanations can take many different shapes. In the case of the emotions, it seems safe to say that the explanation is located at the first-person level. Emotions are phenomenologically salient, and we usually are in a position to know that we are inclined to make judgments because we emote. I have argued that the explanation of our tendency to endorse the contents that the attitude of remembering targets is of the same nature. Both of these explanations contrast with the explanation as to why we endorse the contents we believe, because believing is not a phenomenologically salient attitude.

On Seeming to Remember  341

Notes * I am grateful to Sven Bernecker for his helpful comments on a previous version of this paper. 1 Here and in what follows, I use the term “judgment” where others prefer to talk of occurrent (as opposed to dispositional) belief. Let me emphasize that I do not think of a judgment as a mental activity that critically assesses the evidence for and against a given proposition. In the intended sense, judgments take place whenever we accept or endorse propositions and can be quite passive, as is the case for many perceptual and memory judgments, for instance. 2 The characterization of episodic memory I am about to offer will inevitably be controversial, given the various debates as to what it amounts to. I shall leave this mainly terminological issue aside in what follows. The only point I wish to emphasize is that there is a fundamental difference between the phenomena I describe as episodic memory and those I describe as propositional memory. Naylor (2011) discusses the intricacies facing the definition of episodic memory and defends a position that differs from the one sketched here. 3 Images have a bad press. Here and in what follows, I use the term “image” only to emphasize that remembering something resembles perceiving it. For discussion, see Teroni (2017). 4 Given the way I have characterized episodic memory, one may remember that so-and-so happened in one’s own past without this qualifying as an instance of remembering episodically. In the following, I use examples of propositional remembering of historical and mathematical facts, but my conclusions should hold for any case of propositional memory as I characterize it here. 5 These contents are only typically endorsed, since one can remember that p without judging that p (e.g., when one has conflicting evidence). For a careful consideration of these cases, see Bernecker (2010, chap. 3). 6 This does not mean that sensory imagery never accompanies propositional memory: it obviously sometimes does. This imagery is inessential, however, since it does not constitute one’s memory of the relevant events or objects. See e.g., Martin, 2015. 7 The pervasive use of the expression “propositional attitude” explains why I favor the latter term in the context of a discussion of propositional memory. 8 Three observations are in order. First, we should leave room for some variation between the content of the past representation and that of memory (e.g., Matthen, 2010) and relax the identity constraint I use here (see Bernecker, 2010, pp. 217–229 for a convincing account in terms of content entailment). Second, the causal requirement should avoid deviant causal chains. Third, there is the additional issue, which I leave aside here, of whether only true contents can qualify as memory contents. Acknowledging these issues entails that the characterization I put forward should be significantly refined, but I believe that this does not affect my argument. 9 When I say that a content is a memory content “for the subject,” I do not mean that it is so “from his perspective.” I only mean that this content plays the relevant role in his psychology. The satisfaction of the relevant constraints need not be transparent to the subject. Let me observe in addition that the past awareness constraint simply acknowledges what Campbell (1994) describes as the “stepwise character” of memory. To say that memory presupposes past awareness does not imply that the subject paid attention to this past awareness at the time it occurred. 10 Here and in what follows, I shall call the contents that satisfy these constraints “memory contents.” This should not be read as implying that these contents are exclusive to memory.

342  Fabrice Teroni 11 In favoring this approach to the contrast between content and attitude, I follow suggestions made, among others, by Crane (2003) and Recanati (2007). See also Matthen (2010) and Deonna and Teroni (2012, chap. 7). 12 I criticize some of these options from a slightly different perspective in Teroni (2014). 13 Goldman thinks, and I agree, that this is a consequence of classical variants of functionalism. Many advocates of the past reason theory, which we shall have the opportunity to discuss in Section 4, favor this explanation of our inclination to endorse propositional memory contents. 14 Confronted with the same dialectical situation with respect to perception, Campbell (1984) explains why understanding perception purely in terms of the making of non-inferential judgments is unpersuasive. He does so by drawing attention to the difference between a standard perceiver who judges as she does because of what she sees and a blind yet reliable seer who finds himself with exactly the same judgments popping in his mind. I wish to insist on the existence of a similar difference with respect to memory. 15 I agree here with Werner (2013) that only some dispositions to judge are explained by a specific phenomenology. 16 Incidentally, observe that the contrast between episodic and propositional memory is sometimes drawn by saying that this sort of belief is distinctive of episodic memory (Owens, 1996). 17 Such a view may be inspired by some remarks made by Bernecker (2010, pp. 235–239), who does not explicitly endorse it. 18 For that reason, as Audi (1995) observes, insisting on the idea that propositional memory judgments always trace back to memory images is to pursue an inappropriate analogy with perception. 19 I favor an approach in terms of feelings of familiarity rather than in terms of feelings of pastness for two reasons that I can only sketch here. First, appealing to feelings of pastness may be plausible in relation to episodic memory. But it is much less so in relation to propositional memory. One’s seeming to remember that Caesar was murdered is not a situation in which one feels the pastness of this content. Second, and as Byrne observes, “while the ‘feeling of familiarity’ is, well, familiar, surely the ‘feeling of pastness’ is not.” (Byrne, 2010) 20 This conclusion is akin to Matthen’s (2010) idea that a feeling is characteristic of the attitude of remembering rather than of what is remembered. It differs insofar as Matthen explicitly endorses this claim only as regards episodic memory, appealing to a “feeling of pastness” that he contrasts with a “feeling of presentness” that accompanies perception. 21 The attitude of remembering may also target perceptual contents, as when something we see feels familiar. Discussion of this issue will have to wait for another occasion, however. 22 In what follows, I am exclusively concerned with the justification of these memory judgments. I am not interested in whether memory seemings can justify other judgments, such as the judgments that one has been aware of the content, or that it is because one was aware of the content that it is now available. For what it’s worth, I think that these seemings can justify these beliefs. 23 This is of course not to say that appealing to memory seemings is the only option for internalists (see Teroni, 2014). Yet, they have by and large favored this approach to memory justification. 24 Lackey (2005) contends that memory can generate new justification. Yet, according to her, the situations in which memory does so are quite different from, and more complex than, those I discuss here, and the position she favors cannot in any case salvage the memory seemings account. Moreover, the sorts of situations

On Seeming to Remember  343 she appeals to in support of her claim may fail to support it, as Senor (2007) argues. See also Teroni (2014). 25 See Tucker (2013) for a helpful introduction to the debate. 26 This is a simplification, as the impression may originate in a situation in which, although one could have judged, one did not. As far as I can see, this does not affect the points I want to make here. 27 In his discussion of phenomenal conservatism, which for present purposes can be understood as the claim that all seemings, memory or otherwise, confer prima facie justification, Tooley (2013) distinguishes basic seemings from derived seemings. His distinction is meant to restrict phenomenal conservatism to seemings that do not depend on the subject’s prior cognitive activities, and he does so for reasons that are closely related to those presented here. 28 One may think that appealing to constraints external to the memory seeming goes against the spirit of internalism (Hanna, 2011). This is questionable. For instance, the thought that there is a close relation between justification and the subject’s responsibility for his beliefs, which often underscores internalism, does not support present-tense internalism. After all, we are responsible for distant consequences of what we have culpably done. This is why it is surprising that internalists such as Huemer (2007) never introduce an aetiological condition in order to single out the relevant seemings. For Huemer’s own approach, which consists in distinguishing justified adoption and justified retention, see his (1999). 29 Let me emphasize once more that I am interested in the epistemological role of feelings of familiarity and so, given the earlier claims, in the contribution of the attitude of remembering to justification. I do not claim that endorsing a memory content cannot be justified by something one is aware of when this content crosses one’s mind. I want to insist, though, that in such cases one does not judge because one remembers. 30 Hasan (2013) emphasizes the distinction between the psychological and epistemological roles of seemings, Bergman (2013) that between making intelligible and justifying. 31 This may lead a friend of the option under discussion to distinguish the justification of propositional memory judgments, for which seemings are required, from the conditions under which these judgments constitute knowledge (Audi, 1995). This would amount to saying that, in this area at least, there can be knowledge without justification. I do not discuss this idea here, as it is not relevant for the issues I wish to address. 32 In the same spirit, Conee (2013) writes that “absent defeaters, it is sufficient . . . to believe based on the evidence that prompts the inclination.” 33 Given the observations in the previous paragraph, one should insist that past reasons justify memory judgments only if these judgments causally depend on them. Sam must judge that Caesar was murdered because he heard his teacher say so. This basing relation is a natural addition to the past reasons theory. The relevant causal relation takes place independently of the subject’s access to the past reason and more generally, of any psychological process taking place at the time of memory. 34 For sophisticated versions of this theory, see Naylor (1971, 1982) and Bernecker (2010).

References Annis, D. (1980). Memory and Justification. Philosophy and Phenomenological Research, 40, 324–333.

344  Fabrice Teroni Audi, R. (1995). Memorial justification. Philosophical Topics, 23(1), 31–45. Bergmann, M. (2013). Externalist justification and the role of seemings. Philosophical Studies, 166(1), 163–184. Bernecker, S. (2010). Memory: A philosophical study. New York: Oxford University Press. Burge, T. (1993). Content preservation. The Philosophical Review, 102(4), 457–488. Byrne. A. (2010). Recollection, perception, imagination. Philosophical Studies, 148(1), 15–26. Campbell, J. (1994). Past, space and self. Cambridge, MA: MIT Press. Campbell, K. (1984). Body and mind. Notre Dame: University of Notre Dame Press. Chisholm, R. (1957). Perceiving. Ithaca, NY: Cornell University Press. Conee, E. (2013). Seeming evidence. In C. Tucker (Ed.), Seemings and justification: New essays on dogmatism and phenomenal conservatism (pp. 52–68). New York: Oxford University Press. Crane, T. (2003). The intentional structure of consciousness. In A. Jokic & Q. Smith (Eds.), Consciousness: New philosophical perspectives (pp. 33