Memory: A Self-Referential Account

Citation preview

Memory

P H I L O S O P H Y O F M E M O RY A N D I M AG I NAT IO N Series Editors Amy Kind, Claremont McKenna College Kourken Michaelian, University of Otago * Advisory Board Sven Bernecker, Greg Currie, Christoph Hoerl, Bence Nancy, Kathleen Stock, John Sutton * The philosophies of memory and imagination are two of the most exciting new areas in philosophy. This series exists to publish cutting-​edge work in both areas and to encourage interaction between them.

Memory A Self-​Referential Account J O R D I F E R NÁ N D E Z

1

3 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America. © Oxford University Press 2019 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. CIP data is on file at the Library of Congress ISBN 978–​0–​19–​007300–​8 1 3 5 7 9 8 6 4 2 Printed by Integrated Books International, United States of America

To Zoe and Abby, who help me form beautiful memories every day

Contents Preface Acknowledgments

ix xiii

PA RT I .   T H E NAT U R E O F   M E M O RY 1: Problems of Memory 1.1 Introduction 1.2 Scope of the Project 1.3 Explanandum 1.4 Available Strategies 1.5 Approach 1.6 Summary

3 3 4 8 14 23 27

2: The Metaphysics of Memory 2.1 Introduction 2.2 The Causal Theory of Memory 2.3 The Narrative Theory of Memory 2.4 Functionalism about Memory 2.5 Narratives, Causal Histories, and Functional Roles 2.6 Conclusion

32 32 33 40 47 51 53

3: The Intentionality of Memory 3.1 Introduction 3.2 Mnemonic Content 3.3 Methodology 3.4 The Objective View 3.5 The Subjective View 3.6 The Conjunctive View 3.7 The Causal Self-​Reference Approach 3.8 The Reflexive View 3.9 Conclusion

57 57 58 60 62 64 67 72 78 80

PA RT I I .   T H E P H E N OM E N O L O G Y O F   M E M O RY 4: The Experience of Time 4.1 Introduction 4.2 Temporal Phenomenology

85 85 86

viii Contents

4.3 Mental Time Travel and Re-​Presentation 4.4 Mental Time Travel and Representation 4.5 Temporal Locations and the Feeling of Pastness 4.6 Temporal Distances and the Feeling of Pastness 4.7 Beyond Time: The Experience of Origin 4.8 Conclusion

5: The Experience of Ownership 5.1 Introduction 5.2 Disowned Memory 5.3 Ownership as Identification 5.4 Ownership as Endorsement 5.5 The Appearance of Memory 5.6 The Ownership of Phenomenal States 5.7 Conclusion

89 94 99 103 106 109 112 112 114 117 119 123 127 134

PA RT I I I .   T H E E P I S T E M O L O G Y O F   M E M O RY 6: Immunity to Error through Misidentification 6.1 Introduction 6.2 Memory and IEM 6.3 Quasi-​Memory and IEM 6.4 Observer Memory and IEM 6.5 A Defense of IEM in Memory 6.6 The Perceived Self 6.7 An Account of IEM in Memory 6.8 Conclusion

139 139 140 144 150 157 160 167 168

7: Memory as a Generative Epistemic Source 7.1 Introduction 7.2 Varieties of Preservativism 7.3 Attention in Memory 7.4 Abstraction in Memory 7.5 The Recorder Model of Memory 7.6 Reconstruction in Memory 7.7 Mnemonic Content and Generativism 7.8 Conclusion

171 171 173 177 181 188 191 199 202

Conclusion

206

References Index

209 217

Preface This book is a philosophical investigation of memory. It is, in a nutshell, an attempt to answer the question of what memory is. There are many philosophically interesting aspects of memory. Memory is interesting, as far as the philosophy of mind is concerned, because memory is a mental capacity; a capacity which is different from introspection, perception, and imagination in ways which are difficult to specify. Memory is interesting for epistemology too because, among other reasons, memory seems to provide us with a special knowledge of the past. Memory is also interesting for metaphysics due, for instance, to the connections between memory and personal identity and between memory and time. Furthermore, memories involve some characteristic feelings and, for that reason, memory is interesting for phenomenology. Moreover, there is a normative dimension to memory. We seem to think that, in some cases, we have the obligation to remember some things and the right to forget others. So, memory has important ramifications into ethics as well. It is, in fact, difficult to find an area of philosophy for which memory is irrelevant. The fact that memory has multiple aspects which are interesting from a philosophical point of view presents a challenge for the project in this book. For the project is to offer a philosophical account of memory; an account of the philosophically interesting aspects of memory. And this means that the project should ideally have an all-​encompassing scope. But there is a certain trade-​off between the scope of a philosophical investigation of a mental capacity such as memory and how informative the investigation turns out to be. The broader the investigation, the less informative it tends to be. Thus, in this book, I have tried to select a few philosophical issues concerning memory, and I have attempted to investigate them in some depth. The goal has been to achieve a delicate balance. On the one hand, hopefully the selected issues on memory are all philosophically interesting and sufficiently diverse. On the other hand, hopefully the selection is not so ambitious that one cannot say much, informatively, about each of the issues separately, and about how they are related to each other.

x Preface The different issues on memory that I  will target for explanation are specified in c­ hapter 1. The book discusses one issue on the metaphysics of memory, one issue on the intentionality of memory, two issues on the phenomenology of memory and two issues on the epistemology of memory. In ­chapter 1, I also address some methodological matters. I consider different ways in which one could tackle the project of accounting for the metaphysics, the intentionality, the phenomenology, and the epistemology of memory, and I  highlight the different assumptions that one would need to make, depending on which of those aspects of memory are taken to be basic and which of those are taken to be more derivative. My own approach, outlined in ­chapter 1, is to regard the metaphysics and the intentionality of memory as the basic aspects of memory and to try to shed some light on the phenomenology and the epistemology of memory based on my views about those two basic aspects. Chapter 2 puts forward a view about the metaphysics of memory; a view about the facts in virtue of which a mental state qualifies as a memory. The view is that no intrinsic property of the mental state makes it qualify as such. What makes a mental state qualify as a memory is the functional role of the mental state within the subject’s cognitive economy. Chapter 3 puts forward a view about the intentionality of memory; a view about the content of memories. The view is that memories are self-​referential in the following sense. Memories represent the fact that they have a causal origin in the subject’s past experiences. Put together, these two proposals constitute the core of the account of memory offered in this book. These are, in my view, the main elements of the nature of memory. Chapters 4 and 5 deal with the phenomenology of memory. Specifically, ­chapter 4 deals with the feeling of pastness involved in memory; the feeling that what one is remembering is in the past. It also deals with a certain feeling of familiarity in memory; the awareness of what it was like for one to experience, in the past, what one is remembering in the present. And ­chapter 5 deals with the feeling of ownership in memory; the feeling that the memory that one is having, when one remembers something, is one’s own. I try to explain these feelings by appealing to the content that memories have. In all three cases, the thought is that the relevant feeling is the way in which one experiences some of the things that one is representing when one is having a memory. This account of the phenomenology of memory, then, will rest on the account of the intentionality of memory to be offered in ­chapter 3.

Preface  xi Chapters 6 and 7 are concerned with the epistemology of memory. More specifically, they are concerned with two aspects of the special kind of epistemic justification which is afforded by our memories. Chapter 6 discusses the fact that memory judgments are protected from errors of misidentification, whereas ­chapter 7 discusses the fact that the epistemic justification enjoyed by some of those judgments is generated, and not merely preserved, by the faculty of memory. I try to explain both epistemic aspects of memory judgments by appealing to the content that memories have as well. Thus, in both cases, the thought is that the key to why memory judgments enjoy the kind of epistemic justification which has the relevant feature is to be found in the contents of the memories on the basis of which those judgments are formed. Once more, then, what we will have is an account of the epistemology of memory which rests on the account of the intentionality of memory to be offered in the first part of the book. There are a number of directions one could go from there. Perhaps the functionalist view of the metaphysics of memory to be offered in c­ hapter 2 can tell us something interesting about personal identity, and perhaps the explanation of immunity to error through misidentification to be offered in ­chapter 6 can tell us something interesting about our sense of personal identity over time. Likewise, it is possible that the conception of mental time travel proposed in ­chapter 4 can shed some light on whether memory needs to be oriented toward the past or, by contrast, it can be oriented toward the future as well. And it may be that the proposal about the feeling of ownership of our memories to be put forward in c­ hapter 5 can be generalized to other mental states in ways which illuminate a variety of mental disorders wherein the subject disowns some of their mental states. By the time we conclude the project in this book, all of these avenues will remain open. Nonetheless, the hope is that, at the completion of the project, we will have achieved a sufficiently comprehensive understanding of what memory is and why it matters to us so much. This project began 15  years ago, when I  was a post-​doctoral fellow at Macquarie University, in Sydney. I  was very fortunate to be able to discuss various issues on memory with John Sutton and Tim Bayne while I was there. Thanks to those interactions, I was able to put together some ideas about memory which turned out to be the kernel for c­ hapters 3 and 4 of this book. The rest of the chapters in the book originate in more recent work. Chapter 5 is closely related to an article I have written on the sense of mineness; an article for which I received very valuable feedback from Manuel

xii Preface García-​Carpintero and Marie Guillot. Chapter 2 and c­ hapter 6 are based on articles which greatly benefited from Kourken Michaelian’s comments. And ­chapter 7 is built on an article that I have been able to discuss, very helpfully (for me), with Sven Bernecker. In fact, John, Kirk and Sven belong to a group of philosophers who seem to have been sharing their work on memory informally for the last 10  years or so. This group also includes Felipe De Brigard, Dorothea Debus, Christoph Hoerl, Chris McCarroll, Denis Perrin, Sarah Robins, and Markus Werning, and I am grateful to all of them for their feedback on drafts of the chapters of this book delivered at conferences and workshops over the years. I also want to thank, collectively, my colleagues at the Philosophy Department of the University of Adelaide for the research environment in which the book was written. Many thanks, in particular, to Garrett Cullity, Antony Eagle, and Philip Gerrans for a number of helpful conversations on memory, and to Matthew Nestor and Laura Bottrill for their valuable assistance with various research and editorial tasks. My editor, Peter Ohlin, and the series editors, Kourken Michaelian and Amy Kind, along with the helpful staff at Oxford University Press, made the process of publishing this book a pleasure, for both their patience and their diligence. Three anonymous referees for Oxford University Press (two of whom turned out to be Sven Bernecker and Sarah Robins) read whole drafts of the book and provided remarkably thorough and insightful comments, for which I am very grateful. I particularly want to thank Uriah Kriegel, who has maintained a keen interest in the project over the years. He provided comments on versions of most of the chapters in the book and, on more than one occasion, those comments allowed me to see the way around some serious objection to my account. Other times, the serious objection actually came in his comments. Either way, the process was always really helpful. Many friends, inside and outside of the profession, provided other kinds of support while this book was written, and I cannot mention all of them here. But I do want to thank my brother and my sister. And most of all, I want to thank my two daughters. They are too little to know it now, but the fact is that, without them, this book would never have been written.

Acknowledgments My work on this book was supported by a semester’s Special Studies Leave from the University of Adelaide. This research was also funded by two grants from the Australian Research Council for projects DP130103047 and FT160100313 under the Discovery Scheme. Most of the chapters are based on articles that have appeared elsewhere. Chapter  2 is based on “The Functional Character of Memory,” in (2018) K. Michaelian, D. Debus, and D. Perrin (eds.) New Direction in the Philosophy of Memory. London:  Routledge, 52–​72. Chapter  3 is based on (2006) “The Intentionality of Memory,” Australasian Journal of Philosophy 84: 39–​57. Chapter 4 is based on (2008) “Memory and Time,” Philosophical Studies 141:  333–​356. Chapter  6 is based on (Forthcoming) “Observer Memory and Immunity to Error through Misidentification,” in Synthese, and (2014) “Memory and Immunity to Error through Misidentification,” Review of Philosophy and Psychology 5:  373–​390. Chapter  7 is based on (2016) “Epistemic Generation in Memory,” Philosophy and Phenomenological Research 92: 620–​644. I am grateful to the publishers concerned for their permission to make use of material from these pieces.

PART I

THE NAT U R E OF M E MORY

1 Problems of Memory 1.1  Introduction The goal of this book is to understand a cognitive capacity which is central to our mental lives; the capacity for remembering. Specifically, the goal is to offer an account of four different aspects of memory, all of which are philosophically interesting.1 This chapter is devoted to setting up the discussion that will lead to such an account, and addressing a number of preliminary methodological issues. What kind of memory are we going to focus on exactly? What are the philosophically interesting aspects of the relevant type of memory? What types of explanations of those aspects of memory are available? These are some of the questions that we will need to address before an account of memory can be offered. I will proceed as follows. In section 1.2, I will specify the kind of memory on which we will concentrate. In section 1.3, I will spell out the four aspects of memory that we will target for explanation. In section 1.4, I will distinguish several ways in which we can approach those aspects of memory. In section 1.5, I will make explicit the approach that I will be adopting, and I will offer an overview of how the discussion will be structured over the coming chapters. Finally, in section 1.6, I will review how the discussion has been set up, and I will highlight the assumptions which have been made as part of the set-​up.

1 In what follows, I will use both “remembering” and “memory” non-​factively. Thus, on the use of the two terms that I will adopt here, it is possible to remember something falsely, and it is possible to have a false memory of something. This may be a philosophically unorthodox use of the two terms. But there are some reasons for thinking that “remembering” and “memory” are not factive. After all, we commonly use the expression “false memory,” and this does not strike us as a deviant use of the term “memory.” Which it should, if “memory” was factive. Likewise, claims of the kind “nothing false can be remembered” do not strike us as obviously true. In fact, as pointed out in (Hazlett 2010), they strike us as false. This should not happen if “remembering” was factive either.

4  The Nature of Memory

1.2  Scope of the Project We commonly speak of memory as if it was directed at either facts, events, actions, experiences, objects, or abilities. Thus, we self-​ascribe memories with expressions of the form “I remember that p” (where “p” stands for a fact, as in “I remember that team T lost the game”), “I remember e” (where “e” stands for an event, as in “I remember team T’s defeat”), “I remember ψ-​ing” (where “ψ” stands for an action, as in “I remember playing with team T”), “I remember what it was like to have P” (where “P” stands for a property, as in “I remember what it was like to lose that game”), “I remember x,” or “I remember some x” (where “x” stands for an object, as in “I remember my team T jersey”), and “I remember how to ψ” (where “ψ” stands for an action, as in “I remember how to play the game”). For some of these types of memory, the question of whether the relevant type can be reduced to a combination of some other types of memory arises. One might wonder, for example, whether remembering my team T jersey is something over and above remembering a number of facts about that jersey, remembering what it was like to wear it and remembering actions such as my putting the jersey on. Similarly, one might wonder whether remembering playing with team T is just a combination of remembering my team mates, remembering the playing field, remembering some facts about the games in which team T was involved and remembering what it was like to play with the team. The project of determining which types of memory are, as it were, basic and which types can be reduced to other types is a challenging and interesting project. However, it is not a project in which we will engage here. For our discussion of memory will not be concerned with all of the various types of memory just distinguished. In fact, it will only be concerned with two of them. In this discussion, we will concentrate on memory for facts and memory for perceptual experiences. Memory for facts is the type of memory that we express in propositional form. The account of memory that I will propose below, however, is meant to be neutral on the metaphysics of both facts and propositions. It will be convenient, for expository purposes, to assume that both facts and propositions are identical with sets of possible worlds.2 However, no substantial claim 2 On this view, a proposition is identical with the set of possible worlds with respect to which it is true. For instance, the proposition that my team T jersey is blue and red is identical with the set of possible worlds with respect to which it is true that my team T jersey is blue and red. For this view on the nature of propositions, see (Stalnaker 1976). Since I will be assuming this view, I will speak

Problems of Memory  5 in the coming chapters should hinge on this view about the nature of propositions and facts. For the sake of simplicity, we will be mainly concerned with facts involving objects of a particular kind; objects which are perceivable through the sensory modality of vision.3 However, the account of memory to be proposed is meant to have a broader scope. It should equally apply to memory for facts involving objects perceivable through sensory modalities other than vision. Memory for perceptual experiences, on the other hand, is a type of memory that we express by using locutions of the what-​it-​was-​like form. The account of memory that I  will propose will assume that perceptual experiences are properties of subjects; properties with phenomenal, or qualitative, features. The experience of seeing team T’s jersey, for example, will be construed as a property such that, in virtue of having the property, the subject feels in a particular way; a way which is visually characteristic. The account of memory to be proposed here will also assume that perceptual experiences are properties with intentional, or representational, features. The experience of seeing team T’s jersey, for example, will be construed as a property such that, in virtue of having that property, the subject represents certain things (such as the jersey). In what follows, given the emphasis on memory for facts which involve visible objects, the perceptual experiences that we will often find ourselves discussing will be perceptual experiences of the visual kind. However, the account of memory to be proposed is also intended to apply to memory for perceptual experiences associated with sensory modalities other than vision. Let us specify the scope of the project a bit further. Consider, first of all, propositional memory or memory for facts. Two forms of memory seem to be lumped together into this category. There seems to be a distinction between the type of memory that you have when you remember, for example, that Columbus arrived to America in 1492 and the type of memory that you have when you remember, for example, that the stove in your kitchen was on when you were leaving the house. The distinction is hard to formulate, but it is intuitively clear. In the former case, when you remember, you are having a belief; of possible worlds as “belonging to” propositions, and I will speak of propositions as “containing” possible worlds. Also, I will be somewhat loose in my use of the terms “possible situation” and “possible world,” which I will use interchangeably. Strictly speaking, talk of a possible situation should be understood as referring to a possible world where the situation is the case. 3 Why “mainly”? In ­chapter 3, we will see that facts of a different kind will also be important for the account of memory to be developed here. These are facts about the causal histories of memories themselves.

6  The Nature of Memory a belief which has been preserved from some time in the past, that is, the time at which you originally learnt the remembered fact. Let us call this form of remembering “semantic memory.” In the latter case, when you remember, you are having an experience; an experience that typically originates in a past perceptual experience of the fact. Memory experiences seem to be similar to perceptual experiences in one respect. Memory experiences have phenomenal features. There seems to be such a thing as what it is like to remember, for example, that the stove was on at the time that you were leaving the house. Memory experiences are similar to perceptual experiences in a different respect too. Memory experiences have intentional features. It seems that, in virtue of having the kind of experience that you have when you remember that the stove was on, you are able to represent certain things. You can, for example, visualize your kitchen, the stove, and the flame in it. Since memory experiences allow you to visualize facts or, equivalently, to picture them in your mind, we can think of those experiences as mental images. Let us call the form of remembering which involves such experiences “episodic memory.”4 It is possible for us to remember a fact by having both an experience of, and a belief about, that fact. Suppose that I ask you, for example, whether Mary was at the party last month, and you have an experience wherein you are able to visualize a scene from the party. You may be able to direct your attention at different objects represented by your experience and eventually find Mary among them. In this case, you are remembering that Mary was at the party in the episodic sense. However, it may also be the case that, at the time that you saw Mary during the party, you formed the belief that Mary was there, and your belief has been preserved up to the time at which I am asking my question. In that case, you are remembering that Mary was at the party in the semantic sense. Thus, it is possible for episodic and semantic memory to be in play at the same time. It is also possible for the two kinds of memory to come apart. You may remember that Mary was at the party by having an experience which allows you to visualize Mary even though, at the time that you were at the party, your attention was focused on someone else and, as a result, you saw but did not notice (and therefore did not form the belief) that Mary was there. 4 I am borrowing the terms “episodic” and “semantic” from (Tulving 1972) to refer to two sorts of memory. I will also use “episodic memory” to refer to an experience in virtue of which a subject remembers episodically. Notice that the characterization of episodic memory above makes it a type of memory for facts, just like semantic memory is a type of memory for facts as well. What differentiates episodic remembering from semantic remembering is not the sort of entity being remembered. It is an independence from belief that episodic remembering enjoys and semantic remembering lacks.

Problems of Memory  7 Conversely, you may remember that Mary was at the party in that, at the time that you were at the party, you formed the belief that Mary was there and your belief has been preserved up to the present time. But you may not have an experience which allows you to visualize Mary. Perhaps you did not see Mary during the party, and you formed your belief that Mary was there because, at some point during the party, someone else told you that Mary was there. It is hard to tell whether the distinction between semantic and episodic memory cuts across the distinction between memory for facts and memory for perceptual experiences. For, whereas it seems clear that one can remember one’s past perceptual experiences episodically, it is not equally clear that one can remember them semantically. The episodic sense seems to be the sense in which we understand locutions of the form “I remember what it was like to have P.” And, in particular, it seems to be the sense in which we understand those locutions when “P” stands for, specifically, a perceptual experience. When someone claims to remember what it was like to see the sea for the first time, for instance, it seems natural to understand them as saying that they are having an experience which represents their past perceptual experience; their perceptual experience of the sea. But it is also possible to understand their claim as saying that, at the time that they saw the sea for the first time, they formed, reflectively, a number of beliefs about the perceptual experience that they were undergoing (such as the belief that it was an overwhelming experience), and those beliefs have been preserved up to the present time. Would the latter be semantic memory for past perceptual experiences? In this situation, we may be more inclined to think that the subject is remembering certain facts about their past experience, as opposed to the experience itself. And, if this is correct, then there is no such thing as semantic memory for perceptual experiences. In any case, the issue of whether a subject, in the situation described, is having memories for facts about their past experience or, by contrast, they are remembering their past experience semantically will not be of great significance for our discussion in what follows. The reason is that the object of our investigation will be episodic memory and, thus, the scope of semantic memory will not be our concern here. In what follows, we will be concerned with several aspects of episodic memories for facts and of episodic memories for perceptual experiences.5 Let us turn to the relevant aspects of those memories now. 5 As I  have assumed that the experiences in virtue of which we episodically remember facts involving visible objects are mental images, I will be assuming that episodic memories for such facts

8  The Nature of Memory

1.3  Explanandum We have just seen that remembering facts episodically involves having certain experiences. When the facts being remembered involve visible objects, I have proposed to construe such experiences as mental images. However, the occurrence of mental imagery is not exclusive to memory. When we imagine the fact that the stove is on in the kitchen, for example, we are also capable of visualizing it in much the same way in which we visualize it when we remember it episodically. Similarly, the occurrence of experiences wherein we are able to mentally represent a different experience of ours is not exclusive to memory either. When we imagine having a perceptual experience of the sea, for example, we are capable of representing what it would be like for us to have that perceptual experience, in much the same way in which we represent it when we remember having the experience. An interesting question which arises at this point, then, is what makes certain experiences qualify as episodic memories and what makes other experiences fall outside of the memory domain. This is a metaphysical question. It is a question about the conditions under which an experience qualifies as an episodic memory and not a question about, for example, how we identify certain experiences as episodic memories. The metaphysics of memory includes other interesting questions as well. It is an area which has been explored in detail in connection, for instance, with the topic of personal identity. Consider the question of what makes you the same person over time. In virtue of what fact is a subject, existing at a time, the same person as a subject existing at an earlier time? One might think that it is in virtue of the fact that the former subject remembers experiences, or actions, which belong to the latter subject. And, for that reason, memory has been thought to be central to the topic of personal identity.6 The metaphysics of memory has also been explored in connection with the nature of time. There is, for example, the issue of whether some views about the nature of time sit more easily than others with certain facts about episodic remembering which are commonly accepted by us. Suppose, for

are mental images. I intend to remain neutral on the nature of episodic memories of other types, other than for the above-​made claims that they are experiences with phenomenal and intentional features. 6 The view that memory constitutes personal identity has often been attributed to John Locke based on Locke’s discussion of personal identity in book 2, chapter xxvii, of the Essay (1975). See, however, (Behan 1979) for a different reading of Locke.

Problems of Memory  9 example, that there is an ontological difference between the past and the present and, whereas the present is real, the past is no longer real. Then, past facts are not real. But if past facts are not real, then what makes some of our episodic memories true? It seems prima facie plausible to think that our memories are true or false in virtue of, precisely, past facts. For that reason, memory has been thought to be relevant to the nature of time too.7 In ­chapter 2, we will investigate the metaphysics of memory. However, we will not be pursuing the connection between memory and personal identity. We will not be pursuing the connection between memory and the nature of time either. Instead, we will be concerned with the issue of what conditions an experience must satisfy in order to qualify as an episodic memory. For the sake of simplicity, we will concentrate on memory for facts involving visible objects. (The outcomes of our investigation will easily generalize to memory for perceptual experiences.) Thus, the question which will occupy us while we discuss the metaphysics of memory will be a question about, specifically, mental images. However, the broader formulation of the question that we will try to answer will remain the following: i. Metaphysics What are the conditions under which an experience qualifies as an episodic memory? In section 1.2, we have seen that episodic memories, like perceptual experiences, have intentional features. When we have an episodic memory, we thereby represent certain things. An interesting question which arises, then, is what things we represent in virtue of having episodic memories. This is a question about the intentionality of memory. It is a question about the conditions under which some things are represented by an experience which qualifies as an episodic memory and other things are not. Notice that it is different from the question of how the subject themselves would express their having the relevant experience. Consider, for example, memory for facts. Depending on one’s views about self-​knowledge, one might think that if a subject represents that some proposition p is the case in virtue of having an episodic memory, then this is a fact which must be cognitively available to them. And, for that reason, if the subject’s memory represents that p, then they must be able to express their memory with claims of the sort “I remember 7 On memory and the nature of time, see (Le Poidevin 2017).

10  The Nature of Memory that p.” Conversely, one might think that if a subject represents that some proposition p is the case in virtue of having an episodic memory, then this is a fact about which they cannot easily be wrong. And, for that reason, if a subject expresses one of their episodic memories with a claim of the kind “I remember that p,” then their episodic memory must represent the fact that p. I will not be making either of those two assumptions here. In fact, I will be assuming that both of those conditional claims are false.8 Admittedly, I will sometimes speak of an episodic memory as a memory that the subject would express by saying that they remember such-​and-​such. However, by singling out the memory by reference to the claim that the subject themselves would make in order to express it, I intend to leave the question of what the memory represents open. For what a memory represents does not need to be fully, or manifestly, available to the subject for report. If the notion of representation that we will be employing does not correspond to the claims that the subject would make in order to express their memories, then, one might ask, what will be the operative notion of representation? In ­chapter  3, we will investigate the intentionality of memory. We will pursue the question of what episodic memories represent or, equivalently, what the content of memories is.9 The relevant notion of content will involve the truth-​conditions of memories. For that reason, I will construe the contents of memories as propositions; propositions which capture those truth-​conditions. And, as a result, the question that will concern us with regards to the intentionality of memory will be the question of which types of propositions constitute the contents of our episodic memories. What propositions capture the conditions with respect to which our episodic memories are correct and the conditions with respect to which they are incorrect? Let us highlight this aspect of our explanandum now: ii. Intentionality What are the contents of our episodic memories? In section 1.2, we have seen that, in addition to intentional features, episodic memories have phenomenal features. This is another respect in which 8 On the broad topic of the limits of self-​knowledge, see (Fernández 2013). On some of the ways in which some contents of memory, more specifically, may not be available in self-​knowledge, see (Fernández 2015). 9 Hereafter I  will use several expressions equivalently to refer to the intentionality of memory. These include, for any memory M, “the content of M,” “the intentional object of M,” “what M represents,” and “what M is about.”

Problems of Memory  11 episodic memories and perceptual experiences are similar. When we have an episodic memory, there is such a thing as what having that episodic memory is like. An interesting question which arises, then, is what it is like for us to have episodic memories. This is a question about the phenomenology of memory. The phenomenology of memory is a particularly interesting and puzzling area. On the one hand, it seems quite clear that, in normal circumstances, having episodic memories feels quite differently from having other mental states, such as perceptual experiences or episodes of imagination. On the other hand, it is quite hard to put one’s finger on the specific aspects of the phenomenology of memory which distinguishes it from the phenomenology of perception, or that of imagination. One way in which, when we have episodic memories, we feel differently from the way in which we feel when we have perceptual experiences or episodes of imagination seems to involve an experience of time. More specifically, it seems to involve an experience of the past. It is unclear, however, exactly what is being experienced as being in the past in virtue of having an episodic memory. Consider, for example, the episodic memory that I would express by saying that I remember that Mary was at the party last month. Arguably, the presence of Mary at the party is a fact which is experienced by me as being in the past in virtue of having this episodic memory. But is my past perceptual experience of Mary as being at the party also experienced by me in virtue of having the memory? This is an issue that we will need to address in order to clarify the phenomenology of memory. Another issue that we will need to address is whether (and if so, in what sense) the phenomenology of memory involves some awareness of the self. After all, one might argue that, when I have the episodic memory that Mary was at the party, the presence of Mary at the party is not only experienced by me as being in the past, but it is also experienced by me as being in my past. Relatedly, one might also argue that, when I have the episodic memory that Mary was at the party, the memory at issue is presented to me as being mine, or as belonging to me. These ideas, which seem prima facie plausible, suggest that, in some sense yet to be determined, we are aware of ourselves in memory. This sort of awareness will also need to be addressed while we clarify the phenomenology of memory; a project that will occupy us in the second part of the book. In c­ hapter 4, I will attempt to explain the aspects of the phenomenology of memory which involve the experience of time, whereas in c­ hapter 5, I will address the sense of ownership for our episodic

12  The Nature of Memory memories. Our investigation of memory will therefore be concerned, in those chapters, with the following issue: iii. Phenomenology What is it like for us to have episodic memories? Episodic memories are not only similar to perceptual experiences in that they have intentional and phenomenal features. They are also similar in that, usually, mental states of both types elicit beliefs in us. Just like, in normal circumstances, when we have a perceptual experience, we trust that experience and we are inclined to believe what the experience presents to us, it is also the case that, in normal circumstances, when we have an episodic memory, we trust that memory and we are inclined to believe what the memory presents to us. Let us refer to those beliefs which we form on the basis of our episodic memories as “memory beliefs.” (I will reserve the term “memory judgment” for the narrower category of occurrent memory beliefs.) There are a number of interesting questions about the epistemic status of memory beliefs. The question which most naturally arises is that of whether, in general, memory beliefs are justified. Assuming that justified beliefs are more likely to be true than those which are not justified, the question of whether, in general, memory beliefs are justified turns on that of whether episodic memory is generally reliable. To be clear, the fallibility of memory is not in dispute here. We are all familiar with instances in which we misremember things. In fact, cognitive psychology provides us with relatively specific parameters within which misremembering is more likely to occur.10 But, given this kind of evidence, one might wonder whether we are generally entitled to believe what our episodic memories present to us. Suppose, for a moment, that memory is generally reliable and, thus, we are entitled to believe what our memories present to us. Then, at least two other questions about the epistemic status of memory beliefs seem to arise. One of them is the question of whether, despite the fallibility of memory, there is a certain kind of error from which memory beliefs are protected. As a rough approximation, we may say that this is the kind of error which occurs 10 The so-​called misinformation effect constitutes an example of this. It seems that if, after perceiving a fact, we acquire false information about that fact, then our being in possession of that information can make our episodic memories of the perceived fact less accurate. On this effect, see (Loftus 2005).

Problems of Memory  13 when we accurately remember something happening to someone, but we mistakenly assume that the remembered person is ourselves and, as a result, we wrongly believe that what we remember happened to us. The property of being protected from such errors has been labeled “immunity to error through misidentification” in the philosophical literature.11 The question of whether memory beliefs are immune to error through misidentification is interesting because if memory beliefs do have this peculiar property, then memory beliefs are endowed with a sort of epistemic justification which is special. But the epistemic justification that memory beliefs enjoy may be special in other respects as well. Another interesting question about the epistemic status of memory beliefs is that of whether their justification derives from, or reduces to, the justification that beliefs formed through other faculties, such as perception or introspection, enjoy. If episodic memory merely maintains, over time, the epistemic justification that beliefs formed through other faculties acquired in the past, then the justification of memory beliefs must derive from the justification afforded by those faculties. But if memory is capable of providing justification for a belief regardless of whether, in the past, the subject was justified in forming that belief through some other faculty, then it seems that the justification of memory beliefs does not depend on the justification afforded by any of our other faculties. For that reason, the question of whether the justification enjoyed by memory beliefs derives from the justification that beliefs formed through other faculties, such as perception or introspection, enjoy has been discussed in the philosophical literature as the question of whether episodic memory “preserves” or “generates” epistemic justification.12 In the third part of the book, we will engage with the epistemology of memory. In ­chapter 6, I will try to show that memory beliefs are immune to error through misidentification, and in ­chapter 7, I will contend that memory generates epistemic justification. The issue that will concern us in the third part of the book, therefore, is the broad issue of whether our justification for memory beliefs is special in any way: iv. Epistemology What kind of epistemic justification do our episodic memories afford?

11 On immunity to error through misidentification, see (Prosser and Recanati 2012). 12 For an overview of this debate, see (Senor 2017).

14  The Nature of Memory A question that I will not be addressing, then, is that of whether memory is generally reliable or not. Instead, I will assume that memory is indeed generally reliable. Notice that if the outcomes of our discussion on the epistemology of memory are correct, then making such an assumption will be vindicated as having been the correct strategy all along. For if the epistemic justification enjoyed by our memory beliefs does not reduce to the justification enjoyed by beliefs that we have formed through other faculties, then memory is a basic source of epistemic justification. And it seems that the reliability of basic sources of epistemic justification cannot be established without circularity.13 Any argument for the general reliability of memory will rest on premises which we are only entitled to accept if we assume, in the first place, the general reliability of memory. Thus, the general reliability of episodic memory will be treated as an assumption throughout this discussion.

1.4 Available Strategies To recapitulate, this investigation of memory will touch on four aspects of it; the metaphysics of memory, the intentionality of memory, the phenomenology of memory and the epistemology of memory. The issues to be addressed will be the following: i. Metaphysics What are the conditions under which an experience qualifies as an episodic memory? ii. Intentionality What are the contents of our episodic memories? iii. Phenomenology What is it like for us to have episodic memories? iv. Epistemology What kind of epistemic justification do our episodic memories afford? The project of addressing these four issues can be tackled in different ways, depending on one’s position with regards to which of the four aspects of 13 On the challenge of demonstrating that a basic source of justification is reliable without committing circularity, see (Alston 1986).

Problems of Memory  15 memory concerned is more fundamental. In order to clarify the approach that I  will adopt here, it may therefore be helpful to distinguish several ways in which the relation between the metaphysics, the intentionality, the phenomenology, and the epistemology of memory may be conceived, and how each conception of that relation leads to a different approach to addressing (i–​iv). One way in which the relation between the four aspects of memory can be conceived is by taking the metaphysics of memory to be the fundamental aspect among them. Let us call this conception of the relation between the four aspects of memory, the “primacy of metaphysics.” If one endorses this conception, then the approach that one will take regarding how to address (i–​iv) will be to think that, whatever one’s answer to (i) is, it should account for one’s answers to (ii), (iii), and (iv). The challenge for the advocate of the primacy of metaphysics is that their answer to question (i) cannot rely on any pre-​ conceived answers to questions (ii–​iv). In order to determine which mental states qualify as memories, the advocate of the primacy of metaphysics cannot appeal to the intentional properties, the phenomenal properties, or the epistemic properties that memories typically have. Otherwise, their subsequent answers to questions (ii), (iii), and (iv) will be infected with circularity. The advocate of the primacy of metaphysics may try to avoid the threat of circularity by offering an answer to question (i) which does not appeal to the intentional properties, the phenomenal properties, or the epistemic properties which characterize memories. Instead, they may want to restrict themselves to the use of notions such as causation, realization, constitution, and other purely metaphysical notions. The challenge, of course, is to come up with an answer to question (i) which illuminates questions (ii–​iv) within such constraints. Suppose, for example, that the facts in virtue of which a mental state qualifies as a memory are, let us say, purely causal facts. Explaining how those facts would also account for the kind of intentional, phenomenal, and epistemic properties that memories have seems quite a difficult task. Nevertheless, there is an avenue that the advocate of the primacy of metaphysics may want to pursue. In order to explain the contents that memories have, for example, the advocate of the primacy of metaphysics may want to help themselves to causal theories of mental content.14 If a memory represents a fact in virtue of some causal relations which obtain between the

14 For an account of mental representation which relies on causal relations, see (Dretske 1986).

16  The Nature of Memory fact and the memory, then perhaps the causal relations in virtue of which the mental state in question qualifies as a memory include the relations in virtue of which the memory has a certain content. In this way, the advocate of the primacy of metaphysics may attempt to extract an answer to question (ii) from their answer to question (i). What about question (iii) though? One possible way of understanding the relation between the phenomenology and the intentionality of memory is to think that the phenomenal properties of memories depend on their intentional properties. The thought is that memories have the phenomenal properties that they have because they have certain intentional properties.15 On this picture, it makes sense for one to ask which aspects of the contents of episodic memories explain the fact that memories have a phenomenology of a certain kind. Thus, it seems that if one endorses the primacy of metaphysics, one may appeal to the view that the phenomenology of memory depends on its intentionality in order to address question (iii) based on one’s answer to question (ii). As far as question (iv) is concerned, it seems that at least two strategies are available to the advocate of the primacy of metaphysics. One strategy is to offer an answer to question (iv) based on their answer to question (ii) by appealing to the notion of evidence. The idea would be to appeal to the connection, on the one hand, between content and evidence, and, on the other hand, between evidence and epistemic justification. On some conceptions of epistemic justification, being justified in believing something depends on having evidence for one’s belief.16 If this is correct, then, once we are clear on the kind of content that episodic memories have, this should shed some light on the kind of epistemic justification that memories provide us with. For, at that point, we should have an understanding of the kind of evidence for belief that memories give us. Thus, if one subscribes the primacy of metaphysics, one may address the intentionality of memory first and then use the outcomes of one’s investigation in order to address the epistemology of memory. A different strategy for answering question (iv) is to appeal, not to the connection between evidence and epistemic justification, but to the connection between causation and epistemic justification. On some conceptions of epistemic justification, the issue of whether a belief is justified or not 15 This is a particular instance of the broader view that, for any type of mental state, the phenomenal properties of mental states of that type depend on their intentional properties. Representationalists about qualia endorse the broader view. On representationalism about qualia, see (Dretske 1995). 16 See (Feldman and Conee 1985) for a defence of this view.

Problems of Memory  17 depends on how the belief was produced.17 Essentially, the idea is that a belief is justified if it has been caused by an appropriate causal process; a process which, in normal circumstances, would take the fact that constitutes the truth-​maker of the belief as input and would deliver the belief as output. If this is correct then, once we are clear on which causal facts need to obtain for a mental state to qualify as an episodic memory, this should shed some light on the kind of epistemic justification that memories provide us with. For, at that point, we should have an understanding of whether forming beliefs on the basis of our memories constitutes (given the causal conditions that memories must satisfy to qualify as such) an appropriate process for the production of beliefs or not. An alternative way in which the relation between the four aspects of memory can be conceived is by taking the intentionality of memory to be the fundamental aspect among them. Let us call this conception, the “primacy of intentionality.” If one endorses this conception, then the approach that one will take regarding how to address (i–​iv) will be to think that, whatever one’s answer to (ii) is, it should account for one’s answers to (i), (iii), and (iv). The idea with regards to question (i), for example, would be that the contents of memories are essential to them. And, for that reason, the intentionality of memory should be our guide to the metaphysics of memory. Once we have a clear understanding of what kind of content memories have, we will have a firm grasp on what kinds of mental states qualify as memories: Memories are those mental states which have mnemonic contents. This way of tackling question (i), however, runs into a certain difficulty. The difficulty is that it is hard to see how one could determine which contents memories enjoy if one does not have, in the first place, a firm grasp on the issue of what mental states qualify as memories. What about question (iii)? The advocate of the primacy of intentionality can appeal to a way of understanding the relation between the phenomenology and the intentionality of memory that we have considered above. If one subscribes the primacy of intentionality, it seems natural to think that the phenomenal properties of memories depend on their intentional properties. And if such dependence does obtain, then it should be possible to shed some light on what it is like for us to have memories once we are clear on the kinds of contents that memories have. Thus, it seems that if one endorses the primacy of intentionality, one may appeal to the view that the phenomenology

17 See (Goldman 1979) for an example of this view. We will revisit this view in ­chapter 7.

18  The Nature of Memory of memory depends on its intentionality in order to address question (iii) based on one’s answer to question (ii). A natural way of tackling question (iv) if one endorses the primacy of intentionality is by appealing to the above-​mentioned connections between content, evidence, and epistemic justification. The idea, once again, is that, if we have an understanding of the kinds of contents that episodic memories have, then we will be able to understand what kind of evidence for belief episodic memories give us. And this, in turn, should shed some light on the kind of epistemic justification that memories provide us with. Thus, it seems that if one takes the intentionality of memory to be the fundamental aspect of it, one may address question (iv) based on one’s answer to question (ii) as well. It is also possible to take the phenomenology of memory to be the fundamental aspect of it. Let us call this conception of the different aspects of memory, the “primacy of phenomenology.” If one endorses this conception, then the approach that one will take regarding how to address (i–​iv) will be to think that, whatever one’s answer to (iii) is, it should account for one’s answers to (i), (ii), and (iv). The idea with regards to question (i), for example, would be that the phenomenal features of memories are essential to them. And, for that reason, the phenomenology of memory should be our guide to the metaphysics of memory. Once we have a clear understanding of what it is like for us to have a memory, we will have a firm grasp on the issue of what kinds of mental states qualify as memories: Memories are those mental states which have the phenomenal features which are characteristic of memories. This way of tackling question (i), however, runs into a difficulty. The difficulty is that it is unclear how one could determine what those phenomenal features are if one does not have, in the first place, a clear understanding of what mental states qualify as memories. Consider question (ii) now. A  natural way of understanding the relation between the phenomenology and the intentionality of memory, if one endorses the primacy of phenomenology, is to think that the intentional properties of memories depend on their phenomenal properties.18 The thought is that memories have the intentional properties that they have because they have certain phenomenal properties. On this picture, it 18 This is a particular instance of the broader view that, for any type of mental state, the intentional properties of mental states of that type depend on their phenomenal properties. Philosophers within the phenomenal intentionality research program are sympathetic to the broader view. See (Kriegel 2013), for example.

Problems of Memory  19 makes sense for one to ask which aspects of the phenomenology of having episodic memories explain the fact that memories have contents of a particular kind. Thus, it seems that if one subscribes the primacy of phenomenology, one may appeal to the view that the intentionality of memory depends on its phenomenology in order to address question (ii) based on one’s answer to question (iii). How would one tackle (iv) if one endorses the primacy of phenomenology? One possibility is the following. There is a view according to which some of our feelings can play a grounding role for some of our beliefs. Such feelings are labeled “epistemic feelings” in the relevant literature.19 An example of an epistemic feeling, for example, is the feeling associated with the tip-​of-​the-​tongue phenomenon.20 When we experience this phenomenon, we are unable to produce an answer to some question. And yet, we believe that we know the answer, and we believe that we would be able to produce it if given enough time. On the basis of which grounds do we form such beliefs? It is hard to see what constitutes our grounds for those beliefs if not the characteristic feeling that we experience when we are trying to answer the relevant question. If one endorses the primacy of phenomenology, then, one could help oneself to the notion of epistemic feelings in order to address question (iv) based on one’s answer to (iii). The idea would be that, once we are clear on the phenomenology of memory, we may be able to identify, among the phenomenal features which are characteristic of remembering, some epistemic feelings. And, once we are clear on what the relevant epistemic feelings are, we may be able to account for what kinds of beliefs we are entitled to form on the basis of such feelings and, thus, for what kind of epistemic justification memory provides us with. This strategy, however, has obvious limitations. For one thing, even if we grant that, in the tip-​of-​the-​tongue phenomenon, one remembers the answer to the question, it seems that the kind of memory involved would be semantic, and not episodic, memory. More importantly, the types of beliefs that epistemic feelings are able to support seem to be beliefs about one’s own epistemic state. (This is why epistemic feelings are thought to be relevant for metacognition.) These are beliefs such as the belief that one knows, or does not know, a proposition, the belief that one has committed a mistake, or the belief that one has achieved understanding.21



19 On the nature of epistemic feelings, see (Arango-​Muñoz 2014). 20 For details, see (Brown 1991). 21

See (Arango-​Muñoz and Michaelian 2014) for a helpful inventory of epistemic feelings.

20  The Nature of Memory It is unclear, then, how epistemic feelings could make a contribution to the justification of memory beliefs about past objective facts. It seems more promising, then, for the advocate of the primacy of phenomenology to approach question (iv) through a different route; a route that will be quite familiar by now. Recall the connection, on the one hand, between content and evidence and, on the other hand, between evidence and epistemic justification. If being justified in believing something depends on having evidence for one’s belief, and if memories provide us with evidence for belief in virtue of their contents, then it seems that the advocate of the primacy of phenomenology should try to answer question (ii) first. Once this task has been accomplished, the advocate of the primacy of phenomenology can try to use their answer to question (ii) in order to shed some light on question (iv), just like the advocate of the primacy of intentionality tried to do. There is, finally, the possibility of taking the epistemology of memory as the fundamental aspect of it. Let us call this conception of the relation between the four aspects of memory, the “primacy of epistemology.” If one endorses this conception, then the approach that one will take regarding how to address (i–​iv) will be to think that, whatever one’s answer to (iv) is, it should account for one’s answers to (i–​iii). The idea would be to figure out, first, what kind of epistemic justification memory provides us with. And, then, to tackle questions (i–​iii) by pursuing a sort of transcendental enquiry: What must the metaphysics, the intentionality, and the phenomenology of memory be like in order for memory to deliver the kind of epistemic justification that it does? This is also a challenging project. It is challenging because of a certain threat of circularity; a recurring threat in this discussion. While addressing question (iv), the advocate of the primacy of epistemology may not explain the kind of epistemic justification that memories provide us with by appealing to the fact that memories have, for example, contents of a particular kind or by appealing to the fact that they have certain causal histories. Otherwise, their subsequent answers to either question (ii) or question (i)  will be infected with circularity. The project is also challenging because it is difficult to see how the kind of epistemic justification that memories provide us with could tell us much about the phenomenology or the intentionality of memory. Nevertheless, there is an avenue that the advocate of the primacy of epistemology may want to pursue. In order to address question (ii), they could exploit, once again, the connection between content and evidence and the connection

Problems of Memory  21 between evidence and epistemic justification. The idea would be to assume, first of all, that being justified in believing something requires having evidence for one’s belief. Suppose, now, that one has been able to show that memory provides our memory beliefs with a particular kind of epistemic justification. At that point, one may ask what type of evidence memories must give us in order for them to confer that kind of epistemic justification on our memory beliefs. Since having a certain content seems to be the most natural way for memories to give us evidence, this inquiry should eventually shed some light on the content of memories. By contrast, in order to address question (iii), the advocate of the primacy of epistemology could exploit, once more, the dependence of the phenomenal properties of memories on their intentional properties. The idea would be to assume, first of all, that the phenomenal properties of memories depend on their intentional properties. Suppose, now, that one has been able to show that memories have contents of a particular kind. At that point, one may ask how the phenomenal properties of memories are explained by the fact that memories have contents of that kind. What about question (i) though? Illuminating the metaphysics of memory by an appeal to the epistemic properties that memories have seems especially challenging. There is, however, a conception of the nature of memory which seems quite agreeable with the primacy of epistemology. This is the so-​called epistemic conception of memory.22 The epistemic conception of memory is formulated in terms of knowledge, not epistemic justification. According to it, for every proposition p, one remembers that p if and only if one has a certain knowledge that p: One knows that p in the present because, in the past, one knew that p. In order to address the question of which mental states qualify as memories, the advocate of the primacy of epistemology may put forward an analogous proposal which makes use of the notion of epistemic justification: For every mental state M, M is a memory if and only if there is some proposition p such that, if one has M, then one is justified in believing that p because, in the past, one was justified in believing that p. If the advocate of the primacy of epistemology can make a compelling case for this version of the epistemic conception of memory, then their answer to question (iv) may illuminate the correct answer to question (i). For, once we are clear on the kind of epistemic justification that our memories provide us with, this should allow us to determine which sorts of mental states qualify as

22 For a discussion of the epistemic conception of memory, see (Bernecker 2010).

22  The Nature of Memory memories. Memories are, on this view, those mental states which allow us to form beliefs with the relevant kind of epistemic justification because, in the past, we were able to form those very same beliefs with precisely that kind of epistemic justification. This catalogue of available strategies for addressing the metaphysics, the intentionality, the phenomenology, and the epistemology of memory is, of course, not exhaustive. One could take, for example, two of those four aspects of memory to be more basic relative to the remaining two, which would result in a hybrid approach to questions (i–​iv). The hybrid approach would then inherit some of the virtues and challenges of some of the approaches just considered. To be sure, all four of those approaches face some challenges. For instance, both the primacy of phenomenology approach and the primacy of intentionality approach have trouble with the metaphysics of memory. The former involves, furthermore, the assumption of the dependence of the intentional properties of memories on their phenomenal properties whereas the latter involves, conversely, the assumption of the dependence of the phenomenal properties of memories on their intentional properties. Both the primacy of metaphysics approach and the primacy of epistemology approach have trouble with the intentionality and the phenomenology of memory. The former may involve, depending on how we pursue it, a commitment to causal theories of mental content whereas the latter could involve, on some versions of it, a commitment to the epistemic conception of memory. It is perhaps not surprising that, no matter which approach we eventually decide to follow in order to address questions (i–​iv), there will be a theoretical cost involved. After all, the project of accounting for the metaphysics, the intentionality, the phenomenology, and the epistemology of memory is, despite the limitations of scope stipulated in sections 1.2 and 1.3, an ambitious project. I propose to adopt a hybrid approach which will combine some aspects of the primacy of metaphysics approach with some aspects of the primacy of intentionality approach. Is this the best possible approach? I am inclined to think that we should evaluate a philosophical approach to memory based on the balance of, on the one hand, the explanatory benefits delivered by the approach and, on the other hand, the theoretical cost that we incur by adopting the approach. The approach that I propose to adopt here will be specified in section 1.5. In section 1.6, I will disclose the assumptions which make up the theoretical cost of the proposed approach. The case for the explanatory benefits delivered by it will be made in the coming chapters.

Problems of Memory  23

1.5  Approach In what follows, I will adopt a hybrid approach which is, in some respects, similar to the primacy of intentionality approach. I will take the intentionality of memory to be more basic than the phenomenology and the epistemology of memory. Thus, I will try to account, first, for the intentionality of memory, and then, I will try to use the outcomes of that inquiry to explain the phenomenology of memory and the epistemology of memory. The hybrid approach that I will adopt here is, in other respects, similar to the primacy of metaphysics approach. For I will take the metaphysics of memory to be more basic than the phenomenology of memory and the epistemology of memory. For that reason, I will try to account, first, for the metaphysics of memory, and then, I will use the outcomes of that inquiry to illuminate some aspects of the epistemology of memory. The difference between the strategy to be followed here and both the primacy of metaphysics approach and the primacy of intentionality approach is the following. I will not take the intentionality of memory to be more basic than the metaphysics of memory, and I will not take the metaphysics of memory to be more basic than the intentionality of memory. Accordingly, I will not try to account for either of those two aspects of memory by appealing to the other one. In that sense, the whole investigation of memory to be constructed here is intended to rest eventually on two independent pillars; an account of the metaphysics of memory and an account of its intentionality. Let us look at the architecture of the project more closely now. The first part of the book is concerned with the nature of memory; those aspects of memory which will be assumed to be more basic. For that reason, it deals with the metaphysics of memory and the intentionality of memory. Specifically, in ­chapter 2, I will offer an account of what kinds of mental states qualify as memories. To avoid the threat of circularity which threatens the primacy of metaphysics approach, the account to be proposed will not appeal to the idea that memories have a certain content, it will not appeal to the idea that memories have a certain phenomenology (either involving time or involving ownership), and it will not appeal to the idea that memories provide a special kind of knowledge (either in terms of grounding beliefs which are protected from misidentification errors or in terms of generating epistemic justification). The account will rely, instead, on some ideas borrowed from the framework of functionalism. Accordingly, I will refer to it as the “functionalist theory” of memory. In ­chapter 3, I will offer an account of the

24  The Nature of Memory kinds of contents that memories have. The account of mnemonic content to be proposed will not appeal to the idea that memories have a particular phenomenology (either involving time or involving ownership), and it will not appeal to the idea that memories provide us with a special kind of knowledge (either in terms of grounding beliefs which are protected from misidentification errors or in terms of generating epistemic justification). The account will rely, instead, on some intuitions regarding the truth-​conditions of memories, and it will put forward the idea that memories are, in a certain sense, self-​ referential. Accordingly, I will refer to this account as the “reflexive view” of mnemonic content. The functionalist theory of memory and the reflexive view of mnemonic content are meant to be independent from each other. Suppose, for example, that the functionalist theory of memory turns out to be false, and the kinds of mental states which, in c­ hapter 2, will be claimed to qualify as memories are not, in actual fact, memories. Then, it may still be true that the mental states which do qualify as memories have self-​referential contents. In other words, if the functionalist theory of memory falls, it does not need to take the reflexive view of mnemonic content with it. Conversely, if the reflexive view of mnemonic content falls, it does not need to take the functionalist theory of memory with it either. If memories have contents which, contrary to what will be argued in ­chapter 3, are not self-​referential, then it may still be true that those mental states which qualify as memories are the states which will be characterized along functionalist lines in c­ hapter 2. The second part of the book deals with the phenomenology of memory. Specifically, it deals with the temporal phenomenology of memory and the sense of ownership involved in having memories. The first aspect of the phenomenology of memory will be discussed in c­ hapter 4. First, I will separate two aspects of the temporal phenomenology of memory; the feeling that the remembered fact obtained in the past and the feeling that the subject perceived it in the past. Then, I will relate the latter feeling to the popular notion that memory involves a sort of mental time travel to the moment at which the subject originally perceived the remembered fact. And, finally, I will try to explain both aspects of the temporal phenomenology of memory by appealing to the kind of content that memories have. The broad idea will be that the temporal phenomenology of memory is the way in which we experience some of the things that we represent in memory. The second aspect of the phenomenology of memory will be discussed in ­chapter 5. First, I will make a case for the view that having memories involves the feeling that those

Problems of Memory  25 memories are the subject’s own by appealing to the condition of disowned memory; a condition which can best be explained by the absence of such a feeling.23 Then, I will offer a proposal about the nature of the feeling that our memories are our own by appealing to the kinds of contents that memories have. The broad idea will be, once again, that the phenomenology of ownership in memory is the way in which we experience some of the things that we represent in memory. Accordingly, I  will propose that the lack of the feeling of ownership involved in the condition of disowned memory is the way in which the subject experiences having memories whose contents have been, in a certain sense, disrupted. Other mental disorders in which subjects disown some of their mental states will also be discussed along similar lines. The third part of the book deals with the epistemology of memory. Specifically, it deals with the issue of whether memory beliefs are immune to error through misidentification and with the issue of whether memories generate epistemic justification or merely preserve it over time. In ­chapter 6, I will discuss immunity to error through misidentification with regards to memory. First, I will consider some potential counter-​examples to the view that memory beliefs are immune to error through misidentification, and I will argue that, contrary to first appearances, they do not show that the view is false. In some cases, my argument will rely on the claim that the relevant counter-​examples do not involve, strictly speaking, memories. Thus, the line of reasoning in c­ hapter 6 will partly rely on the functionalist theory of memory offered in ­chapter 2. It will also rely on the reflexive view of mnemonic content defended in ­chapter 3, since it will be on the basis of that view that I will argue, more positively, that memory beliefs are immune to error through misidentification. In the process of discussing the potential counter-​ examples to the view that memory beliefs are immune to error through misidentification, some ideas about the intentionality of perception, and about the relation between the intentionality of perception and that of memory, will emerge. Having made the case that memory beliefs are immune to error through misidentification, I will then try to use those ideas in order to explain why memory beliefs have this peculiar epistemic property. In c­ hapter 7, I will tackle the debate on whether memories generate, or merely preserve, epistemic justification. I will defend a version of “generativism”; the view that memories generate epistemic justification. I will first consider some versions of generativism and some ways in which one could potentially make a case

23 On disowned memory, see (Klein and Nichols 2012).

26  The Nature of Memory for them. As we will see, some of the relevant defenses of generativism are built within a picture of memory according to which memories may not generate content relative to the perceptual experiences in which they originate. I will refer to this picture as the “recorder model” of memory. Other defenses of generativism are built within a more liberal picture of memory according to which memories do have the power to generate content. I will refer to this picture as the “reconstructive” conception of memory. I will not subscribe any of the defenses of generativism considered, but I will suggest that the idea of content generation, to be found within the reconstructive conception of memory, puts us on the right path toward a successful defense of generativism. I will make a case for the view that memory involves a certain kind of content generation by appealing, not to the reconstructive conception of memory, but to the reflexive view of mnemonic content defended in ­chapter 3. And, on the basis of the relevant kind of content generation, I will argue that memories do generate epistemic justification, and not merely preserve it over time. This outline of the investigation that we will pursue reveals an interesting aspect of it. The interesting aspect is that the heavy lifting in the project will be done by the proposal about mnemonic content to be offered in ­chapter 3. The account of the metaphysics of memory to be put forward in ­chapter 2 does play a role in the discussion of the epistemology of memory, but it is a fairly minor role. Mostly, the functionalist theory of memory serves as the background against which questions about the intentionality, the phenomenology and the epistemology of memory will be discussed. It is necessary, since it would be hard to determine, for example, what the temporal phenomenology of memories involves without knowing exactly what we are supposed to count as memories. And it is interesting, since the received view about what counts as a memory is now hotly contested within philosophy. But it does not play a substantial explanatory role in the overall investigation of memory. Thus, even though the investigation is, to continue with the metaphor introduced earlier, built to rest on two main pillars, the weight of the investigation is not equally distributed between the two pillars. It is more like an edifice tilted toward one of them; the account of the intentionality of memory. To that extent, this is a broadly representationalist project. And, for that reason, it involves some crucial assumptions; assumptions which should be clearly disclosed. Let us remind ourselves, therefore, of the plan which has been put in place for the coming discussion and of the assumptions which have been made in the process of designing that plan.

Problems of Memory  27

1.6  Summary In this chapter, we have seen that the investigation of episodic memory that we are about to pursue will cover the following four questions about the metaphysics, the intentionality, the phenomenology, and the epistemology of memory: i. Metaphysics What are the conditions under which an experience qualifies as an episodic memory? ii. Intentionality What are the contents of our episodic memories? iii. Phenomenology What is it like for us to have episodic memories? iv. Epistemology What kind of epistemic justification do our episodic memories afford? We have seen that ­chapter 2 will be concerned with question (i). Accordingly, the discussion in c­ hapter  2 will deliver an account of the metaphysics of memory. Chapter 3 will be concerned with question (ii), and for that reason, it will deliver an account of its intentionality. Chapters 4 and 5 will deal with different elements of an answer to question (iii). Thus, they will offer accounts of different aspects of the phenomenology of memory. Likewise, ­chapters 6 and 7 will deal with different elements of an answer to question (iv), and hence, they will offer accounts of different aspects of the epistemology of memory. The way in which these outcomes are meant to hang together is that if all goes well, then the proposed answers to question (iii) and question (iv) will rest on the proposed answers to both questions (i) and (ii), but especially on the answer to question (ii). Thus, it seems that if the outcomes of our discussion of memory in the coming chapters are correct, the benefits of this investigation will be many. For it will provide an overall picture of several philosophically interesting aspects of memory and of the relations between them. In sum, this investigation will bring them together into a unifying account of memory. The ultimate benefit of the investigation, therefore, is an answer to a question which encompasses (i–​iv) above; the question of what our capacity for remembering episodically amounts to. What about the costs of this inquiry though? In the process of organizing the discussion around questions (i–​iv), I have made a number of assumptions.

28  The Nature of Memory And, while adopting a particular approach for tackling questions (i–​iv), I have committed myself to a number of further assumptions. It will make for clarity, then, if we can highlight those assumptions right from the start, and we can bear them in mind throughout the coming discussion. We may distinguish five assumptions which have been made thus far. To begin with, I  have made two assumptions concerning episodic remembering. The first one is that episodic memories should be identified with experiences and not with beliefs. Experiences have been construed as mental states with intentional and qualitative features. In particular, I have assumed that those experiences have content, in the sense that they can be correct or incorrect. The thought is that one may remember something in virtue of having one of those experiences, and then, it is a separate matter whether one is remembering it correctly or one is remembering it incorrectly. In either of the two possible scenarios, one would episodically remember the same thing by having the same experience.24 The second assumption regarding episodic remembering concerns the nature of these experiences. Consider those cases in which one episodically remembers a fact, and the fact in question involves objects perceivable through the sense modality of vision. These are cases such as that in which you remember that Mary was at the party last month. In those cases, I have assumed, not only that one episodically remembers a fact by having an experience but also that this amounts to holding a mental image in one’s mind. I have assumed, that is, that episodic memory involves mental imagery. No assumption has been made, however, with regards to the nature of those experiences wherein one episodically remembers facts which involve objects perceivable through sense modalities other than vision. Likewise, no assumption has been made with regards to the nature of those experiences wherein one episodically remembers a past experience (and, in particular, a past perceptual experience) of one’s own. Furthermore, I  have made two assumptions regarding the intentional features of episodic memories. The first one is that the content of a memory, or what the memory represents, does not need to correspond to the way in which one would normally express the memory. More precisely, the content of a memory does not need to match the content of the subordinate clause in

24 I  am, in other words, assuming that there is such a thing as remembering falsely, or misremembering. Hence, I am assuming that disjunctivism for memory is false. On disjunctivism, see (Soteriou 2016).

Problems of Memory  29 a claim of the form “I remember that p” that one would ordinarily make in order to express the memory. One of the reasons for this is that there may be, for example, more detail in the content of the memory than one is expressing in one’s claim. This is not to say that we should ignore the way in which one normally expresses one’s memories while trying to determine what the contents of those memories are. As we will see, an important consideration which will serve to evaluate theories of mnemonic content will be the constraint that, whatever the correct theory of mnemonic content turns out to be, it should accommodate the way in which one expresses one’s memories in normal circumstances. The assumption, at this point, is simply that the claim whereby one expresses an episodic memory is not the best guide to the content of that memory and that we should be guided, instead, by our intuitions regarding the truth-​conditions of that memory. The second assumption regarding the intentional features of episodic memories is that the phenomenal features of episodic memories depend on their intentional features. This assumption deserves some further clarification. For there is a sense in which it is a strong assumption, and there is a sense in which the assumption is rather weak. Consider the purposes for which we made the assumption of the dependence of the phenomenology of memory on its intentionality. This assumption was made in order to justify the expectation that we may be able to extract an answer to question (iii) from an answer to question (ii). One might wonder, therefore, whether a commitment to the view that the phenomenal features of episodic memories depend on their intentional features is enough for those purposes. Or, conversely, one might wonder whether it is more than we need. For the purposes of tackling question (iii) based on our answer to question (ii), all we need to assume is that the content of a memory will give us some information about the way in which the memory feels to the subject. To be sure, this will be the case if the way in which the memory feels to the subject depends on its content. If the content of a memory is responsible for the way in which the memory feels to the subject, then pointing to the content of the memory in order to explain what it is like for the subject to have that memory does seem informative. But there does not need to be a relation of dependence between the way in which a memory feels to the subject and the content of the memory in order for the latter aspect of the memory to provide us with information about the former one. It seems that other, weaker, relations between the two aspects of the memory will do. Imagine,

30  The Nature of Memory for example, that we discover that there is a one-​to-​one correlation between contents and sets of phenomenal features of memories. However, it turns out that the reason why such a correlation exists is that, for each pair of contents and sets of phenomenal features of a memory, there is some other property of the relevant memory which acts as a single common cause for both the content and the set of phenomenal features in the pair. In that case, finding out that the memory has a particular content does give us information about the way in which the memory feels to the subject. It tells us that the memory in question will have the phenomenal features that correspond to the relevant content. But, in this scenario, the way in which the memory feels to the subject does not depend on the content of the memory. Instead, it depends on whatever property of the memory acts as the common cause for both the way in which the memory feels to the subject and the content of the memory. Thus, there is a sense in which the assumption that the phenomenal features of memories depend on their intentional features is a strong assumption: It is more than we need for our purposes here. However, there is also a sense in which the assumption that the phenomenal features of memories depend on their intentional features is a weak assumption. For the dependence of the phenomenal features of memories on their intentional features leaves room for a number of possible relations between the two sets of features. The phenomenal features of memories may depend on their intentional features because properties of the former kind are caused by properties of the latter kind, for example. Alternatively, the phenomenal features of memories may depend on their intentional features because properties of the former kind are “realized by” properties of the latter kind. One way of understanding the realization relation is the following: For any properties P and Q, and for any subject S, if S has P and Q, then Q is realized by P in S if there is a condition C such that Q is the prop­ erty of having some property or other which satisfies condition C in subject S, and P satisfies condition C in subject S.25 If the phenomenal features of memories are realized by their intentional features in this sense, then the phenomenal features of memories depend on their intentional features, but they are not caused by them. Yet another possibility is that the phenomenal features of memories depend on their intentional features because they

25 See, for example, (Kim 2003). In c­ hapter  2, we will revisit this relation, not with regards to the intentional and phenomenal features of memories but with regards to the nature of memories themselves.

Problems of Memory  31 stand in the determinable-​determinate relation to each other.26 The thought would be, in this case, that the intentional features of memories are ways of having phenomenal features for them, in the same way in which, let us say, being scarlet is a way of being red. If the intentional features of memories are determinates of their phenomenal features, then the phenomenal features of memories depend on their intentional features, but they are not caused by them. Finally, notice that the dependence of the phenomenal features of memories on their intentional features is consistent with the possibility that properties of the former kind are identical with properties of the latter kind; a scenario in which the phenomenal features and intentional features of memories would be dependent on each other. Thus, there is a sense in which the assumption that the phenomenal features of memories depend on their intentional features is a weak assumption: It is consistent with a number of different pictures about the relation between properties of the two kinds. Finally, an assumption was made regarding the epistemology of memory. The very formulation of question (iv) presupposes that there is such a thing as the epistemic justification that our memories afford. And, for that reason, it presupposes that memory is generally reliable. This assumption is, so to speak, the counterbalance to the first assumption highlighted earlier. By assuming that there is such a thing as misremembering, we allowed for memory to be fallible. Most of the time, however, memory does not seem to mislead us. By and large, memory serves us well. For example, we do find food in those places in which we remember to have encountered it in the past. We succeed in avoiding danger by avoiding those places where we remember to have found it in the past. People who we seem to recognize are people who, in fact, we did meet in the past, and so on. It is hard to see how we could survive, let alone learn anything, if memory was generally unreliable. These five assumptions constitute the price to pay for the outcomes of this investigation of memory which were anticipated earlier. Admittedly, these assumptions, taken together, seem to add up to a pretty steep price. And yet, I will not attempt to defend the assumptions here. Instead, I propose to stipulate them for the sake of the argument and explore how far they can take us in the project of accounting for the metaphysics, the intentionality, the phenomenology, and the epistemology of memory. In the coming chapters, I intend to show that they can take us considerably far in that direction.



26 On the determinable-​determinate relation, see (Funkhouser 2006).

2 The Metaphysics of Memory 2.1  Introduction My aim in this chapter is to determine what is to remember something, as opposed to imagining it, perceiving it, or introspecting it.1 What does it take for a mental state to qualify as remembering? The main issue to be addressed is therefore a metaphysical one. The goal is to specify the conditions that those mental states which qualify as episodes of remembering satisfy and those mental states which do not qualify as episodes of remembering fail to satisfy. If one’s project is to study the intentional, phenomenological, and epistemological properties of memory, then it will be important to have a firm grasp on which mental states qualify as episodes of remembering, and which do not. Otherwise, it will be difficult to define the boundaries of one’s investigation. And this difficulty, in turn, will make it quite challenging for one to isolate the interesting properties of memory. This chapter is intended to make a contribution to the project by specifying exactly what counts as remembering. I will proceed as follows. In sections 2.2 and 2.3, I will discuss the two main existing views about the conditions that a mental state must satisfy to constitute an episode of remembering. The first of these views is backward-​looking. It puts forward conditions that strictly concern the etiology of the mental state concerned. I will argue that the conditions offered by the backward-​looking view are both too strong and too weak: They exclude mental states that, intuitively, count as episodes of remembering while including mental states that, intuitively, do not qualify as episodes of remembering. The second view is forward-​looking. It puts forward conditions that only concern the use that the subject makes of the relevant mental state while forming beliefs about their past. I will argue that the conditions proposed by the forward-​looking view are both too weak and too strong as well. However, the discussion of the two views will allow us to extract some helpful lessons about the constraints that any proposal 1 An earlier version of this chapter can be found in (Fernández 2018).

The Metaphysics of Memory  33 about the nature of memory should respect. An alternative view aimed at incorporating those lessons will be offered in section 2.4; a view which draws on the literature on functionalism. In section 2.5, I  will argue that this view can, on the one hand, accommodate as episodes of remembering those mental states that indicate that the backward-​looking view and the forward-​looking view are too strict and, one the other hand, rule out those mental states that suggest that the two alternative views are too permissive. Accordingly, I will conclude that a functionalist conception of remembering is the most satisfactory view about the metaphysics of memory.

2.2  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 causal 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 proposition p, S remembers that p just in case S is representing that p, S represented that p in the past, and the fact that S represented that p in the past has caused S to represent that p in the present.2 To clarify this formulation of CTM, it is worth elaborating on the notions of representation and causation employed. As far as CTM is concerned, the notion of representation should be understood as covering a number of possible experiences wherein a subject is aware of a fact. A subject may perceptually experience that it is a sunny day, for example. Or they may be imaginatively visualizing that they are at the beach. Alternatively, the subject may be introspectively experiencing that they are afraid of being sunburnt. Or they may be agentively experiencing that they are applying sunscreen on their skin. In all of these cases, the subject counts, on this reading of “representation,” as representing each of the relevant facts. The reason why the advocate of CTM is employing such a broad notion of representation is that we all seem to be able to remember facts of very different types; facts that we witnessed in the past, facts of which 2 The classical version of CTM is due to Charles Martin and Max Deutscher in their (1966). Martin and Deutscher’s formulation of CTM applies to memory for physical objects, to memory for events, and to memory for mental states such as perceptual experiences (Martin and Deutscher 1966, 166). For the sake of our discussion in this chapter, I will assume that CTM is also a theory of propositional memory.

34  The Nature of Memory we were aware subjectively and facts that concerned our actions in the past. The experiences wherein we are, in the present, aware of those past facts only qualify as episodes of remembering, the advocate of CTM claims, if they have been caused by our past representations of those facts, whether those representations were perceptual experiences, episodes of introspection or agentive experiences. Despite being very broad, the notion of representation used to formulate CTM carries a substantive commitment with it. Representation, as the CTM advocate views it, is closed under logical entailment: For any subject S, and any propositions p and q, if S represents that p, and p entails that q, then S represents that q as well. The reason why the notion of representation used in CTM is meant to have this feature is that, otherwise, CTM will not sit very easily with a certain observation about memory. Except for rare cases of so-​ called eidetic memory, we do not remember all the details in those facts that we have represented in the past.3 This raises a difficulty for the classical formulation of CTM. To illustrate the difficulty, consider the following example. Imagine that, as a child, I once went to the beach and saw that there was a white ball with some red writing on it over there. Suppose that I now have a mental image wherein I visualize the ball as being white, but not as having red writing on it. Let us stipulate that I  have that mental image because, in the past, I saw the ball at the beach. And, in virtue of having the mental image, I am inclined to believe that there was a white ball at the beach when I went there as a child. Intuitively enough, I am remembering that there was a white ball at the beach. And yet, CTM seems to preclude us from accepting my current mental state as an episode of remembering, since the fact that I represented in the past (namely, that there was a white ball with red writing on it at the beach) seems to be different from the fact that I am representing in the present (namely, that there was a white ball at the beach). In response to this difficulty, the advocate of CTM will argue that, contrary to what it may have seemed at first glance, in the past I did represent the very fact that I am now representing when I visualize a white ball at the beach. For I represented that there was a white ball with red writing on it at the beach 3 In what follows, I will be speaking of mental images as having “more,” or “fewer,” details than the experiences in which those mental images originate. I will be using those expressions in a non-​ exclusive sense: For any mental states M and P, M contains more details than P just in case there is a detail x such that M contains x, and P does not contain x. And, conversely, M contains fewer details than P just in case there is a detail y such that P contains y, and M does not contain y. In this sense of the two expressions, it is possible for a mental state M to contain both more, and fewer, details than some other mental state P.

The Metaphysics of Memory  35 when I saw it there as a child. And the proposition that there is a white ball at the beach is entailed by the proposition that there is a white ball with red writing on it at the beach. Thus, it seems that CTM can accommodate the intuition that we may remember a fact without representing all the details of that fact which we represented in the past, as long as representation is assumed to be closed under logical entailment. Notice, however, that this feature of the notion of representation delivers some counter-​intuitive results. For the proposition that either it is Tuesday or there is a white ball with red writing on it at the beach is also entailed by the proposition that there is a white ball with red writing on it at the beach. And yet, it seems odd to claim that, by perceptually experiencing that there is a white ball with red writing on it at the beach, I am representing that either it is Tuesday or there is a white ball with red writing on it at the beach. Nonetheless, for the purposes of our discussion of CTM, I will assume that the CTM advocate can make a case for the view that representation is closed under logical entailment. As far as I can see, none of the objections to be raised against CTM later in this chapter hinges on this point.4 The classical formulation of CTM is meant to be neutral on the precise nature of the causal relation. The thought is that no matter what the correct view about the nature of causation turns out to be, that relation must hold between a subject’s past representation of a fact and a current mental state of the subject in order for us to count the subject as remembering the fact in virtue of having the relevant mental state. The motivation for this causal condition concerns, among other reasons, the need to prevent episodes of relearning from qualifying as cases of remembering.5 Suppose I am considering going to the beach. I look at the garden through my window, and I see that it is a sunny day. I decide to do some work first and, in order to concentrate, I close the curtains in the room. I turn on my computer and start working. At some point, while I am lost in thought, I forget that it is a sunny day, but the weather application in my computer informs me that it is a sunny day. As a result, I visualize my garden during a sunny day. In this case, I represented that it is a 4 The commitment to such a notion of representation is, in any case, inessential to CTM. The CTM advocate may choose to start up with a more elaborate formulation of CTM; a formulation which explicitly mentions relations of entailment in the conditions for remembering. The CTM advocate may propose, right from the beginning, that for any subject S and proposition q, S remembers that q just in case S is representing that q, and there is a proposition p such that: q is entailed by p, S represented that p in the past, and the fact that S represented that p in the past has caused S to represent that q in the present. A formulation of this sort for CTM has the advantage of not presupposing that representation is closed under logical entailment. See (Bernecker 2010) for this type of formulation of CTM. 5 On the challenge that cases of relearning pose for CTM, see (Robins 2017).

36  The Nature of Memory sunny day, and I now represent that it is a sunny day but, intuitively enough, I am not remembering that it is a sunny day. The CTM requirement that, for any proposition p, my current representation that p must causally originate in my past representation that p in order for me to qualify as remembering that p is intended to rule out cases such as this one as cases of remembering. Now, even though the advocate of CTM is neutral on the precise nature of the causal relation involved in remembering, they are committed to the view that the causal link between the past representation and the episode of remembering must be “non-​deviant.” 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 fact that a basketball is travelling toward a hoop, and my mental image is at the end of a causal chain which originates in a recent perception of the fact that a basketball is travelling toward a hoop. But, intuitively enough, my mental state does not count as remembering. The reason for this intuition seems to be that, in this case, the causal link between the basketball and my mental image is, in some sense yet to be specified, indirect or deviant. Accordingly, the advocate of CTM requires the causal link between episodes of remembering and the past representations in which they originate to be non-​deviant. Such a move places the burden of specifying what counts as a deviant instance of the causal relation, and what does not, on the CTM advocate’s shoulders.6 However, for the purposes of our discussion of CTM, I will assume that the advocate of CTM can draw the distinction between deviant and non-​deviant 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 hinges on this point either.

6 This is a pressing challenge for the CTM advocate because the causal condition in CTM cannot do the work of excluding cases of relearning as cases of memory unless the CTM advocate can specify what a non-​deviant causal relation is exactly. After all, cases of relearning can be tweaked in order to comply with the causal condition in CTM. Consider again the example, discussed earlier, in which I relearn that it is a sunny day. Let us stipulate, now, that the weather application in my computer is only informing me that it is a sunny day because I forgot that it is a sunny day: My forgetting that it is a sunny day has caused me to look at the weather application in my computer, which has caused me to relearn that it is a sunny day, and to visualize my garden during a sunny day as a result of it. Intuitively enough, this is not a case in which I remember that it is a sunny day. But the CTM advocate will not be able to exclude it unless they can specify what a non-​deviant causal relation is. For there is, in this scenario, a (highly complicated) causal relation which does connect my past perceptual experience of my garden during a sunny day with my current visualization of it.

The Metaphysics of Memory  37 What considerations can be offered in support of CTM? One way in which episodic memory seems to be analogous to semantic memory is that, just like semantic memory preserves the contents of our beliefs over time, episodic memory seems to register and store the contents of those (typically, perceptual) experiences that we had in the past by producing mental images which are aimed at preserving the contents of those experiences.7 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.8 A virtue of CTM is that it accounts for this feature of memory straightforwardly. If CTM is correct, then it is no wonder that episodes of remembering make available to us information that was initially provided by some of our own past experiences. For episodes of remembering must inherit their contents from the experiences in which they originate. After all, according to CTM, a mental state would not count as remembering, in the first place, if it did not represent something that the experience in which it causally originates already represented in the past. CTM enjoys, therefore, the important virtue of capturing the preservative aspect of memory. Unfortunately, CTM also faces two important difficulties. Firstly, CTM seems to be too strict. For, even though it allows an episode of remembering to include less detail than the experience in which it originates, it does not allow it to include more.9 And yet, remembering a past fact may involve 7 Versions of this picture of memory can be found, for instance, in (Aristotle 1972, 28–​32), (Locke 1975, 149–​153), (Hume 2000, 12), (Reid 1969, 326), (Broad 1937, 239–​41), (Malcolm 1963, 208), and (Shoemaker 1984, 19). A comment on some of the terminology to be used here: In this chapter, I will speak of the “preservative” aspect of episodic memory to refer to the idea that episodic memory stores the contents of perceptual experiences that we had in the past. And in ­chapter 7, I will discuss a picture of memory according to which this preservative aspect of memory is an essential feature of it; a picture that I will label “the recorder model” of memory. It is important, however, to distinguish the view that episodic memory stores the contents of perceptual experiences that we had in the past from the view that episodic memory maintains epistemic justification acquired through other sources in the past. In c­ hapter 7, I will discuss the latter view under the label “preservativism.” Hopefully, talk of the “preservative” aspect of memory in this chapter will not cause confusion between the two views. 8 As noted in c­ hapter 1, this ability is 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 these reasons, a crucial feature of it. 9 Martin and Deutscher claim that a memory trace should involve “at least as many features as there are details which a given person can relate about something he has experienced” (1966, 190). Is it accurate to say, then, that CTM does not allow a memory to include more detail than the experience in which it originates? We need to distinguish two views. One is the view (illustrated by the quote) that the mental state that one expresses when one claims to remember something must have as many details, or features, as there are details one can produce about what one claims to remember. Another is the view that the mental state that one occupies when one actually remembers something

38  The Nature of Memory visualizing details about the fact that one’s past experience of it did not originally 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 a mental image wherein I visualize the rabbit, almost exactly as the rabbit appeared to me in the past, except that I now visualize the rabbit as being black. Thus, I am having a mental image wherein I visualize the fact that my father shot a black rabbit. And, accordingly, I am inclined to believe that my father shot a black rabbit. According to CTM, my current experience is not an episode of remembering, since it includes details that were never present in my original experience of the rabbit. This seems highly counter-​intuitive. 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.10 It seems more natural to say that this is a case in which I misremember some fact about my country walks with my father. Let us call cases of this type, “embellishment” cases. Secondly, CTM seems to be too permissive. Imagine that I  am capable of visualizing facts that I experienced in the past but, due to a substantial cognitive deficit of mine, my mental images of those facts do not convey to me the sense that the relevant facts obtained in the past, and they do not convey to me the sense that I experienced them in the past.11 Suppose, furthermore, that I am a painter. I decide to paint a bird on the roof of a house and, at the end of my work, I have painted that scene on my canvas. However, I am not inclined to think that these are a house and a bird which I have ever encountered in the past. I do not think that these are, or were, a real house cannot have more details than those in the experience in which the mental state originates. As far as I can see, CTM advocates endorse both views. 10 The CTM advocate may reply that, nevertheless, I am remembering, in this case, a rabbit being shot. I am not disputing this. It is the further claim that I am not remembering a black rabbit being shot, to which the CTM advocate is committed, that I find counter-​intuitive. The view that episodes of remembering 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 and DeSoto 2015). 11 It is hard to say whether there are actual cases of this kind. The case of patient R.B., discussed by Shaun Nichols and Stanley Klein, might be a candidate (Klein and Nichols 2012). Patient R.B. suffers a remarkable cognitive impairment. He 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 here. However, it is debatable whether the hypothesis is correct. See c­ hapter 5 for a defense of it and (Klein and Nichols 2012) for an alternative hypothesis.

The Metaphysics of Memory  39 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 on the roof of the house is a bird that, in fact, I saw during my visit. And the experience wherein I visualized the scene that I have painted on my canvas causally originates in my past perceptual experience of that bird on the roof of that house.12 According to CTM, in this situation, the experience wherein I visualize the house that I have painted on my canvas does qualify as an episode of remembering. This seems highly counter-​intuitive. After all, the experience that I am having when I engage in the project of painting a bird on the roof of a house does 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 does not make any difference as to which propositions I am inclined to believe regarding my past and which propositions I am inclined to disbelieve. In other words, the experience makes no impact on my beliefs about my past. It is, in that sense, “epistemically irrelevant” to me.13 It seems more natural to describe the case by saying that I am imagining a bird on the roof of a house when I engage in the project of painting such a scene. Let us call cases of this type, cases of “epistemic irrelevance.” In response to the challenge posed by cases of epistemic irrelevance, the CTM advocate might argue that, in the painter case, CTM can account for the remarkable similarity between the scene experienced by me in the past and the scene painted by me in the present: Suppose that having my mental image of the scene that I paint on my canvas amounts to remembering the scene. Then, according to CTM, my mental image of the painted scene must have been caused by my past perceptual experience of the scene. It is no wonder, then, that there is such a remarkable similarity between the scene experienced by me in the past and the scene painted by me in the present. After all, had the scene experienced by me in the past been different in any way, my current mental image would have been different in a corresponding way, and I would have painted a different scene. However, notice that, in this account of the similarity between the scene painted by me in the present and the scene experienced by me in the past, all the explanatory work is done by the notion of causation and not that of memory. For that reason, an analogous explanation of the similarity between

12 This is a variation of a case discussed in (Martin and Deutscher 1966). 13 I am borrowing the term “epistemic irrelevance” from Dorothea Debus, who raises essentially this objection against CTM in (Debus 2010).

40  The Nature of Memory the scene painted by me in the present and the scene experienced by me in the past will be available to us if we construe my current mental state, when I paint the bird on the roof of the house, as an episode of imagination. We just need to concede to the CTM advocate that my mental image of the painted scene must indeed have been caused by my past perceptual experience of the scene. Suppose that the reason why I am able to imagine the bird on the roof of the house, in all the specific detail depicted on my canvas, is that I perceptually experienced that bird on the roof of that house in the past. Then, it will still be true that, had the scene experienced by me in the past been any different, my current mental image would have been different as well, and I would have painted a different scene. It is this counterfactual dependence relation which explains the similarity between the scene experienced by me in the past and the scene painted by me in the present. And such a relation can be in place in imagination as well as in memory. Thus, construing my current mental state as an episode of remembering does not seem to make an explanation of the similarity between the scene painted by me in the present and the scene experienced by me in the past any easier than construing it as an episode of imagination does. It seems, therefore, that CTM captures an important property of memory, namely, its preservative 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 facts, but it can also involve reconstructing that information. Furthermore, remembering, as cases of epistemic irrelevance illustrate, needs to present facts to us as being part of our past. It needs to make a difference as to which propositions we are inclined to believe, and which we are inclined to disbelieve, regarding our past. The outcome of our discussion in this section, therefore, 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, retaining the main virtue of CTM. Let us now turn to an alternative conception which emphasizes precisely those two features of remembering.

2.3  The Narrative Theory of Memory There is an alternative conception of memory wherein memory does not merely record contents. Instead, it is a creative faculty. The main tenet of this

The Metaphysics of Memory  41 “narrative” conception of memory (or “NTM,” for short) is that, in memory, we build stories about our past; stories which include content that we have acquired through our own experience and content from other sources, like imagination or testimony. 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 proposition p, S remembers that p just in case S is representing that p, and S uses their representation of the fact that p as part of a narrative of S’s life.14 To clarify this formulation of the narrative view, it is worth elaborating on the notions of representation and narrative employed. 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 offered as an answer to the question of what it takes for a mental state to qualify as remembering episodically. 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 distinction between autobiographical and non-​autobiographical memory does not map onto the distinction between remembering episodically and remembering semantically neatly. An autobiographical memory is meant to be a memory of a fact regarding one’s past or one’s life, whereas a non-​autobiographical memory is supposed to be a memory of a fact that does not concern one at all.15 Thus, having an autobiographical memory may amount to remembering episodically (consider, for example, the autobiographical memory of one’s first kiss) or it may amount to remembering semantically (consider, for example, the autobiographical memory that one was once seriously ill as a baby). For this reason, when the NTM advocate claims that a mental state representing 14 Versions of this conception of memory can be found, for example, in (Schechtman 1994), (Goldie 2012), and (Brockmeier 2015). NTM is a particular version of a more general conception of memory; a conception that we will discuss in ­chapter  7. According to the more general, “reconstructive” conception of memory, remembering is a reconstructive process in which we build representations of our past by integrating content that we have acquired through our own perceptual experience with content from other sources, such as testimony, inference, and imagination (Michaelian 2016b; De Brigard 2014). NTM is a particular version of the reconstructive conception of memory in that, according to the narrative conception, the relevant reconstructive process is of a particular type: It involves the creation of a narrative of the subject’s life. For this reason, NTM is stricter than the more general, reconstructive conception of memory, which imposes fewer constraints on the relevant reconstructive process. (See note 22 for the relevance of this point.) 15 On autobiographical memory, see (Conway and Pleydell-​Pearce 2000) and (Barsalou 1988).

42  The Nature of Memory some fact counts as remembering the fact just in case the subject makes the represented fact a 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 fact in question was a part of their life.16 The notion of narrative 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 set of facts that the subject remembers. But it seems that this will be a loose structure. It seems plausible to suppose that, for every episodically remembered fact in the subject’s life, a story of that life will include a certain series of facts. The facts in the relevant series will be ordered by the earlier-​ than relation. At the start of the series, we will find the remembered fact and, at the end of the series, we will find the fact that the subject is having a mental image of it.17 But it also seems plausible to suppose that it will not be the case that, for any two episodically remembered facts in the subject’s life, the story will include a series of facts, ordered by the earlier-​than relation, which connects one of the two remembered facts with the other one. Thus, a story of my life, as I remember it, may include a number of facts about a family vacation that took place during my childhood, and it may include a number of facts that concern my early schooldays. But it need not be precise as to whether the facts about the family vacation happened before, or after, those involving my early schooldays.18 Also, stories have authors. The matter of

16 If NTM is a conception of autobiographical memory, then, one might wonder, why is it relevant for our project in this chapter? The reason is that, if the lesson drawn from cases of epistemic irrelevance in section 2.2 is correct, then any instance of remembering episodically counts as having an autobiographical memory. For remembering episodically must present the remembered fact to the subject as belonging to their life. This means that, when the NTM advocate offers a proposal about the nature of memory, the proposal being offered will concern, in particular, the nature of episodic memory, since it concerns, more broadly, the nature of autobiographical memory. 17 How many facts should a series include in order for the series to be a component of the story of someone’s life? Strictly speaking, nothing in our pre-​theoretical notion of a story seems to rule out the possibility that such a series might only include two facts; the episodically remembered fact and the fact that the subject is having a mental image of it. However, it is hard to see how NTM could get any traction by presupposing such a weak reading of “a series” of facts in its conception of a narrative. In what follows, therefore, I will assume that NTM presupposes a more robust notion of a series of facts as a component of the story of someone’s life. 18 The thought here is that the temporal structure of memory, if memory is conceived as a narrative, will be a tree-​like structure. This view is discussed independently of the narrative theory of memory in (Campbell 1997).

The Metaphysics of Memory  43 which facts are part of the story, and that of how those facts are organized, are not random matters. Those aspects of the story are the products of someone’s making. Thus, if NTM is right, then the remembering subject is responsible for the fact that their memory includes some facts and not others. Likewise, the remembering subject can be held accountable for the particular way in which those facts are organized in their memory. The remembering subject is, after all, the author of the narrative that includes those facts. Beyond this, it is difficult to say how much the advocate of NTM may build into the notion of a narrative. A number of issues about this notion remain open for discussion, and it seems that different positions on those issues will yield different versions of NTM. We may wonder, for example, about the phenomenology of narratives, and whether the affective properties of episodically remembering some fact, when that fact is included in a narrative of one’s life, must be identical with the affective properties of the original experience wherein that fact was initially presented to one.19 Similarly, we may wonder about the function of narratives. One plausible view is that the function of including, in episodic memory, a past fact as part of the narrative of one’s life is epistemic: Including the fact 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 plausible view is that the function of including, in episodic memory, a past fact as part of the narrative of one’s life is normative: Including the fact in the narrative is meant to provide one with a standpoint from which one can evaluate one’s past actions, feelings and reactions to significant facts in one’s past.20 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 episodically remembered facts 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 the relevant scenes when one remembers those facts), then is one still a character in the story being narrated? And if one is not, then does the story still count as a story of one’s own life?21 For the purposes of our discussion of NTM, I will 19 Richard Wollheim seems to endorse this view in (1984) whereas Peter Goldie rejects it in (2012). 20 The epistemic view about the function of narratives is put forward in (Schechtman 1994), whereas the normative view is proposed in (Goldie 2012). 21 Peter Goldie, for example, is sympathetic to the idea that a narrative of one’s life may include memories wherein the subject visualizes themselves, 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. Memories of this type are labelled “observer memories” in the psychological literature (Nigro and Neisser 1983). See c­ hapter 6 for a discussion of observer memories.

44  The Nature of Memory assume a notion of narrative which 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 fact 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 facts experienced 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 the fact that a bird is 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 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 facts actually obtained in the past or not. Memory must present facts to us as being part of our past, and it is a significant virtue of NTM that it respects this feature of memory. Unfortunately, NTM also faces two significant difficulties. Firstly, NTM seems to be too strict. It does not allow a mental state to qualify as a memory if the subject cannot fit the fact represented by that mental state into the narrative of their life. And yet, a memory of a past fact in our lives may, intuitively enough, be isolated from all other memories in our

The Metaphysics of Memory  45 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 the fact that I am 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 this fact does convey the sense that it is a part of my life to me. And yet, let us assume, I cannot integrate this fact within any account of the relevant part of my childhood: I cannot remember whether I fell into a public pool or into a pool 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 when I fell into the pool, and I cannot remember who pulled me out of it. In fact, I do not know anything else about the fact that I was sinking to the bottom of the pool, either through memory or through any other source. Thus, I cannot place any other fact about my childhood as obtaining either before or after this fact. On NTM, then, the mental state wherein I visualize the fact that I was 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 fact is completely isolated from all other facts which I take to belong to my past. As a result, it is not part of any narrative that I can construct about my life. And yet, it seems counter-​intuitive to say that, in this instance, I do not remember that I was sinking to the bottom of the pool. Let us call cases of this type, “isolation” cases. Secondly, NTM seems to be too permissive.22 Even though NTM accommodates the requirement that memory must present facts to us as belonging to our past, it does not require memories to have any particular etiology. And yet, it intuitively seems that mental states that do 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 facts that are supposed to have obtained on the previous day, even though the facts that the patients are visualizing never obtained. Furthermore, it is not unusual for patients with Korsakoff ’s syndrome to construct sophisticated stories that include

22 Notice that if this is right, then this particular difficulty for NTM generalizes to the more general reconstructive conception of memory (of which the narrative conception is a particular version). For NTM is stricter than the reconstructive conception of memory.

46  The Nature of Memory those facts 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 that they had an interesting conversation with a fellow traveler in a train, even though they have been in bed for weeks.23 Now, according to NTM, the mental state wherein the patient represents that they had a 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 which includes this fact as a part of it. And yet, it seems counter-​intuitive to say that the patient remembers that they had the conversation in the train in virtue of having this ability. Instead, it seems more natural to say that the patient is imagining the fact that they had the conversation and that the patient’s use of their mental image of that scene while constructing a story of having gone on a trip is an instance of confabulation. Let us call cases such as this one, “confabulation” cases.24 It seems, then, that NTM captures two important properties of memory, namely, the fact that memory presents facts to us as being part of our past and the fact that memory reconstructs the information that it conveys to us regarding those facts. 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 is, as confabulation cases illustrate, that the mental state originates in the subject’s past experiences. 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.

23 For details of this phenomenon in Korsakoff ’s syndrome, see (Talland 1961). 24 Confabulation in memory is sometimes construed as an episode in which the mental state concerned is unreliably formed (Michaelian 2016a, 5–​7). Other times, confabulation in memory is construed as an episode in which the mental state concerned fails to counterfactually depend on a perceptual experience of the event that the subject claims to remember (Bernecker 2017, 9–​10). Cases of confabulation in memory, as characterized here, are indeed cases in which the mental state concerned is unreliably formed. And they are also cases in which the just-​ mentioned relation of counterfactual dependence fails to hold. But, importantly, they are cases which enjoy those two features for a specific reason, namely, they involve amnesia. The issue of whether, in memory, there are cases of confabulation without amnesia is debatable (Hirstein 2006). For the sake of simplicity, I will concentrate on cases, such as that of Korsakoff ’s syndrome, which do involve amnesia.

The Metaphysics of Memory  47

2.4  Functionalism about Memory We have been asking the question of what it takes for a mental state to qualify as remembering some fact. We have found that an answer that is exclusively focused on the etiology of the mental state, such as CTM, presents some challenges. And we have found that an answer that is exclusively focused on the use that the subject makes of the mental state while forming beliefs about their past, such as NTM, faces some difficulties as well. At this point, it seems natural to adopt an approach that includes, within the conditions that a mental state must satisfy to qualify as an episode of remembering, some conditions which concern the etiology of the mental state as well as some which concern the impact that the mental state has on the subject’s beliefs. The framework of functionalism suggests a proposal about the nature of memory which 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 in virtue of the intrinsic properties of the state but in virtue of its association with a certain functional role within the subject. The functional role of a mental state is constituted by the causal relations in which the state tends to stand to perceptual inputs, behavioral outputs, and other states of the subject. Accordingly, different types of mental states are, on the functionalist framework, individuated by reference to the various functional roles that a state can play in a 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 of a subject 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, in order to qualify as a state of the relevant mental type. For example, my being in pain could be functionally construed as the type of mental state which tends to be caused in me by tissue damage, and which tends to produce in me the belief that something is wrong with my body, the desire to abandon that state and behavior such as wincing or moaning. In that case, if burning my hand on the stove tends to cause, let us say, my C-​fibers to fire, and the firing of my C-​fibers tends to cause me to withdraw my hand from the stove while

48  The Nature of Memory wincing, then my having C-​fibers firing qualifies, on the realizer version of functionalism, as my being in pain.25 By contrast, on a different version of functionalism (its “role” version), a state of a subject 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 in order for the subject to occupy the mental state. To continue with the pain example, my being in pain could also be functionally construed as the property of being in some state or other which tends to be caused in me by tissue damage, and which tends to produce in me the belief that something is wrong with my body, the desire to abandon that state, and behavior such as wincing or moaning. In that case, if burning my hand on the stove tends to cause my C-​fibers to fire, and the firing of my C-​fibers tends to cause me to withdraw my hand from the stove while wincing, my having C-​fibers firing does not qualify, on the role version of functionalism, as my being in pain. Nevertheless, I am in pain in virtue of having C-​fibers firing, since being in pain is the property of being in some state or other with the required functional role, and having C-​fibers firing does have the required functional role.26 The difference between realizer functionalism and role functionalism is therefore that, on the realizer version of functionalism, a mental state is a first-​order mental state (the state that actually plays such-​and-​such role; a physical state) whereas, on the role version, it is a higher-​order state (that of being in some state or other which plays such-​and-​such role; a more abstract state). However, both versions of functionalism are similar in their appeal to causal relations for individuating types of mental states. On both versions of functionalism, what matters for whether a subject is having a mental state of some type is not the causal relations that actually hold between that state and other mental states of the subject, as well as the subject’s perceptual inputs and behavioral outputs, but the causal relations that tend to hold between all of those states. This qualification is introduced to make room for possible

25 For an example of this variety of functionalism, see (Lewis 1980). Naturally, the choice of having C-​fibers firing (as opposed to any other type of fibers firing) for this example is arbitrary. 26 Role functionalism is discussed, for example, in (Block 1978).

The Metaphysics of Memory  49 scenarios such as the following: Suppose that the firing of my C-​fibers tends to be caused by damage in my body and tends to produce in me the belief that something is wrong with my body, the desire to abandon that state, and behavior such as wincing or moaning. Suppose, furthermore, that my C-​fibers are firing but, on this particular occasion, the firing of my C-​fibers is due to a rare cause. It so happens that there is a chemical way of stimulating my C-​fibers without causing any damage to my body, and I have inadvertently taken a drug that produces this effect in me. Intuitively enough, it seems that, in this scenario, taking the drug has caused me to be in pain.27 Functionalism needs to accommodate this intuition. Realizer functionalism accommodates it by positing that the firing of my C-​fibers does not need to be caused, in actual fact, by tissue damage in order for it to play the functional role which is required for the firing of my C-​fibers to be identical with my being in pain. By contrast, role functionalism accommodates the intuition by positing that the firing of my C-​fibers does not need to be caused, in actual fact, by tissue damage in order for it to play the functional role which is required for me to be in pain. On both versions of functionalism, it only needs to be the case that the firing of my C-​fibers tends to be produced in me by certain causes and tends to produce in me certain effects (a condition which, by assumption, is being satisfied in the scenario just described). What does functionalism have to do with memory? The main tenet of this chapter is that episodes of remembering are mental states that should be characterized functionally. Or, more precisely, they are mental states that 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 proposition p, S remembers that p just in case S has some mental image i such that i tends to cause in S a disposition to believe both that p and that S experienced that p, and i tends to be caused in S by having experienced that p. If we introduce the term “mnemonic role” to refer to such a functional role, then we can abbreviate FTM as the view that remembering some fact consists in having a mental image that plays the mnemonic role for that fact in the subject.28 27 The intuition can also be raised by disrupting the effect side of the functional role: Suppose that I have burnt my hand on the stove. It turns out, however, that the drug which I have inadvertently taken has paralysed me momentarily, so I do not wince, moan, or display any avoidance behavior. The drug has also (momentarily, but severely) affected my motivational states, so I cannot bring myself to desire to abandon the state that I occupy. Intuitively, it seems that I am in pain, and yet the state that I occupy is not producing the effects that it usually produces in me. 28 Notice that the conditions for remembering put forward in FTM do not include those in CTM. In order for mental image i to play the mnemonic role for the fact that p in subject S, it is not required

50  The Nature of Memory FTM is a functionalist view because according to it, when a subject has a mental image, whether the subject qualifies as remembering or not depends not on the intrinsic properties of that mental image but on 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 remembering is a higher-​ order state. FTM pulls apart, on the one hand, the property of having the specific mental image that the subject is having when they remember, for example, that yesterday was a sunny day and, on the other hand, the subject’s state of remembering that yesterday was a sunny day. According to FTM, having the mental image at issue is different from remembering that yesterday was a sunny day, since the subject could have had the same mental image without remembering that fact: Consider, for example, a possible situation in which the relevant mental image plays a functional role in the subject which is different from the mnemonic role. In that situation, the subject does not remember that yesterday was a sunny day. And yet, the subject is having the mental image in virtue of which, in the actual situation, they remember that fact. Conversely, the subject could have remembered that yesterday was a sunny day without having the mental image that, actually, they are having when they remember that fact: Consider, for example, a possible situation in which a different mental image plays the mnemonic role for that fact in the subject. In that situation, the subject remembers that yesterday was a sunny day. And yet, the subject has a different mental image from that which, in the actual situation, they are having when they remember that yesterday was a sunny day. To help us keep track of the distinction between the two states that a subject occupies when the subject is cognitively related to some fact through episodic memory, we may stipulate a technical use for two familiar expressions. We may refer to the mental image which plays the mnemonic role in the subject for that fact as “a memory of ” the fact, and we may continue to use the expression “remembering” the fact to refer to the subject’s higher-​order state of having some mental image or other which plays the mnemonic role for that fact in the subject.29 Let us examine, next, what considerations can be offered in support of FTM. that S having i has been caused by S having experienced that p in the past. (For the significance of this point, see the later discussion on embellishment cases.) 29 Notice that this technical use of “a memory of ” the fact is neutral on the issue of what the content of the memory is supposed to be. If I have a mental image which qualifies as a memory of the fact that yesterday was a sunny day, for example, then the functional role of that mental image involves the fact that yesterday was a sunny day. But this does not entail that the content of the memory is that yesterday was a sunny day. (See ­chapter 3 for a discussion of the content of memories.)

The Metaphysics of Memory  51

2.5  Narratives, Causal Histories, and Functional Roles It seems that FTM is permissive enough. It does not seem to conflict with our intuitions about those cases which suggested that both CTM and NTM were too strict. Consider, first of all, the case in which I have a mental image of my father shooting a black rabbit even though, in the past, I saw my father shooting a white rabbit. The reason why CTM rules this case out as an episode of remembering is that CTM does not allow a remembered fact to include more details than those which were originally experienced by the subject at the time that they perceived the fact. By contrast, FTM imposes no such constraint on episodes of remembering. 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 fact that my father shot a black rabbit. And it seems that the mental image at issue does play the mnemonic role for that fact 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 by my father, and it tends to cause in me the belief that my perceptual experience was veridical; that the shooting did obtain in the past. On the other hand, my mental image is the type of image that tends to be produced in me by past perceptual experiences of black rabbits being shot. To be sure, on this particular occasion, my mental image was not actually caused by a perceptual experience of a black rabbit being shot since, in the past, I did not have such an experience. Nevertheless, 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, it seems that, unlike CTM, FTM has the resources to accommodate this case as an episode of remembering. Consider, now, the case in which I have a mental image wherein I visualize the fact that I am sinking to the bottom of a pool, even though I am unable to relate this fact to any other fact about my childhood. The reason why NTM rules this case out as an instance of remembering is that NTM requires remembered facts to be embedded in a story that the subject can tell about their own past. By contrast, FTM imposes no such constraint on episodes of remembering. According to FTM, as long as the mental image that I am having when I visualize that I am sinking to the bottom of the pool is an image that plays the mnemonic role for that fact in me, I qualify as

52  The Nature of Memory remembering the fact. And it seems that the mental image concerned does play the mnemonic role in me: It is the type of mental image that tends to be produced in me by past perceptual experiences of sinking to the bottom of a pool, and it is the type of mental image that tends to make me believe that this fact was real and that I experienced it. Thus, it seems that, unlike NTM, FTM can acknowledge this case as an episode of remembering. Furthermore, FTM seems to be strict enough. It does not seem to conflict with our intuitions about those cases which suggested that both CTM and NTM were too permissive. Consider, first, the case in which I have a mental image of a bird 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 the facts that I  visualize were real and that I  experienced them in the past. The reason why CTM accepted this case as an instance of remembering is that CTM did not require mental states to have any particular impact on the subject’s beliefs in order for them to qualify as episodes of remembering. By contrast, FTM imposes a requirement of this kind on episodes of remembering. If my having the mental image of the fact that a bird was on the roof of a house does not tend to produce in me the belief that this was a real fact, and the belief that I perceived it in the past, then, according to FTM, I do not qualify as remembering the fact. 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 tends to produce in me beliefs of those kinds. If I had other mental images of the same type repeatedly, this would continue to have a null effect on my beliefs about the past. Thus, it seems that, unlike CTM, FTM can rule 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 the fact that they were conversing with someone while travelling on a train, and the patient is able to construct a sophisticated story within which they can insert the scene that they are visualizing. The reason why NTM accepted this case as an instance of remembering is that NTM did not require mental states to have any particular etiology in order for those mental states to qualify as episodes of remembering. By contrast, FTM imposes a requirement of this sort on episodes of remembering. If the patient’s mental image of the conversation in the train is of a kind which 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 fact

The Metaphysics of Memory  53 that he was having a conversation in a train. And it seems that the mental image concerned does not play that part of the mnemonic role for that fact 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 mental image that tends to be produced by experiences of such conversations. Had the patient experienced having a conversation with a fellow traveler in a train in the past, the patient would not have retained that experience in the type of mental image that they are having now. Thus, it seems that, unlike NTM, FTM can rule out this case as an episode of remembering.

2.6  Conclusion Role functionalism seems to have provided us with a satisfactory answer to the question of what episodic remembering is: Remembering some fact consists in having a mental image that plays a particular kind of functional role for that fact 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 CTM. They also allow the functionalist proposal to overcome the shortcomings of those proposals about the nature of remembering which are exclusively forward-​looking; proposals such as the narrative theory of memory. For that reason, the functionalist proposal seems to draw the line between those mental states which count as episodes of remembering and those which do not at the right point. It is, unlike the alternative proposals, neither too permissive nor too strict. One may wonder, however, whether the functionalist proposal retains either the virtues of the narrative theory of memory or those of CTM. The main virtue of CTM is that it is rigid enough to accommodate the preservative aspect of memory. FTM incorporates this rigidity. Notice that the functionalist theory regards those cases in which the faculty of memory works properly as cases in which the subject remembers a fact, and the memory that they are having has been caused by its typical cause (an experience of the remembered fact) and produces its typical effects (effects such as the belief that the fact obtained in the past). For this reason, the functionalist theory regards those cases in which the faculty of memory works properly as

54  The Nature of Memory cases in which what the subject remembers is what they experienced in the past. This is the sense in which, when all goes well as far as memory is concerned, memory is preservative according to the functionalist theory. The main virtue of the narrative theory of memory is that it is flexible enough to accommodate the phenomenon of memory reconstruction. Over the course of our lives, we will be able to remember some fact at various different times. And the mental images that we will have, when we remember the relevant fact, will often be different from the original experience that we had of that fact. (This seems particularly likely in the case of facts that we experienced a long time ago.) Some of those times, the mental image that we have when we remember the fact will be different because it is degraded with respect to the original experience of the fact, or contains less detail about that fact. But, other times, the mental image that we have when we remember the fact will be different because it is enriched with respect to the original experience, or contains more detail about that fact. The narrative theory of memory is flexible enough to accept that memory supports the latter type of change and not only the former one: Having a mental image that results from the enrichment of our original experience of some fact counts as remembering, on the narrative theory, just as much as having a mental image that results from its degradation does. FTM incorporates this flexibility. The reason why it has this feature is that, according to the functionalist theory, remembering some fact is identical with having a mental image that plays a certain functional role in the subject, but there is no specific mental image such that remembering the fact is identical with having that image. A subject is therefore allowed to remember the same fact by having different mental images at different times, as long as those mental images play the required functional role in the subject. This is the sense in which memory is reconstructive according to the functionalist theory. Let us not underestimate the significance of integrating the reconstructive aspect of memory. If we claim, as the advocate of the causal theory does, that memory does not support changes of the enrichment type in the mental images that we have when we remember facts, then we will not be able to leave room for the possibility of a certain kind of misremembering.30 30 Sarah Robins raises the objection that the causal theory of memory cannot accommodate misremembering in (Robins 2016). Robins’s use of the term misremembering is narrower than mine. I am using misremembering to refer to an episode wherein one is having a mental image which qualifies as a memory, but the mental image at issue is false. This includes two types of episodes. Firstly, it includes episodes in which the perceptual experience in which the memory originates was correct but one’s memory is, in a certain sense, distorted with respect to it: One’s memory contains details

The Metaphysics of Memory  55 Suppose that, at some point during a subject’s life, a mental image that, in the past, the subject used to have when they remembered some particular fact comes to have more detail about that fact than the subject’s original experience of the fact did. If memory does not support changes of the enrichment type, then this means that the subject no longer remembers the fact in virtue of having that mental image. The enrichment of their mental image has taken the subject from the domain of memory to the domain of some other faculty (presumably, that of imagination). And, clearly, if the subject does not remember the fact in virtue of having the mental image with the added detail, then the subject does not remember the fact veridically, and they do not misremember the fact either. But if changes of the enrichment type in a mental image of some fact do not lead to misremembering the fact, then it is hard to see how a subject can misremember a fact in virtue of having some mental image unless, in the past, the experience in which the subject’s mental image originates was false in the first place. For that reason, there seems to be no room for memory failure, or memory malfunction, within CTM.31 This outcome seems too high a price to pay for capturing the preservative aspect of memory. After all, memory errors of this type are not a rare occurrence. We routinely misremember things that we correctly perceived in the past. We need, therefore, to allow for the possibility of memory errors for which only memory is responsible. And it seems that the best way of doing this is by acknowledging that remembering can involve reconstructing the information about the past which is provided to us by our own past experiences. The way in which the functionalist theory accommodates the phenomenon of memory reconstruction reveals a key feature of the theory. According that the original perceptual experience did not contain. And secondly, it also includes episodes in which one’s memory is not distorted with respect to the original perceptual experience, but the orig­ inal perceptual experience was false to begin with. By contrast, Robins requires that misremembering must “result from the distortion of retained information” (2016, 434). I take this to mean that only episodes of the former type mentioned here qualify, on Robins’s use of the term, as episodes of misremembering. 31 The causal theorist may reply that they are operating with a factive notion of remembering, so there is no need to accommodate cases of misremembering of any kind. This is a reasonable reply, but it runs into a difficulty. Consider a case in which I look at, let us say, a red rose in my garden, and the perceptual experience that I have as a result represents it as being white. Suppose that, later, I have an experience which originates in my perceptual experience of the rose. My current experience perfectly corresponds to my original perceptual experience of the rose, in the following sense. My current experience enjoys all and only the details in my past perceptual experience. In particular, my current experience represents the red rose as having been white. It seems counter-​intuitive to say that, in this case, there is no remembering. After all, my current experience is matching the perceptual experience in which it originates perfectly. It is totally faithful to the past perceptual experience which has given rise to it. And yet, according to the factive version of the causal theory, I am not remembering the rose.

56  The Nature of Memory to the functionalist theory, for any proposition p, if a subject remembers that p, then they are having a mental image that plays the mnemonic role for the fact that p in the subject; a mental image that qualifies as a memory in virtue of playing that functional role. As we have seen, this means that, when a subject remembers that p, the subject normally occupies two different mental states at the same time. One of them is the state of having the relevant mental image. This is a first-​order state of the subject. But the subject also occupies a higher-​order state; the state of having some mental image or other which qualifies as a memory of the fact that p. This is the state of remembering. This distinction is important for the overall picture of memory that we have started to build. The reason why it is important is that the functionalist theory offered in this chapter is a view about the nature of episodes of remembering and not about that of memories. The functionalist theory tells us that episodes of remembering require having mental images which play a certain functional role within the subject. But it does not tell us much about the properties that such mental images are meant to have when they play the functional role that is required for the subject to remember. Specifically, the functionalist view is neutral on the content of such mental images. The functionalist view is also neutral on their phenomenology, and it is neutral on the kind of knowledge that they afford.32 Elucidating these issues is precisely our project for the rest of the book.

32 One benefit of construing memories thus is that it allows for the possibility that one may have a mental state wherein it appears to one that one is remembering something (because the mental state at issue has the content and phenomenology which are characteristic of memory), but the mental state is not a memory. Since we do seem to have apparent memories sometimes, the fact that the functionalist view allows for this possibility is a virtue of the view.

3 The Intentionality of Memory 3.1  Introduction In ­chapter  2, we have seen that memories are mental images which play a certain functional role within the subject, and that a subject qualifies as remembering a fact when they have a mental image that plays the appropriate functional role within them. We have, however, left open the question of what the content of memories is. Thus, the aim of this chapter will be to determine how we should construe the content of memories. Are our memories about facts in the world?1 Are they about our past experiences of those facts? Could our memories be about some combination of those two elements perhaps? These are the issues with which I will be concerned in this chapter. I will proceed as follows. In section 3.2, I will specify the notion of mnemonic content that will be used throughout this discussion. The relevant notion will involve the truth-​ conditions of memories. I  will distinguish two types of truth-​conditions for memories, and I will isolate the type which captures the sort of content that we will investigate. In section 3.3, I will put forward some constraints which will help us evaluate different views of mnemonic content. Those constraints will involve the amount of information that we should attribute to memories. I will assume that a view of mnemonic content should not attribute to a memory a content that makes it too easy for the memory to qualify as being correct and that it should not attribute to a memory a content that makes it too hard for the memory to qualify as being correct either. In section 3.4, I will discuss a view according to which, when one remembers a fact, the content of the memory that one is having only concerns that fact. I will label it the “objective view” of mnemonic content. I will argue that if the objective view is correct, then memories carry too little information. 1 For the sake of simplicity, in this chapter we will concentrate on objective facts of a particular kind, namely, facts which involve visible objects. The view of mnemonic content to be proposed, however, will generalize to memory for facts involving objects perceived through sensory modalities other than vision. An earlier version of this chapter can be found in (Fernández 2006).

58  The Nature of Memory In section 3.5, I will discuss a view according to which, when one remembers a fact, the content of the memory that one is having only concerns a perceptual experience of that fact that one had in the past. I will label it the “subjective view” of mnemonic content. I will argue that if the subjective view is correct, then memories still carry too little information. In section 3.6, I will discuss a view according to which, when one remembers a fact, the content of the memory that one is having concerns an objective fact as well as a perceptual experience of it that one had in the past. I will call this view the “conjunctive view” of mnemonic content. I will argue that if the conjunctive view is correct, then memories carry, once again, too little information. Thus, in section 3.7, I will explore a different approach to mnemonic content; the “causal self-​reference approach.” The main tenet of the causal self-​reference approach is that memories represent certain facts about themselves. Specifically, memories represent their own causal histories. I  will distinguish four versions of this approach depending on how those causal histories are conceived. In section 3.8, I will defend one of the versions of the causal self-​reference approach by arguing that it attributes the right amount of information to memories. Accordingly, in section 3.9, I will conclude that the relevant version of the causal self-​reference approach delivers the right view of mnemonic content.

3.2 Mnemonic Content The basic idea about memories having content is that memories can be evaluated as correct or incorrect. For each memory, there are conditions with respect to which the memory is correct, or true, and conditions with respect to which the memory is incorrect, or false (for short, “truth-​conditions” of it). Thus, it is natural to think that if you want to know what the content of a memory is, you should ask yourself what the truth-​conditions of the memory are. What are, in other words, the possible situations with respect to which the memory is true? This question, however, needs to be refined. For the notion of truth with respect to a possible situation can be understood in two different ways and, as a result, the truth-​conditions of a memory can be construed in two different ways as well. For the purposes of capturing the relevant distinction, it will be convenient to represent the truth-​conditions of memories by means of propositions and to construe propositions as sets of possible worlds.

The Intentionality of Memory  59 An intentional state, such as a belief, a perceptual experience, or a memory, can be true with respect to a possible situation in two senses. In one sense, an intentional state is true with respect to a possible situation if, in that situation, it exists and is true. (Let us call this way of being true with respect to a possible situation, being “true in” it.) In another sense, an intentional state is true with respect to a possible situation if, in that situation, a certain fact obtains. Which fact? The fact that needs to be the case in the actual situation for the intentional state to be true. (Let us call this way of being true with respect to a possible situation, being “true of ” it.)2 To illustrate the distinction, consider a possible situation in which nobody believes anything. If I believe, in the actual situation, that nobody believes anything, then my belief is not true in that possible situation, since my belief does not exist there; no belief does. And yet, my belief is true of that possible situation precisely because no belief exists there. The possible situation is, in that sense, accurately represented by my belief.3 Since there are two ways of understanding the notion of truth with respect to a possible situation, there are also, for any intentional state, two types of truth-​conditions for the state. One may distinguish, on the one hand, the truth-​conditions captured by the proposition that contains those situations in which the state is true and, on the other hand, the truth-​conditions captured by the proposition that contains those situations of which the state is true. We may call them, respectively, the “in-​proposition” and the “of-​proposition” associated with the intentional state.4 The distinction between two types of truth-​conditions has an obvious implication for the approach which identifies the content of a memory with its truth-​conditions. If this is how mnemonic content needs to be understood, then we should acknowledge that a memory has two different kinds of content.5 One of them 2 Robert Adams draws this distinction in his (1974). Notice that, when the possible situation with respect to which an intentional state is evaluated is the situation in which the state is taking place, then the intentional state is true of the situation if and only if it is true in the situation. For that reason, when we discuss, in what follows, a case in which a memory is evaluated with respect to the possible situation in which the memory is taking place, I will omit the distinction between the two ways of being true with respect to the situation, and I will simply speak of the memory as either being true or being false. 3 Analogous considerations would apply to a picture depicting the non-​existence of all pictures; an example offered in (Pollock 1985). The picture is not true in a situation where no pictures exist, because the picture itself does not exist there; no picture does. But the picture is true of that situation, precisely because there are no pictures there. The possible situation is, in that sense, accurately represented by the picture. 4 The idea that there are different kinds of truth-​conditions for intentional states is not new. It can be found in the literature on indexicals. See (Perry 1999) for instance. 5 The idea that there are different kinds of content for mental states is not new either. It can be found in the two-​dimensionalism literature. See, for example, (Chalmers 2002).

60  The Nature of Memory is captured by the in-​proposition associated with the memory whereas the other one is captured by the of-​proposition associated with the memory. In this chapter, we seek a view about the content of memories in the sense of “content” that concerns the truth-​makers of memory. We are trying to determine what kinds of entities our memories are about or, equivalently, what kinds of entities make our memories true or false.6 We want to know whether those entities are facts, out there, in the world, or they are facts involving mental properties of some kind, or perhaps some combination of those two elements. For that reason, our focus in this discussion will be on the latter kind of content; the content captured by the of-​proposition associated with a memory. Thus, we will try to determine, for any memory, what the possible situations of which the memory is true are. In what follows, I will represent the content of a memory M as a set of possible worlds which meet a condition C, where C is such that, for any possible world W, M is true of W just in case W meets C. I will refer to such sets with expressions of the form “{W: C takes place in W}.” The project of putting forward a view about the content of memories becomes, on this approach, the project of producing a template that yields the relevant condition C for each memory.

3.3  Methodology How should we go about determining what the content of a memory is? If the relevant notion of content is that captured by the of-​proposition associated with the memory, then the following seems to be an appropriate strategy to pursue. First, we can entertain a number of possible situations in connection with the memory being investigated. Then, we can consult our intuitions on whether the memory is true of those situations or not. And, finally, we can consider what the entertained situations accurately represented by the memory have in common. The hope is that this exercise will provide us with some insight into what is shared by all the possible situations of which the memory is true; the condition that a possible situation must meet in order for the memory to be true of it. Once we reach that conclusion, we will have an answer to the question of what the content of the memory is. 6 As mentioned in c­ hapter 1, I will be using a number of semantic notions indistinctly. These include what a memory “refers to,” what the memory “is about,” what the memory “represents,” and what its “intentional object” is. Admittedly, these notions are not equivalent but, as far as I can see, the relevant distinctions will not matter for the purposes of this discussion.

The Intentionality of Memory  61 In this process, it will be useful to have some criterion which allows us to evaluate different candidate propositions for the content of a memory. A reasonable way of evaluating a candidate for the content of a memory seems to be by checking whether the candidate proposition meets a certain requirement: Any view about the content of a memory should attribute the right amount of information to it. That is, it should draw the line separating those possible situations which are accurately represented by the memory from those which are not at the right point. A certain test can guide us while deciding whether a view of mnemonic content meets this requirement or not. We can call it the “right amount of information” test, or “RAI.” Let me elaborate on what RAI requires exactly. If, according to our pre-​theoretic intuitions about memory, a memory is accurately representing, or is true of, some possible situation, then that situation should belong to the content that our candidate view of mnemonic content attributes to the memory. Suppose, for example, that we are enquiring what the content of a certain memory is, and we consider a view according to which the content of the memory is {W: W meets C1} for some condition C1. Suppose, furthermore, that the memory is, intuitively enough, representing some possible situation correctly. Then, C1 should be met in that situation. Otherwise, we can conclude that the content that the view being considered attributes to the memory is, in a sense, too thick. The attributed content is too thick in the sense that it is making it too difficult for possible situations to count as being accurately represented by the memory. After all, there is a possible situation that, intuitively, is well represented by the memory, and yet the situation does not meet the condition posited by the view that we are contemplating. Conversely, if it is the case that, according to our pre-​theoretic intuitions about memory, a memory does not accurately represent, or is not true of, some possible situation, then that situation should not belong to the content that our candidate view of mnemonic content attributes to the memory. Imagine, once again, that we are enquiring what the content of a memory is, and we consider the view that its content is {W: W meets C2} for some condition C2. Imagine, furthermore, that the memory is, intuitively enough, not representing some possible situation correctly. Then, C2 should not be met in that situation. Otherwise, we can conclude that the content that the view being considered attributes to the memory is, in a sense, too thin. The attributed content is too thin in the sense that it is making it too easy for possible situations to count as being accurately represented by the memory.

62  The Nature of Memory After all, there is a possible situation that, intuitively, is not well represented by the memory, and yet the situation meets the condition posited by the view that we are contemplating. In the next three sections, we will consider three types of propositions which seem plausible candidates for the content of our memories. In order to evaluate the three candidates, we will focus on whether the views proposing those contents attribute the right amount of information to memories or not. We will focus, in other words, on how the relevant views fare vis à vis the RAI test.

3.4  The Objective View Consider the following example. Suppose that I am looking at a red apple in front of me, and my perceptual experience presents it to me as being red. Let us stipulate that I have this perceptual experience on 7/​2/​1987. Suppose that, years later, I have a memory that originates in that perceptual experience; one that I would express by saying that I have a memory of being in front of a red apple. Let us stipulate that I have this memory on 7/​2/​2018 and, in the meantime, I have not received any testimony about the red apple. Suppose that my faculties of perception and memory are reliable. Let us call this possible situation “W0,” and let us refer to my past perceptual experience and my memory in W0 as “P” and “M” respectively.7 How should we construe the content of M? Perhaps the most natural candidate for the content of M is the following proposition, to which we may refer as the “objective proposition” (or, for short, “OBJ”): OBJ: {W: In W, I was in front of a red apple}

The view that OBJ is the content of M can be motivated by considering a number of possible situations in which I was not in front of a red apple and consulting our intuitions on the veridicality of M with respect to those situations. We may consider, for example, a situation in which I was in front of a banana, a situation in which I was in front of an orange, and a situation

7 The stipulation that P takes place on 7/​2/​1987 and M takes place on 7/​2/​2018 will not be important, and can therefore be ignored, for our purposes in this chapter. It will, however, become relevant in the discussion on the temporal phenomenology of memory in c­ hapter 4.

The Intentionality of Memory  63 in which I was in front of a plum. Intuitively enough, M is false of all three situations. It seems to represent all those situations inaccurately. We may also consider some possible situations in which I was in front of a red apple, but other details in the room, outside of my field of vision, were different from what they were in the actual situation. In one of those situations, for example, there is a tiny ant hiding behind the red apple. In another, there is a cat sleeping on the mat behind me; and so on. Intuitively, M is true of those situations. It represents them accurately. Given that the possible situations in the former group have in common the absence of the red apple, and those in the latter group have in common the presence of the red apple, it seems prima facie plausible that the red apple is the main component of the content of M. The view that OBJ is the content of M is a specific instance of a broader view about the intentionality of memory. If the content of M is OBJ, then M is about something objective; something that took place independently of what my mental states were at the time. Let us call the view that our memories are about past objective facts, the “objective view.” The objective view is an attractive position. It seems to fit with the way in which we ordinarily express our episodic memories. Consider the fact that I express M by saying that I have a memory of being in front of a red apple. This way of expressing M suggests that our memories put us in cognitive contact with past objective facts. The thought that our memories put us in cognitive contact with past objective facts seems to be a natural thought about memory; a thought that any view of mnemonic content should respect.8 And it is a virtue of the objective view that it vindicates this thought. For if the content of M is OBJ, then my memory does indeed put me in cognitive contact with a past objective fact, namely, the fact that I was in front of a red apple. There is, however, a difficulty for the view that OBJ is the content of M. The difficulty concerns the following possible situation: W1: The apple is red, and it is located exactly in the same position with respect to me as it is in W0. It is also located in front of me at the same time as it is in W0, that is, 7/​2/​1987. I, however, am blind so I do not have P, 8 I will use the term “cognitive contact” as a placeholder for whatever relation allows a subject to think about, and talk about, a fact. For our current purposes, it will not matter what that relation is exactly. Whatever the precise nature of the relation turns out to be, it will remain true that a view of mnemonic content should accommodate the thought that our memories allow us to think, and talk, about objective facts in the past.

64  The Nature of Memory and I do not have M. Furthermore, nobody else is in a position to see the red apple so no perceptual experience of the apple takes place.

If the content of M is OBJ, then I should be representing W1 accurately when I have M in W0. After all, if the objective view is correct, then what matters for the accuracy of M in W0 with respect to any possible situation is whether, in the situation concerned, the apple in front of me is red. Which, by assumption, is the case in W1. And yet, it intuitively seems that, by having M in W0, I  am not representing W1 accurately. What does this intuition suggest? It suggests that my memory M in W0 is not only about an apple that, as a matter of fact, was red in the past. In addition, M concerns my own past perceptual experience of the apple. The view that OBJ is the content of M, however, does not factor any of my past perceptual experiences into the content of my memory. For that reason, it seems that the sort of content that this view attributes to memories does not pass the RAI test. The attributed content is too thin, or carries too little information.

3.5  The Subjective View The moral drawn from the shortcomings of the objective view points us in the direction of an alternative candidate for the content of memory M. If the issue is that the objective view does not square with the fact that our memories seem to carry information about our own past perceptual experiences, then perhaps a better candidate for the content of M is the following proposition, which we may call the “subjective proposition” (or, for short, “SUBJ”): SUBJ:  {W: In W, I seemed to perceive a red apple in front of me by having P}

The view that SUBJ is the content of M can be motivated by considering a number of possible situations in which I did not have P and consulting our intuitions on the veridicality of M with respect to those situations. We may consider, for example, the situation in which I was in front of a banana, the situation in which I was in front of an orange, and the situation in which I  was in front of a plum. Let us stipulate that my faculties of perception and memory are working correctly in those situations. As a result, in those situations, I do not seem to perceive a red apple in front of me by having

The Intentionality of Memory  65 P. Instead, the object in front of me appears to me to be, respectively, yellow, orange, and purple.9 Intuitively enough, M is false of all three situations. It seems to represent all those situations inaccurately. We may also consider other possible situations in which my faculties of perception and memory are working correctly. In those situations, there was a red apple in front of me, and I seemed to perceive it by having P. But other details in the room, outside of my field of vision, were different from what they were in the actual situation at the time that I seemed to perceive the apple. Take, once again, the situation in which there is a tiny ant hiding behind the red apple or the situation in which there is a cat sleeping on the mat behind me. Intuitively enough, M is true of those situations. It represents them accurately. Given that the possible situations in the former group have in common the absence of my perceptual experience of the red apple and those in the latter group have in common its presence, it seems prima facie plausible that my past perceptual experience P is the main component of the content of M. The view that SUBJ is the content of M is a specific instance of a more general position about the intentionality of memory. If the content of M is SUBJ, then M is about something subjective, namely, one of my own past perceptual experiences. Let us call the view that our memories are about our own past perceptual experiences, the “subjective view.”10 There are two features of the subjective view which make it an attractive position. Firstly, the subjective view overcomes the difficulty that we raised for the objective view. The subjective view accounts for the intuition that M is false of W1. After all, if the subjective view is correct, then what matters for the accuracy of M in W0 with respect to any possible situation is whether, in the situation concerned, I had P and, as a result, it appeared to me that an apple in front of me was red. By assumption, this is not the case in W1. So the subjective view accommodates the intuition that M does not accurately represent W1. This virtue of the subject view is unsurprising, though, since the view was precisely designed to overcome the limitations of the objective view. The more interesting virtue of the subjective view is that it vindicates another natural thought about memory. The thought concerns those cases in which our memories match the past perceptual experiences in which they originate, but it turns out that those perceptual experiences were, in the first instance, false. 9 Notice that this stipulation is harmless as far as our evaluation of OBJ in section 3.4 is concerned. The reason is that, according to the objective view, whether I had P or not in some possible situation is irrelevant for the issue of whether M is true of that possible situation. 10 The subjective view is endorsed, for example, by Wolfgang Von Leyden in (1961, 61).

66  The Nature of Memory Imagine, for example, a situation in which we discover that the apple was green when I had P in the past. That is, we discover that I originally misperceived the apple as being red and therefore that a false perceptual experience has led me to have M. A natural thought to have about the relation between perception and memory in this situation is that my faculty of memory has not malfunctioned. This is the thought which seems to be behind the strong intuition that, whatever may go wrong in this situation as a result of having misperceived the apple in the past, my faculty of memory is not to blame for it. If I form, for example, the belief that I was in front of a red apple on the basis of M, then my belief will turn out to be false. Furthermore, it will be a false belief that I formed through memory; a false memory belief. And yet, it does not seem that my faculty of memory is responsible for my mistake. For the sake of brevity, let us refer to this feature as the “innocence” of memory. It is a virtue of the subjective view that it vindicates the innocence of memory. Notice that if the content of M is SUBJ, then my memory M in W0 is not misrepresenting those possible situations in which, in the past, I misperceived the apple as being red when I had P. To illustrate this aspect of the subjective view, consider the following possible situation: W2: The apple is green. I, however, have P when I look at it. A temporary malfunction of my visual apparatus has caused me to misperceive the green apple as being red. Years later, I have M as a result of having had P.

According to the subjective view, since I seemed to perceive the apple as being red when I had P in W2, my memory M in W0 is true of W2. For what matters for the accuracy of M in W0 with respect to any possible situation, according to the subjective view, is whether, in the situation concerned, I seemed to perceive a red apple in virtue of having P; not whether the apple was in fact red. And W2 is, by assumption, a situation in which I did seem to perceive a red apple by having P. How exactly does this show that the subjective view vindicates the innocence of memory? Suppose that, while having M in W0, I form the belief that I was in front of a red apple on the basis of M. The thought about the innocence of memory is the thought that if we happened to discover that W0 is, like W2, a situation in which I misperceived a green apple as being red when I had P, then we would not blame M for the falsity of my belief about the apple. If the subjective view is correct, then this thought makes sense. For if the content of M in W0 is SUBJ and, for that reason, M in W0 is true of W2,

The Intentionality of Memory  67 then M in W0 is also true of W0. It is true of W0 regardless of whether, in W0, the apple was red in the past or not. This means that if I form, in W0, a false belief about the apple being red on the basis of M because, as it happens, I misperceived the apple in the past by having P, then I am certainly making a mistake, but M has not misled me. According to the subjective view, my mistake in believing that the apple was red is, in this scenario, not due to the fact that M is false. It is presumably due to the fact that I have assumed (incorrectly, it turns out) that my perceptual experience was veridical at the time that I experienced the apple. It is no wonder, then, that we think that my memory is not at fault in this scenario. The difficulty for the subjective view is the following. Granted, the view that SUBJ is the content of M does account for our intuitions regarding the innocence of memory. However, it does not square with our intuitions regarding W2. If the content of M is SUBJ, then I should be representing W2 accurately when I have M in W0. And yet, it intuitively seems that, by having M in W0, I am not representing W2 accurately. M in W0 seems to be false of W2. A parallel consideration applies to the scenario in which we discover, in W0, that I misperceived a green apple as being red when I had P in the past. If it turns out that, when I seemed to perceive a red apple by having P, I was actually looking at a green apple, then the memory whereby it now seems to me that there was a red apple in front of me is, intuitively enough, just as false as my past perceptual experience P. What do these intuitions suggest? They suggest that memory M in W0 is not only about an apple that appeared to me to be red. In addition, M concerns the fact that the apple was in fact red. M provides me with information about an objective fact in the past. This is, after all, the very thought which motivated the objective view. The objective view was able to vindicate the thought that our memories put us in cognitive contact with past objective facts by factoring those facts into the contents of our memories. By contrast, the subjective view only builds past perceptual experiences into those contents. For that reason, it seems that the sort of content that this view attributes to memories does not pass the RAI test either. The attributed content is, once more, too thin. It carries too little information.

3.6  The Conjunctive View Let us take stock. We have considered two views about mnemonic content; the view that our memories are about past objective facts and the view

68  The Nature of Memory that our memories are about our own past perceptual experiences. We have found both views to be problematic. However, we have also encountered two thoughts about memory which motivated those two views, and both of those thoughts seem worth preserving. The first thought is that memory puts us in cognitive contact with past objective facts. This thought is worth preserving because it accounts for how we ordinarily express the content of our memories. We express it by talking about past objective facts. The second thought is that memory puts us in cognitive contact with our own past perceptual experiences. This thought is worth preserving because it accounts for why we do not blame our faculty of memory for the false beliefs that we form on the basis of memories which originate in false perceptual experiences. What we need, then, is a view of mnemonic content which accommodates these two thoughts about the sorts of things that our memories put us in cognitive contact with. The lessons learnt from our discussion of the objective view and the subjective view suggest another prima facie plausible candidate for the content of memory M. If the difficulty with the two views is that neither of them builds enough facts into mnemonic content to account for the idea that memory puts us in cognitive contact with both objective facts in the past and past perceptual experiences, then perhaps we should build perceptions into mnemonic content. This would allow us to engage objective facts and perceptual experiences in the contents of memories.11 Consider, for example, the following proposition as a candidate for the content of M. We may call it “O&S”: O&S:  {W: In W, I perceived a red apple in front of me by having P}

The view that O&S is the content of M is a specific instance of a more general position about the content of our memories. According to the “conjunctive” view, our memories are about something objective (namely, objective facts in the past) and they are about something subjective (namely, our own past perceptual experiences) as well. The main consideration in support of the view that O&S is the content of M is that it accommodates our intuitions about the veridicality of M in W0 with respect to both W1 and W2. If the content of M is O&S, then what matters for the accuracy of M in W0 with 11 I will take perceptions to be veridical perceptual experiences. Accordingly, I will use the verb “perceiving” factively in what follows.

The Intentionality of Memory  69 respect to any possible situation is partly whether, in the situation concerned, the apple in front of me is red. After all, a perception of a red apple requires the apple to be, in fact, red. And, by assumption, this is not the case in W2. Thus, the conjunctive view delivers the outcome that M in W0 is false of W2. Likewise, if the content of M is O&S, then what matters for the accuracy of M in W0 with respect to any possible situation is partly whether, in the situation concerned, the apple in front of me appears to me to be red in virtue of having P. For a perception of a red apple also requires the apple to be perceptually experienced as being red. And, by assumption, this is not the case in W1. Thus, the conjunctive view delivers the outcome that M in W0 is false of W1 as well. Clearly, the reason why the conjunctive view is able to accommodate our intuitions regarding both W1 and W2 is that, according to the view, our memories are about past objective facts, and they are also about our own past perceptual experiences. One might think that, for that reason, the view should preserve the virtues of the objective view as well as those of the subjective view. And it does preserve them, but not as straightforwardly as one might have expected. Take, for example, the main virtue of the objective view; the fact that the view captures the way in which we ordinarily express our memories. The objective view accounts for the fact that I would express, for instance, M by saying that I have a memory of being in front of a red apple. It easily accounts for this fact because, according to the objective view, the memory being expressed by me is simply a memory of the fact that I cite in my report (that is, the fact that I was in front of a red apple). The conjunctive view also accounts for the fact that I would express M by saying that I have a memory of being in front of a red apple, but it does so less straightforwardly. It accounts for this fact because, according to the conjunctive view, the memory being expressed by me is partly a memory of the fact that I cite in my report. However, according to the conjunctive view, the memory is partly about something else as well, namely, my having seemed to perceive that I was in front of a red apple. For that reason, the advocate of the conjunctive view cannot take my report completely at face value when I express M by saying that I have a memory of being in front of a red apple. This kind of talk needs to be interpreted as shorthand for saying, more strictly, that I have a memory of being in front of a red apple and seeming to perceive it. Since, according to the conjunctive view, our memories partly represent past objective facts, the view can explain why we cite past objective facts in our reports of our

70  The Nature of Memory memories. But the involvement of past objective facts in the content of our memories is only partial. And, for that reason, the explanation offered by the conjunctive view is not as simple as, and therefore is less preferable than, that offered by the objective view. Take, now, the main virtue of the subjective view; the fact that it accounts for the innocence of memory. The subjective view has no trouble accounting for the fact that, in the scenario in which I form the belief that I was in front of a red apple in the past on the basis of M, and my belief turns out to be false because, as it happens, I misperceived a green apple as being red when I had P in the past, we would not blame my faculty of memory for my false belief. The subjective view accounts for this fact because, according to it, M is only providing me with the information that I had a perceptual experience of a red apple in the past. And this would remain true even if we found out that I misperceived the apple when I had P. On the subjective view, my memory was never meant to reach, as it were, beyond the confines of my own mind. It is not surprising, then, that we do not hold it responsible for a mistaken belief about the objective world. Unfortunately, this account of the innocence of memory is not available to the advocate of the conjunctive view. The reason is that, according to the conjunctive view, our memories are partly about past objective facts. Thus, according to the conjunctive view, if I misperceived the apple when I had P in the past, then my resulting memory M is also false. After all, the apple was not red, and M is partly about the apple being red. Does this mean, then, that the conjunctive view cannot account for the innocence of memory? Interestingly, it does not. The advocate of the conjunctive view can appeal to a distinction between two kinds of misremembering (a distinction introduced in ­chapter  2). A memory may be false because, even though the perceptual experience in which the memory originates is true, the memory does not match that experience. As a result, some of the information carried by the memory is not that provided by the past perceptual experience. We could call this “an error of transmission.” If M, for example, did not originate in a past perceptual experience of a red apple but in, let us say, a perception of a yellow banana in front of me, then M would be false as a result of an error of transmission. By contrast, a memory can be false because, even though the memory matches the perceptual experience in which it originates, the original perceptual experience was false to begin with. As a result, the information provided by the past perceptual experience about a purported objective fact does make it into the content of the memory, but the information is incorrect. We could call

The Intentionality of Memory  71 this the “transmission of an error.”12 As the advocate of the conjunctive view sees it, the scenario in which we discover that memory M originates in a false perceptual experience because, when I had P, I was actually looking at a green apple is precisely a scenario in which I have a false memory due to the transmission of an error. Now, the important point for the purposes of explaining our intuition about the innocence of memory is that, if the conjunctive view is right, then we should certainly hold our faculty of memory responsible for errors of transmission, but we should not hold our faculty of memory responsible for the transmission of errors. If M turns out to be false because M does not originate in P, but in a perception of a yellow banana, then surely my faculty of memory is to blame for the fact that M is false. But if M turns out to be false only because P was false to begin with, then it does not seem that my faculty of memory should be blamed for the fact that M is false. After all, on the conjunctive view, M is doing what, in part, it is supposed to do. It is preserving the information originally provided to me by the past perceptual experience in which the memory originates. Thus, according to the conjunctive view, if I form the belief that the apple was red in this scenario, I will have formed a false belief about a past objective fact, and my belief will have been formed on the basis of a false memory. But, since the relevant memory is false due to the transmission of an error, we should not blame my faculty of memory for the fact that the memory is false. And, for that reason, it makes sense that we do not blame my faculty of memory for the mistake that I have made in believing, on the basis of that memory, that the apple was red.13 There is, however, a difficulty for the view that O&S is the content of M. The difficulty concerns the following possible situation: W3: The apple is red, and it is located exactly in the same position with respect to me as it is in W0. I have P, but I quickly forget about the apple and I never have M.

If the content of M is O&S, then I should be representing W3 accurately when I have M in W0. After all, when I had P in W3, the apple was, and appeared to me 12 A similar distinction between two types of error in memory is drawn in (Von Leyden 1961). 13 Notice that this explanation of the innocence of memory is available to the advocate of the objective view, but it is not available to the advocate of the subjective view. Within the objective view, there is room for the transmission of errors, and there is room for errors of transmission. Within the subjective view, however, there is only room for errors of transmission.

72  The Nature of Memory to be, red. And yet, intuitively, it does not seem that I am correctly representing W3 when I have M in W0. Granted, I did have a perception of the red apple by having P in W3. But not just any perception of a red apple will do. It seems that part of what M informs me of is the fact that the perception that I remember having in the past is the very experience which has given rise to M. More generally, it seems that our memories inform us of the fact that they have originated in our remembered perceptions, as opposed to originating in instances of testimony or reasoning. In the case of M, for example, in virtue of having M in W0, I am provided with the information that M originates in my having perceived a red apple in the past. This explains why we have the intuition that, by having M in W0, I am not correctly representing W3. But the view that O&S is the content of M does not factor any facts about the causal history of M into its content. For that reason, it seems that the sort of content that this view attributes to memories fails to pass the RAI test as well. The attributed content is, yet again, too thin. It carries too little information.

3.7  The Causal Self-​Reference Approach The moral to draw from our discussion in the last three sections seems to be that our memories provide us with information about a number of different things. They provide us with information about past objective facts. They provide us with information about our own past perceptual experiences. And they provide us with information about their own connection to those facts and to those perceptual experiences. In this section, I will explore an approach to mnemonic content which tries to accommodate these three lessons about memory. We may call it the “causal self-​reference approach” to memory. The main tenet of the causal self-​reference approach is that memories represent their own causal history. Memories wear, in that sense, their causal origin on their sleeves. In order to convey this idea, the causal self-​reference approach attributes to memories a very thick content. To begin with, the content of a memory is, on this approach, token-​reflexive in that a memory represents itself, among other things. (More precisely, the memory exists in all the situations which belong to its content.)14 Also, a memory represents a past perceptual experience and an objective fact. And, finally, a memory 14 One might wonder whether the “causal” and “self-​reference” elements in the causal self-​ reference approach can be pulled apart. There is a view according to which a memory involves the feeling of a causal history (Perrin 2018). One might argue that if such a feeling exists, then both the

The Intentionality of Memory  73 represents a certain combination of relations involving the three mentioned elements, namely, the memory itself, a past perceptual experience, and an objective fact in the past. At least some of those relations are causal relations.15 This characterization of the causal self-​reference approach is rather minimal, since there are different ways of conceiving the connection between the three just-​mentioned elements of mnemonic content within the approach. As a result, slightly different views of mnemonic content will count as causally self-​referential views. In this section, I will distinguish four such views. The views differ from each other depending on the position that each view takes on two issues. What is the relation between the past objective fact and the past perceptual experience represented by a memory? And what is the relation between the past perceptual experience represented by a memory and the memory itself? There is, first of all, a causally self-​referential view of mnemonic content according to which a memory represents the fact that it has been caused by a perceptual experience of the subject which, in turn, has been caused by an objective fact. This view has been advocated, for example, by John Searle. Thus, Searle writes: The memory of seeing the flower represents both the visual experience and the flower and is self-​referential in the sense that, unless the memory was caused by the visual experience which in turn was caused by the presence of (and features of) the flower, I didn’t really remember seeing the flower.16

Two points in Searle’s remark are worth highlighting. Firstly, Searle talks here about a memory of seeing a flower, as opposed to a memory of the seeing of a flower (or a memory of a seeing of a flower). This suggests that, according to cause of the memory and the causal relation in which the memory stands to that cause are part of the content of memories (assuming, that is, that the phenomenology of memories depends on their intentionality in the way proposed in ­chapter 1). But, the thought continues, this does not mean that the memory itself is part of its content. Thus, one might argue, a memory can represent a causal history without being self-​referential. It seems to me, however, that, in order for a memory to represent its own causal history, it must represent what that history is a history of. 15 It may seem that the causal self-​reference approach to memory removes memories from the world, in the following sense. Since, on the causal self-​reference approach, memories represent themselves as having a certain causal history, it may appear as if, on this approach, memories were only about themselves, and not about the world. The idea that the content of a memory involves a certain combination of relations between the memory itself, a past perceptual experience, and an objective fact in the past is meant to address this worry. As we are about to see, however, some ways of conceiving this combination of relations are preferable to others. 16 (Searle 1983, 95).

74  The Nature of Memory Searle, the remembered perceptual experience is represented by the memory as a past perceptual experience of the subject who is having the memory. Thus, in order for the memory to accurately represent the perceptual experience of the flower, it is necessary that the remembered perceptual experience is a perceptual experience of the subject who is having the memory. Secondly, Searle makes reference to two causal relations. The first causal relation is the relation between the memory and the past perceptual experience. The second causal relation is the relation between the past perceptual experience and the presence (and features) of the flower. The second causal relation suggests that, according to Searle, the remembered perceptual experience is represented by the memory as a past perceptual experience “of ” the flower in a very specific sense, namely, it is represented as having been caused by the flower. This means that, in order for the memory to accurately represent the perceptual experience of the flower, it is not necessary for the remembered perceptual experience to have been veridical, as long as the perceptual experience has been caused by the flower. Searle’s remarks about memory suggest, then, a very different candidate for the content of memory M from those which we have considered thus far. We may call the proposition put forward by this causally self-​referential view “CSR1”: CSR1: {W: In W, M is caused by my having P, which in turn was caused by the presence of a red apple}

The view that the content of M is CSR1 conveys the main idea in the causal self-​reference approach; the idea that memories represent their own causal origin.17 After all, M has a certain causal origin in all of those situations which belong to CSR1, that is, the situations accurately represented by M. But this is not the only way in which the main idea in the causal self-​reference approach can be developed. There are at least three ways in which one could formulate a causally self-​referential view differently. One might propose, for example, that the remembered perceptual experience is represented by the memory as a past perceptual experience which 17 One might think that Searle’s view attributes contents to memories which are too complicated. After all, it seems too strong to require that I must be able to spell out something like CSR1 as the content of my memory in order for me to remember in virtue of having M. But notice that the claim that the content of memory M is CSR1 is not the claim that I would be able to produce CSR1 as the content of my memory. It is only the claim that CSR1 captures the truth-​conditions of my memory M.  Analogous considerations apply to other versions of the causal self-​reference approach to be considered below.

The Intentionality of Memory  75 may or may not have belonged to the subject of the memory. This variation from Searle’s version of the causal self-​reference approach would yield the following candidate for the content of M: CSR2: {W: In W, M is caused by P, which in turn was caused by the presence of a red apple}

Alternatively, one might propose that the remembered perceptual experience is represented by the memory as bearing some relation to an objective fact which differs from causation, such as a truth-​making relation. This variation from Searle’s version of the causal self-​reference approach would yield the following candidate for the content of M: CSR3:  {W: In W, M is caused by my having perceived a red apple through P}

Finally, one might propose both variations from Searle’s version of the causal self-​reference approach at the same time, which would yield the following candidate for the content of M: CSR4:  {W: In W, M is caused by a perception of a red apple through P}

All four versions of the causally self-​referential approach square with our intuitions on whether M accurately represents W1, W2, and W3 or not. For example, notice that, since P does not take place in W1, W1 does not belong to CSR1, to CSR2, to CSR3, or to CSR4. This means that, on all four versions of the causal self-​reference approach, M turns out not to represent W1 accurately, which seems to fit our intuitions. Likewise, since a red apple does not exist in W2, W2 does not belong to CSR1, to CSR2, to CSR3, or to CSR4. This means that, on all four versions of the causal self-​reference approach, M turns out not to represent W2 accurately, which seems to fit our intuitions too. And, finally, since M itself does not take place in W3, W3 does not belong to CSR1, to CSR2, to CSR3, or to CSR4. As a result, all four versions of the causal self-​ reference approach deliver the outcome that M does not represent W3 accurately. And this outcome seems to fit our intuitions as well. In retrospect, it was to be expected that the four candidates for the content of M within the causal self-​reference approach would accommodate our intuitions about those three possible situations. After all, each of the four candidates preserves the characteristic tenet of the causal self-​reference approach; the idea that

76  The Nature of Memory memories represent their own causal origin. And this idea was suggested, in the first place, by our intuitions about the veridicality of M with respect to, precisely, possible situations W1, W2, and W3. Are all four versions of the causal self-​reference approach equally plausible then? They do not seem to be. It seems that, even though all four versions do accommodate our intuitions about possible situations W1, W2, and W3, not all of them have the resources to accommodate our intuitions about two other possible situations. Consider, for example, the following situation: W4: The apple is red. In normal circumstances, red objects are perceptually experienced by subjects who look at them as being green, and vice versa. This, however, does not happen to me when I look at the apple. Instead, a temporary malfunction of my visual apparatus causes me to misperceive the red apple by having P. Years later, I have M as a result of having had P.

The possible situation that we are entertaining, then, is a situation which takes place against the background of an inverted-​spectrum type of world.18 Imagine that, in a possible world in which red objects normally appear to those subjects who look at them as being green, and vice versa, I look at the red apple and, oddly, I come to have P. As a result, the red apple appears to me to be red. Imagine, furthermore, that M originates in P in this situation. If the content of M is CSR1, then I should be representing this situation accurately when I have M in W0. After all, if Searle’s version of the causal self-​reference approach is correct, then what matters for the accuracy of M in W0 with respect to any possible situation is whether, in the situation concerned, the red apple in front of me has caused me to have a perceptual experience wherein the apple appears to me to be red, and whether the perceptual experience has caused me to have M. Which, by assumption, is the case in W4. And yet, it intuitively seems that, by having M in W0, I am not representing W4 accurately. What does this intuition suggest? It suggests that my memory M in W0 is not only about a perceptual experience which may or may not have been veridical in the past. It seems that memory is, so to speak, more opinionated than that. The way in which M informs me of P is not neutral on whether the remembered experience was veridical. M informs me of P as having been

18 For a discussion of the possibility of the inverted spectrum, see (Shoemaker 1982). The first formulation of an inverted-​spectrum scenario seems to be due to John Locke, in book 2, chapter xxxii, of the Essay (1975).

The Intentionality of Memory  77 veridical (whether it was veridical or not). The view that CSR1 is the content of M, however, does not factor any facts about the veridicality of my past perceptual experience into the content of my memory. For that reason, it seems that the sort of content that this view attributes to memories does not pass the RAI test. For the attributed content is too thin, or carries too little information. Analogous considerations apply to the view that CSR2 is the content of M. For all the possible situations which belong to CSR1 also belong to CSR2, which means that CSR2 is a less demanding content than CSR1. Thus, if our intuitions about the veridicality of M in W0 with respect to W4 indicate that CSR1 is too thin a content for M, then they also indicate that CSR2 is too thin a content for M. Let us turn our attention, now, to the two candidates for the content of M which include facts about the veridicality of my past perceptual experience in the content of my memory. In all the possible situations which belong to CSR3, and in all the possible situations which belong to CSR4, my past perceptual experience of the red apple P qualifies as a perception, and is therefore veridical. This means that, on both of these versions of the causal self-​reference approach, M in W0 turns out not to represent W4 accurately. Thus, it seems that both CSR3 and CSR4 square with our intuitions on whether, in W0, M represents W4 accurately or not. However, not both of these propositions seem equally good candidates for the content of M. To illustrate this point, consider the following possible situation: W5: The apple is red. In normal circumstances, perceptual experiences elicit memories in subjects who were not the bearers of those perceptual experiences originally. Someone other than me perceives the red apple by having P. Their having P causes me, and not them, to have M. I myself have never experienced the red apple perceptually.

The possible situation that we are entertaining, then, is a situation which takes place against the background of a world in which a subject’s memories are normally caused by the past perceptual experiences of other subjects. Imagine that, in that world, someone other than me has P when they perceive the red apple, and that perception leads me, and not them, to have M later. If the content of M is CSR4, then I should be representing this situation accurately when I have M in W0. After all, if this version of the causal self-​ reference approach is correct, then what matters for the accuracy of M in W0 with respect to any possible situation is whether, in the situation concerned,

78  The Nature of Memory the apple is perceived by someone as being red, and whether that fact has caused me to have M. Which, by assumption, is the case in W5. And yet, it intuitively seems that, by having M in W0, I am not representing W5 accurately. What does this intuition suggest? It suggests that my memory M in W0 is not only about a perceptual experience which may or may not have been mine in the past. It seems that memory is more opinionated than that in this respect as well. The way in which M informs me of P is not neutral on whether the remembered experience was mine. M informs me of P as having been mine (whether it was mine or not).19 The view that CSR4 is the content of M, however, does not factor any facts about the ownership of the original perceptual experience into the content of my memory. For that reason, it seems that the sort of content that this view attributes to memories does not pass the RAI test. For CSR4 turns out to be, like all the previous candidate propositions, too thin a content for M. By contrast, CSR3 rules out W5 as a situation accurately represented by M in W0. The reason is that, on that version of the causal self-​reference approach, what matters for the accuracy of M in W0 with respect to any possible situation is whether, in the situation concerned, the apple is perceived by me as being red when I have P, and whether that fact has caused me to have M. Which, by assumption, is not the case in W5. Thus, it seems that, even though both CSR3 and CSR4 are reasonably good candidates for the content of M, only the former accommodates our intuitions about the veridicality of M with respect to W5. Accordingly, the candidate that I wish to put forward as the content of M is proposition CSR3.

3.8  The Reflexive View My contention is that M represents the fact that M causally originates in my having perceived a certain objective fact in virtue of having P, namely, the presence of a red apple in front of me. Let us call this version of the causal 19 For an alternative defense of the view that remembered facts are presented to us as having been formerly experienced by ourselves, see (Rowlands 2016). The claim that, in memory, we are informed of past perceptual experiences, and of perceptions in particular, as having been our own has been the subject of some controversy. One might view the conceptual possibility of q-​memory (Shoemaker 1970) as being in conflict with this claim, for example. Interestingly, one might view the actual phenomena of disowned memory (Klein and Nichols 2012) and observer memory (Nigro and Neisser 1983) as providing empirical evidence against this claim too. Disowned memory will be discussed in some detail in ­chapter 5. The issues raised by q-​memory and observer memory will be discussed in ­chapter 6.

The Intentionality of Memory  79 self-​reference approach, the “reflexive” view. We can formulate the reflexive view more generally as follows: RV  Reflexive view For any subject S, memory M and proposition q: If S has M and S would express M by saying that they remember that q, then there is a perceptual experience P that S would express by saying that they perceive that q, such that the content of M is the proposition {W: In W, M is caused by S having perceived that q through P}

We have seen that, like other versions of the causal self-​reference approach to mnemonic content, the reflexive view accounts for our intuitions regarding the veridicality of M with respect to possible situations W1, W2, and W3. And we have seen that, unlike other versions of the approach, the reflexive view does not conflict with our intuitions regarding either W4 or W5. Let us now consider whether the reflexive view can preserve the main virtue of the objective view of mnemonic content as well as that of the subjective view. The main virtue of the objective view, let us recall, is that it accommodates the fact that we ordinarily express our memories by making reference to objective facts. The main virtue of the subjective view, by contrast, is that it accounts for the fact that we do not blame our faculty of memory for the false beliefs that we form on the basis of memories which originate in false perceptual experiences. The reflexive view can preserve the main virtue of the objective view. It can do it in an analogous way to that in which the conjunctive view of mnemonic content did. The reason why the conjunctive view accounted for the fact that I would express M, for example, by saying that I have a memory of being in front of a red apple was that, according to the conjunctive view, the memory that I would be expressing by making such a report is partly a memory of the objective fact that I cite in my report. Similarly, the reflexive view accounts for the fact that I would express M by saying that I have a memory of being in front of a red apple because, if the reflexive view is right, then my memory M is, in part, about the presence of a red apple in front of me. It makes sense, therefore, that I cite that objective fact in my report. As we saw in our discussion of the conjunctive view, however, this account of why we cite objective facts in our reports of our memories comes at a cost. According to the reflexive view, there is an involvement of past objective facts in the content of our memories, but their involvement is only partial. For that reason, on the reflexive view, we cannot take my report of M completely at face value

80  The Nature of Memory when I say that I have a memory of a red apple. Instead, my talk needs to be interpreted as shorthand for saying, more strictly, that the memory that I am having comes from my having perceived a red apple. The reflexive view can preserve the main virtue of the subjective view as well. It can do it, once again, in an analogous way to that in which the conjunctive view of mnemonic content did. How did the conjunctive view account for the fact that we do not blame our faculty of memory for the false beliefs that we form on the basis of memories which originate in false perceptual experiences? The advocate of the conjunctive view appealed to a distinction between two types of misremembering. A memory can be false as a result of the transmission of an error (when the memory originates in a false perceptual experience) or as a result of an error of transmission (when the memory carries information not provided by the perceptual experience in which it originates). The relevance of this distinction for the innocence of memory was that, on the conjunctive view, it makes sense to hold our faculty of memory responsible for errors of transmission, but it makes little sense to hold our faculty of memory responsible for the transmission of errors. An analogous strategy is available to the advocate of the reflexive view. Why is it that we do not blame my faculty of memory for a false belief that I have formed on the basis of a memory when that memory originates in a false perceptual experience? The reason, the advocate of the reflexive view will claim, is that, even though I am forming a belief on the basis of a false memory, the memory is false as a result of the transmission of an error. And, on the reflexive view too, it makes little sense to hold our faculty of memory responsible for the transmission of errors. After all, according to the reflexive view, if a memory happens to be false because the perceptual experience in which it originates was false in the past, then that memory has done what, at least in part, it is supposed to do. It has delivered the information originally provided to me by the past perceptual experience in which the memory originates. It is no wonder, then, that if M turns out to be false because P was originally false and, on the basis of M, I form the false belief that there was a red apple in front of me, then we do not blame my faculty of memory for my mistake. This is precisely what we should expect if the reflexive view is right.

3.9  Conclusion To recapitulate, it seems that the reflexive view has a number of important virtues. Firstly, the reflexive view accommodates the two thoughts about

The Intentionality of Memory  81 memory which motivated the objective view and the subjective view. The reflexive view accommodates the thought that memory puts us in cognitive contact with past objective facts because it attributes to memories a type of content which involves objective facts. More precisely, if the reflexive view is right, then, for each memory, there is an objective fact such that any possible situation accurately represented by the memory is a situation in which the fact takes place. (Every situation accurately represented by M, for example, is a situation in which there is a red apple in front of me.) The reflexive view also incorporates the thought that memory puts us in cognitive contact with our past perceptual experiences because it attributes to memories a type of content which involves our past perceptual experiences. More precisely, if the reflexive view is right, then, for each memory, there is a past perceptual experience of the subject such that any possible situation accurately represented by the memory is a situation in which the perceptual experience takes place.20 Furthermore, the reflexive view captures the thought about memory which pushed us toward the causal self-​reference approach and away from the conjunctive view of mnemonic content. This is the thought that memories represent their own causal origin. Finally, the reflexive view captures this thought in a manner which is consistent with our intuitions about the relation that holds between remembered perceptual experiences and remembered objective facts. And it captures it in a manner which is also consistent with our intuitions about the relation that holds between remembered perceptual experiences and the bearers of the relevant memories. Given these virtues of the reflexive view, it seems reasonable to conclude that we have reached the most plausible view of mnemonic content. The outcome of our discussion in this chapter is therefore that, when we remember, the memories that we have not only carry the information that some objective facts happened in the past, but they also carry information about themselves and how they are connected to those facts. What remains 20 This feature of the reflexive view accounts for a certain aspect of memory that it shares with perception, namely, its “particularity” (Teroni 2017). In memory, we are aware of particular scenes in the past just like, in perception, we are aware of particular scenes in the present. The reflexive view accommodates the particularity of memory by building specific perceptual experiences into the contents of our memories. By having my memory M, for example, I am not only remembering that I perceived a red apple in the past. I am also remembering a specific perceptual experience P through which I perceived the red apple. However, the fact that P is involved in the content of M does not mean that, in order for M to have the content that it has, P must have actually taken place in the past. It only means that every possible situation accurately represented by M is a situation in which I seemed to perceive a red apple by having P. This is consistent with the actual situation in which I am having M being a situation in which P did not take place in the past.

82  The Nature of Memory to be seen, however, is whether the view that memories carry such a great amount of information is of any explanatory value. Our project in the next four chapters is precisely to explore this question. In c­ hapters 4 and 5, we will consider whether the reflexive view of mnemonic content can shed some light on two features of what it is like for us to have memories. The project in ­chapters 6 and 7, by contrast, will be to determine whether the reflexive view of mnemonic content can also help us settle two interesting controversies in the epistemology of memory.

PART II

THE PHE NOME NOLO G Y OF M EMORY

4 The Experience of Time 4.1  Introduction Remembering is a rich type of mental state from a phenomenological point of view. There are many aspects of what it is like for a subject to remember a fact episodically.1 For that reason, we tend to think of our capacity to remember episodically not only as a capacity which is useful in our lives, but also as a capacity that enhances the quality of our lives. Remembering seems to involve a special form of consciousness; a form of consciousness without which our daily lives would be considerably impoverished. The goal of this chapter, however, is not to account for all of the phenomenological properties which make up the special form of consciousness involved in remembering. That would be a far too ambitious project. Instead, I will concentrate on two features of what it is like for a subject to remember a fact which seem to involve time. I will refer to them as the “feeling of pastness” and the “awareness of previous experience” in memory. The goal of the chapter is to propose an account of these two phenomenological features by appealing to the content that memories have. I will proceed as follows. In section 4.2, I will describe the feeling of pastness and the awareness of previous experience in memory, and I  will highlight the connection between them. The awareness of previous experience in memory will be addressed in sections 4.3 and 4.4. In section 4.3, I will discuss an attempt to account for the awareness of previous experience by making use of a different notion; the notion of mental time travel. I will argue that the account at issue fails because the conception of mental time travel employed

1 In this chapter too, we will concentrate, for the sake of simplicity, on memory for facts which involve visible objects. Accordingly, when we consider our awareness of our past perceptual experiences in memory, the focus will be on our past visual experiences. No substantial point in the discussion that follows, however, should hinge on the fact that we will be, for the most part, paying attention to cases involving vision. An earlier version of this chapter, not including the material on mental time travel, can be found in (Fernández 2008).

86  The Phenomenology of Memory conflicts with the fact that memories enjoy a feeling of pastness. A  different conception of the nature of mental time travel in memory will be proposed in section 4.4; a conception suggested by the reflexive view of mnemonic content. I will argue that the reflexive view, and the notion of mental time travel suggested by it, can help us explain why memories enjoy an awareness of previous experience. The feeling of pastness in memory will be addressed in sections 4.5–​4.7. In sections 4.5 and 4.6, I  will discuss what I  will call “temporal” views of mnemonic content. As we will see, temporal views incorporate the temporal positions of remembered facts into the contents of the relevant memories. I will explore the possibility of explaining the feeling of pastness by appealing to two such views. I  will argue that both temporal views attribute the wrong kind of content to memories and have trouble explaining the feeling of pastness. In section 4.7, I will argue that the reflexive view overcomes the difficulties of both temporal views and explains the feeling of pastness. The task of illuminating the connection between the awareness of previous experience and the feeling of pastness in memory will be addressed in section 4.8. My contention will be that the reflexive view can shed light on this connection as well. Thus, I will conclude that the reflexive view of mnemonic content defended in ­chapter 3 has significant explanatory power with regards to the phenomenology of memory.

4.2 Temporal Phenomenology Remembering has two interesting phenomenological features. Both of these features seem to involve, at least at first glance, an experience of time. However, what is being experienced as having a certain position in time is different in each case. Let us consider the two phenomenological features in order. Commonly, when a subject remembers some fact, the way in which the fact is presented to them in memory is not neutral on the temporal position of the remembered fact. The fact seems to be presented to the subject as happening in the past. Thus, when I remember my apartment being on fire, the fire does not appear to me as taking place now (as it would if I were perceiving it). It does not appear to me as taking place at no particular time either (as it would if I were imagining it). The fire appears to me as taking place in the past. The phenomenology of memory seems to involve, then,

The Experience of Time  87 some form of temporal awareness which concerns the objective world. We can try to formulate this idea in the following “feeling of pastness” principle (or “PAST,” for short):2 PAST  Feeling of pastness

For any subject S and proposition p: If S remembers that p, then S is aware of the fact that p as obtaining in the past. Furthermore, when a subject remembers some fact, ordinarily, the way in which the fact is presented to them in memory is not neutral on whether, or on how, the fact was perceptually experienced by them. The fact seems to be presented to the subject as having been perceptually experienced by them in a particular way. Its having been perceptually experienced, and how it was perceptually experienced, seems to be part of what the subject is being presented with in their memory of the fact. Thus, when I remember my apartment being on fire, the way in which the fire appears to me is not neutral on the issue of whether I perceptually experienced the fire, and it is not neutral on the issue of what it was like for me to experience the fire perceptually either. In virtue of remembering the fire, I seem to be aware not only of the fact that I perceptually experienced the fire in the past but also of what it was like for me to have such an experience. I am aware that it felt hot, that it looked bright, that it made a crackling noise, and so on. The phenomenology of memory seems to involve, therefore, an awareness of our own past experiences. We can express this idea in the following “awareness of previous experience” principle (or, simply, “APE”):3 APE  Awareness of previous experience

For any subject S and proposition p: If S remembers that p, then S is aware of what it was like for S to perceptually experience that p.

2 I am taking this term from (Russell 1921, 161–​162). References to the pastness of memory can also be found, for example, in (James 1890) and (Bergson 1911). 3 There are some positions in the psychology of memory which are in the vicinity of APE. Hermann Eddinghaus, for example, claimed that remembered facts are recognized as formerly experienced (1885, 1). Notice that APE is a stronger claim. APE is not only the claim that, when a subject remembers a fact, they are aware that they have previously experienced it. It is the claim that, when a subject remembers a fact, they are aware of a particular way in which they experienced the fact.

88  The Phenomenology of Memory The feeling of pastness and the awareness of previous experience are not independent features of remembering. In fact, it seems that the two forms of awareness involved in PAST and APE are closely related to each other. They seem to be bound together, in the following sense. If we try to focus our faculty of memory on some objective facts, then our resulting memories will make us aware of those facts as being in the past. But our memories will also make us aware of what it was like for us to have perceptual experiences of those facts. Conversely, if we try to focus our faculty of memory on our past perceptual experiences of some objective facts, then our resulting memories will make us aware of what it was like for us to have those perceptual experiences in the past. But they will also make us aware of the perceptually experienced facts as being, themselves, in the past. It seems, therefore, that the conscious experience of the past provided by memory somehow ties our awareness of our own past perceptual experiences together with our awareness of objective facts in the past. Both APE and PAST should play a constraining role in our theorizing about the intentionality of memory. It will be an important virtue of any theory about the nature of mnemonic content that it sits easily with the fact that episodically remembered facts are presented to us as obtaining in the past, and with the fact that they are presented to us as having been experienced in a particular way. It will also be an important virtue of any theory about the nature of mnemonic content that it can shed some light on the fact that memory connects our awareness of our own perceptual experiences with our awareness of objective facts in the past. For methodological purposes, this means that we are entitled to require from any proposal about the content of episodic memories that it needs to be consistent with PAST and APE, as well as with the connections between them. Furthermore, we can arbitrate between different proposals about the content of episodic memories, based on which have some prospects of accounting for PAST, for APE, and for the connections between the two, and which do not. In what follows, I will propose that the reflexive view of mnemonic content, offered in c­ hapter 3, can account for both the feeling of pastness and the awareness of previous experience in memory, as well as for the connection between them. To build a case for this proposal, I will highlight, first, a contrast between the account of the awareness of previous experience offered by the reflexive view and an alternative account suggested by an interesting notion in the psychology of memory; the notion of mental time travel. Then, I  will highlight a contrast between the account of the feeling of pastness

The Experience of Time  89 offered by the reflexive view and the accounts offered by two alternative views of mnemonic content. Both of these views build the temporal positions of remembered facts into the contents of our memories. My contention will be that, with regards to both APE and PAST, the account offered by the reflexive view sidesteps the difficulties of alternative accounts while preserving some of their virtues. Let us begin, then, by considering the awareness of previous experience in memory.

4.3  Mental Time Travel and Re-​Presentation There is a certain view about the phenomenology of episodic memory in the psychological literature which is very thought-​provoking. This is the view that a key characteristic of the phenomenology of episodic memory is that remembering involves the experience of mentally projecting oneself into the past; an experience that has come to be known as “mental time travel.”4 To be clear, the view that mental time travel is characteristic of memory is not the (wildly implausible) view that, in virtue of remembering a fact in one’s past, one actually travels back to the time at which one experienced the fact. The thought is rather that the phenomenology of remembering a fact in one’s past is, in some respect yet to be specified, similar to the experience of being transported to that moment in the past. When Endel Tulving, for example, characterizes episodic memory partly by reference to its phenomenology, this is precisely the feature of memory on which he focuses: It [episodic memory] makes possible mental time travel through subjective time—​past, present, and future. This mental time travel allows one, as an “owner” of episodic memory (“self ”), through the medium of autonoetic

4 The claim that the phenomenology of memory typically involves the experience of mental time travel is often associated with a larger picture about the nature of memory. According to this picture, memory is just a specific instance of a more general human capacity; the capacity to simulate (past, future, or possible) experiences. For two interesting developments of the simulationist picture, see (Michaelian 2016b) and (De Brigard 2014). The idea that the experience of mental time travel is a key characteristic of the phenomenology of episodic memory, however, does not commit us to the simulationist picture of memory. It is possible for memory to involve the experience of mental time travel and for it to remain a distinct capacity which can only be oriented toward the past. For the next two sections, I will explore the prospects of the idea that memory involves, in some sense, the experience of mental time travel while remaining neutral on whether the simulationist picture of memory is correct.

90  The Phenomenology of Memory awareness, to remember one’s own previous “thought-​about” experiences, as well as to “think about” one’s own possible future experiences.5

There is something very appealing about the idea that remembering feels like travelling back in time. It seems right to say that the experience of remembering is like that of being projected into the past. Notice, for example, that we care about preserving our episodic memories for emotionally significant facts in our past. When we come to lose such memories, this is an event which upsets us very deeply. If a time comes when one semantically remembers, for instance, that so-​and-​so was present at one’s wedding, that people were gathered inside such-​and-​such building, and that this or that music was playing in the background, but one no longer remembers any of those facts episodically, then one is likely to experience that change in one’s memory as a painful loss. And the reason why we care about preserving our episodic memories for emotionally significant facts in our past seems to be that once we lose those memories, we feel like we have been cut off from some important parts of our lives. Only episodic memories allow us to access, in some sense, those times in our lives.6 It seems natural to describe the relevant type of access by using metaphors such as mentally “travelling,” or being mentally “transported,” back to the time at which those facts in our past obtained. Thus, the idea that remembering involves an experience like that of time travel makes some sense of why we care about losing our episodic memories for emotionally significant facts in our lives. The notion of mental time travel seems to be, then, a helpful metaphor. But it is only a metaphor. What exactly is to travel, mentally, back to the time at which some fact in one’s past obtained? A reasonable answer to this question is offered by Tulving himself, while he glosses the definition of episodic memory mentioned earlier. Commenting on his use of the notion of mental time travel in that definition, Tulving makes the following remarks (my emphasis):

5 In (2005, 9). A reference to the experience of “travelling back into the past in one’s mind” can also be found in an earlier definition of episodic memory by Tulving in (1983, 1). Interestingly, references to the experience of mental time travel are missing from Tulving’s original definition of episodic memory in (1972, 385). 6 The point can also be made that, conversely, when a past fact in our lives is emotionally significant because it has been traumatic and painful, then it may be liberating for us to stop remembering it. But notice that the kind of memory of the traumatic fact that we care about losing in this case is episodic, and not semantic, memory. For it is episodic memory that allows us to access the relevant time in our lives.

The Experience of Time  91 First, episodic memory’s function is to enable mental time travel or remembering, that is, to make it possible for the rememberer to travel back in his or her mind to an earlier occasion or situation in the rememberer’s life, and to mentally relive the experienced and thought-​about happenings. Semantic memory cannot do so, at least not with a comparable efficacy.7

The way in which Tulving spells out the notion of mental time travel is very economical from a conceptual point of view. Mental time travel is not explained in terms of other notions which are, themselves, in need of explanation (such as that of subjective time, or that of autonoetic awareness, alluded to on p. 89). The concept of mental time travel is not explained by appealing to a complex array of several other notions either. Mentally travelling to the time at which one experienced a fact, Tulving seems to be telling us, is simply to re-​experience the fact.8 Another way of expressing the same idea would be the following. Suppose that, in the past, one had a certain experience and, in virtue of having that experience, some fact was presented to one. Suppose, now, that one remembers that fact. What does it mean to say that, by remembering the fact, one mentally travels back to the moment at which one had the original experience of the fact? It means that one is having the experience wherein the fact was presented to one all over again. The remembered fact is, in other words, being re-​presented to one. There seems to be nothing mysterious in the idea of undergoing, for a second time and in a different situation, an experience that one had in the past. The notion of mental time travel seems, therefore, simple enough if we construe it thus. And, once we adopt this reading of it, the view that the phenomenology of remembering a fact is that of mentally travelling back to the time at which one experienced the fact no longer seems obscure. It becomes the view that the phenomenology of remembering a fact is that of re-​experiencing the fact. Let us abbreviate this view about the phenomenology of remembering as the “re-​presentation view” of mental time travel. Can the re-​presentation view of mental time travel accommodate the awareness of previous experience in memory? The re-​presentation view of mental time travel is interesting for our purposes in this chapter because it does seem to have the potential to account 7 In (2005, 14). 8 The idea that mentally travelling to the time at which one experienced some fact amounts to re-​ experiencing the fact can be found throughout Tulving’s work. See, for example, (1983, 1), (2001, 21), (2002a, 6), and (2002b, 313).

92  The Phenomenology of Memory for APE. The advocate of the re-​presentation view can propose that the reason why, when we remember a fact, we are aware of what it was like for us to experience that fact in the past is the following: Remembering the fact feels like mentally travelling back to the time at which we experienced that fact. Mentally travelling back to the time at which we experienced the fact amounts to re-​experiencing the fact all over again. And, surely, when we re-​ experience some fact, we are aware of what it is like for us to undergo, in the present, our experience of that fact for the second time. But, presumably, what it is like for us to undergo, in the present, the experience of the fact for the second time is the same as what it was like for us to have, in the past, the experience of the fact for the first time. (After all, we are supposed to be having the very same experience all over again.) It is not surprising, then, that when, in the present, we remember a fact, we are aware of what it was like for us to experience that fact in the past. All we need in order to obtain such an awareness of the past is our current awareness of what it is like for us to remember the fact. It seems, therefore, that our awareness of our previous experience in memory is precisely what we should expect if the representation view of mental time travel is correct. On closer inspection, though, a difficulty emerges. It does not seem phenomenologically accurate to say that what it is like for us to undergo, in the present, our memory of some fact is the same as what it was like for us to experience the fact in the past.9 If I remember who was present at my wedding, for example, remembering that scene does not feel like a repetition of my past perceptual experience of the scene, for reasons that we have already encountered. My memory presents the scene as taking place in the past, whereas my past perceptual experience presented it as taking place in the present. My memory enjoys a feeling of pastness, which my past perceptual experience lacked on my wedding day. In short, the difficulty for the account of APE sketched above is that it conflicts with PAST. Where exactly did the re-​presentation view of mental time travel go wrong then? The fact that memories enjoy a feeling of pastness suggests that one of three claims put forward by the advocate of the re-​presentation view should be dropped: One might drop the claim that remembering a fact feels like mentally travelling back to the time at which we experienced the fact. If what PAST shows is that this claim is false, then remembering is nothing 9 This point is raised in (Matthen 2010, 8) and (Byrne 2010, 21).

The Experience of Time  93 like mental time travel, and the account of APE given earlier went wrong at its very beginning. Alternatively, one might drop the claim that mentally travelling back to the time at which we experienced a fact amounts to re-​ experiencing the fact all over again. If what PAST shows is that this claim is false, then we misunderstood what mental time travel is. However, it may still be true that, on a suitably reconstructed notion of mental time travel, memory is like mental time travel in some phenomenologically interesting respect. Finally, one might drop the claim that re-​experiencing some fact all over again requires the sameness of phenomenology on both occasions of experience. If this is where the account of APE went wrong, then we misunderstood what re-​experiencing some fact requires. However, it may still be true that, on a suitably reconstructed notion of re-​experiencing a fact, memory is like mental time travel in the sense that memory amounts to re-​ experiencing a past fact all over again. The first option seems to take us too far. If there is no phenomenologically interesting sense in which remembering is like mental time travel, then it is hard to see why, as noted above, we feel cut off from important parts of our lives when we lose our episodic memories for emotionally significant facts in those parts of our lives. The advocate of the re-​presentation view of mental time travel may have gone wrong in their account of APE, but they do seem to be highlighting, with the metaphor of mental time travel, a genuine phenomenological aspect of memory. By rejecting the idea that memory is like mental time travel on the grounds that the re-​presentation view of mental time travel conflicts with PAST, we make it quite difficult for us to capture this aspect of memory. The third option does not seem very promising either. Let us suppose that, in order for a subject to re-​experience a past fact, the subject does not need to have an experience which feels like their original experience of the fact. In what sense, then, does the subject re-​experience a fact when, in virtue of remembering it, they mentally travel back to the time at which they experienced the fact? It seems that, once we give up the idea that re-​experiencing a fact requires having an experience that feels like the subject’s original experience of the fact, Tulving’s helpful gloss of the notion of mental time travel as re-​experience becomes uninformative. For that reason, the third option does not seem to get us far enough. By rejecting the idea that re-​experiencing some fact all over again requires the sameness of phenomenology on both occasions of experience, we are not really relieving the pressure that PAST generates for the account of APE offered by the re-​presentation view of

94  The Phenomenology of Memory mental time travel. We are simply transferring that pressure to the claim that mentally travelling back to the time at which we experienced a fact amounts to re-​experiencing the fact all over again. In light of these difficulties, I  propose to pursue the second option sketched earlier, that is, to drop the claim that mental time travel to the moment at which the subject experienced a fact consists in re-​experiencing the fact. Attractive as it may seem, this notion of mental time travel is in tension with PAST. What we need, then, is an alternative way of spelling out what mental time travel consists in. The alternative notion of mental time travel must, on the one hand, account for the intuition that, by remembering some facts in our past episodically, we are hanging on to the relevant parts of our lives in a way in which remembering those facts semantically does not allow us to do. On the other hand, the alternative notion of mental time travel must be consistent with the observation that a fact is presented to us as happening in the past in virtue of undergoing the sort of mental time travel which is associated with memory. My suggestion is that the causal self-​reference approach to the content of memories, discussed in ­chapter 3, can provide us with such an alternative notion of mental time travel.

4.4  Mental Time Travel and Representation In our discussion of the causal self-​reference approach to the content of memories, I argued that one version of this approach had some advantages over alternative versions of it. According to the “reflexive” view of mnemonic content, if a subject has an episodic memory that they would express by saying that they remember some fact, then what their memory represents is that it causally originates in the subject’s perception of that fact. We can formulate the view more precisely as follows: RV  Reflexive view

For any subject S, memory M and proposition q: If S has M and S would express M by saying that they remember that q, then there is a perceptual experience P that S would express by saying that they perceive that q, such that the content of M is the proposition {W: In W, M is caused by S having perceived that q through P}

The Experience of Time  95 Let us recall, once more, the example that we used to illustrate the reflexive view. Suppose that I had a perceptual experience of a red apple on 7/​2/​1987 and, as a result, I am having, on 7/​2/​2018, a memory that I would express by saying that I remember that there was a red apple in front of me. Let us call this possible situation “W0,” and let us call my perceptual experience and my memory in W0, respectively, “P” and “M.” The reflexive view is the view that the content of M is captured by the following proposition: CSR3:  {W: In W, M is caused by my having perceived a red apple through P}

Notice that, on the reflexive view, the possible situations which belong to the content of my memory M are such that, in those situations, I perceptually experienced the red apple by having P. In this sense, my past perceptual experience P is part of what my memory M represents. The thought is that, by having M, I am not only presented with the fact that a red apple was in front of me, and I am not only presented with the fact that I seemed to perceive it either. In addition to those facts, I am presented with the particular experience that I had when I seemed to perceive the red apple in the past. Part of what my memory represents is that I seemed to perceive the red apple by having a specific perceptual experience; an experience with such-​and-​such qualitative properties. This feature of the reflexive view suggests a notion of mental time travel which is different from the notion offered by Tulving and discussed earlier. My proposal is that, in memory, mental time travel to the moment at which we experienced a fact in the past does not consist in re-​experiencing the fact. The reflexive view suggests that, in memory, mental time travel to the moment at which we experienced a fact in the past consists, instead, in representing the past experience of the fact. When, in memory, we mentally travel to the moment at which we experienced a past fact in our lives, we do not repeat our past experience of the fact. Instead, we step back from that experience, as it were. We represent that experience and some of its properties in our minds. Let us call this view, the “representation view” of mental time travel. What exactly is the relation between the representation view of mental time travel and the reflexive view of mnemonic content? The representation view is a view about the nature of mental time travel to the past. It is the view that mental time travel to the past consists in the representation of past perceptual experiences. And the view that memories represent past

96  The Phenomenology of Memory perceptual experiences is weaker than the reflexive view of mnemonic content. The reflexive view entails the view that memories represent past perceptual experiences, but it is not entailed by it. The subjective view of mnemonic content discussed in ­chapter 3, for example, is consistent with the view that memories represent past perceptual experiences, but it is not consistent with the reflexive view. Thus, the reflexive view about the content of memories leads us to the representation view of mental time travel straightforwardly, but the converse is not true. The representation view of mental time travel preserves an important virtue of the re-​presentation view. The re-​presentation view of mental time travel makes sense of the intuition that, by remembering some facts in our past episodically, we are hanging on to the relevant parts of our lives in a way in which remembering those facts semantically does not allow us to do. The intuition, in other words, is that we can somehow access the parts of our lives in which episodically remembered facts took place, whereas we feel cut off from those parts of our lives for which we only remember facts semantically. One way of bringing this intuition to light is by focusing, as we did earlier, on emotionally significant facts in our past; facts which are important for us to remember in an emotional sense. If one was given a choice between episodically remembering, let us say, the facts that took place during the day in which one’s child was born, and remembering those facts semantically, one would surely choose to remember those facts episodically. Semantically remembering that the birth happened during the early hours of the morning, or that it did not take a long time, or that so-​and-​so was present, will not do. If one merely preserves such beliefs about that significant day in one’s life but can no longer remember the day episodically, then one will feel that this part of one’s life is lost; lost in some important sense. The challenge, needless to say, is to specify what the relevant sense is with some clarity. The re-​presentation view of mental time travel was able to meet this challenge. If, in memory, we mentally travel back to the time at which we experienced some fact in our lives in that, by having a memory of the fact, we re-​experience the fact, then there is a clear sense in which episodically remembered facts are accessible to us and semantically remembered facts are not. Episodically remembered facts will be experienced by us. (In fact, according to the re-​presentation view, episodically remembered facts will be experienced by us in the same way in which they were originally experienced by us in the past.) By contrast, semantically remembered facts do not need to be experienced by us in any way. It is then no wonder, the advocate of

The Experience of Time  97 the re-​presentation view may argue, that we have the intuition that episodically remembered facts are accessible to us in a way in which semantically remembered facts are not. The representation view of mental time travel also accounts for the intuition that we feel cut off from those parts of our lives that we no longer remember episodically, although it accounts for this intuition differently. If the representation view is correct, then an important contrast between episodic and semantic memory is that, in virtue of remembering a fact episodically, we represent our past experience of the fact, whereas this is not the case when we remember the fact semantically. There is a clear sense, then, in which we can hang on to those parts of our lives which contain facts that we remember episodically, but we are cut off from those parts of our lives which only contain facts that we remember semantically. In the former case, in virtue of remembering the relevant facts, we can represent some experiences that we had in the past. Specifically, we can represent what it was like for us to have experiences of those facts. In the latter case, in virtue of remembering the relevant facts, we are not able to represent any experiences that we may have had of those facts in the past.10 It makes sense, therefore, that we have the impression that we are able hang on to those facts in our lives that we remember episodically, whereas we feel cut off from those facts in our lives that we can only remember semantically. The reason why we feel, in the episodic case only, that we can “revisit” a past fact by remembering it is that remembering a fact episodically allows us to contemplate what it was like for us to experience the fact in the past. Remembering the fact semantically, by contrast, does not give us that ability. Likewise, the reason why we feel, in the episodic case only, that we can “project” ourselves into the past by remembering a past fact is that we can entertain our past experience of that fact by remembering it episodically, but not by remembering it semantically. It is to be expected, then, that we have the impression that episodically remembered facts are, despite being in the past, accessible to us in a sense in which semantically remembered facts (including those which belong to our lives) are not. The representation view of mental time travel retains a further virtue of the re-​presentation view. The re-​presentation view was able to offer an 10 For this reason, the representation view of mental time travel suggests that, as far as the phenomenology of remembering is concerned, there is no difference between semantically remembered facts which are parts of our lives and semantically remembered facts which are not; a suggestion which seems correct.

98  The Phenomenology of Memory account of APE, and the representation view can offer one too. The account offered by the representation view flows from its approach to the intuition that episodically remembered facts are accessible to us in a way in which semantically remembered facts are not. Suppose that the representation view of mental time travel is right, and the reason why we sense that episodically remembered facts are accessible to us in a way in which semantically remembered facts are not is that our episodic memories represent, among other things, our perceptual experiences of past facts. Then, it does not seem surprising that, when we remember a fact episodically, we are aware of what it was like for us to experience the fact in the past. After all, what it was like for us to experience the fact in the past is part of what our memory represents when we remember that fact episodically. The reason why APE holds, then, is not that, by remembering the fact, we re-​run a past perceptual experience of the fact in our minds. The reason is that, by remembering the fact, we have a new experience; one which takes our past perceptual experience of the fact as part of its intentional object. Where does this leave us? We have seen two competing accounts of why remembering a fact episodically makes us aware of what it was like for us to experience the fact. Both accounts appeal to the idea that memory involves mental time travel, but this idea is understood differently in each account. One account of APE relies on a reading of the idea that memory involves mental time travel according to which remembering a fact involves re-​ running a past experience of it. The alternative account relies on a reading according to which remembering a fact involves representing a past experience of it. We have seen that the account of APE offered in terms of re-​running past experiences conflicts with the fact that memories enjoy a feeling of pastness. The question now arises as to whether the explanation offered in terms of representing past experiences also conflicts with the feeling of pastness. We will take a small detour to address this question. The account of APE in terms of mental time travel, understood as the representation of past experiences, was suggested by the reflexive view of mnemonic content. In sections 4.5 and 4.6, I will discuss two approaches to the feeling of pastness in memory which rely on alternative views of mnemonic content. And, in section 4.7, I will argue that the reflexive view offers an account of PAST which is preferable to both of those approaches. If the case for such an account of PAST can be made successfully, then it will turn out that the reflexive view can explain both APE and PAST. As the formulation of the

The Experience of Time  99 reflexive view seems to involve no contradiction, it will be reasonable to conclude, at that point, that there is no conflict between PAST, on the one hand, and the account of APE suggested by the reflexive view on the other hand. Let us turn, therefore, to the task of explaining the feeling of pastness in memory.

4.5  Temporal Locations and the Feeling of Pastness If our main goal is to explain PAST in terms of the content that memories have, then a strategy that suggests itself rather naturally is the following: We can try to build a reference to the temporal position of a remembered fact into the content of the relevant memory. The basic idea in this approach is that, when we have a memory that we would express with a claim of the form “I remember that p,” the fact that p is presented to us as having obtained within a certain period of time in the past. The proposal, then, is to construe the content of the memory as the proposition that contains those possible worlds where the fact that p obtains within that period of time. Let us call this general approach to mnemonic content, the “temporal approach.” The temporal approach comes in two main varieties, depending on how we think of the relevant period of time. A distinction in the psychology of memory may be useful to differentiate them. This is the distinction between temporal “locations” and temporal “distances.” A temporal location is a point on a conventional or natural time pattern whereas a temporal distance is the amount of time between some fact and the present.11 We may use this distinction to separate two groups of views which fall within the temporal approach. I will refer to them as “subject-​independent” views of memory and “subject-​dependent” views of memory. As a first approximation, we can say that subject-​independent views use temporal locations to specify the content of memories whereas subject-​dependent views use temporal distances for that purpose. I will discuss the former group of views in what remains of this section. The latter group of views will be discussed in the next section. According to subject-​independent views, when a subject remembers a fact episodically, the fact in question is presented to them as being the case at a certain point in time, independently of the time at which the subject has

11 See (Friedman 2001) for this distinction.

100  The Phenomenology of Memory their memory.12 More precisely, according to the subject-​independent theorist, the content of a memory that one would express by saying that one remembers a certain fact is determined by a particular condition about the time at which the fact obtains. The condition is that the fact must obtain within a certain period of time; a period of time which, as a matter of fact, is earlier than the time of remembrance. The subject-​independent view (or “IND,” for short) can then be formulated as follows:13 IND  Subject-​independent view

For any subject S, memory M and proposition p: If S has M and S would express M by saying that they remember that p, then there is a period of time T earlier than the time at which S has M such that the content of M is the proposition {W: It is the case that p within T in W}. Notice that, according to IND, the requirement that period of time T must be earlier than the time at which S has M is not part of the content of M. This is not, in other words, a feature of period T that the subject represents in virtue of having their memory. Nonetheless, it is necessary for T to be earlier than the time of remembrance in order for T to be involved in the content of M. The thought is that a memory represents a fact as obtaining at such-​and-​ such time, no matter what the time of remembrance is. According to IND, facts may be remembered as having been the case, for instance, at noon (as opposed to an hour ago), on Monday (as opposed to yesterday), or in 1987 (as opposed to thirty-​one years ago). The IND view, as characterized here, has several gaps. The advocate of IND will need to address, then, a number of questions. It is not clear, for instance, whether all the periods of time involved in the contents of our memories must have the same length or not. In addition, it is unclear how precise the boundaries of a period of time involved in the content of a memory should be. Finally, it is not clear whether the period of time involved in the content 12 This seems to have been the guiding idea behind so-​called time-​tagging theories of memory in psychology. Thus, in (Tzeng 1976), it was hypothesized that the outputs of some organic pacemaker might be associated with perceived events, which could encode temporal information for later retrieval in memory. On time-​tagging, see (Glenberg and Swanson 1986) as well. 13 For any memory, there are many possible subject-​ independent views about its content depending, among other factors, on which conventional or natural time pattern is used to specify the relevant period of time. In what follows, I will speak of “the” subject-​independent view for the sake of simplicity. However, IND is, strictly speaking, a template for generating subject-​independent views of mnemonic content.

The Experience of Time  101 of a memory whereby we claim to remember some fact can be determined by the temporal position of other remembered facts. It therefore seems that further work needs to be carried out within IND. Depending on the position that the IND advocate takes on the just-​mentioned issues, they will formulate slightly different, more refined versions of the subject-​independent view. However, I propose to set aside these gaps in IND for the sake of the argument. This will allow us to appreciate that IND faces some difficulties which are independent of how exactly those gaps are filled. The main difficulty for IND is that the view seems to attribute the wrong kind of content to memories. There seem to be memories and possible situations such that, intuitively, we would not count the situations as being accurately represented by the memories, and yet IND predicts that they are. Consider, once again, two possible situations discussed in c­ hapter 3. Take, first of all, possible situation W0, in which I have a perceptual experience P of a red apple in the past and, as a result, I am now having a memory M that I would express by saying that I remember that there was a red apple in front of me. Recall that, in W0, P occurs on 7/​2/​1987 and M occurs on 7/​2/​2018. Now consider W1, in which the apple is red, and it is located exactly in the same position with respect to me as it is in W0. I, however, am blind so I do not have P, and I do not have M. Furthermore, nobody else is in a position to see the red apple so no perceptual experience of the apple takes place in W1. Recall that, in W1, the apple is in front of me on 7/​2/​1987 as well. In ­chapter 3, I argued that M in W0 does not represent W1 correctly, which raised some difficulties for one particular view about the intentionality of memory, namely, the objective view. My contention in this section is that the subject-​ independent view faces a similar predicament with regards to W1. One reason why M in W0 does not seem to represent W1 correctly is that no fact about the presence of the apple in front of me in W1 corresponds to a certain piece of information that, intuitively enough, M seems to be carrying. This is the information that the remembered fact obtained in the past. Admittedly, in W1, the red apple is in front of me on 7/​2/​1987. And that day is certainly in the past in W0. But it is hard to see why that day should count as being in the past in W1. Clearly, no fact about 7/​2/​1987 makes that day (as opposed to a day, let us say, a week earlier) a past day. It seems reasonable to think that any day, in any possible situation, counts as a past day only relative to the current time in that situation. The current time in W0 is 7/​2/​2018; the time at which I have M. But it seems arbitrary to choose 7/​2/​2018 as the current time in W1. There does not seem to be,

102  The Phenomenology of Memory more generally, a principled point of reference relative to which 7/​2/​1987 counts as a past day in W1. (We could, in fact, stipulate that the current time in W1 is the time at which the red apple is in front of me, that is, 7/​2/​ 1987.) It seems, therefore, that nothing in W1 corresponds to a salient piece of information that M is carrying in W0, namely, that the fact that I am remembering happened in the past. To that extent, M in W0 does not seem to be representing W1 accurately. Consider, now, whether IND manages to accommodate this intuition. There are a variety of propositions which may count as the content of M consistently with IND. Thus, the IND theorist may put forward, as the content of M, propositions such as {W: A red apple is in front of me on 7/​2/​1987 in W}, {W: A red apple is in front of me during 1987 in W} or, perhaps, {W: A red apple is in front of me some time in the eighties in W}. The IND theorist needs to aim for a proposition which contains W0 since, intuitively enough, M is representing the situation in which it is taking place correctly. But, for the purposes of the present objection, it does not really matter how exactly the IND theorist chooses to delimit the period of time which determines the content of M. The trouble for IND is that if W0 belongs to the content that the view attributes to M, then W1 will belong to it too. Since the red apple is in front of me at the same time in W0 and W1, whatever period of time the IND theorist chooses to determine the content of M, that content cannot contain W0 without containing W1 as well. However, W1 is not, intuitively, a possible situation that M represents correctly. Therefore, IND does not attribute the right content to M. In the terminology introduced in ­chapter 3, the view fails to pass RAI. For it attributes a kind of content to memories which is too permissive. This is the reason why, in addition, IND does not seem to be able to account for PAST. What the intuition raised by the W1 case suggests is that there is more to representing a fact as obtaining in the past than representing its temporal position absolutely, that is, independently of the time of remembrance. There seems to be a relational aspect to the temporal phenomenology of memory. This is an aspect of the phenomenology of memory that the subject-​independent view does not seem to capture appropriately. The natural alternative to turn to, then, is a “subject-​dependent” view. This kind of temporal view builds the time of remembrance into the content of a memory. It seems reasonable, therefore, to expect that such a view may allow us to sidestep the difficulties that threaten IND. Perhaps, one might think, it will also allow us to shed some light on PAST. Let us therefore explore this avenue next.

The Experience of Time  103

4.6  Temporal Distances and the Feeling of Pastness According to subject-​dependent views, when a subject remembers a fact, the fact in question is presented to them as obtaining at a certain point in time relative to their own.14 More specifically, according to the subject-​ dependent theorist, what determines the content of a memory that one would express by saying that one remembers a certain fact is a condition about the time at which the fact obtains. The condition is that the fact must obtain a certain amount of time earlier than the time of remembrance. The subject-​dependent view (or “DEP,” for short) can then be formulated as follows.15 DEP  Subject-​dependent view

For any subject S, memory M and proposition p: If S has M and S would express M by saying that they remember that p, then there is a period of time T such that the content of M is the proposition {W: It is the case that p T-​earlier than M in W} Notice that, according to DEP, it is part of the condition which determines the content of a memory that the time at which the remembered fact obtains is earlier than the time at which the memory takes place. The idea is that a memory represents a fact as being separated from the instant of remembrance by a certain amount of time. Basically, the memory represents that the fact obtained this much or that much time ago, or earlier than now. Thus, according to DEP, remembered facts may be represented to have obtained, for instance, an hour ago (as opposed to at noon), yesterday (as opposed to on Monday), or thirty-​one years ago (as opposed to in 1987).

14 There are some intimations of this view in Edmund Husserl’s writings on memory. Thus, in (1964, 82), he writes (my emphasis): “I remember the lighted theatre of yesterday . . . . Accordingly, the theatre hovers before me in the representation as something actually present. I mean this, but at the same time I apprehend this present as lying back in reference to the actual present of perceptions now extant. . . . What is remembered appears as having been present, that is, immediately and intuitively. And it appears in such a way that a present intuitively appears which is at an interval from the present of the actual now.” 15 For any memory, there are many possible subject-​dependent views about its content depending, among other factors, on which conventional or natural time pattern is used to specify the relevant period of time. In what follows, I will speak of “the” subject-​dependent view for the sake of simplicity. However, DEP is, strictly speaking, a template for generating subject-​dependent views of mnemonic content.

104  The Phenomenology of Memory The advocate of DEP has some gaps to fill; gaps which are analogous to those in IND. The DEP theorist will need to specify, for instance, whether the periods of time involved in the contents of our memories must be precisely delimited or whether their limits can be vague. Similarly, the DEP theorist will need to specify whether those periods of time may be determined by the temporal position of other remembered facts. It is then quite clear that DEP is, at best, incomplete as it stands. Depending on the position that the DEP advocate takes on the just-​mentioned issues, they will formulate slightly different, more refined versions of the subject-​dependent view. Nonetheless, let us, once again, set aside these gaps in DEP for the sake of the argument. We will then be able to appreciate other difficulties for DEP; difficulties which are independent of how exactly those gaps are filled. Before we proceed, though, notice that DEP constitutes genuine progress over IND in two respects. First of all, DEP can avoid the difficulties that possible situation W1 raised for IND. According to DEP, M in situation W0 correctly represents W1 just in case the red apple is in front of me before I have M in W1. But, by assumption, I do not have M in W1. For that reason, if DEP is correct, M does not accurately represent W1. Thus, DEP squares with the intuition which conflicted with IND; the intuition that M in W0 does not represent W1 accurately. A related advantage of DEP over IND is that, unlike IND, DEP seems to have good prospects of accounting for PAST. If a memory is representing that some fact obtains within a certain period of time earlier than the time at which the memory takes place, then it does not seem surprising that the remembered fact is presented to us, in virtue of having that memory, as happening in the past. After all, it seems natural to think that the temporal region which counts as the past for a subject who has a memory is the temporal region earlier than the time of remembrance. Thus, if the content of their memory is, as the DEP theorist claims, that a fact obtained earlier than the time of remembrance (no matter how much earlier), then it seems that the remembered fact is represented to have obtained in the temporal region which qualifies as the past. So PAST seems to emerge as a quite natural phenomenon if the content of episodic memories is indeed the content that DEP attributes to them. DEP seems to enjoy, therefore, some advantages over IND. Unfortunately, it also seems that DEP fails to pass RAI, for reasons which are converse to those why IND failed. There seem to be memories and possible situations such that, intuitively, we would count the situations as being accurately represented by the memories, and yet the DEP theorist should claim that

The Experience of Time  105 they are not. Let us keep in mind possible situation W0, in which I had a perceptual experience P of a red apple on 7/​2/​1987 and, as a result, I  am having, on 7/​2/​2018, a memory M that I would express by saying that I remember that there was a red apple in front of me. Imagine a different possible situation now: W6: The red apple is in front of me on 7/​2/​1987, I  am looking at it, and I come to have P. Later, I travel back in time. In fact, I travel to a time before 1987. And, at some time earlier than 7/​2/​1987, I have M as a result of having P.16

Now, consider me having M in W0. Am I representing W6 correctly when I have that memory or not? Intuitively enough, when I have M in W0, W6 is one of the possible situations that I represent accurately. After all, in W6, the episode in which I perceive the red apple is a part of my life. It has not become a figment of my imagination just because I travelled back in time.17 However, W6 does not belong to the content that DEP attributes to M in W0. There are a number of propositions which may count as the content of M consistently with DEP. The DEP theorist could offer propositions such as {W: there is a red apple in front of me 31 years earlier than M in W}, {W: there is a red apple in front of me more than 20 years earlier than M in W} 16 It is sometimes argued that time travel is not logically possible (in which case, there is no such world as W6). The prima facie challenge for the logical possibility of time travel concerns the paradoxes of time travel discussed, for example, in (Lewis 1976). These paradoxes are sometimes taken to show that the very notion of time travel is incoherent. Chief among these paradoxes are the following two. If you could travel in time, you could kill your grandfather before, let us say, your father was conceived. But if you killed him at that point, you would not be born and, therefore, you would not be there to kill him in the first place. Thus, it seems that you could only succeed in the killing if you failed, which is absurd. Also, if you could travel in time, you could meet yourself at some earlier time of your life. But if you found yourself at some earlier time of your life, then you would be wholly in two places at the same time, which is equally absurd. Whether or not these paradoxes indeed reveal that time travel is logically impossible, it seems that the difficulties that they raise for us here can be largely avoided by considering a slightly modified time travel scenario, which we may call “recurrence.” Suppose that, in W7, I am given the option to re-​live a portion of my life starting some time before 1987. I can make it the case that the current time becomes, let us say, 1/​1/​1979. However, it will not be up to me to change anything in my life. I can only, so to speak, “rewind” time. So I just get to re-​experience the past events of my life all over again. In the recurrence scenario, I cannot meet myself in the past. Similarly, I cannot kill my grandfather. This allows us to sidestep the two time travel paradoxes just described. Nonetheless, it seems conceivable that, when I start to re-​live my childhood days before 1987, I could have experiences such as M. This is all we really need in order to make the point that W6 is meant to illustrate. For the sake of simplicity, though, I will ignore these complications and keep using the traditional variety of time travel presented in W6 (as opposed to the “recurrence” variety presented in W7). 17 In the terminology from (Lewis 1976), my perception of the apple is in my “personal past” even though it is not in the “objective past.”

106  The Phenomenology of Memory or simply {W: there is a red apple in front of me earlier than M in W}. The present objection does not rely on the exact period of time that the DEP theorist decides to use in order to specify the content of M. The difficulty for DEP is that, in W6, the red apple is not in front of me before the time at which I have M. The red apple is in front of me on 7/​2/​1987, which is, by assumption, later than the time at which I have M. Thus, the advocate of DEP is committed to the claim that I do not represent W6 accurately when I have M in W0, which seems counter-​intuitive. This case suggests that the subject-​dependent view fails to pass the RAI test as well. Like IND, DEP attributes the wrong kind of content to episodic memories. But, unlike IND, it attributes a type of content which is too strict. We seem to have exhausted the temporal approach. It seems that neither the subject-​dependent view nor the subject-​independent view can offer good candidates for mnemonic content, since they attribute a kind of content to memories which is either too strict or too permissive. A question that naturally arises at this point, then, is whether we might be able to design a view that incorporates those features of the subject-​dependent view which allowed it to shed some light on PAST while, at the same time, meeting the demands of the RAI test.

4.7  Beyond Time: The Experience of Origin We are seeking a view about mnemonic content which passes the RAI test and accounts for the feeling of pastness in memory. Our discussion so far suggests that the following might be a promising strategy: In order to deal with the W1 and W6 scenarios, as well as accommodating PAST, it may be a good idea to build the property of having been caused by a perceptual experience of the subject into the content of a memory. After all, the feature of IND which caused it to conflict with W1 was its disregard for any kind of relation that may exist between my perceptual experience of the apple and my memory. And the feature of DEP which caused it to conflict with W6 was its focus on a temporal relation between my perceptual experience of the apple and my memory, as opposed to a causal relation between them. These considerations seem to lead us to the causal self-​reference approach to the content of memories, since all four versions of this approach

The Experience of Time  107 discussed in ­chapter 3 precisely focus on the causal histories of memories; causal histories which involve past perceptual experiences of remembered facts. In section 4.4, we saw that one particular version of this approach could make sense of the appealing notion of mental time travel in a way which illuminated APE; the awareness of previous experience in memory. The view at issue was the reflexive view of mnemonic content, or RV: RV  Reflexive view

For any subject S, memory M and proposition q: If S has M and S would express M by saying that they remember that q, then there is a perceptual experience P that S would express by saying that they perceive that q, such that the content of M is the proposition {W: M is caused by S having perceived that q through P} Given that RV can shed some light on one of the main phenomenological features of memory, it seems reasonable to consider whether it might also be able to explain the other main feature, that is, the feeling of pastness. If RV can achieve this while overcoming the challenges of IND and DEP, then it will deliver a substantial account of the phenomenology of memory. Exploring RV further seems to be, therefore, a worthwhile strategy at this stage. Let us begin by noticing that RV does overcome the challenges of IND and DEP. Recall that the content that RV attributes to my memory M in W0 is the following: CSR3:  {W: In W, M is caused by my having perceived a red apple through P}

In ­chapter 3, we saw that RV squares with the intuition that I represent W1 inaccurately when I have M in W0. What RV requires of any possible situation W in order for W to be correctly represented by M in W0 is that, in W, M is caused by my perception of a red apple. In W1, I do not have a perceptual experience, let alone a perception, of a red apple. So RV, like DEP, can accommodate the intuition that M, in W0, does not correctly represent W1. This is a virtue that RV shares with DEP over IND. Unlike DEP, however, RV also squares with the intuition that I represent situation W6 accurately when I have M in W0. RV builds the causal history of a memory into its content.

108  The Phenomenology of Memory This allows RV to accommodate the intuition that time travel does not falsify the time traveler’s memories: What RV requires of any possible situation W in order for W to be correctly represented by M in W0 is that, in W, M is caused by my perception of a red apple. And, in W6, M is indeed caused by my perception of a red apple. Thus, RV can accommodate the intuition that I am correctly representing W6 when I have M in W0. What about the feeling of pastness? Does RV throw any light on why episodic memories enjoy that phenomenological feature? The reflexive view is different from both IND and DEP in that RV does not take the feeling of pastness associated with an episodic memory to be the experience of a temporal property of the fact that we claim to remember when we express the memory. Instead, RV construes the feeling of pastness as the experience of a different property; a property of the memory itself. This is the property of having been caused by a perception of the fact that we claim to remember. In other words, according to RV, the feeling of pastness is not an experience of time, but an experience of causal origin. If RV is correct, then, we do not experience temporal properties of past facts when those facts are presented to us in memory. Consider, once again, the example in which I had a perceptual experience P of a red apple on 7/​2/​1987 and, as a result, I am having, on 7/​2/​2018, a memory M that I would express by saying that I remember that there was a red apple in front of me. If RV is correct, the reason why I am experiencing a feeling of pastness when I have M is not that my perception of the red apple is in the past. The reason is that my memory M was caused by that perception. This is the property of M that I am experiencing when, in virtue of having M, I have a feeling of pastness. So why does it seem to us as if, in memory, we experienced temporal properties of past facts even though, in fact, we do not experience such properties? Suppose that RV is correct, and one experiences one’s memories as having been caused by perceptions of objective facts. Arguably, it is nomologically necessary that causes precede their effects. In other words, laws of nature guarantee that the time at which a cause happens is earlier than the time at which its effect happens.18 If this is correct, then having played a certain role in the causal history of the memory that one is having goes hand in 18 The advocate of RV does not need to endorse this claim. Even if it turned out to be nomologically possible that some effects precede their causes, it seems that, as a matter of fact (if not as a matter of natural law), causes precede their effects. For the purposes of explaining our inclination to treat the feeling of pastness as an experience of temporal properties of objective facts, this is all the RV advocate really needs.

The Experience of Time  109 hand with having a certain position in time, namely, being in the past. This explains, the RV advocate may argue, our inclination to identify the feeling of pastness with the experience of being in the past. In the red apple example, for instance, RV tells us that M represents the fact that it has been caused by my perception of the red apple in front of me. But there is a strong correlation between this property of M and a certain property of the fact that I was in front of a red apple, namely, the property of being in the past. So it is not surprising that we take the feeling of pastness associated with the memory to be the experience of the latter property even though, when we have that memory, we are actually experiencing the former property. Thus, RV can shed some light on the reasons why episodic memories enjoy the phenomenological feature that we have been calling the “feeling of pastness.” It can do it by re-​constructing what that feature of memories is really a feeling of.

4.8  Conclusion Let us take stock. In section 4.2, we highlighted two phenomenological features of memories; the awareness of previous experience and the feeling of pastness. We pointed out that it would be a virtue of a theory of mnemonic content that it could accommodate both the fact that episodically remembered facts are presented to us as obtaining in the past and the fact that they are presented to us as having been experienced in a particular way. In section 4.4, we discussed an account of the awareness of previous experience in memory which rests on the reflexive view of mnemonic content, and in section 4.7, we have discussed an account of the feeling of pastness which also rests on that view. It seems, therefore, that we have reached a view of mnemonic content which is explanatorily powerful with regards to the phenomenology of memory. However, there is an issue which remains to be addressed. In section 4.2, we also saw that it would be a virtue of a theory of mnemonic content that it could account for the fact that the consciousness of the past which is afforded by memory connects our awareness of our own perceptual experiences with our awareness of objective facts in the past. One might wonder, therefore, whether the reflexive view illuminates the connection between the feeling of pastness and the awareness of previous experience or not. The connection between the feeling of pastness and the awareness of previous experience in memory has two directions. The first direction of the

110  The Phenomenology of Memory connection consists in the following fact: If we aim our faculty of memory at some objective facts, then our resulting memories will make us aware of those facts as being in the past. But our memories will also make us aware of what it was like for us to experience the relevant facts perceptually. If the reflexive view of mnemonic content is right, then this fact makes sense. According to the reflexive view, the reason why, when we remember a fact, we are aware of it as being in the past is that the memory that we are having represents itself as originating in our perception of the fact. After all, the feeling of pastness is, on this view, the experience of the causal relation between the perception of the fact and our current memory. It seems, therefore, that episodic memory does not allow us to represent a fact in a way which makes us aware of it as being in the past without representing, in addition, our original perceptual experience of the fact. It does not seem surprising, therefore, that when we have a memory which carries a feeling of pastness, our memory carries an awareness of previous experience as well. The second direction of the connection between the feeling of pastness and the awareness of previous experience in memory consists in the following fact: If we aim our faculty of memory at some of our own past perceptual experiences, then our resulting memories will make us aware of what it was like for us to have those experiences. But our memories will also make us aware of the perceptually experienced facts as being themselves in the past. If the reflexive view of mnemonic content is correct, then this fact makes sense too. According to the reflexive view, remembering what it was like for us to have a perceptual experience of some fact amounts to being in a state the content of which is that we are in that state because we perceived the fact by having the experience at issue. It seems, therefore, that an episodic memory does not allow us to represent a perceptual experience of some fact in a way which makes us aware of what it was like for us to have the experience in the past without also representing that we are having our memory because the perceptual experience of that fact was veridical. But if our memory represents that we are having the memory because the perceptual experience of the fact was veridical, then our memory represents the relevant objective fact as being in the past. Thus, it is no wonder that, when we have a memory which carries an awareness of previous experience, our memory carries a feeling of pastness as well. The outcome of this chapter is that the view of mnemonic content defended in c­ hapter 3 has significant explanatory power in the area of the phenomenology of memory. For the reflexive view seems to be capable of illuminating

The Experience of Time  111 two significant aspects of the temporal phenomenology of memory. There is, however, an interesting dimension of the phenomenology of memory which, on the face of it, does not seem to concern time. Instead, it seems to concern, at least at first glance, the self. It would be interesting to know whether the explanatory power of the reflexive view of mnemonic content extends to this further aspect of the phenomenology of memory. Let us turn, therefore, to this issue now.

5 The Experience of Ownership 5.1  Introduction In ­chapter 4, we have discussed two aspects of what it is like for a subject to remember a fact episodically. These two aspects concerned the subject’s experience of time, and they were specific to memory. We saw that one characteristic phenomenal feature of episodic memories is that, when a subject remembers some fact episodically, the subject is aware of that fact as being in the past. And we saw that another characteristic phenomenal feature of episodic memories is that, when a subject remembers some fact episodically, the subject is aware of their own past experience of that fact. In this chapter, we will concentrate on a phenomenal feature of episodic memories which does not seem to involve the experience of time, and which memories seem to share with other types of phenomenal states.1 Normally, when a subject has an episodic memory of some fact, the subject is aware of the fact that the memory in question is theirs. The subject is not only aware of the fact that the memory is being instantiated, but they are also aware of the fact that they themselves are having that memory. We can abbreviate this idea by saying that the experience of having an episodic memory involves a “sense of mineness” or an “experience of ownership.”2 One reason for thinking that the experience of a memory as being one’s own is a genuine experience, over and above the experience of the memory as being instantiated, is the following. Typically, when a subject has an episodic memory, the subject takes that memory to be their own; the subject attributes the memory to themselves. If you episodically remember, 1 I will use the term “phenomenal state” to refer to mental states with phenomenal properties; states for which there is such a thing as what it is like for the subject to be in them. I will also assume that there is a phenomenology of agency, and that there is a phenomenology of thought. See (Bayne 2008) for a discussion of the former and (Bayne and Montague 2011) for a discussion of the latter. Accordingly, I will include actions and thoughts in the category of phenomenal states. 2 The idea that our phenomenal states carry with them a sense or mineness, or an experience of ownership, is discussed, for example, in (Gallagher 2005), (Zahavi 2008), and (Kriegel 2009).

The Experience of Ownership  113 for example, that it rained yesterday, then you will attribute to yourself the memory that it rained yesterday; you will take yourself to be the subject who is having that memory. The view that the experience of having an episodic memory involves a sense of mineness, over and above the experience of the memory as being instantiated, explains why our episodic memories lead to the self-​attribution of those memories. If episodic memories carry with them a sense of mineness, then it does not seem surprising that, when a subject has an episodic memory, the subject attributes the memory to themselves. After all, if memories carry with them a sense of mineness, then what the subject is doing by self-​attributing the memory is simply trusting the experience that they undergo when they have that memory. The objective of this chapter is to clarify the nature of this sense of mineness, or experience of ownership, for memories. The question that will concern us, then, is what qualifies as experiencing a memory as being one’s own. I will begin by highlighting, in section 5.2, a phenomenon that gives us a further reason for believing in the existence of the sense of mineness in episodic memory. This is the condition of “disowned memory,” wherein a subject claims to experience that they have memories which are not their own. In section 5.3, I will consider a proposal about the nature of the sense of mineness for episodic memories which is based on a particular diagnosis of disowned memory. According to the “identification model” of the sense of mineness for memory, the experience of a memory of some scene as being the subject’s own is the feeling of being identical with a past person; a person involved in the remembered scene. I will argue that, while this model does account for some details in the available reports of disowned memory, it is also in tension with other details in those reports. Accordingly, I will put forward, in section 5.4, an alternative proposal. According to the “endorsement model” of the sense of mineness for memory, the experience of a memory as being the subject’s own is the experience of the memory as matching the past. I will argue that this model squares with those details in the available reports of disowned memory which are in tension with the identification model, as well as with those which are explained by it. A potential objection will arise for the endorsement model at that stage; an objection that will be addressed in section 5.5. Finally, in section 5.6, I will suggest a template for generalizing the proposed model of the sense of mineness for memory to accommodate other cases of disowned phenomenal states, such as disowned thoughts, disowned impulses, disowned feelings, and disowned actions.

114  The Phenomenology of Memory

5.2 Disowned Memory An important consideration in favor of the idea that our experience of having episodic memories does involve a sense of mineness concerns a condition wherein the subject reports to feel that they have a memory while, at the same time, claiming that the memory in question does not feel like it is theirs. I will refer to this condition as “disowned memory.” Disowned memory is extremely rare. As a matter of fact, the only relatively clear case of disowned memory which is reported in the philosophical and psychological literatures seems to be that of patient R.B., a case investigated in detail by Stanley Klein.3 Shaun Nichols and Stanley Klein have argued, more specifically, that the case of patient R.B. has interesting philosophical implications for the connection between memory and personal identity.4 Patient R.B. suffers, due to head trauma sustained during a bicycle accident, various cognitive deficits including, it seems, a remarkable memory impairment. Patient R.B.  can have, we are told, accurate memories of scenes from his past. And yet, for some of those memories, he also claims that the memories at issue are not “his” and that he does not “own” them.5 Here are some of the claims that patient R.B. makes, and that Klein and Nichols take to be reports of episodic memories:6 Report 1 I was remembering scenes, not facts  .  .  .  I  was recalling scenes  .  .  .  that is . . . I could clearly recall a scene of me at the beach in New London with my 3 In (Klein 2015), (Klein 2013), and (Klein 2012). 4 In (Klein and Nichols 2012). 5 One might be worried that patient R.B. seems to be the only actual case of disowned memory. Furthermore, the information that we possess on this patient (who seems to have recovered from this condition) beyond his verbal reports is very limited. Some caution while leaning on this case is therefore warranted. Nonetheless, there is one consideration in favor of this chapter’s methodology; the methodology of relying on patient R.B.’s reports. The consideration at issue has to do with the structure of patient R.B.’s reports. This is a structure of the kind “I have such-​and-​such mental state, but it is not mine.” There are other mental disorders wherein states which differ from memories, states such as thoughts and actions, are disowned. These disorders are admittedly very different from disowned memory. (We will consider some of them in section 5.6.) But if we consider reports from patients who suffer these disorders, we will find that the reports through which they disown some of their thoughts, and some of their actions, have the same structure as that of the reports through which patient R.B. disowns some of his memories. This strikes me as sufficient reason for taking patient R.B.’s reports seriously, that is, for taking them as expressions of a robust feature of his experience. 6 For the sake of brevity, not all of R.B.’s reports cited in the literature are reproduced in this list. Reports 4 and 5, for example, are part of a longer exchange between Klein and patient R.B. in (Klein and Nichols 2012). As far as I can see, however, those reports which are not reproduced here are neutral on whether the correct interpretation of R.B. is that discussed in section 5.3, or it is the interpre­ tation to be proposed in section 5.4.

The Experience of Ownership  115 family as a child. But the feeling was that the scene was not my memory. As if I was looking at a photo of someone else’s vacation.7 Report 2 Things that were in the present, like my name, I continue to own. Having been to MIT had two different issues. My memories of having been at MIT I did not own. Those scenes of being at MIT were vivid, but they were not mine. But I owned “the fact that I had a degree from MIT.” That might have simply been a matter of rational acceptance of fact.8 Report 3 I can picture the scene perfectly clearly . . . studying with my friends in our study lounge. I can “relive” it in the sense of re-​running the experience of being there. But it has the feeling of imagining, [as if] re-​running an experience that my parents described from their college days. It did not feel like it was something that really had been a part of my life. Intellectually I suppose I never doubted that it was a part of my life. Perhaps because there was such continuity of memories that fit a pattern that lead up to the present time. But that in itself did not help change the feeling of ownership.9 Report 4 RB: I can see the scene in my head. I’m studying with friends in the lounge at my residence hall. I am able to re-​live it. I have a feeling . . . a sense of being there, at MIT, in the lounge. But it doesn’t feel like I own it. It’s like I’m imagining, re-​living the experience but it was described by someone else.10 Report 5 RB: I can recall memories [from the non-​ownership period of his life] at will. I have normal control over remembering facts and scenes from my past. But 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.11



7 (Klein and Nichols 2012, 686). 8 (Klein and Nichols 2012, 686). 9 (Klein and Nichols 2012, 686).

10 (Klein and Nichols 2012, 687). 11 (Klein and Nichols 2012, 687).

116  The Phenomenology of Memory Report 6 What happened over the coming months was interesting: every once in a while, I would suddenly think about something in my past and I would “own” it. That was indeed something “I” had done and experienced. Over time, one by one I  would come to “own” different memories. Eventually, after perhaps eight months or so, it seemed as if it was all owned. As if once enough individual memories were owned, it was all owned. For example, the MIT memory, the one in the lounge . . . I now own it. It’s clearly part of my life, my past.12 Report 7 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.13

It seems that R.B. is having a highly unusual experience. R.B. claims, on the one hand, to have certain memories and, on the other hand, not to own those memories. It is hard to know how to make sense of these reports. But if we try to take R.B.’s reports at face value, then it seems natural to attribute to him an awareness of the instantiation of certain memories without the awareness that those memories are his own. If this interpretative approach is correct, then R.B.’s case seems to illustrate that the awareness of an episodic memory as being the subject’s own is dissociable from the subject’s awareness of the occurrence of that memory. And if the two experiences are dissociable, then they are different experiences. Thus, it seems that the case of patient R.B. suggests that the sense of mineness is a genuine experience; an experience which, ordinarily, is part of the characteristic phenomenology of having episodic memories, even though it turns out to be separable from it. The phenomenon of disowned memory may give us a reason for thinking that there is such a thing as a feeling of mineness for episodic memories but, so far, it does not seem to shed much light on the nature of the feeling at issue. At this point, we do not have a firm grasp on what R.B. means when he claims not to “own” some of his memories, and that those memories are not “his.” One would like to know what feeling R.B. is supposed to be lacking when he experiences the episodic memories that he disowns. And, more generally,

12 (Klein 2013, 6).

13 (Klein 2015, 18).

The Experience of Ownership  117 one would like to know what it takes for a subject to experience a memory as being their own. Let us therefore turn to these questions now.

5.3  Ownership as Identification Stanley Klein and Shaun Nichols have put forward a proposal regarding the nature of the sense of mineness for episodic memory. What patient R.B. lacks, they tell us, is “a sense of numerical personal identity with the past person.”14 Since the main idea in this proposal concerns R.B.’s identity with a past person, let us abbreviate this view as the “identification model” of the sense of mineness for memory. Notice that Klein and Nichols’s proposal that R.B. lacks the sense of being identical with a person in the past can be understood in at least two ways. On one version of this proposal, what R.B. is trying to express when he claims not to own some of his memories is that he lacks the sense of being identical with the person who had the remembered perceptual experiences in the past. Let us call this view the “experiencer version” of the identification model. On a different version of this proposal, what R.B. is trying to express when he claims not to own some of his memories is that he lacks the sense of being identical with the person who was part of the remembered scenes in the past. Let us call this view the “object version” of the identification model. What are the virtues and shortcomings of each version of this model? Both versions of the identification model square with a number of details in R.B.’s reports. In report 1, for example, R.B.  claims to have felt that the remembered scene was “not his memory.” And, in report 2, R.B. claims to have experienced that the remembered scenes were “not his.” We can make sense of these claims if the experiencer version of the identification model is right and, when R.B. has the relevant memories, he does not feel that the person who, in the past, experienced the remembered scenes was him. The claims make sense, too, if the object version of the model is correct, and R.B. does not feel that he is the person who was part of the remembered scenes. Some details in R.B.’s reports sit more easily with the experiencer version of the identification model. In report 7, for example, R.B. claims that he feels as if the scene is “something he didn’t experience.” One would certainly expect claims of this type if the experiencer version of the identification model were

14 (Klein and Nichols 2012, 689).

118  The Phenomenology of Memory right, and R.B. did lack the sense of being the person who, in the past, experienced the remembered scenes. And, conversely, in those instances in which R.B. did enjoy the feeling of being the person who, in the past, experienced the remembered scenes, one would expect R.B. to claim, as he does in report 6, that he felt that the relevant scene “was something ‘I’ had done and experienced.” Other details in R.B.’s reports sit more easily with the object version of the identification model. In report 3, for example, R.B. claims that the relevant scene did not feel like it was something that really had been “a part of his life.” Along similar lines, R.B. claims, in report 5, that the scenes remembered did not feel as if they had “happened to him.” If R.B. did lack the feeling of being the person who was part of the remembered scenes, one would expect him to claim that he does not have the sense that the remembered scenes happened to him. And, in those instances in which R.B. did enjoy the feeling of being the person who was part of the remembered scenes, one would expect R.B. to claim, as he does in report 6, that the scene felt like it was part of his life. Thus, there seems to be some support for both versions of the identification model in R.B.’s reports. One can see, then, the motivation for reading R.B. as saying that he lacks the sense of being identical with a person in the past. However, the identification model of the sense of mineness faces some difficulties as well. For it seems to be in tension with a number of details in R.B.’s reports. Let us consider, first, the object version of the identification model. The view that R.B. does not have the sense of being identical with the person who was part of the remembered scenes does not sit easily with some references that R.B. makes to himself while describing the content of his memories. In report 1, for example, R.B. refers to the remembered scene as a scene “of me” at a certain beach, which does not seem to be neutral on who R.B. remembers to have been at that beach in the past. In that report, R.B. seems to be describing a memory which presents a scene that involves a beach in New London, R.B.’s family, and R.B. himself. This is perhaps clearer in report 2, where he refers to a memory of “having been” at MIT, and in report 3, where he refers to a memory of “studying with my friends.” This kind of talk does not seem to be neutral on who R.B. remembers to have been at MIT at the remembered time. It does not seem to be neutral on who R.B. remembers to have been studying with his friends at that time either. Notice that R.B. does not claim that he remembers “someone being at MIT,” and he does not claim that he remembers “someone studying with his friends.” And yet, one would expect him to use locutions of that kind if he did not feel that he was the person who was part of the scenes

The Experience of Ownership  119 to which he refers in reports 2 and 3. Similarly, in report 7, R.B. claims to remember “myself walking into the room.” In this case too, R.B.’s references to himself give the strong impression that R.B.’s memory is presenting him with some scenes which appear to R.B. to have involved him specifically. Let us consider, now, the experiencer version of the identification model. The view that R.B. lacks the sense of being identical with the person who experienced the remembered scenes does not square with some references that R.B. makes to a certain phenomenology which seems to be associated with his memories. In reports 3 and 4, for example, R.B. talks about “reliving” and “re-​running” a past experience in memory. This experience is described by R.B., in report 4, as involving “a sense of being there, at MIT, in the lounge.” It is hard to make sense of this talk if R.B. does not have the feeling that the person who he remembers to have experienced the scene at the MIT study lounge is R.B. himself. If R.B. lacks the sense of being the person who originally experienced the scene, then in what sense is R.B. reliving, or re-​running, that scene when he has the relevant memory? After all, that scene is, by assumption, not remembered by R.B. as having been experienced by him in the first place. An analogous worry applies to R.B.’s talk of having “a sense of being there.” If the experiencer version of the identification model is correct, then one would expect R.B. to claim, in order to describe his memory of the scene at the study lounge, that his memory conveys a sense of someone being there, not a sense of being there. R.B.’s talk of having a sense of being there when he remembers the scene mentioned in reports 3 and 4, and his talk of reliving and re-​running the experiences of that scene in memory, strongly suggest that R.B.’s memories do carry the feeling that he is the person who experienced the remembered scenes in the past. What we need, then, is a reading of R.B.’s reports which, on the one hand, makes sense of R.B.’s remarks about the reported memories not being his own while, on the other hand, accommodating both R.B.’s remarks about them being memories of himself and R.B.’s comments that the relevant memories involve an experience of reliving the remembered scenes. Let us turn, therefore, to a proposal which is aimed at satisfying those constraints.

5.4  Ownership as Endorsement It seems clear that at least some of R.B.’s memories do lack some salient phenomenal feature that episodic memories normally enjoy. It also seems clear

120  The Phenomenology of Memory that, due to that fact, R.B. feels, in some sense, estranged or alienated from those memories. Thus, if we manage to specify the form of estrangement that R.B. is experiencing, this case should provide us with some clues about the nature of the sense of mineness for episodic memories which is present in the non-​clinical population. The question, then, is what phenomenal feature is missing from some of R.B.’s memories; those memories which R.B. disowns. Usually, when we have an episodic memory of some fact, the memory appears to us as fitting, or matching, the past. Thus, if I have a memory of me giving a lecture while facing a theatre with some students in it, it will thereby seem to me as if the memory matches the fact that I was giving a lecture in a theatre with some students in it. Let us abbreviate the idea that a subject has the sense that the memory that they are having matches the past by saying that the subject “endorses” the memory. The suggestion that I wish to put forward is that R.B. does not endorse those memories to which he refers in reports 1–​5 and 7, but he endorses those to which he refers in report 6; hence his disowning the former and claiming to own the latter. By contrast, in normal circumstances, we are aware of our episodic memories as being our own in that we have the sense that those memories match the past. Let us abbreviate this view as the “endorsement model” of the sense of mineness for episodic memory. What reasons are there to think that the endorsement model of the sense of mineness is correct? In order to describe the experience of the memories that R.B. disowns, he uses two telling analogies; an analogy with the experience of imagination and an analogy with the experience of looking at a photograph. In reports 3 and 4, in which R.B. tries to describe what it is like for him to have a memory of the scene at the MIT study lounge, he compares his experience with the feeling of imagining a scene that is being described by someone else.15 Suppose that, when R.B. has a memory of the scene at the MIT study lounge, it does not feel to him as if the scene of him studying with his friends at that lounge really took place. Suppose that, by having an episodic memory of that scene, R.B. can picture being at the study lounge with his friends. And yet, when R.B. pictures that scene, the fact that he was there, studying with his friends, does not seem to R.B. to have actually happened. Then, it makes sense that he tries to express this experience by saying that having a memory of the scene at the study lounge feels like an episode of 15 I take it that this is also the experience to which R.B. is referring when, in report 7, he claims that the scene concerned feels like something he was told about by someone else.

The Experience of Ownership  121 imagination. For if R.B. was imagining the scene as described by someone else, then R.B.’s relevant episode of imagination would certainly not present the scene to R.B. as having been the case in the past. R.B.’s other analogy is revealing as well. In report 1, in which R.B. tries to describe what it is like for him to have a memory of the scene at the beach in New London, he compares his experience with that of looking at a photograph; a photograph of someone else’s vacation.16 Suppose that, when R.B. has a memory of the scene at the beach in New London, it does not feel to him as if the scene actually happened. Then, it makes sense that he tries to express this experience by saying that having a memory of the scene at the beach in New London feels like looking at a photograph. For if R.B. was looking at a photograph of the scene, the scene would not appear to R.B. as having really happened. To be sure, R.B. would be able to visualize the scene by looking at the photograph. But visualizing the scene in this way would not convey to R.B. the sense that the scene being visualized had in fact taken place in the past. The endorsement model of the sense of mineness, like the identification model, can account for why R.B. disowns some of his memories. Notice that if the endorsement model is right, then R.B. believes that he is having certain memories, but he does not have the feeling that the scenes represented in those memories happened in the past. And if he does not have the feeling that the scenes represented happened in the past, then he should not have the feeling that he is remembering those scenes.17 The suggestion, then, is that what R.B. is trying to express, by disowning some memories, is that he does not feel like he is remembering some scenes in spite of the fact that he thinks he is remembering those scenes. If, for example, R.B. does not feel like the scene of him and his family at the beach in New London really took place when he remembers it, then it seems natural for him to claim that the remembered scene “is not his memory.” For R.B. will not feel like he is remembering the scene at the beach. Similarly, if R.B. does not feel like the scene of him and his friends at the MIT study lounge did actually happen when he remembers the scene, then it seems natural for him to claim that he feels that the remembered scene is “not his.” After all, R.B. will not feel like he 16 In (Klein 2012, 493), R.B. is cited, instead, as saying “As if I am looking at a movie of someone else’s vacation” in that report. Similar considerations to those which follow will apply whether R.B. actually used an analogy with the experience of looking at a photograph or he used an analogy with the experience of looking at a movie. 17 Interestingly, Klein agrees that, when R.B. has an episodic memory that he disowns, having that memory is not experienced by him as an act of recollection (Klein 2015, 19).

122  The Phenomenology of Memory is remembering the scene at the study lounge, even though R.B. thinks that he is remembering it. It is also no wonder that R.B. claims, with regards to some of the scenes that he remembers, that they do not feel like they were something that had really been “a part of his life,” or that 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. The contrast that R.B. is drawing with those memories to which he refers in report 6 can also be accounted for if R.B. has managed to endorse the relevant memories. Suppose that what R.B. feels, for each of the memories that he claims to have come to “own,” is that the remembered action or experience did take place in the past. Then, it is not surprising that R.B. describes what it is like for him to have those memories by saying that, at that point, what he remembers seems to be something that he had indeed done and experienced; or by saying that what he remembers seems to have been part of his life, his past. Unlike the identification model of the sense of mineness, however, the endorsement model can accommodate R.B.’s references to himself while he describes his memories in reports 1–​3 and 7, as well as R.B.’s talk of “reliving” and “re-​running” a past experience in memory in reports 3 and 4. Suppose that R.B.’s memories represent him as having experienced certain scenes in the past even though, oddly enough, R.B. does not have the sense that the remembered scenes actually ever happened. Then, it makes sense that he refers to those memories as memories “of him” at a certain beach, memories of “having been at MIT” (as opposed to memories of “someone being at MIT”), memories of “studying with his friends” (as opposed to memories of “someone studying with his friends”) and memories “of himself ” walking into the room in reports 1, 2, 3, and 7 respectively. After all, R.B.’s memories do represent him as having experienced those scenes. It is just that, in virtue of having those memories, R.B. is aware of himself as having experienced those scenes in a similar way to that in which he would be aware of himself as having experienced those scenes if R.B. was imagining that he had experienced them in the past. Thus, the endorsement model allows us to take the references that R.B. makes to himself while describing the content of his memories in reports 1–​3 and 7 at face value. What about R.B.’s talk of “re-​running” and “reliving” the experience of being at the MIT study lounge with his friends, and his talk of having “a sense of being there” in reports 3 and 4? Suppose that, when R.B. has a memory of the scene at the study lounge, R.B.’s memory represents him as having

The Experience of Ownership  123 experienced that scene, even though R.B. does not have the sense that the scene actually ever happened. Then, the fact that R.B. himself is the person who experienced the scene will be part of what R.B. is aware of in virtue of having his memory. That is, R.B. will be aware of the scene as having been experienced by him. But if R.B. is aware of the scene as having been experienced by him, then it seems natural for R.B. to talk of having a sense of being there, and reliving the experience of being there, when he has a memory of the scene. After all, R.B. will then be presented with his own past perceptual experience of the scene in virtue of having his memory.18 It seems, therefore, that the reading of R.B.’s reports, according to which he does not experience some of his memories as matching the past, preserves the virtues of Klein and Nichols’s interpretation of those reports while, at the same time, sidestepping its difficulties.

5.5  The Appearance of Memory We have seen that the case of patient R.B. supports the view that the feeling of a memory as being one’s own and the sense that the memory is matching the past are one and the same experience. For the hypothesis that R.B. lacks the sense that some of his memories match the past accounts for a number of details about the way in which R.B. disowns those memories. There is, however, a salient detail in R.B.’s reports which requires explanation if R.B. is indeed lacking the sense that his disowned memories match the past. The relevant detail is that R.B. refers to his disowned memories as memories. He takes himself to be the bearer of memories and not of some other kind of mental state. Why is this particular detail significant as far as the endorsement model of memory ownership is concerned? Notice that, in the account of why R.B. disowns some of his memories offered earlier, I  have argued that, since R.B.  lacks the sense that some of his memories match the past, R.B.  is missing the feeling that he is remembering the scenes represented by those memories. Likewise, I have argued that, since R.B. lacks the sense that some of his memories match the past, R.B. is aware of himself as having experienced the scenes represented 18 In ­chapter 4, we discussed the phenomenology of mental time travel in a subject’s memory, and I proposed that this phenomenology is due to the representation of the subject’s own past perceptual experiences in memory. The line of argument here is that we should account for R.B.’s talk of “re-​ running” and “reliving” an experience in memory along the same lines.

124  The Phenomenology of Memory by those memories in a way which is similar to that in which he would be aware of himself as having experienced those scenes if R.B. was imagining that he had experienced them in the past. One might be concerned that, at first glance, those two claims are in conflict with the fact that patient R.B.  refers to his disowned memories as memories. If the endorsement model of memory ownership is correct, would it not be more natural for R.B. to refer to his disowned memories as episodes of imagination instead? This is a legitimate concern. The purpose of this section, then, is to address it by offering a conjecture about why R.B. may be referring to the memories that he disowns as memories. For most of this chapter, we have been focusing on the phenomenology of R.B.’s memories. But memories have, as we have discussed in previous chapters, features of other kinds. In particular, memories have a very particular kind of content. In c­ hapter 3, I put forward the “reflexive view” of mnemonic content, according to which the content of memories has the following structure: RV  Reflexive view

For any subject S, memory M and proposition q: If S has M and S would express M by saying that they remember that q, then there is a perceptual experience P that S would express by saying that they perceive that q, such that the content of M is the proposition {W: In W, M is caused by S having perceived that q through P} The main idea in the reflexive view of mnemonic content is that a memory represents itself as coming from a perception of a certain fact, namely, the fact that we claim to remember in virtue of having that memory. Now, it seems reasonable to assume that if a memory has a content of this kind, then, when a subject remembers a fact in virtue of having the memory, there are some features of the subject’s experience which correspond to the different elements in the content of their memory.19 Consider, for example, my memory of a red apple. In c­ hapter 4, we discussed the fact that, when I have 19 As discussed in ­chapter 1, this claim is consistent with different views about the relation between intentional and phenomenal properties of episodic memories. However, the current investigation of memory assumes a picture of the relation between intentional and phenomenal properties of memories according to which properties of the former type are responsible for properties of the latter type.

The Experience of Ownership  125 that memory, I am aware of the red apple as being in the past, and I am also aware of what it was like for me to perceptually experience the apple. We saw that those phenomenal features of my experience could be explained if the reflexive view of mnemonic content was correct. If what I represent, strictly speaking, in virtue of having my memory is the fact that my memory comes from a perception of the red apple, then it makes sense that I am aware of the apple as being in the past when I have that memory. For the awareness of the apple as being in the past can be construed as the awareness of the apple as being at the causal origin of my memory. And it also makes sense that I am aware of what it was like for me to perceptually experience the apple. For this phenomenal feature of my memory can be construed as the way in which I experience the representation of my past perceptual experience of the apple. Now, what does all of this have to do with patient R.B. and the experience of ownership? The thought is that perhaps a different component in the content of our memories may be responsible for our experience of ownership when we have those memories. A certain candidate suggests itself rather naturally given the proposal about the nature of this experience offered in section 5.4. If the feeling of a memory as being one’s own and the sense that it matches the past are one and the same experience, then it seems natural to think that the reason why, in normal circumstances, we have this experience when we have a memory is that the memory not only represents a past perceptual experience of the subject as having caused the memory, but it also represents that perceptual experience as having been veridical. It represents it as a perception. The suggestion, then, is that, when I have my memory of the red apple, I experience my memory as being my own because the feeling of mineness associated with my memory is identical with the sense that my memory matches the past. And the sense that my memory matches the past is the way in which I experience one of the things represented by my memory, namely, the fact that the memory has been caused by my having a perception (and not just any perceptual experience) of the red apple. This thought about the relation between the intentionality and the phenomenology of memory has an interesting implication for R.B.’s case. If patient R.B. lacks, as argued in section 5.4, the experience of his disowned memories as matching the past, then you would think that the content of R.B.’s disowned memories has been, in a certain sense, disrupted. Consider, for example, R.B.’s memory of the scene at the MIT study lounge. (Analogous considerations will apply to the rest of R.B.’s disowned memories.) Suppose

126  The Phenomenology of Memory that the reason why, in normal circumstances, recalling a scene of the study lounge brings with it the experience of the relevant memory as being one’s own is that the memory represents a perceptual experience of the remembered scene as being veridical; a perceptual experience at the causal origin of the memory. Then, it seems natural to think that, when R.B. recalls a scene of the study lounge, and yet lacks a feeling of mineness for that memory, what happens is that R.B. is representing the fact that the mental image that he is having causally originates in a past perceptual experience of the scene at the study lounge, but not necessarily a veridical experience. If our discussion of the feeling of pastness in ­chapter 4 is correct, then the hypothesis that the content of R.B.’s memory includes the causal origin of the memory explains why R.B. refers to the remembered scene as being a scene from the past. And if our discussion of the experience of mental time travel in c­ hapter 4 is correct, then the hypothesis that the content of R.B.’s memory includes a past perceptual experience of the scene at the MIT study lounge explains why R.B. claims to “re-​live,” or “re-​run,” the experience of being at the study lounge when he has his memory. What is missing, I suggest, in the content of R.B.’s memory is the fact that the remembered perceptual experience of the scene at the MIT study lounge was veridical. The content of R.B.’s memory is, in other words, neutral on that fact. How does this suggestion help us with our original worry? Our original worry was that if R.B.  is missing the sense that those memories which he disowns are matching the past, then it is hard to see why he refers to those memories as memories in the first place. The proposal is that the content of R.B.’s disowned memories may have been disrupted, and those memories may no longer be representing themselves as originating in past perceptions. This would square with the fact that R.B. refers to those memories as memories. For the content of R.B.’s disowned memories would be sufficiently similar to that of R.B.’s owned memories in order for R.B. to recognize them as memories. Specifically, the content of the memories that R.B. disowns would include their own causal origin, which would mark them as memories. This is, of course, not to say that memories could not possibly enjoy a type of content which is not causally self-​referential. As discussed in ­chapter 2, neither the content nor the phenomenology of memories is essential for those mental states to qualify as memories. (And, in fact, report 6 from patient R.B. gives us some reason to believe that changes in both the content and the phenomenology of his memories have taken place over time.) However, we do not need to assume that the content of memories is necessarily self-​referential in order

The Experience of Ownership  127 to address our original worry. We only need to assume that, normally, it is self-​ referential. Suppose that the content of R.B.’s disowned memories is neutral on whether those perceptual experiences which R.B. represents as being at the causal origin of his disowned memories match the past or not. Then, the content of R.B.’s disowned memories remains causally self-​referential. That is, R.B.’s disowned memories still represent their own causal origin. Given that this is the type of content that memories (and, in particular, R.B.’s owned memories) enjoy, it does not seem puzzling that patient R.B.  refers to the mental states that he is disowning as memories, as opposed to episodes of imagination. After all, the content of R.B.’s disowned memories is sufficiently dissimilar to that of episodes of imagination for R.B. to be able to tell them apart.

5.6  The Ownership of Phenomenal States In ­chapter 4, we discussed two phenomenal features of our memories which involved our experience of time, and which only memories seem to enjoy. The sense of mineness, by contrast, does not seem to be exclusive to memory. Phenomenal states of other types, such as thoughts and actions, usually carry with them a sense of mineness as well. Typically, if I have some thought, then I will experience that thought as being mine. Likewise, if I perform some action, then I will normally experience the action as being mine. It would be surprising, therefore, if the nature of the sense of mineness in our episodic memories was very different from that of the sense of mineness in our phenomenal states of other types. Thus, it seems that the advocate of the endorsement model of memory ownership owes us some story about how the model could be applied to phenomenal states of other types. Outlining the main plot of such a story is the purpose of this section. How could we investigate, first of all, the sense of mineness in phenomenal states of types other than memory? The condition of disowned memory provided us with some insight into the sense of mineness for our episodic memories. It seems worth considering, then, whether there are analogous pathological conditions in which the subject disowns some of their phenomenal states of other kinds; conditions that we could explore in order to illuminate the sense of mineness for mental states of those kinds. And, in fact, it does seem that we can find some disturbances of this sort in two kinds of mental disorders. Disturbances wherein the subject disowns some of their own phenomenal states can be found, on the one hand, in psychiatric

128  The Phenomenology of Memory disorders such as schizophrenia. And they can be found, on the other hand, in some neurological disorders which involve the subject’s inability to inhibit their own actions. Let us consider the two types of mental disorders in order. In certain delusions, subjects with schizophrenia seem to be able to report some of their phenomenal states while, at the same time, disowning those states.20 These are so-​called “passivity” symptoms such as the thought insertion delusion and delusions of “made” feelings, impulses, and actions. The following reports illustrate, respectively, delusions of thought insertion, delusions of made feelings, delusions of made impulses, and delusions of made actions: Report 8 As I walked along, I began to notice that the colors and shapes of everything around me were becoming very intense. And at some point, I began to realize that the houses I was passing were sending messages to me: Look closely. You are special. You are especially bad. Look closely and you shall find. There are many things you must see. See. See. I didn’t hear these words as literal sounds, as though the houses were talking and I were hearing them; instead, the words just came into my head—​ they were ideas I was having. Yet I instinctively knew they were not my ideas. They belonged to the houses, and the houses had put them in my head.21 Report 9 I cry, tears roll down my cheeks and I look unhappy, but inside I have a cold anger because they are using me in this way, and it is not me who is unhappy, but they are projecting unhappiness onto my brain. They project upon me laughter, for no reason, and you have no idea how terrible it is to laugh and look happy and know it is not you, but their emotions.22 Report 10 The sudden impulse came over me that I must do it. It was not my feeling, it came into me from the X-​ray department, that was why I was sent there for implants yesterday. It was nothing to do with me, they wanted it done. So I picked up the bottle and poured it in. It seemed all I could do.23 20 The diagnostic manual of the American Psychiatric Association DSM-​5 characterizes a delusion as “a fixed belief that is not amenable to change in light of conflicting evidence” (2013, 87). I will assume this conception of delusions for the purposes of the present discussion. 21 In (Saks 2007, 27). 22 In (Mellor 1970, 17). 23 In (Mellor 1970, 17).

The Experience of Ownership  129 Report 11 When I reach my hand for the comb it is my hand and arm which move, and my fingers pick up the pen, but I don’t control them. . . . I sit there watching them move, and they are quite independent, what they do is nothing to do with me. . . . I am just a puppet who is manipulated by cosmic strings. When the strings are pulled my body moves and I cannot prevent it.24

There is a certain analogy between, on the one hand, reports illustrating the four delusions above and, on the other hand, patient R.B.’s reports of disowned memory. Just like, in the disowned memory case, R.B. reports to remember certain things even though R.B. claims that the relevant memories are not his, in these delusions, patients claim to have certain thoughts, feelings and impulses which are not theirs. Similarly, they claim to be the proprietors of the bodies in which certain actions are taking place while, at the same time, rejecting the claim that they are the agents of those actions.25 The reason why this analogy is interesting concerns the issue of what methodology to adopt while interpreting reports 8–​11. We approached the case of disowned memory by assuming that a charitable way of interpreting R.B.’s disownment of some of his memories is by attributing to R.B. the lack of a sense of mineness in his awareness of those memories. Similarly, then, it seems that a charitable way of reading reports of patients with the thought insertion delusion and delusions of made feelings, impulses, and actions is by assuming that they, too, lack a sense of mineness in their awareness of the relevant states. It seems to be a natural way of reading report 8, in which the patient claims that a certain thought is not “their idea.” It seems to be a reasonable way of reading report 9 as well, in which the patient claims that “it is not them” who experiences a certain feeling. It appears to be a sensible way of reading report 10, in which the patient claims that the impulse was not “their feeling.” And it also seems to be a natural way of reading report 11, in which the patient claims that what happened in their disowned action is that “their bodies” (as opposed to the patient themselves) moved. But if the reason why patients with the thought insertion delusion and delusions of made feelings, impulses, and actions disown those states is that they are not aware of the relevant states as being theirs, then this diagnosis suggests that most of us, in 24 In (Mellor 1970, 18). 25 I do not mean to suggest that the sources of these conditions are also analogous. In R.B.’s case, for example, there is no suggestion of any form of mental illness being present. The analogy that I am drawing only concerns the structure of the patients’ reports.

130  The Phenomenology of Memory the non-​pathological condition, are indeed aware of our thoughts, feelings, impulses, and actions as being our own. It suggests, in other words, that the sense of mineness is a genuine feature of our awareness of our phenomenal states generally, and not only of our awareness of our episodic memories. Can we, then, generalize the endorsement model of the sense of mineness for memory to all of our phenomenal states? This is quite a complicated issue. It seems, on the one hand, that there are some reasons for thinking that the endorsement model of memory ownership can be generalized to the ownership of phenomenal states of other types. But it also seems, on the other hand, that such a generalization will require additional conceptual resources and may, in any case, suffer from some limitations of scope. The reason why it seems that the endorsement model can be generalized is that the notion of endorsement does not only apply to memories. It applies more generally, since endorsement is simply an experience wherein some phenomenal state is being presented to the subject as fitting, merited or appropriate. In the specific case of memories, the relevant type of fitting consists in matching the past. But a mental state can be fitting, or appropriate, without matching the past. Consider, for example, those phenomenal states which we regard as being subject to reasons, or grounds, that we have in our possession. Such mental states include, for example, thoughts and feelings. Suppose that I step out of the subway into the street, and I come to think that I am in the middle of a storm when I feel the heavy rain falling on me, and I hear the thundering and see the lightning in the sky. There is an intuitive sense in which I will find reasons for regarding my thought as being appropriate. I  will regard it as being appropriate in the sense of matching the current state of the world. And if I, furthermore, think that people can get struck by lightning during a storm and come to experience fear, then there is also an intuitive sense in which I will find reasons for regarding my fear as being appropriate. I will regard it as being appropriate in the sense of fitting a situation which warrants fear; a fearsome situation. Finally, if I think that I will be protected from lightning while being indoors, and I begin to run in the direction of a nearby building, or simply have the impulse to run in that direction, then there is an intuitive sense in which I will find reasons for regarding my action, or my impulse, as being appropriate as well. I will regard my action, or my impulse, as being appropriate in the sense of matching a goal which is worth pursuing for me. Thus, we can think of a subject as endorsing a thought when the subject finds reasons for regarding the content of that thought as being correct.

The Experience of Ownership  131 Likewise, we can think of a subject as endorsing a feeling when the subject finds reasons for regarding the feeling as being warranted by the circumstances. And, finally, we can think of a subject as endorsing either an impulse or an action when the subject finds reasons for regarding the goal of that impulse, or of that action, as being worth pursuing for them. For, in all of those cases, the subject will experience the relevant state as being appropriate in virtue of the fact that they find reasons, or grounds, for occupying that state. This broader notion of endorsement allows us to put forward a generalized version of the endorsement model of memory ownership. According to the generalized version of the model, a subject is aware of their phenomenal states as being their own just in case the subject has the experience of endorsing those states; the experience of finding them to be appropriate given the reasons for them in the subject’s possession.26 But what evidence is there to suggest that the experience of endorsing, in this sense, a phenomenal state is indeed the experience of the phenomenal state as being one’s own? The main reason is that the experience of endorsement seems to be the experience which has gone missing in those conditions that we have used to motivate the existence of a sense of mineness, namely, delusions of thought insertion and delusions of made feelings, impulses, and actions.27 The hypoth­esis that patients with the thought insertion delusion do not endorse their “inserted” thoughts accounts for some references that patients make to those thoughts as being representationally neutral. The patient in report 8, for example, refers to their inserted thought as an “idea.”28 The expression suggests that the thought is not being experienced as the type of mental state which needs to match the world. After all, entertaining an idea will not bring with it the feeling that there are reasons for regarding the content of that idea as being correct. It seems, then, that if thought insertion patients experience their inserted thoughts as being representationally neutral, they will not find reasons for regarding the content of those thoughts as being correct; they will 26 Since the relevant notion of endorsement does not apply to those mental states that we do not regard as being subject to reasons, the model cannot explain our experience of ownership for mental states such as sensations, dreams, and generalized emotions (that is, feeling happy as opposed to feeling happy about such-​and-​such thing). Whether this feature of the model is a limitation in scope or not will depend on whether there are, in fact, cases of disowned mental states of those types. 27 For a discussion of the idea that the experience of endorsement is missing in cases of disowned thoughts, see (Fernández 2010). The idea that it is missing in cases of disowned action is explored, for example, in (Graham and Stephens 2000). 28 Other patients with thought insertion refer to their inserted thoughts as “pictures” (Mellor 1970, 17) and “pieces of information” (Hoerl 2001, 190).

132  The Phenomenology of Memory not endorse them. Furthermore, the hypothesis that patients with delusions of made feelings do not endorse their disowned feelings squares with the fact that, in report 9, the patient claims to behave as if they felt happy “for no reason.” If the patient is not endorsing their feeling of happiness, then they are not regarding the feeling as warranted by the circumstances. And if they are not regarding the feeling as warranted by the circumstances, then you would expect them to claim that they find no reason for having the disowned feeling. Likewise, the hypothesis that patients with delusions of made impulses do not endorse their disowned impulses squares with the fact that, in report 10, the patient claims to have felt an impulse to pour the bottle because “they wanted it done.” If the patient is not endorsing the impulse to pour the bottle, then you would expect the patient to say that someone else, and not themselves, wanted the bottle to be poured. For, in that scenario, the patient would not find reasons of their own which could motivate their impulse. And, finally, the hypothesis that patients with delusions of made actions do not endorse their disowned actions accounts for the fact that, in report 11, the patient claims to experience a bodily movement that “has nothing to do with them.” If the patient is not endorsing the action to comb their hair, for example, then they are not regarding the state of affairs wherein their hair has been combed as a goal worth pursuing for them. But if they are not regarding that state of affairs as a goal worth pursuing for them, then it is no wonder that the patient says that the combing action has nothing to do with them. After all, in that scenario, the patient would not be able to produce any reasons which recommend the action of combing their hair. Let us now turn to a different type of mental disorder. The lack of the experience of ownership for action is also manifested in a neurological disorder known as “anarchic hand syndrome.”29 A subject with anarchic hand performs well-​executed movements apparently aimed at some specific goal with one of their hands, and the subject acknowledges that the hand in question belongs to them. However, a subject with anarchic hand syndrome typically claims that the relevant hand is performing its own actions.30 The subject is unable to inhibit the relevant movements (even if the movements are perceived by them to be unsafe), and they may try to interrupt them by 29 See (Della Sala et al. 1991). 30 Specifically, claims referring to the patient’s hand include “it will not do what I want it to do” (Goldberg et al. 1981), “it has a mind of its own” (Feinberg et al. 1992), and the claim that the hand “does what it wants to” (Giovannetti et al. 2005). For discussion of anarchic hand syndrome, see (Goldberg and Bloom 1990) as well.

The Experience of Ownership  133 stopping the “anarchic hand” with their other hand. The significance of anarchic hand syndrome for our discussion in this section lies in the fact that the generalized version of the endorsement model of ownership can account for a certain contrast between anarchic hand and a different disorder of action which also involves the inability to inhibit one’s own actions. “Utilization behavior” is a type of behavior constituted by actions of reaching out and automatically using objects in a manner which is well executed though contextually inappropriate. Putting on multiple pairs of sunglasses when those are repetitively placed in front of the subject is an example often mentioned in the literature.31 Anarchic hand syndrome and utilization behavior are different mental disorders; different in a number of ways.32 What is, then, the interesting contrast for our purposes here? It seems that patients who suffer anarchic hand syndrome do not endorse the actions that they disown. This is suggested by one of the defining features of anarchic hand syndrome. In the neurological literature, we are told that it is characteristic of anarchic hand syndrome that the actions performed by the “anarchic hand” are, according to the subject’s verbal reports, “unintended” or “unwanted.”33 Thus, it seems that patients who suffer anarchic hand syndrome are not aware of their intention, or their desire, to pursue the goal at which the hand’s action is aimed. This suggests that they do not find reasons for regarding that goal as worth pursuing for them. Interestingly, patients with utilization behavior are, by contrast, able to produce explanations for why they have performed those actions. It is natural to think that those explanations are, in actual fact, post hoc rationalizations. Nevertheless, the explanations appeal to the patients’ intentional states; states such as thoughts and desires. The patient’s explanation for putting on multiple pairs of sunglasses may be, for example, that they want to make sure that they are protecting their eyes from the sun. It seems reasonable to infer from those explanations that patients with utilization behavior do find reasons for regarding their actions as being aimed at goals which are worth pursuing for them. That is, patients with utilization behavior endorse the actions that, 31 For details on utilization behavior, see (Lhermitte 1983). 32 For a helpful discussion of the various ways in which the two disorders differ from each other, including the contrast to be highlighted later, see (Pacherie 2007). 33 These terms are used in (Della Sala et al. 1991, 1113) and (Della Sala 2009, 37) respectively. There are important differences between intention and desire. However, the point that patients with anarchic hand do not endorse their disowned actions can easily be formulated in terms of either intention or desire. I will therefore ignore the differences between the two notions for the purposes of this discussion.

134  The Phenomenology of Memory nonetheless, they are unable to control. The reason why this contrast is interesting for our present purposes is that, whereas patients with anarchic hand disown the actions performed by their “anarchic hand,” patients with utilization behavior do not disown their actions. If the generalized version of the endorsement model of ownership is correct, then this contrast between anarchic hand and utilization behavior is to be expected. It is no wonder that patients with anarchic hand, who do not endorse their actions, disown those actions, whereas patients with utilization behavior, who do endorse their actions, do not disown them. For if the generalized version of the endorsement model of ownership is correct, the experience of ownership and the experience of endorsement are one and the same experience. The extension of the endorsement model to the ownership of thoughts, impulses, feelings, and actions will, however, require further work, for reasons which are analogous to those discussed in section 5.5 with regards to the special case of memory. Notice that, while making the case that the experience of endorsement has gone missing in delusions of thought insertion and delusions of made feelings, impulses, and actions, I have been assuming that, nevertheless, the relevant patients are aware of their disowned thoughts as thoughts, they are aware of their disowned feelings as feelings, they are aware of their disowned impulses as impulses, and they are aware of their disowned actions as actions. After all, what they claim to have is mental states of those types (as opposed to, let us say, sensations or dreams). This means that, in spite of the fact that these patients lack the experience that the mental states that they are aware of having are appropriate, the patients still enjoy some kind of awareness of those states which has not been disturbed. For they seem to be able to identify the type of mental state that they are having correctly. What a complete generalization of the endorsement model of ownership requires, therefore, is an account of what this type of awareness amounts to. As it stands, the generalized account of ownership remains incomplete, even though there are reasons for thinking that the account can be developed in the direction of explaining our felt ownership of phenomenal states other than memories.

5.7  Conclusion Let us take stock. We started by noticing a certain phenomenal feature of memories. The feature at issue was the property of being such that, when

The Experience of Ownership  135 a subject has a memory, they experience the memory as being theirs, over and above experiencing the instantiation of the memory in them. We have been referring to this feature as a sense of mineness for memories. We considered a reason for believing that a subject who remembers has a sense of mineness for their memory which is different from their sense that the memory is taking place. The reason was that, in the condition that we referred to as disowned memory, the two experiences seemed to be dissociated from each other. Patient R.B. seemed to lack the former experience for some of his memories while retaining the latter one. To clarify the nature of the sense of mineness for memory, we focused on those reports from patient R.B. which seem to illustrate this dissociation in some detail. We considered the issue of what R.B. may mean when he disowns the relevant memories, and we ruled out a reading of those reports according to which R.B. lacks the sense of being identical with a past person; a person remembered in those memories. An alternative reading of patient R.B.’s reports was proposed. According to it, R.B. lacks the sense that the memories that he disowns are matching the past. A suggestion was made, furthermore, that the reason why R.B. still refers to his disowned memories as memories is that their content is sufficiently similar to that of typical memories for him to recognize them as memories. They do not represent themselves as coming from perceptions of past facts (which arguably accounts for why patient R.B. lacks the sense that they match the past), but they still represent themselves as coming from past perceptual experiences (which arguably accounts for why R.B. refers to them as memories). Finally, we have seen some reasons for thinking, more generally, that the distinction between experiencing that one is the bearer of some mental state and experiencing that the state is one’s own can be drawn for phenomenal states of other types, such as thoughts, feelings, impulses, and actions. And we have seen some reasons for being cautiously optimistic about the prospects of generalizing the view that a subject who disowns a memory is lacking the sense that the memory matches the past to disowned phenomenal states of those other types. Let us now step back from our discussion of the sense of mineness and, as it were, zoom out for a moment. We will then be able to appreciate an interesting point about the methodology used in both ­chapter 4 and the present chapter; a point briefly alluded to in section 5.5. The approach that we have adopted in this chapter in order to tackle the sense of mineness is analogous to that taken in c­ hapter 4 to account for the feeling of pastness and for the experience of mental time travel in memory. In both cases, we have

136  The Phenomenology of Memory considered a certain phenomenal feature of memories, and we have tried to identify some component in the content of memories which could explain why memories enjoy that phenomenal feature. In ­chapter 4, for example, the fact that memories represent past perceptual experiences was meant to explain why we have a sense of mentally travelling to some past scene when we remember that scene. Likewise, the fact that memories represent themselves as coming from those perceptual experiences was meant to explain why we have the sense that the experiences, as well as the relevant experienced scenes, took place in the past. And, in this chapter, it has been proposed that the fact that memories represent those perceptual experiences as being veridical explains why we have the sense that those memories match the past. What this approach reveals, then, about the overall structure of the proj­ ect, is that the reflexive view about the nature of mnemonic content offered in c­ hapter 3 is a fundamental part of the overall philosophical investigation of memory that we have been pursuing. In the last two chapters, we have seen that it is in virtue of that view about the intentionality of memory that it can explain several aspects of the phenomenology of memory. This outcome constitutes some progress toward a theory of episodic memory of the global kind envisaged in c­ hapter 1. To achieve further progress, it will be useful to consider whether the reflexive view of mnemonic content defended in ­chapter 3 can also deliver some interesting results in the area of the epistemology of memory. In c­ hapters 6 and 7, we will consider two controversies in this area; the debate on whether or not our memory judgments are immune to error through misidentification, and the debate on whether our memories only preserve epistemic justification over time or they can also generate it. We will examine whether the reflexive view can help us resolve either of the two controversies. Let us turn our attention, then, to the epistemology of memory.

PART III

T HE E PIST E MOLO G Y OF M EMORY

6 Immunity to Error through Misidentification 6.1  Introduction In ­chapters 2–​5, we have discussed the metaphysics of remembering, as well as the intentional and phenomenal properties of memories. But memories have, in addition to intentional and phenomenal properties, interesting epistemic properties. The judgments that we make on the basis of our memories (hereafter “memory judgments”) seem to enjoy a special kind of epistemic justification.1 In this chapter, I will explore a particular way in which the epistemic justification afforded by memories seems to be special.2 As a first approximation, we can try to formulate what the relevant aspect of this justification consists in by using the notion of liability to error. Normally, when we make a judgment on the basis of one of our mental states, and our judgment is justified, there is still room for our judgment to be wrong. After all, epistemic justification does not require truth. The issue that will concern us in this chapter is whether memory judgments are, nonetheless, not vulnerable to a particular kind of error. This feature of the special kind of epistemic justification afforded by memories has come to be known in the philosophical literature as “immunity to error through misidentification” (IEM).3 In this chapter, we will pursue the issue of whether memory judgments are IEM or not. The chapter has three parts. The introductory part, which is comprised of sections 6.1 and 6.2, is devoted to framing the project of the chapter. Specifically, in section 6.2, I  will identify the variety of IEM that will occupy us in this discussion, and I will motivate its significance. Sections 6.3 1 As pointed out in c­ hapter 1, a memory judgment will be construed as a type of memory belief, namely, an occurrent memory belief. 2 A  version of this chapter, not including the material on observer memory, can be found in (Fernández 2014). The material on observer memory can be found in (Fernández, forthcoming). 3 In what follows, I will use “IEM” to abbreviate both the noun “immunity to error through misidentification” and the adjective “immune to error through misidentification.”

140  The Epistemology of Memory and 6.4 make up what we may think of as the negative part of the chapter; a part which is devoted to discussing objections to the view that memory judgments are IEM (for short, the “IEM view”). Thus, in section 6.3, I will discuss a challenge to the IEM view raised by the possibility of so-​called “quasi-​memory,” and I will argue that such a possibility does not show that the IEM view is false. Similarly, in section 6.4, I will discuss a challenge to the IEM view based on the phenomenon of so-​called “observer memory.” I will argue that the occurrence of observer memories does not show that the IEM view is false either. The discussion of the two challenges to the IEM view posed by quasi-​ memory and observer memory will bring to light some ideas about memory, and about the IEM phenomenon, which will prove quite useful in the positive part of the chapter. The positive part of the chapter is comprised of sections 6.5–​6.8. Section 6.5 is devoted to a defense of the truth of the IEM view; a defense which will be grounded on the reflexive view of mnemonic content put forward in ­chapter 3. Sections 6.6 and 6.7, on the other hand, are devoted to an account of the truth of the IEM view. The account at issue will rest on two main ideas. One of them is a corollary of the reflexive view which concerns the relation between the intentionality of memory and that of perception. The other one is a view about the intentionality of perception which will arise from our discussion of observer memory. Finally, in section 6.8, a lesson regarding self-​awareness will be drawn from the fact that memory judgments are IEM by analogy with similar conclusions about the epistemic status of judgments based on introspection and proprioception.

6.2  Memory and IEM The notion of immunity to error through misidentification is introduced by Sydney Shoemaker in some discussions of self-​reference that mainly concern introspection. In those discussions, Shoemaker puts forward several formulations of IEM. In some of those formulations, IEM has to do with knowledge whereas, in others, it seems to have more to do with rationality.4 4 For formulations of IEM in terms of knowledge and rationality, see (1968, 557)  and (1996, 210)  respectively. According to Shoemaker (1968, 556), the latter is meant to capture a criterion used by Ludwig Wittgenstein to distinguish two uses of the first-​person pronoun (1958, 66–​67); a use “as subject” and a use “as object.” There may be other interesting notions of IEM to be found in Shoemaker’s work. Thus, James Pryor’s (1999) distinguishes two varieties of IEM involving the

Immunity to Error through Misidentification  141 Interestingly, though, in some discussions of memory, Shoemaker introduces IEM in yet a different way; a way which appeals to neither the notion of knowledge nor that of rationality. The following two passages illustrate the type of IEM which concerns Shoemaker when it comes to memory: Consider a case in which I say, on the basis of my memory of a past incident, “I shouted that Johnson should be impeached,” and compare this with a case in which I say, again on the basis of my memory of a past incident, “John shouted that Johnson should be impeached.” In the latter case it could turn out that I do remember someone who looked and sounded just like John shouting that Johnson should be impeached, but that the man who shouted this was nevertheless not John—​it may be that I misidentified the person as John at the time I observed the incident . . . But this sort of misidentification is not possible in the former case. My memory report could of course be mistaken, for one can misremember such incidents, but it could not be the case that I have a full and accurate memory of the past incident but am mistaken in thinking that the person I remember shouting was myself. I shall speak of such memory judgments as being immune to error through misidentification with respect to the first person pronouns, or other “self-​referring” expressions, contained in them.5 Thus if I claim on the strength of memory that I saw John yesterday, and have a full and accurate memory of the incident, it cannot be the case that I remember someone seeing John but have misidentified that person as myself; my memory claim “I saw John” is subject to error through misidentification with respect to the term “John” (for it could have been John’s twin or double that I saw), but not with respect to “I.”6

In this discussion of memory, the notion of IEM is spelled out in terms of truth or, more precisely, truth-​conditions for memory. The issue that seems to concern Shoemaker in the two passages above is the extent to which an accurate memory may leave room for error when a subject judges, on the basis of that memory, that they had a certain experience or performed a certain action in the past.7 Is it possible for the subject to be wrong in notion of epistemic justification which, Pryor argues, can be attributed to Shoemaker; “immunity to which-​object-​misidentification” and “immunity to de re misidentification.” 5 In (1970, 269–​270). 6 In (1970, 270). 7 As we have seen in previous chapters, we seem to have memories which concern our own past propositional attitudes, our experiences, our actions, and our physical states (states such as being at

142  The Epistemology of Memory claiming that they had the relevant property because the person who they correctly remember to have had the property is someone else, and they have misidentified that person as themselves? If this is possible, Shoemaker claims, the subject’s judgment that they once had the relevant property is not IEM relative to their memory. Otherwise, their judgment is IEM relative to it.8 More generally, we may formulate this “truth version” of IEM (or “IEMτ,” for short) as follows: IEMτ  For any property P and grounds G:



If I judge that I have P on the basis of G, then that judgment is IEMτ relative to G just in case it is impossible that there is a subject S such that: i. G represents S as having P. ii. G is fully accurate. iii. I mistakenly think that I am identical with S. iv. My judgment that I have P is false as a result of (iii).

What is meant by “impossible” in this formulation of IEM? In discussions of IEM concerning introspection and proprioception, Shoemaker distinguishes two varieties of IEM. Judgments made on the basis of proprioception are meant to be IEM because, given one’s biological constitution, one’s proprioceptive experiences will always provide one with information about one’s own body. This fact about one’s constitution rules out, as a matter of fact, the possibility that one mistakenly thinks of a different person (a person whose body one is experiencing through proprioception) as being oneself. But it does not rule it out in principle, since it remains logically possible for one’s brain to be appropriately connected to someone else’s body in such a way that proprioception provides one with information about that body. By contrast, judgments made on the basis of introspection are supposed to be IEM in a a certain location, or moving in a certain direction). For the purposes of this discussion, it will be helpful to use an umbrella term which refers to states of all of those kinds. I will be referring to them as “properties.” 8 Notice that the question which concerns IEM is different from that of whether it is possible for the subject to be wrong in claiming that they remember having had the relevant property because, even though someone correctly remembers that the subject had the relevant property, the subject has misidentified the person who is remembering as themselves. The question which concerns IEM is also different from that of whether it is possible for the subject to be wrong in claiming that they had the relevant property because, even though the person who they remember having had the relevant property is themselves, the subject’s memory of having that property is incorrect. As far as I can see, the advocate of the IEM view can answer both of these questions in the affirmative.

Immunity to Error through Misidentification  143 stronger sense. It is meant to be logically impossible for one to mistakenly think of a different person, a person whose mental states one is introspecting, as being oneself. Thus, Shoemaker writes: In being aware that one feels pain one is, tautologically, aware not simply that the attribute feels pain is instantiated, but that it is instantiated in oneself.9

Following Shoemaker, we may refer to the former type of IEM as “de facto” IEM and to the latter type as “logical” IEM. This distinction yields two varieties of IEMτ depending on how one chooses to specify the modal force of “impossible” in the formulation offered above: If the circumstances in the actual world guarantee that I will never make the mistake specified in (i–​iv) when I judge that I have property P on the basis of grounds G, then my judgment that I have P is de facto IEMτ relative to G. By contrast, if there is no possible world in which I  make that mistake, then my judgment is logically IEMτ relative to G. In this chapter, I will concentrate on the question of whether memory judgments are logically IEMτ relative to the memories on the basis of which they are made. In what follows, therefore, talk of IEM in memory, and talk of whether memory judgments are IEM, should be understood as referring, respectively, to logical IEMτ in memory and to whether memory judgments are logically IEMτ. The philosophical interest of IEM with regards to memory concerns the topic of self-​awareness. If memory judgments are IEM, then that suggests that there is some sense in which, in memory, one is aware of the subject who is remembered to have instantiated such-​and-​such properties as being oneself.10 In that case, it seems important to examine what that form of self-​ awareness consists in.11 By contrast, if memory judgments are not IEM, then that suggests that, in memory, there are some criteria that one is using to infer that one is the subject who is remembered to have instantiated such-​and-​ such properties. In that case, it seems important to determine what kind of 9 In (1968, 563–​564). 10 Notice that, even though this is true of logical IEMτ with regards to memory, the same cannot be said about de facto IEMτ: The fact that one’s memories will always originate, as a matter of fact, in properties that one had in the past seems to suffice for one’s memory judgments to qualify as being de facto IEMτ. But that fact is also consistent with the possibility that, in memory, one is only aware that certain properties were instantiated in the past (as opposed to being aware of who, in the past, was the bearer of those properties). 11 This seems to be, for example, part of the project in which Gareth Evans is engaged in his (1982, 205–​267) discussion of self-​identification.

144  The Epistemology of Memory “identification criteria” one is using to perform such inferences, since criteria of that sort could arguably make up a concept of oneself.12 Let us consider, then, what type of approach we should pursue in order to determine whether memory judgments are IEM.

6.3  Quasi-​Memory and IEM The debate on whether memory judgments are IEM has concentrated on the notion of “quasi-​memory,” or “q-​memory.” In this section, I will argue that, ultimately, this is not a productive approach. However, the reason why the appeal to the notion of q-​memory is fruitless will turn out to be illuminating. Let us turn, therefore, to the role of q-​memory in discussions of IEM. A q-​memory is a mental state which is intended to be phenomenologically indistinguishable from a memory. In that sense, it is an “apparent memory” or, equivalently, a state wherein it appears to the subject that they are remembering something, whether or not they actually are. A q-​memory is also supposed to be causally derived from an earlier experience (typically, a perceptual experience). This causal connection is stipulated to be similar to the causal connection normally involved in memory, except for the fact that it is consistent with the possibility that the subject of the original experience is not the subject of the q-​memory. Thus, on this construal of q-​memory, all memories are particular instances of q-​memory; instances in which the subject of the q-​memory and the subject of the earlier experience happen to coincide. We may refer to those instances of q-​memory which are not memories as “mere q-​memories.” What is, then, the relevance of mere q-​memories for IEM? The possibility of having mere q-​memories may be used to challenge the view that memory judgments are IEM. Thus, Sydney Shoemaker argues that, in a situation in which an apparent memory qualifies as a mere q-​memory, the subject would be wrong in judging that they are the person who had the q-​remembered property, even if their mere q-​memory was accurate and they were forming their judgment on the basis of it. The fact that such a situation is logically possible, Shoemaker claims, suggests that memory judgments are not IEM: 12 The possibility of such criteria is discussed in ­chapter  4 of (Shoemaker 1963)  as well as in (Shoemaker 1987, 111).

Immunity to Error through Misidentification  145 Now whereas I can remember an action from the inside only if it was my action, a world in which there is quasi-​remembering that is not remembering will be one in which it is not true that any action one quasi-​remembers from the inside is thereby an action he himself did. So—​assuming that ours may be such a world—​if I quasi-​remember an action from the inside, and say on this basis that I did the action, my statement will be subject to error through misidentification; it may be that my quasi-​memory of the action is as accurate and complete as it could be, but that I am mistaken in thinking that I am the person who did it.13

Derek Parfit makes a similar remark in the context of a discussion of a non-​ circular psychological criterion for personal identity. Parfit claims that, when a subject judges that they had a certain experience in the past on the basis of one of their memories, the identity between them and the subject of the remembered experience is not, strictly speaking, part of what they remember. Nevertheless, they are justified in assuming such an identity in their judgment because, as a matter of fact, we do not have mere q-​memories: When I seem to remember an experience, I do indeed seem to remember having it. But it cannot be a part of what I seem to remember about this experience that I, the person who now seems to remember it, am the person who had this experience. That I am is something I automatically assume. (My apparent memories sometimes come to me simply as the belief that I had a certain experience.) But it is something that I am justified in assuming only because I do not in fact have q-​memories of other people’s experiences.14

Parfit’s remark could be used to challenge the IEM view in an analogous way to that in which Shoemaker challenges it. Suppose that Parfit is right in claiming that, when I judge that I had some experience in the past based on one of my memories, the fact that I was the subject of that experience is not, strictly speaking, part of what I remember. Then, it seems that if the memory on the basis of which I form my judgment was a mere q-​memory, my judgment would be incorrect even if my mere q-​memory was accurate. But it seems logically possible for any of my memories to be a mere q-​memory.

13 In (1970, 273). 14 In (1971, 15).

146  The Epistemology of Memory Thus, this line of reasoning suggests, once again, that our memory judgments are not IEM. Several authors have taken issue with an aspect of the notion of q-​memory which is implicit in the use of it that we have just contemplated. The relevant aspect is the presupposition that a subject’s q-​memories do not present their corresponding q-​remembered properties as being the subject’s own. Gareth Evans, for example, points out that, in the scenario depicted by Shoemaker, it has not been shown that my q-​memory of the past action is accurate or correct. We may stipulate, Evan concedes, that a q-​memory does not need to causally originate in a property which is the subject’s own (in Shoemaker’s example, an experience of a past action). And yet, despite our stipulation, the content of the q-​memory may still carry the commitment that the quasi-​ remembered property was the subject’s own: Given the notion thus introduced, we are able to say that a subject q-​ remembers an event that he did not witness, and in consequence that he q-​remembers witnessing an event, being in front of a tree, etc., when he did not witness the event, was not in front of a tree, etc. (Of course introducing such a definition leaves the question of the content of memory states quite untouched; it can still be right to say, as I have, that an apparent memory of ϕ-​ing is necessarily an apparent memory of oneself ϕ-​ing.)15

Interestingly, John McDowell makes essentially the same point in response to Parfit. Parfit’s appeal to the notion of q-​memory is intended to show that, when I remember one of my past properties, the fact that I, the person who seems to remember it, am the person who had the remembered property is not part of what I remember. It is, in that sense, an assumption that we can separate from the content of my memory. But q-​memory can only do the work of separating that assumption, McDowell claims, if a q-​memory is a state whose content is silent on the identity of the subject of the q-​remembered property. And this aspect of q-​memory does not follow from the characterization of q-​memory as an apparent memory which may originate in a property of a different subject: Once one has the notion of mere quasi-​memory, made intelligible in terms of suitable abnormal aetiologies for memory impressions, one can entertain the supposition that an apparent memory is a mere quasi-​memory, thereby

15 In (1982, 248).

Immunity to Error through Misidentification  147 distancing oneself from the belief that it was oneself who lived through the recalled state or occurrence. In that sense the belief is indeed “separable.” But this does not equip the memory impression with an identity-​neutral content:  The supposition one would be entertaining is that an impression whose content is not identity neutral, because it is that of an ordinary memory, is illusory in respect of that aspect of its content.16

I side with Evans and McDowell in this controversy. To illustrate what is problematic about arguing that memory judgments are not IEM by appealing to q-​memory, it may be useful to spell out the argument against the IEM view condensed in Shoemaker’s passage just quoted. Let M be a memory, and let P be the property instantiation which is remembered in virtue of having M. Let us suppose, furthermore, that, on the basis of M, I form the judgment that I had P. The argument seems to have the following structure: 1. If we can have mere q-​memories, then M can be a mere q-​memory. 2. If M can be a mere q-​memory, then M can be an accurate mere q-​memory. 3. Necessarily, if M is an accurate mere q-​memory, then M has been caused by the fact that a subject S, other than myself, had P. Therefore, 4. If we can have mere q-​memories, it is possible for M to be accurate, and to have been caused by the fact that a subject S, other than myself, had P. Given the generality of the premises, conclusion 4 suggests that, for any memory judgment, it is logically possible for clauses (i–​iv) in the characterization of IEM to be fulfilled, which renders the relevant judgment vulnerable to error through misidentification. Shoemaker’s argument seems prima facie compelling. If we can have mere q-​memories, then there seems to be no reason to rule out the possibility that M itself might originate in the past experiences of a different subject while it appears to me to be a memory. Which suggests that premise 1 is correct. And if mere q-​memories can be either accurate or inaccurate (which appears plausible enough once we accept that memories, in particular, can be either accurate or inaccurate), then

16 In (1997, 241).

148  The Epistemology of Memory there seems to be no reason to rule out the possibility that M itself might be accurate. Which suggests that premise 2 is correct. Finally, premise 3 seems to hold true just in virtue of our definition of “mere q-​memory.” Thus, each of the premises in the argument seems quite reasonable on the face of it. What is wrong with this argument, then? The concern raised by Evans and McDowell can be expressed in the form of a dilemma. The characterization of q-​memory as an “apparent memory” leaves open the issue of what exactly one apparently remembers when one has a q-​memory. And the argument outlined here seems to trade on this ambiguity. Either a q-​memory is a state wherein one apparently remembers having a certain property, or it is a state wherein one apparently remembers the having of a certain property. In the former case, a difficulty for premise 2 arises. If what I apparently remember in virtue of having M is having P, then, in order for M to be accurate, it must be the case that I myself had P in the past. But if M also qualifies as a mere q-​memory, then I am not the person who had P; someone else was the bearer of it. Thus, mere q-​memories can never be accurate, and premise 2 turns out to be false. In the latter case, a difficulty for premise 1 arises. The reason why premise 1 initially seemed plausible is that memories and mere q-​memories were only meant to differ in their causal origins. However, once we stipulate that, in virtue of having a q-​memory, a subject only appears to remember the having of a certain property, it is no longer clear that any memory can be a mere q-​memory. It is not clear, that is, unless we assume that subjects who have memories only appear to remember the having of certain properties as well. (In that case the only difference between memories and mere q-​memories will indeed be a difference in causal origin.) The trouble with making that assumption, though, is that the question of whether memory judgments are IEM precisely hinges on whether memories are silent on the ownership of remembered properties or not. Suppose that, on the basis of my memory of this morning’s walk with my dog, I form the judgment that I saw a brown dog at the local park this morning. If what I remember, strictly speaking, is having a certain perceptual experience this morning, then it seems that my memory judgment will be IEM relative to my memory of this morning’s walk. For if my memory is informing me that I was the owner of the remembered perceptual experience, then it seems logically impossible for my memory to be accurate and for me to go wrong when I judge, on the basis of it, that I had the remembered experience. By contrast, if what I remember, strictly speaking, is the instantiation of that perceptual experience, and I somehow infer that the bearer of

Immunity to Error through Misidentification  149 the experience was me, then it seems that my memory judgment will not be IEM relative to my memory of this morning’s walk. For if my memory is neutral on who the bearer of the remembered experience was, then there is room for the logical possibility that my memory is accurate, and yet I am wrong in judging that the person who had the remembered experience was me. It seems, therefore, that the issue of whether my memory judgment is IEM relative to my memory of this morning’s walk hangs on whether the content of my memory is neutral on who the bearer of the past perceptual experience was. Thus, it would be question-​begging to assume, first, that memories are neutral on the ownership of remembered properties and, then, to use that assumption about the content of memories in order to argue, through the notion of q-​memory, that memory judgments are not IEM. The upshot is that tackling the issue of whether memory judgments are IEM by appealing to the notion of q-​memory is not a promising approach. In order to show that the judgment that one instantiated a certain property in the past, made on the basis of an episodic memory, is not IEM relative to it, one needs to show that it is possible for the judgment to be false even though the memory is accurate. The fact that it is logically possible for any of our memories to be a mere q-​memory does show that it is possible for one’s judgment to be false. But it does not show that, in the situation in which one’s judgment is false, the apparent memory on which it is based remains accurate. To make a case for this, one needs to frame the discussion in terms of a notion of q-​memory which makes a substantial assumption about the content of memories. And, as we have just seen, the nature of mnemonic content is that on which the IEM phenomenon seems to be grounded in the first place. Thus, the strategy of appealing to q-​memory to determine whether memory judgments are IEM is bound to prejudge the issue under discussion. The moral that we can draw from our discussion of q-​memory, then, is that a more promising approach should tackle the question of whether memory judgments are IEM by addressing the more fundamental question of whether a subject is present in the content of their own memories or not.17 17 There is an approach to IEM according to which what protects a judgment which has been formed on the basis of a mental state from any error of misidentification is, instead, that the subject is not a component of the content of that mental state. François Recanati (2007), for example, argues that, when a subject’s judgment about the position of their limbs formed on the basis of one of their proprioceptive experiences is IEM relative to that experience, it is not because the content of the proprioceptive experience is that the subject instantiates a certain property; having limbs in a certain position. The content of the proprioceptive experience is only that a certain property is being instantiated; the having of limbs in a certain position. But the subject’s judgment is IEM relative to the experience because the domain of properties experienced through proprioception only includes

150  The Epistemology of Memory

6.4  Observer Memory and IEM There is a phenomenon discussed in the psychological literature on memory which is relevant to the issue of whether a subject is present in the content of their own memories or not. The phenomenon concerns the possibility of having so-​called observer memories. This possibility arises from the fact that memory may present a past fact to its subject from two types of visual perspectives. One of them is the type of perspective from which the subject would have perceptually experienced the fact if the subject had been a part of it, or had gone through it, in the past. By having a memory that presents a past fact from a perspective of this type, the subject visualizes the past fact, so to speak, from the inside. The subject visualizes the fact, but they do not visualize themselves as a part of it. Let us call memories that present past facts from this type of perspective, “first-​person” or “field” memories. A  memory may also present a past fact from the type of perspective that a different observer would have had to occupy in the past in order to witness the remembered fact with the subject as a participant of it. By having a memory that presents the past fact from a perspective of this type, the subject visualizes not only the past fact but they also visualize themselves, as it were, from the outside. Let us call memories that present past facts from this type of perspective, “third-​person” or “observer” memories.18 In order to illustrate the distinction between field and observer memories, consider the following example. Suppose that, years ago, I suffered a traffic accident while sitting directly behind the driver of a car. I  may now remember the position of each of the occupants of the car by visualizing the accident from the perspective of the person who was sitting directly behind the driver. In virtue of having this mental image, I visualize the back of the driver’s head, for example, as well as the back of their seat. In this case, I am having a field memory of the scene. But I may also remember the position properties of the subject who is exercising that faculty. Our discussion of q-​memory suggests that the view that the subject is not present in the content of their memories might yield the result that memory judgments are de facto IEMτ. But it falls short of establishing that those judgments are logically IEMτ. For it does not seem logically necessary for the domain of properties experienced through memory to include only properties of the subject themselves. 18 For seminal work on this distinction, see (Nigro and Neisser 1983). Further research on the distinction between field and observer memories has suggested that a subject is more likely to have a field memory of some fact when the remembered fact has a strong emotional significance for them (Talarico, LaBar, and Rubin 2004). It seems that facts surrounding traumatic events, for example, tend to be remembered from the field perspective (Porter and Birt 2001). For philosophical discussions of observer memory, see (McCarroll 2018) and (Sutton 2010).

Immunity to Error through Misidentification  151 of each of the occupants of the car by visualizing the scene from a different perspective, let us say, the perspective of the person sitting directly behind the front passenger’s seat. In virtue of having this mental image, I still visualize the driver in the car, as well as their seat. But I also visualize, among other things, the person sitting directly behind the driver (a person who, by assumption, is myself). In this case, I am having an observer memory of the scene. What is the relevance of observer memories for our discussion of immunity to error through misidentification? At first glance, the possibility of having observer memories may seem to suggest that memory judgments are not IEM. The train of thought which may lead one, reasonably enough, to such a conclusion is composed of two considerations. The first consideration is that if some of the memories that one has are observer memories, then it seems that, in order to judge that one had some property in the past on the basis of a memory, it may sometimes be necessary for one to identify oneself in the remembered scene. How would such a process of identification work exactly? Suppose, for example, that I judge that I was sitting behind the driver during the traffic accident on the basis of an observer memory wherein I picture the position of the occupants of the car from the point of view of the person sitting behind the front passenger’s seat. Then, it seems that I must have arrived at the judgment that the person behind the driver was me through a process of inference. I must have thought that the person behind the driver had certain properties (for example, that he appeared to be short and unshaven, and that his hair appeared to be brown and disheveled). I must have thought that, among the occupants of the car, I was the only person with the properties in question. And, finally, I must have concluded from those premises that I am the person who was sitting behind the driver. The second consideration is that, when one needs to identify oneself in the scene that one is remembering in this way, one may make a mistake in the identification process through no fault of one’s memory. How would such a mistake arise then? It seems possible for one to be correct in remembering, on the one hand, that someone in the scene that one is remembering had such-​and-​ such properties, and to be wrong in thinking, on the other hand, that one had the relevant properties at the time. Thus, I could be wrong in believing that I appeared to be unshaven on the day of the accident. (Perhaps I shaved that morning, and later I forgot about it.) Alternatively, I could be wrong in thinking that I was the only person in the car who appeared to be short and unshaven and whose hair appeared to be brown and disheveled. (Perhaps

152  The Epistemology of Memory someone who bears an uncanny resemblance to me, in those and many other respects, was travelling in the car with me that day.) It seems to follow, then, that if some of the memories that one has are observer memories, then memory judgments are not immune to error through misidentification. For it seems possible, in cases of observer memory, for one to judge that one had some property in the past on the basis of a memory and for one to be wrong because the person who one correctly remembers to have had the property is someone else, and one has misidentified that person as oneself. In what remains of this section, I will argue that this train of thought is misguided. Some of the reasons why it is misguided, though, will reveal an interesting lesson regarding the presence of the self in memory. Thus, even though it will eventually become evident that the possibility of having observer memories does not threaten the IEM view, it will be productive to examine why the phenomenon of observer memory does not provide a counter-​example to that view. Let us consider, then, the idea that observer memory makes memory judgments vulnerable to error through misidentification in more detail. It will be convenient to organize the discussion of observer memory around a particular case. Take, for example, the case in which I have mental state wherein I visualize the position of the passengers in a car during a traffic accident in the past. There is a person sitting directly behind the driver’s seat, but I do not visualize the scene from that passenger’s perspective. Instead, I visualize the scene from the perspective of the passenger sitting directly behind the front passenger’s seat. (Let us stipulate that we are in a country where cars are driven on the left side of road and, thus, the front passenger’s seat is placed on the left side of the car.) The mental state that I occupy when I visualize this scene appears to me to be an observer memory. As a result, I judge, on the basis of that apparent observer memory, that it was me who was sitting directly behind the driver’s seat. Next, let us remind ourselves of the conditions that need to be met, in this case, for an error of misidentification to take place. For such an error to take place, there needs to be a subject S such that (i) my observer memory represents S as sitting behind the driver’s seat, (ii) my observer memory is fully accurate, (iii) I mistakenly think that I am identical with S, and (iv) my judgment that I was sitting behind the driver’s seat is false as a result of (iii). Finally, let us drop an assumption that we made earlier while this example was introduced for the purposes of illustrating the distinction between field and observer memories. This is the assumption that I  was, in fact, sitting

Immunity to Error through Misidentification  153 behind the driver’s seat. Instead of making such an assumption, let us distinguish three possible scenarios; the scenario in which, at the time of the accident, I was sitting behind the driver’s seat, the scenario in which I was sitting neither behind the driver’s seat nor behind the front passenger’s seat, and the scenario in which I  was sitting behind the front passenger’s seat. After all, the details of the case just sketched could be filled in according to each of those three scenarios.19 It is easy to appreciate that, in the first scenario, the case is a non-​starter as a counter-​example to the IEM view. The reason is simply that condition (iv) is not met. If I was, by assumption, sitting behind the driver’s seat, then my memory judgment is in fact true. Let us therefore set this possibility aside and consider the scenario in which I  was sitting neither behind the front passenger’s seat nor behind the driver’s seat, and the scenario in which I was sitting behind the front passenger’s seat. Consider the scenario in which I was sitting neither behind the driver’s seat nor behind the front passenger’s seat. My contention is that, in this scenario, condition (ii) fails to be met; my observer memory is not fully accurate. My observer memory represents someone sitting behind the driver’s seat and, as we stipulated above, there is indeed someone sitting behind the driver’s seat. My observer memory is, to that extent, accurate. But the presence of a passenger behind the driver’s seat is not all that my observer memory represents. My observer memory seems to represent my own position within the remembered scene as well. Why is that? Notice that my observer memory represents the passenger sitting behind the driver’s seat from a certain perspective. After all, it is only a certain side of their face, and more generally of their body, that I visualize in virtue of having my observer memory. Since we stipulated that the driver’s seat is on the right side of the car, it is the left side of their body that I am visualizing. Now, why is that side of the remembered passenger’s body the side that I visualize in virtue of having my observer memory? The answer to this question seems to be that the left side of their body is the side that I would have been able to see, had I been sitting behind the front passenger’s seat, given their position in the car relative to mine. This feature of my observer memory suggests a more general point about what is involved in visually representing an object from

19 As we are about to see, these details do have a bearing on the issue of whether or not the mental state that I occupy when I visualize the inside of the car qualifies as an observer memory. It will therefore be important to keep the three scenarios separate.

154  The Epistemology of Memory a certain perspective. It suggests that, when a subject visually represents an object from a certain perspective, the subject thereby represents that they are in a certain position relative to that of the object. Thus, when I visualize the passenger sitting behind the driver’s seat from the perspective of the person sitting behind the front passenger’s seat, I am thereby representing that I occupy a certain position in the car relative to that of the passenger sitting behind the driver’s seat. Specifically, I am representing the fact that the passenger is located to my right. This seems to be part of what it is for me to visualize the passenger behind the driver’s seat from the perspective of the person who is sitting behind the front passenger’s seat. But this particular aspect of what my observer memory represents is, of course, not accurate. After all, we have been assuming that, in the scenario that we are considering, I was sitting neither behind the driver’s seat nor behind the front passenger’s seat at the time of the accident. This is why, in this scenario, the observer memory that we have been entertaining would not constitute a counter-​example to the IEM view either. What about the scenario in which I was, in fact, sitting behind the front passenger’s seat and I  visualize the person sitting behind the driver’s seat from that perspective? It is tempting to think that, in this scenario, the case just sketched does show that memory judgments are vulnerable to misidentification errors. We have been assuming that there was someone sitting behind the driver’s seat at the time of the accident. To simplify matters, let us also assume that my current mental image originates in my past perception of that passenger. Then, I am correctly visualizing that there was someone sitting behind the driver’s seat. I am visualizing that person from a certain perspective and, in virtue of this fact, I am representing my own position in the car relative to that of the person behind the driver’s seat. As it happens, I am representing my own position in the car correctly. The person behind the driver was, in fact, located to my right. But I believe, wrongly, that I am the person who I visualize as having been sitting behind the driver’s seat. Thus, when I judge that I was sitting behind the driver’s seat on the basis of my mental image of the scene, it turns out that I am making a false judgment. Have we not reached, then, a counter-​example to the view that memory judgments are IEM if we construe the traffic accident case along these lines? The traffic accident case presents us with an interesting type of mental state when we formulate it thus. Notice that the mental image that I have when I visualize the person sitting behind the driver’s seat from the perspective of the person sitting behind the front passenger’s seat no longer qualifies as an

Immunity to Error through Misidentification  155 observer memory. Once we assume that I was in fact sitting behind the front passenger’s seat, the perspective from which I now visualize the scene turns out to be the perspective that I occupied while I was travelling in the car at the time of the accident. Thus, if the mental image that I have in this case does constitute a counter-​example to the IEM view, it is not in virtue of enjoying any property which is characteristic of observer memories. Whether the mental image that I have in this case does constitute such a counter-​example depends on whether it qualifies as a memory in the first place. After all, the thesis that memory judgments are IEM only concerns those judgments that we make on the basis of our memories. We have already encountered some reasons to be suspicious of the idea that the mental state that I occupy in the traffic accident case, when we assume that I was sitting behind the front passenger’s seat at the time of the accident, qualifies as a memory. Let us remind ourselves of a certain theory about the nature of memory offered in ­chapter 2. According to the functionalist theory of memory, for any subject S and proposition q, S remembers that q just in case S has some mental image i such that i plays the mnemonic role for the fact that q in subject S. What does playing such a role involve? What is required is that image i tends to cause in subject S a disposition to believe both that q and that S experienced that q, and that image i tends to be caused in S by having experienced that q. Recall that this view distinguishes the mental image which, when the subject remembers a fact, plays the mnemonic role for that fact in them from the subject’s higher-​order state of having some mental image or other that plays the mnemonic role for the relevant fact in them. According to the functionalist theory of memory, memories are mental states of the former type whereas remembering episodes are mental states of the latter type. Consider the traffic accident case, and let us continue to assume that, at the time of the accident, I was sitting behind the front passenger’s seat. Am I  having a memory of the passenger sitting behind the driver’s seat when I visualize the inside of the car from the point of view of the passenger sitting behind the front passenger’s seat? The functionalist theory of memory tells us that there is no intentional or phenomenological feature of the mental image that I am having such that my mental image qualifies as a memory just in case it enjoys that feature. Instead, what matters for the issue of whether my mental image qualifies as a memory is whether, in my cognitive economy, the image plays the mnemonic role for the fact that there was a passenger sitting behind the driver’s seat or not. Admittedly, my mental image satisfies

156  The Epistemology of Memory some parts of that mnemonic role. By assumption, my mental image does originate in a perceptual experience of the person sitting behind the driver’s seat. Furthermore, my mental image does predispose me to believe that there was a passenger sitting behind the driver’s seat. And yet, there is an important part of the mnemonic role for the fact that there was a passenger sitting behind the driver’s seat that my mental image fails to satisfy. Notice that, in virtue of having this mental image, I have no inclination to believe that I perceived, in the past, the passenger sitting behind the driver’s seat. Clearly I must have no such inclination since, otherwise, I would not be making, now, the judgment that it was me who was sitting behind the driver’s seat at the time of the accident on the basis of that mental image. For this reason, my mental image does not fully play, in me, the mnemonic role for the fact that there was a passenger sitting behind the driver’s seat. And this means, according to the functionalist theory, that my mental image does not qualify as a memory.20 Where does this leave us? We have seen that, contrary to what it may have seemed at first glance, the possibility of having observer memories does not threaten the IEM view. It does not threaten this view because either observer memories do not yield false memory judgments, or they are not themselves fully accurate, or they do not qualify as genuine memories. In the process of figuring out whether the phenomenon of observer memory posed a threat to the IEM view, however, we have learnt an interesting lesson regarding what is involved in visually representing an object from a certain perspective. The traffic accident case, construed along the lines of the second scenario described, has taught us that part of what it is for a subject to visually represent an object from a certain perspective is for the subject to represent that they themselves are in a certain spatial position relative to that of the object. This feature of perspectival representation illustrates one of the ways in which a subject may be present in the content of their own mental states; an idea that we will explore in further detail while explaining the source of IEM in memory judgments.

20 This argument could be used to motivate the broader view that observer memories are not memories at all. (I am grateful to Kourken Michaelian for pointing this out to me.) If this is correct, then one might appeal to the functionalist theory of memory proposed in ­chapter 2 in order to exclude observer memories from the category of memories. Notice that this is independent of whether or not memories have causally self-​referential contents; an idea that will come up again in section 6.5.

Immunity to Error through Misidentification  157

6.5  A Defense of IEM in Memory Let us take stock. In sections 6.3 and 6.4, we have considered two possible challenges to the idea that memory judgments are IEM; one based on the notion of quasi-​memory, another on the notion of observer memory. We have seen that neither challenge is ultimately successful, but we have also drawn an interesting moral from the discussion of each challenge. On the one hand, our discussion of quasi-​memory indicated that the issue of whether memory judgments are IEM turns on the question of whether a subject is present in the content of their own memories or not. On the other hand, our discussion of observer memory has brought to light one particular instance in which a subject is present in the content of one of their mental states, namely, when the subject has a visual experience of a fact and, in virtue of having such an experience, represents the fact from a certain perspective. Both of these ideas will prove useful while we develop the positive part of the project in this chapter. In what remains of this section, I will offer a defense of the view that memory judgments are IEM. In order to argue that the IEM view is true, I will pursue the question of whether a subject is present in the content of their own memories or not. In sections 6.6 and 6.7, I will try to locate the source of immunity to error through misidentification in memory. In order to account for the truth of the IEM view, I will explore, in further detail, the idea that, when a subject visually represents an object from a certain perspective, the subject represents that they themselves are in a certain spatial position relative to that of the visualized object. Let us turn, therefore, to a defense of the IEM view. Consider the question of whether one is present in the contents of one’s memories. There is a reason for thinking that one is present in them. It concerns the reflexive view of mnemonic content put forward in ­chapter 3. Recall that, according to the reflexive view, our memories represent causal relations that involve factive mental states, thus engaging the world as well as our own minds. More precisely, the main idea in the reflexive view was that the content of a memory is that the memory itself has been caused by a past perception of its subject. The view was formulated as follows: RV  Reflexive view

For any subject S, memory M and proposition q: If S has M and S would express M by saying that they remember that q, then there is a perceptual experience P that S would express by

158  The Epistemology of Memory saying that they perceive that q, such that the content of M is the proposition {W: In W, M is caused by S having perceived that q through P} The reflexive view, or RV, attributes a very complex content to memories. On RV, the content of a memory involves the memory itself, it involves a causal relation, and it involves a factive mental state in the past. For the purposes of our discussion in this section, however, the most interesting component of the content that RV attributes to a memory is a component which is different from those three, namely, the very subject of the memory. Notice that, according to RV, the content of a memory is not only that the memory itself was caused by a perception of the fact that the subject claims to remember in virtue of having the memory, but also that the perception at issue was the subject’s own. What is the relevance of this component of mnemonic content for the IEM view? Consider what it takes for one of my memory judgments to be vulnerable to a misidentification error. Suppose that I have a memory M and, on the basis of M, I form the judgment that I had a certain property P in the past. My judgment will only be vulnerable to a misidentification error if it is possible that there is a subject S such that (i) M represents S as having had P, (ii) M is fully accurate, (iii) I mistakenly think that I am identical with S, and (iv) my judgment that I had P is false as a result of (iii). For example, imagine that the relevant property P is, as in our example from section 6.3, the property of seeing a brown dog (a prop­ erty that I instantiated while walking my dog at the local park). And imagine that, accordingly, the judgment that I make on the basis of my memory M is the judgment that I saw a brown dog. Then, my judgment will only be vulnerable to a misidentification error if the following is possible: There is a subject S, other than myself, such that my memory M represents S having seen the brown dog, and M is fully accurate. But I mistakenly think that I am identical with S and, as a result, I make the false judgment that it is me who saw the brown dog. It does not seem that this state of affairs could ever obtain if RV is correct. How could I, on the one hand, have a full and accurate memory of the past scene at the park and, on the other hand, be mistaken in thinking that the person who I remember to have seen the brown dog was myself? If RV is correct, my memory M can only be fully accurate if M causally originates in my experience of seeing the brown dog.21 And if the remembered perceptual 21 Notice that this claim about RV only says that M carries, among other pieces of information, the information that the experience of seeing the brown dog was mine. It is not the claim that M is factive with respect to that part of its content.

Immunity to Error through Misidentification  159 experience was indeed my experience, then I am not mistaken in thinking that I am identical with the person who I remember to have seen the dog. It seems, therefore, that a misidentification error is not possible if memories have the kind of content that is attributed to them by RV. The upshot, then, is that memory judgments are IEM if the reflexive view of mnemonic content is correct. This defense of the view that memory judgments are IEM crucially relies on the fact that, according to RV, the content of a memory is not only that the memory itself was caused by a perception of the fact that the subject claims to remember in virtue of having the memory, but also that the perception at issue was the subject’s own. Recall that, in c­ hapter 3, this feature of RV was motivated through the use of the following thought experiment.22 Suppose that I am looking at a red apple in front of me, and my perceptual experience presents it to me as being red. Suppose that, years later, I have a memory which originates in that perceptual experience; one that I would express by saying that I have a memory of being in front of a red apple. Let W0 be this possible situation, and let us refer to my past perceptual experience and my memory in W0 as “P” and “M” respectively. Consider, now, a different possible situation W5. In W5, the apple is red. Also, in W5, perceptual experiences normally elicit memories in subjects who are different from the bearers of those perceptual experiences. Thus, someone other than me looks at the apple, and they have P. But their having P causes me, and not them, to have M years later. I myself never experience the red apple perceptually. Intuitively, it seems that, by having M in W0, I am not representing W5 accurately. This is the intuition which suggested to us, in ­chapter 3, that my memory M in W0 is not only about a perceptual experience which may or may not have been mine in the past. Instead, M informs me of P as having been mine. And that, in turn, is how we arrived at the conclusion that, in this case, the content of my memory M should be construed as the set of possible worlds in which M is caused by my having perceived a red apple through P (a proposition to which we referred as “CSR3”). Thus, the defense of the view that memory judgments are IEM just offered is, ultimately, grounded on our intuitions regarding W5 and other similar situations. The fact that the proposed defense of the IEM view is grounded on the intuitions regarding the intentionality of memory discussed in ­chapter  3 22 For a different defense of the view that remembered facts are presented to us as having been formerly experienced by ourselves, see (Rowlands 2016).

160  The Epistemology of Memory raises a difficulty. As I see it, the intuition that my memory M would inaccurately represent a possible situation in which the memory originates in a perception which belongs to a different subject does support the idea that my memories represent past perceptions as having been mine. I also believe that the idea that my memories represent past perceptions as having been mine does support the view that memory judgments are IEM. Where is the difficulty then? The trouble for this defense of the IEM view is that this defense tells us nothing about where immunity to error through misidentification comes from. It tells us that the IEM phenomenon takes place in memory, but it does not tell us why. What we need is an account of why we have the intuition that my memory M would falsely represent a possible situation in which the memory originates in a perceptual experience which belongs to someone else. In the absence of such an explanation, we seem to be forced to conclude that our project in this chapter is unfinished. It is one thing to establish that memory judgments are IEM. We have, I submit, achieved this outcome. And it is a substantial achievement, given the challenges to this view posed, for example, by the notion of quasi-​memory and by that of observer memory. But locating the source of immunity to error through misidentification in memory is an entirely different thing. And that outcome, we have not achieved quite yet. What is required, at this point, is an account (and not simply a defense) of the truth of the view that memory judgments are IEM.

6.6  The Perceived Self The reflexive view of mnemonic content has an important corollary for our discussion of IEM. It concerns the relation between the content of our memories and the content of the perceptual experiences in which they typically originate. Suppose that, as the reflexive view states, the content of an episodic memory always concerns a causal relation between the memory itself and a past perception of the subject. Then, when we remember, in particular, what it was like for us to perceptually experience some fact in the past, what we are aware of is, in part, a perception of that fact. But perceptions require their contents to, actually, be the case. Thus, the reflexive view yields the result that, when we have a memory of what it was like for us to perceptually experience some fact in the past, the perceptually experienced fact becomes part of what we remember by having the memory. An episodic

Immunity to Error through Misidentification  161 memory inherits, as it were, part of its content from the perceptual experience at which it is directed. Why is this corollary important for locating the source of immunity to error through misidentification in memory? In section 6.5, we have seen that a reason for thinking that memory judgments are IEM is that memory presents those perceptual experiences that one remembers to have been instantiated in the past as having been one’s own. In that sense, one is present in the content of one’s memories. The issue which remained unresolved at the conclusion of section 6.5, however, was why one is present in the content of one’s memories. The just-​mentioned corollary of the reflexive view suggests that the question of whether one is present in the content of one’s perceptual experiences may throw some light on this issue, since memories receive part of their contents from the perceptual experiences at which they are directed. If this suggestion is correct, then an explanation of the IEM phenomenon in memory might be found in the intentionality of perception. Let us consider, therefore, whether one is present in the content of one’s perceptual experiences or not. A  hypothesis about the content of perceptual experiences should respect and, ideally, account for three features of such experiences. Two of these features concern the, so to speak, input side of perception, whereas the third feature has to do with the output side of perception. Consider, first of all, a fact about perception that our discussion of observer memory highlighted in section 6.4. This is the fact that perception is perspectival. If one seems to perceive some scene, then that scene is presented to one from a perspective, or point of view. This phenomenological aspect of one’s perceptual experience depends on the spatial position that one occupies with regards to the objects which are involved in the perceived scene and, thus, it changes as one changes one’s spatial relation with regards to those objects. Right now, for example, I seem to see a wall and a window to my right, and I am aware of the smell of coffee as coming from the cup in front of me. But I turn 180 degrees, then I will seem to see a wall and a window to my left, and I will be aware of the smell of coffee as coming from behind me. Secondly, a perceptual experience does not present a scene to one in such a way that one feels separated from that scene, or as if one was a spectator of it. In virtue of having a perceptual experience, one feels that one is part of the scene that the experience is making manifest to one. The relevant contrast here is with the experience of looking, for example, at a monitor or at a cinema screen. In such an experience, one feels that a scene is being manifest

162  The Epistemology of Memory to one. The objects of that scene, and their features, are before one’s mind in virtue of having the experience at issue. But one will not thereby feel as if one is present in the relevant scene. In perceptual experience, by contrast, there is a felt immediacy which puts one inside of the scene that the experience is making manifest to one, among the objects in that scene. If I look around me and I see, for example, the wall and the window in my room, and the cup of coffee on my desk, then these objects are presented to me as part of my surroundings. They are presented to me, in other words, as parts of a scene in which I am also present. I will refer to this phenomenological feature of perceptual experiences as their “immediacy.”23 Thirdly, perception directly feeds into action. A subject who has a perceptual experience will be poised to perform certain actions just in virtue of the fact that they have that experience. Thus, if one apparently sees a large object approaching at a high speed, then one will step out of its way. Likewise, if one seems to see the object flying at one’s face, then one will duck, and so on. Perception is immediately salient to action in the sense that one does not need to form any particular belief about one’s spatial position in order to start moving. Admittedly, unless one has a background of beliefs and desires, such as the belief that there is no reason to distrust one’s vision and the desire not to get hit or run over, one might not be poised to move when one perceptually experiences the approaching object. Nevertheless, it does not seem that any specific belief regarding one’s spatial location is required in order to initiate one’s movement. What must be true of the contents of perceptual experiences in order for them to have these three interesting features? The main proposal in this section is that these three features of perceptual experiences point us toward a certain view about the intentionality of perception. We may call it the “extrinsic” view of perception. According to the extrinsic view of perception, for any proposition q and subject S, if S has a perceptual experience that they would express by saying that they perceive that q, then there is an extrinsic property R such that R is a property of S, and the content of S’s perceptual experience is that S is related through R to the fact that q.24 Thus, if I have a 23 The idea that perceptual experience involves a sense of immediacy is not new. In (2017), David Chalmers compares the experience of virtual reality with our experience of our perceived environment partly by reference to a sense of “presence,” that is, the sense of being present at the relevant environment. 24 The view is similar to James Gibson’s “ecological optics” in (1979). To be precise, the view offered here is that if a subject has a perceptual experience that they would express by saying that they are perceiving a certain scene, then the content of their experience is their being related in some way to

Immunity to Error through Misidentification  163 perceptual experience that I would express by saying that I see a small cup of coffee on my hand, my experience makes me aware of the fact that the cup on my hand is small by making me aware of the difference between my own size and that of the cup. Likewise, if the coffee then spills on my hand, and I have a perceptual experience that I would express by saying that the coffee from the cup is hot, my experience makes me aware of the temperature of that liquid by making me aware of the difference between the temperature on the relevant region of my skin and the temperature of the portion of the liquid in contact with it. The basic thought in the extrinsic view of perception, then, is that the content of a perceptual experience is always the subject of the experience having a certain extrinsic property. In perception, we are aware of objective facts; objective in that they obtain independently of our existence and our perceiving them. However, we are only aware of those facts by being aware of the relations in which we, the perceivers, stand to them. One may develop the extrinsic view of perception in at least three different directions, depending on one’s views on the nature of perceptual content. The notion of mnemonic content that we have been operating with is one according to which mnemonic contents are sets of possible worlds. Analogously, one may propose that the content of a perceptual experience is a set of possible worlds as well. Specifically, it is the set of possible worlds in which the object which is presented to the subject in virtue of having the perceptual experience has the property which appears to the subject to be instantiated in that object in virtue of having that experience. What does the extrinsic view of perception look like if perceptual contents are sets of possible worlds? Within this framework, the extrinsic view becomes the view that the content of a perceptual experience is a set of possible worlds. It is the set of possible worlds in which the subject of the experience has the property of being related in some way to the object that is presented to the subject in virtue of having the experience. Suppose, for example, that a subject S has a visual experience E and, in virtue of having E, it appears to S that, let us say, the cup of coffee on their hand is small. The extrinsic view, then, is the view that the content of S’s visual experience E is the set {W: In W, S is considerably bigger than the coffee cup}.

an object which is part of that scene. For the sake of convenience, however, I will sometimes talk of the properties which, according to the extrinsic view, are the intentional objects of our perceptual experiences as relations to scenes, or facts, and I will sometimes talk of those properties as relations to objects which are parts of those scenes. Hopefully this will cause no confusion.

164  The Epistemology of Memory The extrinsic view of perception may, however, take other forms if one conceives perceptual contents differently. An alternative view about the content of perceptual experiences is that those experiences have “Russellian” contents. Roughly, the Russellian view is the view that the content of a perceptual experience is an ordered pair composed, on the one hand, of the very object which appears to the subject to have some property in virtue of having the relevant perceptual experience, and, on the other hand, of the property that the object appears to have.25 What does the extrinsic view of perception look like if perceptual contents are Russellian? Within this framework, the extrinsic view becomes the view that the content of a perceptual experience is an ordered pair composed of an object and a property. The first component of this pair is the subject of the experience. The second component is an extrinsic property; the property of being related in some way to the object that is presented to the subject in virtue of having the experience. In the example regarding the size of the cup of coffee, for instance, the view is that the content of S’s visual experience E is the pair . A third view about the content of perceptual experiences is that those experiences have “Fregean” contents. The Fregean view is the view that the content of a perceptual experience is an ordered pair composed, not by an object and a property, but by modes of presentation of them. The relevant pair is composed, on the one hand, by the mode of presentation under which an object is presented to the subject in virtue of having the relevant perceptual experience, and, on the other hand, by the mode of presentation under which a property appears to the subject to be instantiated in that object in virtue of having that experience.26 What does the extrinsic view of perception look like if perceptual contents are Fregean? Within this framework, the extrinsic view becomes the view that the content of a perceptual experience is an ordered pair composed of two modes of presentation. The first component of this pair is a mode of presentation of the subject of the experience.27 The second component is a mode of presentation of an extrinsic property; the property of being related in some way to the object that is presented to 25 Why “roughly”? The need to account for the content of perceptual experiences in hallucination scenarios may require existentially quantified contents instead. For discussion, see (Chalmers 2004). 26 On perceptual content and modes of presentation, see (Burge 1991). 27 The main challenge for the Fregean view is to specify what modes of presentation are exactly. For this challenge, see (Thau 2002). In the context of putting forward the extrinsic view of perception, this challenge is especially pressing. For the Fregean advocate of the extrinsic view is committed to the claim that, in perception, there is such a thing as a mode of presentation of oneself.

Immunity to Error through Misidentification  165 the subject in virtue of having the experience. In the cup of coffee case, for instance, the view is that the content of S’s visual experience E is an ordered pair composed, on the one hand, by a mode of presentation of S and, on the other hand, by a mode of presentation of the property of being considerably bigger than the cup. The precise details of the extrinsic view of perception will differ, therefore, depending on whether we take perceptual contents to be Russellian, Fregean, or possible-​world contents. However, for the purposes of our discussion of memory and IEM, we do not need to adopt a position on this issue. The reason is that, as we are about to see, the main motivation for the extrinsic view of perception is that it accounts for the observation that perceptual experiences are perspectival, for the observation that they are immediate, and for the observation that they directly feed into action. And these three observations seem to be neutral on the exact framework that we choose to adopt in order to characterize perceptual contents. Let us turn, therefore, to the relation between the extrinsic view of perception and the fact that perception is perspectival, is immediate, and directly feeds into action. My contention is that if the extrinsic view of perception is correct, then it is natural for perception to have these three features.28 Let us take each of these features of perceptual experience in order. Firstly, if perceptual experiences always make one aware of a fact by making one aware of an extrinsic property of one’s own which involves that fact, then it makes sense that facts are always perceived from one’s perspective. The reason for this is that the perceptual awareness of a fact from one’s perspective just is the awareness of the relations between one’s own properties (properties such as one’s spatial position, size, or temperature) and those of the object involved in the relevant fact. The thought here is not only that there is a correlation between one’s perceptual awareness of a fact from a certain perspective and the representation of certain extrinsic properties of 28 Notice that the extrinsic view of perception is different from the “relational” view of perception defended, for example, in (Brewer 2007), (Campbell 2010), and (Martin 2004). The relational view of perception is the view that perceptual states are relations to objects and properties of the perceived environment (Campbell 2010, 202), whereas the extrinsic view of perception is the view that perceptual states represent those relations. To illustrate the contrast, suppose that there is in fact no small cup of coffee on my hand. Then, according to the relational view of perception, I do not perceive such a cup at all whereas, according to the extrinsic view of perception, I perceive it incorrectly. Notice that perceptual experiences are perspectival, they are immediate, and they directly feed into action whether they are correct or not. Thus, if the extrinsic view of perception can in fact account for these three features of perceptual experience, then it seems that this aspect of the extrinsic view will constitute a reason to prefer it over the relational view of perception.

166  The Epistemology of Memory one’s own. More strongly, the thought is that the representation of extrinsic properties of one’s own is all there is to perceiving a fact “from one’s point of view.” Hence, the reason why the perspectivity of our perceptual experiences varies as we change some of our relations to the objects involved in the facts that we claim to perceive is simply that perceiving those facts from our perspective and perceiving those relations is one and the same thing. Secondly, if it is correct that, when one has a perceptual experience, one is always aware of the fact that one is holding some relation to an object, then it also makes sense that perceptual experiences have an immediate character. Suppose that, when I see the wall and the window in my room, for example, and the coffee cup on my desk, what I am being aware of is, among other things, the fact that I stand in certain spatial relations with respect to that wall, that window, that cup of coffee, and that desk. It does not seem surprising, then, that, in virtue of having those perceptual experiences, I feel as if I am part of the scene which contains those objects. After all, part of what I am perceiving is, precisely, the fact that I am located among those objects. It is to be expected, then, that I would feel as if I am present in the scene which contains those objects when I have the relevant perceptual experiences. Thirdly, if it is the case that, when one perceives a fact, one is aware of some of the relations which hold between one’s properties and those of the objects involved in the relevant fact, then it also makes sense that perceptual experiences are immediately salient to action. Recall the scenario in which one seems to see a large object approaching at a high speed and the scenario in which one seems to see an object flying at one’s face. If the extrinsic view is correct, then, if I am in either of the two scenarios, what I am aware of is the relation between my own spatial position and that of the perceived object. After all, I seem to see an object flying at me or an object approaching me. This means that my own spatial location relative to that of the object is directly presented to me in perception. It is not surprising, then, that I do not need to form any particular belief about my own spatial position in order to duck or step away. If the extrinsic view is correct, then, when I start moving, I am just trusting perception or taking at face value what I seem to see. There seem to be some considerations, then, which lend support to the idea that, in perception, one is aware of objective scenes in the world by being aware of the fact that one bears certain relations to objects which are part of those scenes. The upshot, then, is that there is a sense in which one is always present in the content of one’s perceptual experiences: What one perceives, when it appears to one that one is perceiving an objective scene, is

Immunity to Error through Misidentification  167 always the fact that one is instantiating a certain extrinsic property, namely, a relation in which one stands to some object in that scene. With this outcome in mind, let us now turn to the issue of what accounts for the fact that memory judgments are IEM.

6.7  An Account of IEM in Memory In section 6.6, we started by noticing a certain corollary of the reflexive view of mnemonic content. This view implied that the content of a perceptual experience is part of what one remembers when one has a memory of what having that perceptual experience was like. Then, we considered some reasons for thinking that one is always present in the content of one’s perceptual experiences. The combination of those two points yields an interesting result. If one is present in the content of one’s perceptual experiences, and part of what one remembers when one has a memory of a past perceptual experience is the content of that experience, then it seems to follow that what one remembers, when one has a memory of a past perceptual experience, is partly oneself. The conclusion, in other words, is that one can be present in the content of one’s episodic memories. This is not a surprising conclusion. After all, we had already encountered the view that the subject is present in the content of their own memories in section 6.5. There we saw that this view follows from the reflexive view of mnemonic content, and we saw that the view suggests that memory judgments are IEM. In that discussion, we were also reminded of the reason why the reflexive view is committed to this view. The reflexive view takes subjects to be present in the contents of their own memories in order to accommodate a certain intuition. This is the intuition that a memory of a subject which originates in one of the subject’s own past perceptual experiences would represent inaccurately a possible situation in which, even though the memory originates in the same perceptual experience, the perceptual experience belongs to someone else in the past. What, then, have we learnt in section 6.6 which is new? Our discussion in section 6.6 has brought to light the source of this intuition. It has taught us that the intuition stems from, on the one hand, the idea that a perceptual experience represents its own subject and, on the other hand, the idea that, when a subject remembers a perceptual experience, the subject thereby remembers what that perceptual experience represented in the past.

168  The Epistemology of Memory To illustrate this point, let us revisit an example of a memory judgment discussed in sections 6.3 and 6.5. Suppose that, this morning, I walked my dog at the local park, and I  saw a brown dog there. Imagine that, on the basis of my memory of this morning’s walk, I form the judgment that I saw a brown dog in the park. In section 6.5, we saw that my memory judgment is IEM because my memory is making me aware of, among other facts, the fact that I myself had a perception in the past (namely, a perception of a brown dog), as opposed to making me aware of the mere fact that the perception took place in the past. But why do we have the intuition that my memory is presenting to me a past perception as having been mine? Consider what happens when I  have my memory of this morning’s walk. According to the reflexive view, I remember a perceptual experience of a brown dog in the park as having been veridical. As a result of its being presented to me as veridical, when I remember that perceptual experience, I also remember its content. And if our discussion in section 6.6 is correct, then the content of that experience is the fact that I instantiated an extrinsic property, namely, the relation between my spatial position and that of the dog. Thus, just like, this morning at the park, I perceptually experienced a certain spatial position relative to that of the dog as being mine, when I have my memory hours later, I remember that spatial position as having been mine. I remember, in other words, being the subject whose spatial position relative to the dog was perceived in that experience. But it seems that if I remember being the subject whose spatial position relative to the dog was perceived through that experience, then I remember being the subject of that experience. That is, I remember that experience as having been my own. It therefore seems that the intentionality of perception is ultimately responsible for why, in this example, I am remembering the perceptual experience of a brown dog in the park as having been my own. More generally, the intuition that a memory presents to me a past perceptual experience as having been mine, and not only as having been instantiated in the past, seems to be grounded on the intentionality of perception. This is the reason why the intentionality of perception constitutes the source of the IEM phenomenon in memory.

6.8  Conclusion If the account presented here of why memory judgments are IEM is correct, then the moral that can be drawn from their immunity is that this feature

Immunity to Error through Misidentification  169 of memory judgments reveals a fact about our faculty of memory. It reveals the fact that, in memory, we are presented to ourselves in a certain way, namely, as the bearers of extrinsic properties which were perceived in the past; properties such as occupying a certain spatial position or having a particular size relative to that of another object. After all, if these considerations are right, we always remember being the subjects whose extrinsic properties were represented by some past perceptual experiences. What is the significance of this fact? Consider our “first-​person conception” of ourselves. This is the conception of ourselves that we form through our use of faculties such as introspection, proprioception, and memory; faculties the deliverances of which are only available to ourselves. It seems that certain epistemically special features of the faculties that contribute to our first-​person conception might reveal some interesting facts about that conception. Thus, Gareth Evans suggests that the IEM feature of judgments based on the deliverances of proprioception, for example, shows that, by exercising that faculty, we are presented to ourselves in a certain way, namely, as the bearers of properties such as having a body and occupying space. And this outcome is, in turn, meant to tell us something important about our first-​person conception of ourselves. According to Evans, it tells us that this conception not only includes the fact that we are thinking things, or the bearers of mental properties, but it also includes the fact that we are the bearers of physical properties: The considerations of this section tell against the common idea that our conception of ourselves “from the first-​person perspective” is a conception of a thinking, feeling, and perceiving thing, and not necessarily a physical thing located in space. . . . Thus the cases of immunity to error through misidentification that we have considered in this section reveal that our conception of ourselves is firmly anti-​Cartesian: our “I”-​ideas are Ideas of bearers of physical no less than mental properties.29

Now, if the reason why memory judgments are IEM is that we always remember being the subjects whose extrinsic properties were perceived in the past, then we may draw an analogous lesson from the fact that those judgments enjoy such an immunity. The lesson is that our first-​person conception of ourselves does not only include the fact that we are thinking things,

29 In (1982, 224).

170  The Epistemology of Memory or bearers of mental properties, as Descartes may have suggested upon reflecting on the nature of introspection. And our first-​person conception of ourselves does not only include the fact that we are the bearers of physical properties such as being extended in space, as Evans suggests upon reflecting on the nature of proprioception, either. Our first-​person conception of ourselves also includes the fact that we are the bearers of temporal properties. Our first-​person conception of ourselves, in other words, is the conception of an object which has a history. For if memory judgments are IEM for the reasons discussed in this chapter, memory does not only give us evidence of the past. Memory gives us evidence of our own existence in the past as well. Ultimately, that is the significance of the immunity to error through misidentification phenomenon in memory.

7 Memory as a Generative Epistemic Source 7.1  Introduction In this chapter, I seek to clarify the epistemological role of memory further by addressing the question of whether memory only preserves epistemic justification over time or can also generate it.1 The received view on this issue is that memory is a source of justification, but only a preservative source. Memory, we are told, cannot produce a subject’s justification for a belief, just maintain it over time. Thus, if a subject is justified in believing some proposition on the basis of memory, it is because they have been, at some point in the past, justified in believing it through some source other than memory. Let us call this view “preservativism.”2 The opposing view is that a subject may be justified in believing some proposition on the basis of memory regardless of whether, in the past, they were justified in believing it through some other source. Memory can, in that sense, create a subject’s justification for a belief. Let us refer to this view as “generativism.”3 My aim in this chapter is to offer a defense of generativism based on a certain view about mnemonic content; the reflexive view defended in c­ hapter 3. I will proceed as follows. I will begin by distinguishing, in section 7.2, several varieties of preservativism and the corresponding varieties of generativism. I will then isolate the variety of generativism which will occupy us here and motivate its significance. In sections 7.3–​7.6, I will consider three defenses of generativism. The first defense, to be explored in section 7.3, employs the 1 An earlier version of this chapter, not including the material on the abstraction defense of generativism and the material on the reconstruction defense of it, can be found in (Fernández 2016). 2 In this context, a “source of justification” is understood as a faculty the deliverances of which can provide its subject with justification for belief. Thus, by calling memory a preservative source of justification, the preservativist is granting that our memories can provide us with justification for beliefs formed on their basis. The preservativist is not denying that memory can “produce” or “generate” justification in that sense. Preservativism with regards to epistemic justification is endorsed, for example, in (Plantinga 1993, 61), (Owens 2000, 157), and (Senor 2007). With regards to knowledge, preservativism is endorsed, for example, in (Dummett 1994, 262) and (Audi 1997, 410). 3 Generativism with regards to epistemic justification is defended, for example, in (Lackey 2005, 2007), (Michaelian 2011), and (Bernecker and Grundmann 2017).

172  The Epistemology of Memory notion of attention and, thus, I will refer to it as the “attention defense” of generativism. The second defense, to be explored in section 7.4, employs the notion of abstraction and, accordingly, I will refer to it as the “abstraction defense” of generativism. We will see that both the attention defense and the abstraction defense are limited in that they only establish that memory is a source of belief, not justification. In section 7.5, I will suggest that this limitation is due to the fact that both defenses of generativism are built within the constraints of a certain conception of memory. According to this conception, memory records and reproduces, sometimes in a simplified way, the contents of experiences that we had in the past. Our diagnosis of the limitations of the attention defense and the abstraction defense will point us in the direction of challenging this conception of memory. Thus, the view that memory generates, and not simply preserves, content over time will be explored at that stage. In section 7.6, I will consider a defense of generativism built on the idea that memory is not bound by the contents of our past experiences. Instead, memories are the result of editing, or reconstructing, the contents of the past experiences in which the memories originate. I will refer to the defense of generativism based on this idea as the “reconstruction defense.” We will see that the reconstruction defense of generativism does not succeed in establishing generativism any more than the attention defense or the abstraction defense does. However, my objections to the reconstruction defense of generativism will only concern the reasons why, according to the reconstruction defense, content gets generated in memory. I will not take issue with the idea of content generation per se. In fact, I myself will explore the general strategy of establishing generativism about epistemic justification by appealing to the idea that content is generated, and not only preserved, in memory. In section 7.7, I will rehearse the “reflexive” version of the causally self-​ referential approach to memory, according to which our memories represent their own causal histories. The proposal, as the reader will recall, is that our memories represent themselves as originating in past perceptions of objective facts. I will argue that if this conception of memory is correct, then what we may believe on the basis of memory always includes something that we were not in a position to believe before we utilized that capacity. The reflexive version of the causally self-​referential approach to memory is, in that sense, a content-​generating view. For that reason, it suggests that memory produces new grounds, or evidence, for belief through the process of remembering. It is on that outcome that the case for generativism will eventually rest. However,

Memory as a Generative Epistemic Source  173 a certain concern about the type of generativism being established will arise at that point. The concern is that the relevant type of generativism might not capture the spirit of the original debate. This concern will be addressed in section 7.8.

7.2  Varieties of Preservativism Preservativism comes in different flavors depending on how we choose to specify some of the notions involved. As a result, there are a number of possible readings of the opposing generativist view as well. It will therefore make for clarity if we begin our discussion by differentiating our project here from other projects on the generative role of memory in the literature. Preservativism has been introduced as the view that if a subject is justified in believing some proposition on the basis of memory, then they are justified because they have been, at some point in the past, justified in believing that proposition through some source other than memory. To specify the reading of this view that will concern us in the coming discussion, it will be useful to draw three distinctions regarding the notions of memory, belief, and justification. The first distinction that may help us situate our project is a distinction that we drew at the onset of this investigation, in ­chapter 1. It concerns two types of propositional memory. One may remember some proposition in the sense that, at some point in the past, one believed that it was the case and that belief has been preserved up to the present time by memory. We have been calling this type of remembering “semantically remembering.” One may also remember some proposition in the sense that one is having a quasi-​perceptual, or imagistic, experience wherein the relevant proposition is presented to one as having been the case. We have been calling this type of remembering “episodically remembering.” The distinction between semantic and episodic memory yields two senses in which one may believe a proposition on the basis of memory. One may believe a proposition on the basis of memory in the sense that one remembers the proposition semantically and, therefore, believes it. (We may call this believing a proposition “on the basis of semantic memory.”) Alternatively, one may believe a proposition on the basis of memory in the sense that one remembers the proposition episodically and, on the basis of one’s memory experience, one forms a belief in that proposition. (We may call this believing a proposition “on

174  The Epistemology of Memory the basis of episodic memory.”)4 In this investigation of memory, we have been focusing on episodic remembering. Thus, the type of preservativism that will concern us in this chapter is, roughly speaking, the view that if a subject is justified in believing some proposition on the basis of episodic memory, then they are justified because they have been, at some point in the past, justified in believing that proposition through some source other than episodic memory. Notice, now, that there are two ways in which one may construe the notion of believing. Believing may be understood as an act; the act of forming a belief. It may also be understood as a state; the state of holding, or having, a belief. The distinction yields two senses in which one may read “justified in believing” in the formulation of preservativism for episodic memory offered above. A subject may be justified in believing a proposition in the sense that they are justified in forming the relevant belief, whether or not they actually form it. A subject may also be justified in believing a proposition in the sense that they are justified in having the relevant belief, which requires them to have actually formed it. In this chapter, I will concentrate on the type of justification that applies to belief formation. Thus, I will be concerned with the variety of preservativism according to which a subject who is justified in forming a belief on the basis of episodic memory is justified because, at some point in the past, they were justified in forming that belief through some source other than episodic memory. This characterization of preservativism for episodic memory is more precise than its initial formulation, but it does not yet distinguish two sorts of epistemic justification. One may be justified in forming a belief in the sense that one has adequate grounds, or good evidence, for forming the belief in question. And one may be justified in forming a belief in the (stronger) sense that one has good evidence for forming the belief and, furthermore, there 4 This is the distinction drawn, for instance, by Matthew McGrath in (2007, 6). As noted in ­chapter 1, episodic and semantic memory can come apart and, as a result, the two ways of believing a proposition on the basis of memory can come apart as well. Suppose that, while in school, I learnt that James Cook charted the eastern coast of Australia in 1770, and my memory has preserved the belief that he did until now. In this case, I now believe that James Cook charted the eastern coast of Australia in 1770 on the basis of semantic, but not episodic, memory. Conversely, suppose that my parents take me to the zoo as a child, and I see a Tasmanian devil eating there. However, I lack the concept of a Tasmanian devil, so I do not believe that there is a Tasmanian devil eating at the zoo during my visit. Years later, however, I have a memory experience that originates in my past perception of the Tasmanian devil there and, having acquired the concept of a Tasmanian devil in the meantime, I form the belief that there was a Tasmanian devil eating at the zoo during my visit on the basis of that memory experience. In that case, I believe that there was a Tasmanian devil eating at the zoo during my visit on the basis of episodic, but not semantic, memory.

Memory as a Generative Epistemic Source  175 is no proposition, either that one believes or that one should believe, which undercuts such evidence. Let us call the two types of justification, respectively, “prima facie” and “ultima facie” justification.5 My main interest in this chapter concerns prima facie justification. In what follows, therefore, I will focus on the issue of whether episodic memory only preserves a subject’s prima facie justification for forming beliefs over time, or whether it can also create such justification. More specifically, the reading of the preservativist thesis that will occupy us in this discussion can be spelled out as follows (where “EP” stands for “epistemic preservation”): EP  For any subject S, proposition p and time t



If S is prima facie justified in forming the belief that p on the basis of episodic memory at t, then there is a time t* earlier than t such that: (i) S is prima facie justified in forming the belief that p at t*. (ii) S’s prima facie justification for forming, at t*, the belief that p is due to a source other than episodic memory. (iii) S is prima facie justified in forming the belief that p at t because of (i) and (ii).6

The distinction between prima facie and ultima facie justification and the distinction between justification for forming beliefs and justification for having them cut across each other. This suggests that, in addition to the EP variety of preservativism, there are three other varieties of preservativism for episodic memory to be distinguished. Generativism for episodic memory can in turn be conceived as the rejection of either of those four varieties of preservativism. What is, then, the significance of the particular variety of generativism which amounts to a rejection of EP? The significance of this variety of generativism concerns the issue of whether memory is a basic source of justification or not. A source of justification is basic if the justification that it yields for forming a belief does not depend on whether the subject possesses justification for forming beliefs 5 For more on this distinction, see (Senor 1996). 6 Notice that EP is formulated under the assumption that epistemic justification is an all or nothing affair. On a framework wherein epistemic justification is conceived of as a matter of degree, preservativism will be the view that episodic memory does not generate any amount of prima facie justification for forming beliefs: If a subject has some degree of justification for forming a belief through episodic memory, it is because they had at least the same degree of justification for forming it through a different source earlier. As far as I can see, the argument against preservativism to be developed in section 7.7 threatens that view as well.

176  The Epistemology of Memory through some other sources. A non-​basic source of justification, on the other hand, depends for its capacity to yield justification on the operation of some of the subject’s other sources of justification.7 For instance, on a reductive conception of testimonial justification, according to which a hearer is justified in forming a belief on the basis of a speaker’s report only if the hearer has positive reasons for accepting the report (reasons, that is, which are obtained independently of testimony), testimony is not a basic source of epistemic justification. By contrast, on a non-​reductive conception of testimonial justification, according to which a hearer is justified in forming a belief on the basis of a speaker’s report so long as there is no evidence against accepting the report, testimony is a basic source of epistemic justification.8 It is important to determine which of our sources of justification are basic and which are not because basic sources can, as it were, make justification on their own whereas non-​basic sources seem to be parasitic on basic sources. It seems, therefore, that work on the metaphysics of justification (including, for example, naturalizing projects on how the normative property of justification might arise from non-​normative properties of the subject) should focus on those faculties which constitute basic sources of justification. That is, after all, where justification comes into being. The thesis that episodic memory generates prima facie justification for forming beliefs tells us something interesting about the epistemic powers of episodic memory. Imagine, for a moment, that we were able to show that EP needed to be rejected. Imagine, that is, that we were able to show that, on the basis of episodic memory, we can be prima facie justified in forming certain beliefs regardless of whether we were prima facie justified in forming those beliefs through another source prior to remembering episodically. Such an outcome would suggest that episodic memory is capable of providing us with some grounds for belief which were not available to us before we utilized that capacity.9 After all, the fact that episodic memory makes some specific contribution in the form of evidence seems to be, in the scenario that 7 A parallel distinction regarding the sources of knowledge (as opposed to justification) is discussed in (Audi 2002). 8 On the epistemology of testimony, including reductionism, see (Coady 1992). 9 The two varieties of generativism according to which episodic memory generates ultima facie justification for having and forming beliefs do not allow us to draw the same conclusion about the epistemic powers of episodic memory. The reason is that ultima facie justification can be created by the fact that past defeating factors are removed from the subject’s environment over time without any need for the subject’s memory to be involved in that removal. (For discussion of an analogous point on semantic memory, see (Senor 2007).) The significance of the variety of generativism according to which episodic memory generates prima facie justification for having beliefs will be addressed in section 7.3.

Memory as a Generative Epistemic Source  177 we are contemplating, the best explanation of why, based on our episodic memories, we can be prima facie justified in forming some beliefs irrespective of whether we were prima facie justified in forming those beliefs through a different source in the past. But if it did turn out that episodic memory can provide us with some grounds for belief which were not available to us before we utilized that capacity, then that upshot would in turn suggest that episodic memory is not parasitic on other sources of justification, such as perception, introspection, or intuition. The conclusion would eventually be that episodic memory is a basic source of epistemic justification. Such a conclusion would constitute a significant result in the epistemology of memory. Let us turn, therefore, to the considerations that one might offer against the EP reading of preservativism.

7.3  Attention in Memory In order to evaluate preservativism, it may be helpful to revisit a certain point about the distinction between episodic and semantic memory. The point is that the two forms of memory can come apart. Episodic memory, in particular, does not need to be accompanied by semantic memory. You may, for instance, seem to perceive that your car keys are in the driver’s door while you are leaving your car parked and not think that they are in the door. Minutes later, you may have a memory that originates in your past perceptual experience of the keys and run back to your car to retrieve them. In that scenario, you episodically remember that your car keys were in the driver’s door when you left your car parked. But you do not remember it semantically, since that requires having formed the belief that they were in the door before you had your memory experience, and you did not form such a belief. (If you had formed it, you would have run back to retrieve the keys by now.) Let us call cases of this sort, cases of “inattentive remembering.”10 One might think that cases of inattentive remembering reveal that preservativism is wrong. Jennifer Lackey, for example, has pursued this idea in an effort to defend generativism. Lackey formulates the particular version of preservativism that constitutes her target thus:11

10 The term “inattentive remembering” is Sven Bernecker’s. See his (2010, 85). 11 In (Lackey 2005, 637). Lackey’s main interest concerns semantic memory, even though her attack on preservativism is meant to generalize to episodic memory as well.

178  The Epistemology of Memory PVM  For any subject S, proposition p, and times T1 and T2:

S knows (justifiedly believes/​rationally believes) that p on the basis of memory at a time T2 only if (i) S knows (justifiedly believes/​rationally believes) that p at an earlier time T1, and (ii) S acquired the knowledge that p (justification with respect to p/​rationality with respect to p) at T1 via a source other than memory. As a counter-​example to this version of preservativism regarding episodic memory, Lackey offers the following “Overloaded Driver” case of inattentive remembering:12 Overloaded Driver Yesterday morning was like most others for Clifford: He spent it drinking coffee, listening to the radio, and driving in his car during his hour-​and-​a-​half commute to work. As was typical for these commutes, Clifford’s attention was divided between the other cars on the road, the surrounding environment, the discussion and music on the radio, and his thoughts about the day’s work. Because of this perceptual and cognitive overload, Clifford found himself, as he often did on these drives, taking in more pieces of information than he actually processed at that time. Indeed, this was made apparent earlier this morning, when Clifford bumped into his friend, Phoebe, at the bakery and started talking about his commute. During this conversation, Phoebe asked him whether construction had begun on I55. Though this is not the way that Clifford takes to work, he does pass it every day and, moreover, it is the route that he occasionally takes to a nearby shopping center. Upon being asked this question by Phoebe, Clifford paused, called to mind passing I55 on his drive to work yesterday, and correctly remembered seeing construction work being done on this freeway. He, therefore, responded affirmatively to Phoebe’s question, adding that he will be sure to map some alternate routes so as to avoid the traffic delays inevitably brought by construction. Prior to the recollection of the visual image triggered by this question, however, Clifford would have continued taking I55 to the shopping center and wouldn’t have made even minor efforts to avoid this freeway.

12 Lackey introduces this case in (2005, 650). She labels it “Overloaded Driver” in (2007, 217).

Memory as a Generative Epistemic Source  179 How exactly is Overloaded Driver meant to refute PVM? The idea seems to be that, due to Clifford’s divided attention during his drive, he happens to take in some information which does not become the content of any of his beliefs at the time. However, during his conversation with Phoebe, Clifford’s memory allows him to form a belief with that information as its content. Assuming that Clifford’s faculties of perception and memory are reliable, then, Clifford justifiedly believes something on the basis of memory which, previous to his remembering, he did not justifiedly believe. Thus, Lackey concludes, PVM is false. In what follows, I will refer to the type of strategy that challenges preservativism by appealing to cases of inattentive remembering as the attention defense of generativism. The attention defense succeeds to a certain extent, but the sense in which it establishes that memory functions generatively seems to be uninteresting. Lackey’s pursuit of the attention defense illustrates this point quite well. Let us assume that, thanks to the reliability of Clifford’s faculties of perception and memory, he justifiedly believes that there was construction work on I55 on the basis of an episodic memory triggered by his conversation with Phoebe. The crucial question for evaluating PVM, then, is whether Clifford justifiedly believed, on the basis of some source other than memory, that construction work had begun on that freeway before he remembered it. Notice, however, that justifiedly believing requires believing. As a result, all it takes for the answer to that question to be negative is that Clifford did not believe that construction work had begun on I55 until he remembered it. Since, by assumption, this is a case of inattentive remembering, Clifford did not have that belief until he spoke to Phoebe and his memory was triggered. Thus, PVM should certainly be rejected. But in what sense does the rejection of PVM establish that memory is generative? Lackey’s use of the notion of justifiedly believing a proposition suggests that Lackey is interested in the kind of justification that a subject has for having beliefs, not for forming them.13 It seems uncontroversial that Overloaded Driver does succeed in showing that episodic memory generates prima facie justification for having beliefs. But the reason why Overloaded Driver succeeds in showing it is simply that the case shows that episodic memory is 13 This reading of “justifiedly believing” is debatable. For if one takes the bearers of epistemic justification to be beliefs, rather than subjects, then an alternative reading is available. According to it, a subject justifiedly believes a proposition just in case she believes it, and her belief is justified. The alternative reading, however, also requires that a subject must believe those propositions which she justifiedly believes. Thus, the reply to the attention defense offered here will still apply if Lackey uses “justifiedly believing” in that sense.

180  The Epistemology of Memory capable of generating beliefs. What explains why Clifford satisfies the antecedent of PVM when he forms his belief about I55 after talking to Phoebe, but he does not satisfy clause (i) in its consequent, is not that Clifford’s epistemic status vis à vis the content of his belief improves after the conversation triggers Clifford’s memory. Instead, the explanation lies in the fact that Clifford has no belief about the construction work on I55 until his conversation with Phoebe takes place (together, that is, with some assumptions about the reliability of Clifford’s faculties). The lesson that can be extracted from the fact that episodic memory is a generative source of prima facie justification for having beliefs, therefore, is that episodic memory is doxastically generative. Unfortunately, that lesson is not very exciting. Once it has been noticed that a subject may form a belief on the basis of episodic memory without forming it on the basis of semantic memory, the view that episodic memory is doxastically generative should be obvious.14 What about the prima facie justification that a subject has for forming beliefs, then? Perhaps Overloaded Driver could be used to show not only that Clifford is prima facie justified in having a belief that has been formed on the basis of episodic memory; a belief that he lacked before he spoke to Phoebe. Perhaps Overloaded Driver could also be used to show that, after their conversation triggers Clifford’s memory, Clifford has prima facie justification for forming a belief that he was not prima facie justified in forming before he spoke to Phoebe. The possibility is worth considering. For if Overloaded Driver can be used in this way, then the attention strategy may be all we need to refute EP. It does not seem, however, that this line of attack against preservativism will succeed. The natural way for the advocate of the attention strategy to use Overloaded Driver as a counter-​example to EP is to claim that, after talking to Phoebe, Clifford is justified in forming the belief that there is construction work on I55 on the basis of episodic memory, so he satisfies the antecedent of EP. But, the thought continues, there is no time prior to that conversation at which Clifford is justified in forming his belief through another source, so he does not satisfy clauses (i) and (ii) in its consequent. This seems highly counter-​intuitive. Notice that the advocate of the attention defense needs to assume that, at the time of his drive, Clifford’s senses are reliable. (Otherwise, it will be challenging to establish that Clifford does satisfy the antecedent of EP.) But if Clifford’s senses are assumed to be reliable at the time of his drive,

14 This seems to be the concern raised, for example, in (Bernecker 2010, 99) and (Senor 2007, 199).

Memory as a Generative Epistemic Source  181 then it is hard to see why we should not grant to Clifford that the perceptual experience that he is having when he sees the works on I55 does prima facie justify him in forming the belief that construction work has begun there. Intuitively enough, it seems that Clifford was prima facie justified in forming the belief that construction work had begun on I55 when he drove past that freeway, and it seems that he acquired his justification through perception. The upshot is that we cannot rely on Overloaded Driver, or any similar case of inattentive remembering, to challenge EP. And yet, considering the attention defense as a potential strategy has not been fruitless. Notice an important feature of Overloaded Driver. This is a feature which was highlighted earlier in order to explain why, despite succeeding as a counter-​example to PVM, the case did not succeed in establishing an interesting type of generativism. It was pointed out that the reason why Clifford satisfied the antecedent of PVM when he formed his belief about I55 after talking to Phoebe, but he did not satisfy its consequent, was not that Clifford’s epistemic status vis à vis the content of his belief had improved after the conversation triggered Clifford’s memory. This is why we had the intuition that no epistemic justification for forming the belief about I55 had been generated in this case. It seems natural, then, to turn to a case in which the subject’s epistemic position vis à vis the content of their belief improves through the use of memory as a strategy for challenging preservativism. Let us explore this kind of strategy now.

7.4  Abstraction in Memory There is an interesting aspect of episodic memory, as far as the relation between memory and perception is concerned, which may be of some help for the generativist. Episodic memory seems to have the power to simplify the amount of information that we registered when, in the past, we perceived some fact while conserving, at the same time, the most salient details about that fact. Suppose, for example, that I am standing in front of a building, and I am looking at its façade. As it happens, the building is exactly 26 stories high, it has 140 windows, and it is made of crimson brick. Suppose now that, weeks later, I episodically remember the façade of the building. The memory image that I entertain while I remember the building may not be detailed enough to allow me to remember that the building was 26 stories high, or that it had 140 windows, or that it looked a particular shade of red to me (namely, crimson). It may simply be the memory image of a big red building

182  The Epistemology of Memory which has more windows than I can easily count. This is not an unusual scenario. In fact, the loss of detail in our memories with respect to the perceptual experiences in which the memories originate seems to be the rule rather than the exception, as we saw in c­ hapter 2. Let us refer to those cases in which a memory of some fact contains fewer details about the fact than the experience in which the memory originates as abstraction cases. It could be argued that the mnemonic operation of simplifying the amount of information originally taken in through a perceptual experience, or abstracting from the details of that experience, has an epistemic benefit for the subject. And, given this benefit, memory makes the epistemic position in which a subject stands to a remembered fact better than the epistemic position that the subject occupied at the time that they originally perceived the fact. Thus one might argue that, in abstraction cases, the subject’s memory has generated epistemic justification for forming their belief about the relevant fact by performing this operation of simplification or abstraction. Sven Bernecker and Thomas Grundmann, for example, have pursued this thought in an effort to motivate generativism. Bernecker and Grundmann formulate the particular version of generativism which concerns them thus:15 Robust Epistemic Generativism There is a subject S, proposition p and times T1 and T2, such that: S can justifiedly believe (know) at time T2 that p on the basis of memory even if the prima facie justification (warrant) of S’s belief at T2 is not solely due to evidence (justificatory factors) that S already possessed at T1, and that was provided by a source other than memory.

In support of this version of generativism, Bernecker and Grundmann offer the following “Seat Counter” abstraction case:16 Seat Counter Jim is sitting in the middle of a large, oddly shaped movie theater waiting for the show to start. To pass time he decides to figure out how many people can be seated in the theater. He counts the number of seats in the front row and the number of rows. Due to the dim lighting in the theater Jim can’t be certain

15 In (Bernecker and Grundmann 2017, 4). 16 In (Bernecker and Grundmann 2017, 5).

Memory as a Generative Epistemic Source  183 that he correctly counted the number of seats in the front row and that every row has the same number of seats as the front row. Then he multiplies the two numbers. On the basis of a rough calculation Jim comes to believe that the theater has 660 seats. As a matter of fact there are 660 seats. But Jim is lucky in that he could have easily miscounted the seats in the front row as well as the number of rows. Also it just so happens that every row has the same number of seats as the front row (the row he based his rough calculation on). A few weeks later, Jim has forgotten the exact number of seats in the theater but still remembers that the theater has more than 500 seats. On the basis of his memory, Jim comes to believe that the theater has more than 500 seats.

How exactly is Seat Counter meant to illustrate Robust Epistemic Generativism? Bernecker and Grundmann assume, first of all, a reliabilist conception of justification, according to which a belief is justified if it has been produced by a reliable belief formation process. A belief formation process is, in turn, taken to be reliable if it tends to produce true beliefs in the actual world, and in close possible worlds.17 Now, it seems that Jim’s belief formation process at the time that he is sitting in the theater (that is, counting poorly and rough calculation) is, by these standards, not reliable. After all, consider what happens in a close possible world in which not every row has the same number of seats as the front row. And consider, furthermore, what happens in those close possible worlds in which Jim miscounts either the number of seats in the front row or the number of rows. In all of those close possible worlds, Jim still calculates that the theater has 660 seats, and he is wrong. Thus, Bernecker and Grundmann conclude, Jim’s belief that the theater has 660 seats is not justified. By contrast, take Jim’s belief formation process at the time that he remembers the theater. The process is that of forming a belief on the basis of an abstract, or simplified, memory. In this particular case, the memory operation of simplifying information has increased, Bernecker and Grundmann argue, the reliability of Jim’s belief formation process. It acts as a corrective for errors that Jim could have easily committed when he counted the seats while sitting in the theater. After all, consider, once again, what happens in the close possible world in which not every row has the same number of seats as the front row. And consider, once again, what happens in those close possible worlds in which Jim miscounts either the number of seats in

17 In (Bernecker and Grundmann 2017, 6).

184  The Epistemology of Memory the front row or the number of rows. When, in those close possible worlds, Jim remembers the theater, he has an incomplete memory of it. And, on the basis of that memory, Jim forms the belief that the theater has more than 500 seats, which is correct. Thus, it seems that Jim has a justified belief on the basis of memory even though, in the past, Jim’s belief was not justified on the basis of other sources. Hence, Bernecker and Grundmann conclude, Robust Epistemic Generativism is correct. In what follows, I will refer to the type of strategy which motivates generativism by appealing to abstraction cases as the abstraction defense of generativism. The abstraction defense succeeds to a certain extent, but the sense in which it establishes that memory functions generatively does not seem to be any more substantive than the sense in which the attention defense does. Before we can appreciate this, however, there is a complication in Seat Counter that we need to address. If we are going to discuss Seat Counter as an example of how abstraction cases may be used to motivate generativism, it should be pointed out, first of all, that Seat Counter does not necessarily concern episodic memory. It could concern semantic memory instead. Notice that, when Jim remembers the number of seats in the theater, he may not be remembering a fact about those seats in the sense of having a quasi-​perceptual experience wherein the theater appears to him to have had such-​and-​such many seats. Jim may be remembering a certain fact about the seats in the theater in a different sense. Perhaps Jim believes that fact to have been the case, and Jim’s belief originates in a previous belief of his about the theater.18 For that reason, even if we grant that Seat Counter can be used to motivate Robust Epistemic Generativism, this case may not provide us with a counter-​example to EP. After all, EP only concerns episodic memory. The interesting question for our purposes, then, is whether there is an abstraction case, essentially structured like Seat Counter, which can be used to motivate Robust Epistemic Generativism by appealing to the subject’s episodic memory unequivocally. The question is interesting because, if we can find such a case, we will have made some progress toward 18 This is in fact my reading of Bernecker and Grundmann’s claim that Jim “still remembers that the theater has more than 500 seats.” Jim’s episode of remembering seems to have originated in one of his past beliefs, the belief that the theater has 660 seats, and not in one of his past perceptual experiences. Which suggests that semantic memory, and not episodic memory, is at work there. It is possible that Bernecker and Grundmann intend their claim that Jim still remembers that the theater has more than 500 seats to be read as referring to Jim’s episodic memory, but nothing in the description of the case shows that Jim is having an episodic memory. Thus, I think it will be helpful to have, in front of us, a version of the case which makes Jim’s use of episodic memory explicit.

Memory as a Generative Epistemic Source  185 challenging EP and establishing the type of generativism which concerns us in this chapter. As far as I  can see, the following “Seat Counter*” abstraction case maintains Bernecker and Grundmann’s use of the notion of abstraction in memory while restricting the alleged epistemic benefits of this operation to Jim’s episodic memory: Seat Counter* Jim is sitting in a large, oddly shaped movie theater. He decides to figure out how many people can be seated in it. Due to the dim lighting in the theater, Jim can’t be certain that he is correctly perceiving each row of seats in the theater. Furthermore, Jim is not particularly good at making fine-​grained distinctions as to how many items his perceptual experience is presenting to him when he perceives large collections of items of a single type. Despite his lack of skill in this area, and despite the poor lighting conditions, Jim estimates, on the basis of his perceptual experience of the theater, that the theater has 660 seats. As a matter of fact, there are 660 seats in the theater. But Jim is lucky in that he could have easily misperceived some row of seats. And he is lucky in that, even if he had not misperceived anything, he could have easily misjudged how many seats he was perceiving in the theater. A few weeks later, Jim has forgotten the exact number of seats in the theater but he conserves a simplified memory of what the theater looked like at the time. This memory lacks the detail of Jim’s original experience, but it preserves information about the rough arrangement of rows of seats in the theater, and the approximate size of those rows. On the basis of this memory, Jim comes to believe that the theater has more than 500 seats.

It seems that Seat Counter* could be used to make a case for Robust Epistemic Generativism analogously to how Seat Counter was used by Bernecker and Grundmann for that purpose. The line of reasoning would go as follows. Jim’s belief formation process at the time that he is sitting in the theater (that is, perceiving in poor lighting conditions and discerning how many items one appears to be perceiving) is not reliable. After all, consider what happens in a close possible world in which one row in the theater has an extra seat, but this seat is not clearly visible to Jim in the dim lighting. In that close possible world, Jim’s perceptual experience is not noticeably different. Thus, he still estimates that the theater has 660 seats, and he is wrong. Likewise, consider what happens in a close possible world in which one row in the theater

186  The Epistemology of Memory has an extra seat, and this seat is clearly visible to Jim. In that close possible world, Jim’s perceptual experience is different, but he is, by assumption, not good at noticing such small differences in his perceptual experiences. Thus, he still estimates that the theater has 660 seats, and he is wrong. Assuming Bernecker and Grundmann’s reliabilist conception of justification, therefore, Jim’s belief that the theater has 660 seats is not justified. By contrast, take Jim’s belief formation process at the time that he remembers the theater. The process is that of forming a belief on the basis of an abstract, or simplified, memory of the theater. The memory operation of simplifying information has increased, the thought goes, the reliability of Jim’s belief formation process. It acts as a corrective for errors that Jim could have easily committed when he estimated the number of seats in the theater on the basis of his perceptual experience of it. After all, consider, once again, what happens in those two close possible worlds in which one row in the theater has an extra seat (not clearly visible to Jim in one of the two worlds; clearly visible to Jim in the other one). When, in those close possible worlds, Jim remembers the theater, he has an abstract memory of it. And, on the basis of that memory, Jim forms the belief that the theater has more than 500 seats, which is correct. Thus, it seems that Jim has a justified belief on the basis of memory even though, in the past, Jim’s belief was not justified on the basis of other sources. Hence, one might conclude, Robust Epistemic Generativism is correct. One difficulty for this line of reasoning regarding Seat Counter* is the following. Let us grant, for the sake of the argument, that Jim has, in the present, a justified belief on the basis of episodic memory. The belief that Jim has formed on the basis of his abstract episodic memory of the theater is the belief that the theater has more than 500 seats. What it takes for this case to constitute a counter-​example to Robust Epistemic Generativism is that Jim’s justification for his belief that the theater has more than 500 seats is not solely due to evidence that Jim already possessed through another source when he was sitting in the theater. The crucial question, then, is whether Jim possessed, through a source other than memory, justification for the belief that there were more than 500 seats in the theater when he was sitting there. And, intuitively enough, it seems that he did. Consider the perceptual experience on the basis of which Jim formed his belief that the theater has 660 seats. This perceptual experience of the theater seems to be providing him with justification for forming the more conservative belief that the theater has more than 500 seats. After all, had he formed

Memory as a Generative Epistemic Source  187 that belief, Jim would not have easily gone wrong. Notice that, in the two close possible worlds in which a row in the theater has an extra seat (not clearly visible to Jim in one of the two worlds; clearly visible to Jim in the other one), it is false that there are 660 seats in the theater, but it remains true that there are more than 500. The intuition that, when Jim is sitting in the theater, Jim’s perceptual experience is providing him with justification for forming the belief that the theater has more than 500 seats is perhaps not surprising. For we are conceding to the generativist that, when Jim remembers the theater, his episodic memory of the theater provides him with justification for forming the belief that the theater has more than 500 seats. And that episodic memory is, by assumption, less detailed than the perceptual experience in which it originates. It is hard to see, then, how the less detailed experience (that is, Jim’s memory) can justify Jim in forming the belief that the theater has more than 500 seats, while the more detailed experience (that is, Jim’s past perceptual experience) failed to justify Jim in forming the very same belief. Let us consider, then, what an abstraction case like Seat Counter* shows, and what it does not show. It seems that Seat Counter* does not establish Robust Epistemic Generativism, for the reasons just mentioned. And, for the same reasons, it seems that the case fails as a counter-​example to EP. If these considerations are correct, then Jim has prima facie justification for forming the belief that the theater has more than 500 seats at the time that he is sitting there. And the source of Jim’s prima facie justification for forming this belief is different from memory. Thus, Seat Counter* is consistent with the EP claim that if a subject has prima facie justification for forming a belief on the basis of episodic memory, it is because the subject had, at some point in the past, justification for forming that belief through another source. The case does not show, therefore, that memory can provide us with prima facie justification for forming beliefs. As we saw in section 7.3, however, justifiedly believing requires believing, and Jim does not believe that the theater has more than 500 seats while he is sitting there. Thus, Jim has, in the present, a belief justified on the basis of episodic memory which he did not justifiedly have in the past. For that reason, Seat Counter* does succeed in showing that memory can provide us with prima facie justification for having beliefs. The outcome of the discussion in the last two sections, therefore, seems to be that both the attention strategy and the abstraction strategy can motivate a certain variety of generativism. But the relevant variety of generativism only teaches us that we have the ability to form beliefs on the basis of our episodic memories; something that

188  The Epistemology of Memory we knew all along. To the extent that our interest in the preservativism versus generativism debate concerns the question of whether memory is a basic source of epistemic justification, it seems that we should continue to strive for a more substantial variety of generativism.

7.5  The Recorder Model of Memory Our discussion of the attention defense and the abstraction defense of generativism brings to light an interesting aspect of the two strategies. In inattentive remembering cases, what the subject believes on the basis of memory is the same as what they perceptually experienced, albeit inattentively, in the past. In abstraction cases, what the subject believes on the basis of memory is less than, or is included in, what they perceptually experienced in the past. In both inattentive remembering and abstraction cases, therefore, the subject has a belief with a content that they had already entertained prior to their episode of remembering; just not through the attitude of belief. The content of the subject’s memory belief is, either partly or entirely, the content of one of their past perceptual experiences. In Overloaded Driver, for example, Clifford’s relevant belief is the belief that construction work has begun on I55, and Clifford’s relevant past experience is the perceptual experience that he undergoes when he looks at that freeway during his drive. Likewise, in Seat Counter*, Jim’s relevant belief is the belief that the theater has more than 500 seats, and Jim’s relevant past experience is the perceptual experience that he has when he looks at the theater while he is waiting for the movie to start.19 In both cases, what the subject believes on the basis of memory is something that they had already experienced in the past. This feature of both inattentive remembering cases and abstraction cases seems to be the aspect of the two strategies which hinders their prospects of establishing the generative role of memory: The reason why we have the intuition that Overloaded Driver, for example, is a case in which, previous to the subject’s episode of remembering, he was prima facie justified in forming the belief that he later forms on the basis of memory is that, before Clifford’s memory is triggered, he had apparently perceived that which he 19 Admittedly, the description of Seat Counter* does not specify what the content of Jim’s perceptual experience is when he looks at the theater. But it seems reasonable to assume that if Jim’s perceptual experience is leading him to believe that the theater has 660 seats, then the content of his perceptual experience is approximate enough to there being 660 seats in the theater.

Memory as a Generative Epistemic Source  189 later remembers episodically. It is precisely because we think that Clifford seems to perceive the construction work on I55 when he drives past that freeway that we are inclined to attribute to Clifford, at that point, prima facie justification for forming the belief that construction work has begun on I55. Similarly, the reason why we have the intuition that Seat Counter* is a case in which, previous to the subject’s episode of remembering, he was prima facie justified in forming the belief that he later forms on the basis of memory is that, before Jim acquires his memory belief, Jim had apparently perceived that which he later remembers episodically. It is precisely because we think that Jim seems to perceive the theater as having more than 500 seats when he is sitting there that we are inclined to attribute to Jim, at that point, prima facie justification for forming the belief that the theater has more than 500 seats. If this is correct, then there is a heuristic lesson that we can learn from our intuitions about inattentive remembering cases and abstraction cases. The lesson is that, for a strategy to succeed in vindicating generativism, it must offer a counter-​example to EP which involves a memory belief with a particular type of content; a content that the subject did not entertain previous to their episode of remembering. On reflection, it is easy to see why such counter-​examples will not come cheap. There is a reason why the attention strategy and the abstraction strategy rely on cases in which the subject believes, on the basis of memory, something that they had already experienced before they remembered it. According to an influential picture of memory, episodic memories store the information that, in the past, we acquired through some of our past experiences, namely, those experiences in which our memories originate. For that reason, the contents of our episodic memories are supposed to be copied from their corresponding past experiences (with a varying degree of detail, that is, depending on the vivacity of each episodic memory). Let us refer to this picture as the “recorder model” of memory.20 What exactly is meant by “copying” and “storing” within this picture? The thought is that anything present in the content of an episodic memory must be present in the content of the subject’s corresponding past experience. As a result, the 20 It is important to distinguish the view that episodic memory only maintains epistemic justification provided by other sources over time from the view that episodic memory stores the contents of our experiences over time. In this chapter, I  have been referring to the former view as “preservativism.” Hereafter I will refer to the latter view as “the recorder model” of episodic memory, to avoid confusion between the two views. Notice that the two views are not logically equivalent. In particular, the recorder model of memory does not entail preservativism: The abstraction defense of generativism, for example, is consistent with the recorder model of memory.

190  The Epistemology of Memory content of the memory cannot exceed the content of the corresponding past experience. An episodic memory may provide its subject with less information than the experience in which it originates, but not with more. We may refer to this idea, which constitutes the main tenet of the recorder model, as the “content preservation constraint.” The view that memory must respect the content preservation constraint enjoys a long tradition in the philosophy of memory. It is subscribed, for example, by Thomas Reid when he writes: 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. Our first acquaintance with any object of thought cannot be by remembrance.21

If the recorder model of memory is correct, then the fact that, in both inattentive remembering cases and abstraction cases, the subject who is forming a belief on the basis of one of their episodic memories has entertained the content of that belief through some other experience previous to their remembering is not a peculiar feature of those cases. It is an essential feature of any instance of episodic memory. When a subject forms a belief on the basis of an episodic memory, that memory must, on this picture, receive its content from one of the subject’s past experiences, namely, the experience in which the memory originates. This requirement is supposed to apply whether the subject’s situation qualifies as a case of inattentive remembering, or as an abstraction case, or as neither. It is no wonder, then, that inattentive remembering cases and abstraction cases always turn out to be cases in which, previous to the subject’s remembering, the subject had already entertained the content of their episodic memory (and, thus, the content of the belief that they form on the basis of it) through an experience of a different kind. Quite simply, as Reid puts it, the subject could not have remembered it otherwise. The diagnosis of why the attention defense and the abstraction defense of generativism are limited that emerges from these considerations is that both defenses are built within the constraints of the recorder model of memory. This picture restricts the possible scenarios available to the advocates of both

21 Reid (1969, 326). Presumably, Reid is not distinguishing between episodic and semantic memory in this passage; hence the claim that things remembered must be formerly perceived (in the episodic case) or known (in the semantic case).

Memory as a Generative Epistemic Source  191 defenses to those cases in which the subject forms, on the basis of an episodic memory, a belief about something that they had already experienced, typically through perception, before they remembered it. And, as we have seen, such cases are not likely to refute EP. Notice that the implications of this diagnosis reach further than the attention defense, and further than the abstraction defense, of generativism. What this diagnosis tells us is that any defense of generativism which incorporates the view that episodic memory only stores the contents of past experiences over time will face the same difficulties as those of the attention defense and of the abstraction defense. Thus, the moral to draw from our discussion of both strategies is that refuting EP is going to require abandoning some assumptions that seem deeply entrenched in our philosophical thinking about memory. Specifically, it will take a departure from the idea that the content of an episodic memory is inherited from the experience in which the memory originates. Let us explore, therefore, the possibility of rejecting this idea now.

7.6  Reconstruction in Memory There is a popular conception of memory within psychology according to which memory is not a passive device for registering and replicating contents. It is instead a faculty akin to imagination in its productive capacity. The main tenet of this “reconstructive” picture of memory is that, in memory, we are engaged in an inventive project wherein we build representations of our past by integrating content that we have acquired through our own perceptual experience with content from other sources, such as testimony, inference, and the imagination. Daniel Schacter and Donna Addis, for example, write that memory “is not a literal reproduction of the past, but rather is a constructive process in which bits and pieces of information from various sources are pulled together.”22 Similarly, Elizabeth Loftus describes this picture as “a new paradigm of memory, shifting our view from the video-​recorder model, in which memories are interpreted as the literal truth, to a reconstructionist model, in which memories are understood as creative blendings of fact and fiction.”23 The reference to an element of fiction in the integration process is

22 In (Schacter and Addis 2007, 773). 23 In (Loftus and Ketcham 1994, 5). On the reconstructive picture of memory, see (Barclay and De Cooke 1988) as well.

192  The Epistemology of Memory telling. For memory, on this picture, does not need to represent the past as we experienced it. As a result, on the reconstructive picture of memory, episodic memories do not need to comply with the content preservation constraint. On this picture, an episodic memory may provide its subject with less information than the experience in which it originates in some respects, and it may provide its subject with more information than the original experience in others. The reconstructive picture of memory can be developed in a number of ways. In c­ hapter 2, we discussed the costs and benefits of having a conception of memory which rejects the content preservation constraint. We saw that one particular version of the reconstructive picture (that is, the narrative conception of memory) is both too liberal and too restrictive in its judgments of what qualifies as a memory. It is worth, therefore, being cautious about how one fills in the precise details of the reconstructive picture of memory. For the purposes of this section, however, I propose that we set these metaphysical worries aside. Let us concentrate, instead, on the question of whether the reconstructive picture of memory can deliver some cases which constitute counter-​examples to EP or not. To be sure, if the cases delivered by the reconstructive picture turn out to be prima facie counter-​examples to EP, then we will need to address the question of whether the relevant cases involve episodic memories or not, given the criteria for episodic remembering put forward in ­chapter 2. But if no counter-​example to EP is eventually found, then we will not need to concern ourselves with the question of whether the mental states involved in those cases qualify as episodic memories. On the face of it, the fact that, on the reconstructive picture of memory, the content preservation constraint can be dropped invites optimism about the prospects of finding a counter-​example to EP. Recall that the reason why both inattentive remembering cases and abstraction cases fail as counter-​examples to EP is that, in both types of cases, the subject has a belief with a content that they had already entertained prior to their episode of remembering. The content of the subject’s memory belief is, or is part of, the content of one of their past experiences, namely, the experience in which the relevant memory originates. As a matter of fact, the content preservation constraint guaranteed that this would always be true of those beliefs that we form on the basis of our memories. If the content preservation constraint is dropped, however, we may be able to find cases in which the content of the subject’s memory belief is not identical with (and is not part of) the content of the past experience in which the memory originates. Let us call such cases,

Memory as a Generative Epistemic Source  193 “reconstruction cases.” Given this diagnosis of why inattentive remembering cases and abstraction cases fail as counter-​examples to EP, the possibility that reconstruction cases may provide us with such counter-​examples seems, at this point, worth exploring. Kourken Michaelian has exploited reconstruction cases as part of an effort to challenge preservativism about epistemic justification.24 Michaelian understands preservativism as the view that “the level of justification of the output of memory cannot exceed the level of justification of its input.”25 Michaelian’s main strategy for rejecting this view relies on two points. The first point concerns the intentionality of memory whereas the second point concerns the nature of epistemic justification. The first point is that reconstruction cases are real: In some instances of remembering, Michaelian tells us, new mnemonic content is generated. That is, the memory that the subject is having (the “output” of memory) is representing something that the subject’s past perceptual experience in which the memory originates (the “input” of memory) did not represent. Now, the point that reconstruction cases are real will not, by itself, defeat preservativism. Let us suppose, for the sake of the argument, that there are cases in which, when a subject forms a belief on the basis of one of their episodic memories, the memory at issue represents something that the subject’s corresponding past perceptual experience did not represent. Why should we conclude, from this, that the subject is justified in forming their memory belief to a higher degree than the degree to which, in the past, they were justified in forming the same belief through sources other than memory? We may conclude this, Michaelian tells us, in those instances in which the reconstruction process through which the new mnemonic content has been generated is reliable. Like Bernecker and Grundmann, then, Michaelian espouses a reliabilist conception of epistemic justification. Thus, the second point in this defense of generativism is that a belief is justified just in case it has been generated through a reliable process; a process that tends to generate true beliefs.26 And there are reconstruction cases, we are told, in which the process through which the memory belief has been formed is reliable. For the process through which the new content of the relevant memory has been generated is itself reliable. In such reconstruction cases, Michaelian claims, the level of justification of the output of



24 In (Michaelian 2011, 2013, and 2016b). 25 In (Michaelian 2016b, 94).

26 In (Michaelian 2011, 337).

194  The Epistemology of Memory memory may exceed the level of justification of its input, which would show that generativism is correct. Michaelian offers several examples of such reconstruction cases. Some of them are examples involving the so-​called information effect, in which the subject’s memory of some fact includes information acquired, not through the perceptual experience in which the memory originates, but through the collateral contribution of other sources such as testimony or reasoning.27 In this kind of case, the idea is that, provided that the additional source of information is reliable, the subject may be justified in forming a belief on the basis of their memory even though, in the past, they were not justified in forming the same belief on the basis of the corresponding perceptual experience, which suggests that generativism is correct. Other examples of reconstruction cases offered by Michaelian to motivate generativism are cases of “source monitoring,” in which the subject forms a belief about the origin of the mental state that they are having; a belief about whether the mental state at issue is, for example, a memory versus an episode of imagination. In this kind of case, the thought is that a subject may form the correct belief that a mental state that they are having is a memory based on certain features of that memory; features such as the amount of detail in the memory, its embeddedness in beliefs regarding the spatial and temporal context of the scene being remembered, and its embeddedness in supporting memories. And if the features at issue are reliable indicators of remembering, then the subject may be justified in forming the belief that the mental state that they are having is a memory. And yet, at the time that the subject was having the perceptual experience in which the memory originates, they were not justified in forming the belief that they were having a memory, which suggests, once again, that generativism is correct.28 These two types of cases run into a certain difficulty. They do not seem to be cases in which the subject is, strictly speaking, forming a belief on the basis of one of their episodic memories. In cases in which the information effect takes place, for example, memory is not the only faculty at play. Reasoning or testimony are also bound to be involved. This is, after all, how the subject remembers some information about a fact which was not perceived by them 27 In ­chapter 1, I referred to the effect whereby, after a subject perceives some fact, the acquisition of false information about that fact can make the subject’s memories of it less accurate as the misinformation effect. Instances of the misinformation effect are therefore instances of the information effect. For discussion of the information effect, see (Michaelian 2013). 28 On source monitoring and generativism, see (Michaelian 2011, 328–​329).

Memory as a Generative Epistemic Source  195 at the time that they experienced the fact in the past. It is thanks to the operation of reasoning or testimony that such information is supposed to get built into the subject’s memory of the fact. Likewise, in source monitoring cases, memory is not the only faculty at play. Introspection is involved. After all, it is not through the use of the faculty of memory that the subject gains information about the features of their memories which mark them as such. (The subject is certainly not remembering that, for example, the mental state that they are having has a considerable amount of detail.) For this reason, it does not seem that cases of either type will serve as compelling counter-​examples to a version of preservativism such as EP. Faced with cases involving either the information effect or source monitoring, the preservativist will reply that these are not cases in which the antecedent of EP is being satisfied, since the subject is not forming their belief in some proposition on the basis of episodic memory exclusively. Is this a fair criticism of the reconstructive strategy though? After all, one key feature of reconstructivist views is a refashioning of the boundaries around the memory system.29 These boundaries are usually weakened, and advocates of the reconstructive conception of memory typically consider processes involved in reasoning, introspection, and testimony as parts of the memory system. Thus, advocates of the reconstructive conception of memory could reply that, in the reconstruction cases just considered, we should accept that the subject is forming their belief in some proposition on the basis of episodic memory, provided that our conception of the faculty of memory is appropriately broadened. This is a reasonable response, but it seems under-​developed in an important way. Suppose that memory is a separate faculty from reasoning, testimony, and introspection. Then, one can see the significance of the claim that memory is epistemically generative: Epistemically generative faculties must be basic sources of justification, so the claim that memory is epistemically generative puts memory in the category of basic sources of justification (along with faculties such as perception, for example) and outside the category of derivative sources of justification (which contains faculties such as reasoning). By contrast, suppose that the boundaries of the faculty of memory are expanded so as to include those faculties which are involved in the reconstruction cases discussed earlier. Then, the outcome is a faculty that encompasses, at least, reasoning, 29 This aspect of the reconstructive conception of memory is highlighted in (Michaelian 2016b) and (De Brigard 2014), for example.

196  The Epistemology of Memory testimony, and introspection along with memory. Let us call it “memory+.” Let us suppose, now, that the reconstructivist can show that memory+ is epistemically generative. What is the significance of this claim? The significance of the claim that memory is epistemically generative does not extend to the claim that memory+ is epistemically generative. (Memory+ includes basic sources of epistemic justification so, naturally, it will itself be basic. A discussion of the generativism versus preservativism debate is not necessary for reaching such a conclusion.) The reconstructivist may have a different kind of significance in mind for the claim that memory+ is epistemically generative. But, in the absence of an explanation of what that significance amounts to, it seems prudent not to broaden the boundaries of memory. Now, there is a type of reconstruction case offered by Michaelian which sidesteps the difficulty that threatens source monitoring cases and cases involving the information effect. This type of case involves the memory phenomenon of “boundary extension.”30 Boundary extension cases are cases in which, by remembering some scene, we visualize parts of the scene which were just outside of our visual field when we saw the scene in the past. The evidence suggesting that this is a real phenomenon is, essentially, the following. Subjects are presented with, for example, pictures of a scene with some simple structure in it. The scene could consist in a line of rubbish bins placed against the background of a picket fence, for instance. Later, subjects are asked to draw the scene in the picture as they remember it. It seems to be quite common for subjects to draw details in the scene which are outside of the edges of the picture; details such as the contours of some of the rubbish bins or the continuation of the fence at both ends of the line.31 This kind of evidence suggests that, when we remember a scene with a simple structure in it, we may visualize parts of the scene which were just outside of our visual field at the time that we saw the scene. It is as if, when we episodically remembered a simply structured scene that we perceived in the past, what we did is, so to speak, zoom out slightly with respect to the point of view that we had occupied when we originally perceived the scene. What is the relevance of this phenomenon for generativism? One might think that boundary extension cases could be used to make a case for generativism. The line of reasoning would go as follows. The process

30 Michaelian discusses this type of reconstruction case in (2011, 333–​334). 31 On the phenomenon of boundary extension, see (Intraub and Richardson 1989) and (Intraub, Bender, and Mangels 1992).

Memory as a Generative Epistemic Source  197 through which, when we remember episodically, we fill in the details of the remembered scene which were just outside of our visual field seems to be reliable. The simple structure in the scene, and in the objects which compose it, is the key to this reliability. Suppose that I look at a line of rubbish bins placed against the background of a picket fence. Suppose that, when I look at the bins, the bin placed at the end of the line, on the right side, is not entirely within my visual field, but I can see that it has certain features. The left side of the bin has, let us say, a lined silver surface and a round lid on top. Suppose that, later, I episodically remember the bins, and I visualize the whole bin located at the far right as having those features. Accordingly, I form the belief, on the basis of my memory, that a whole bin with those features was at the end of the line. Now, notice that if an object with a simple structure, such as a rubbish bin, has a visible side with certain features, then, more often than not, the rest of the object will have the same features. In normal circumstances, we are not surrounded by oddly designed objects whose surfaces change randomly and abruptly. Thus, it is likely that my belief that there is a whole bin at the end of the line with a lined silver surface and a round lid on top will turn out to be correct. More generally, it seems that forming a belief about objects that we partly perceived in the past on the basis of episodic memories of those objects which have undergone the boundary extension effect will tend to produce true beliefs. But if this is correct, then my belief that there was a whole bin at the end of the line with a lined silver surface and a round lid on top is, assuming reliabilism, justified. And yet, one might argue, I never saw the whole bin in the past. Thus, perception has not provided me with all the justification that I have, in the present, for forming the belief that there was a whole bin at the end of the line with those features. Does this not show that generativism is correct? I believe it does not. Essentially, the difficulty is that, even though I never saw the whole rubbish bin in the past, it does not follow that, in the past, I was not justified in forming the belief that there was a whole bin at the end of the line with a lined silver surface and a round lid on top on the basis of perception. It does not follow, that is, if we assume reliabilism. Consider the perceptual experience that I was having when I looked at the line of rubbish bins. The bin placed at the end of the line, on the right side, was not entirely within my visual field, but I could see that the left side of the bin had a lined silver surface and a round lid on top. Was I justified, at that point, in forming the belief that the whole bin had those features? Consider the process that consists in forming a belief about the features of a simply structured object

198  The Epistemology of Memory on the basis of a partial perceptual experience of the object. This seems to be a belief formation process which tends to produce true beliefs, that is, a reliable belief formation process. The reason for this is the same as the reason why forming a belief about simply structured objects that we partly perceived in the past on the basis of episodic memories of those objects which have undergone the boundary extension effect tends to produce true beliefs. In normal circumstances, objects with a simple structure which is visible on one side of the object tend to continue to have the same structure on their other sides. Now, if this is correct, it seems that, at the time that I was looking at the line of rubbish bins, I was justified in forming the belief that there was a whole bin at the end of the line with a lined silver surface and a round lid on top on the basis of my partial perceptual experience of the bin. For there was a reliable belief formation process available to me, at that point, through which I was in a position to acquire that belief. Granted, I did not make use of the relevant process. But this is, for the purposes of challenging EP, irrelevant. As we have seen in our discussion of inattentive remembering cases and abstraction cases, what a counter-​example to EP requires is that the subject must not have been justified in forming, through sources other than episodic memory, the very belief that, in the present, they are forming on the basis of memory. And, for the reasons mentioned earlier, boundary extension cases do not seem to meet this requirement. Boundary extension cases are such that, in the past, there was justification for the subject to form the very belief that they are now forming on the basis of memory; justification made available to the subject by sources other than episodic memory. Where does this leave us? Inattentive remembering cases and abstraction cases have taught us that, in order to make a case for generativism, we may need to reject the content preservation constraint on episodic memories. This consideration pushed us away from the recorder model of memory, on which the content of memories cannot exceed that of the perceptual experiences in which they originate, and toward the reconstructive picture of memory, on which it can. However, the lesson from the kind of reconstruction cases that we have considered in this section is that not any sort of content generation will do. If the generation of mnemonic content is due to a process which involves the operation of faculties other than memory, then the generation of such content is not likely to bring with it the generation of epistemic justification in memory. What we need, then, is an alternative form of mnemonic content generation. We need to find

Memory as a Generative Epistemic Source  199 alternative reasons why the content of a subject’s memories can go beyond the content of those perceptual experiences in which the memories originate. In order to avoid the difficulties that we have encountered in this section, the required type of mnemonic content generation will need to satisfy two conditions. On the one hand, it will need to make a certain kind of epistemic justification available to the subject; justification for forming beliefs on the basis of their episodic memories. On the other hand, the relevant memory beliefs will need to be beliefs that the subject was not in a position to form through any other source in the past. Fortunately, our discussion of the nature of mnemonic content in ­chapter 3 has already highlighted this form of content generation for us.

7.7  Mnemonic Content and Generativism Let us remind ourselves of the causally self-​referential approach to the intentionality of memory. Recall that, according to the preferred version of this approach, the intentional objects of our memories are causal relations which involve factive mental states. The main idea in the “reflexive view” of mnemonic content was that if a subject has an episodic memory that they would express by saying that they remember some fact, then what their memory represents is that it causally originates in the subject’s perception of that fact. The view was formulated as follows: RV  Reflexive view

For any subject S, memory M and proposition q: If S has M and S would express M by saying that they remember that q, then there is a perceptual experience P that S would express by saying that they perceive that q, such that the content of M is the proposition {W: In W, M is caused by S having perceived that q through P}. My contention in this section will be twofold. Firstly, I will claim that if episodic memories are, as argued in c­ hapter 3, causally self-​referential, then there is content in our episodic memories over and above that of our corresponding past experiences; content which is specifically generated in memory. Accordingly, I will suggest that the recorder model of memory, discussed in section 7.5, should be rejected. Furthermore, I will argue that if

200  The Epistemology of Memory our episodic memories generate content in this way, then they give us justification for forming certain beliefs which we were not in a position to form through any other source previous to remembering episodically. For that reason, I will suggest, the EP variety of preservativism should be rejected as well. On the view that episodic memories are causally self-​referential, it seems that our memories present to us certain facts which were not initially presented to us in our corresponding past experiences. There is, in other words, more information in the contents of our episodic memories than there was in the contents of the experiences in which those memories originate. To illustrate this point, let us return to the example that we have been using to illustrate the preferred version of the causal self-​referential approach, that is, RV. Suppose that I am looking at a red apple in front of me, and my perceptual experience presents it to me as being red. Suppose that, years later, I have a memory which originates in that perceptual experience; one that I would express by saying that I have a memory of a red apple. Finally, suppose that I have not received any testimony about the red apple before having my memory. Let us refer to my past perceptual experience and my memory as “P” and “M” respectively. How should we construe the content of M? The proposal in RV is that the content of M is the proposition {W: In W, M is caused by my having perceived a red apple through P}. Consider a fact about the causal history of M, namely, the fact that M causally originates in a perception of a red apple. If the content of M is the proposition offered by RV, then this is a fact that I represent in virtue of having M. Is this also a fact that, in the past, I represented in virtue of having the experience in which M originates? Surely not. It is hard to imagine why a fact that partly concerns my future mental life would be presented to me when, upon looking at the red apple, I come to have P. It seems, therefore, that the content of my episodic memory M exceeds that of my perceptual experience P, since the fact that M originates in P is part of the content of M, but not of P. (Analogous considerations will apply to the contents of memories with different causal histories from that of M.)32 Thus, the upshot seems to 32 The fact that M causally originates in a perception of a red apple is not the only fact which is part of the content of M, but not of P. The fact that I am having this very experience, M, is one other such fact. Arguably, the fact that I had P is another, although this may depend on whether perceptual experiences are themselves causally self-​referential. For if perceptual experiences are causally self-​referential, then the fact that I had P was part of the content of P when I had that experience in the past. (On the causally self-​referential approach to perception, see (Searle 1983).) For the sake of simplicity, I will focus on a fact about the causal origin of M in order to challenge the recorder model

Memory as a Generative Epistemic Source  201 be that our episodic memories violate the content preservation constraint, since their contents are not restricted by those of our corresponding past experiences. What does this mean for the recorder model of memory? If our episodic memories carry information which was not included in the contents of the experiences in which those memories originate, then it seems reasonable to conclude that memory does not merely store information acquired through experiences that we had in the past. Instead, memory generates new information in each instance of remembering, which suggests that the recorder model of memory is misguided. The view that episodic memories are causally self-​referential also has damaging consequences for preservativism. According to the EP variety of preservativism, if one is prima facie justified in forming a belief on the basis of episodic memory, it is because, at some point in the past, one was prima facie justified in forming that belief through some other source. The type of counter-​example that I wish to raise for this view is built on the idea that our episodic memories generate, and not simply preserve, content over time. Hence, the counter-​example will qualify as a reconstruction case. However, it is unlike previous reconstruction cases in an important respect. The case to be raised as a counter-​example to EP is not a case in which the generation of mnemonic content is due to the fact that mnemonic content integrates content that we have acquired through our own perceptual experiences with content from other sources, such as testimony, inference, and imagination. It is a case in which the generation of content in memory is simply part of the nature of mnemonic content. To specify the relevant counter-​example, let us consider, once more, the situation in which I have episodic memory M. If M is causally self-​referential, then I seem to have some grounds, or evidence, for forming the belief that my having that experience is due to my having perceived a red apple in the past. Episodic memory M itself is providing me with those grounds, since the fact that M originates in a perception of a red apple is a fact which is presented to me in virtue of having M. Assuming that my capacity for remembering episodically is reliable, then, it seems that, when that capacity delivers M, I am prima facie justified in forming, on the basis of M, the belief that that very experience originates in a perception of a red apple. Thus, at the time at which I have M, I seem to satisfy the antecedent of EP for a certain proposition; of memory, and in order to make a case for generativism, namely, the fact that M causally originates in a perception of a red apple.

202  The Epistemology of Memory the proposition that M originates in a perception of a red apple. But it does not seem that, with regards to that proposition, I manage to satisfy the consequent of EP. For there does not seem to be any previous time at which, through a source other than episodic memory, I was prima facie justified in forming that belief about the causal history of M. Which source of epistemic justification could have allowed me to form such a belief? It is hard to make sense of the idea that M’s causal history could have been either perceived or introspected by me before M took place. It also seems that at no time before I had M could I have inferred, from any of my other beliefs, the proposition that, in the future, I would be episodically remembering the red apple by having M. Finally, by assumption, I did not receive any testimony about the red apple before having episodic memory M either, which rules out the possibility that I might have received testimony about the fact that, in the future, I would be remembering the red apple by having M. It seems, therefore, that we can conclude that, prior to my having M, I did not have any grounds for forming a belief about what the causal history of M was going to be. Which strongly suggests that, before I had episodic memory M, I did not enjoy prima facie justification for forming the belief that M would originate in a perception of a red apple. The conclusion, then, is that the EP variety of preservativism should be rejected and, therefore, generativism regarding prima facie justification for forming beliefs on the basis of episodic memory should be upheld.

7.8  Conclusion The picture of memory with which we are left at this point is an interesting one. It is a picture wherein the epistemically generative role of memory turns out to be grounded on its intentionally generative role. What we have seen is that episodic memory is epistemically generative, but the reason why it is epistemically generative is perhaps not the reason that one might have expected. The reason is that, when we remember facts that we perceived in the past episodically, memory also puts us in cognitive contact with new facts. And this, in turn, allows us to form beliefs which we were not able to form through any other source before we utilized memory. Now, a certain concern may arise about this reason for memory being epistemically generative. One may worry that, even though the conclusion of our discussion is that EP needs to be rejected, the reason why EP needs to be rejected appears to

Memory as a Generative Epistemic Source  203 be consistent with the spirit of preservativism. And if they are indeed consistent, then we have not shown that preservativism is wrong by rejecting EP. In that case, we have only been fighting a strawman all along. Let us return, then, to the thesis that memory does not generate epistemic justification; only preserves it. Admittedly, there is a plausible reading of this thesis which does differ from EP. It is the reading according to which if a subject was in a position to form a belief through some source other than memory in the past, then their current justification for forming the same belief on the basis of memory cannot have improved with respect to the justification that they had in the past through the relevant source. This reading of the preservativist thesis could be spelled out more precisely as follows: EP*  For any subject S, proposition p and time t:

If S is prima facie justified in forming the belief that p on the basis of episodic memory at t, and there is a time t** earlier than t such that S entertained that p at t**, then there is a time t* earlier than t such that: (i) S is prima facie justified in forming the belief that p at t*. (ii) S’s prima facie justification for forming, at t*, the belief that p is due to a source other than episodic memory. (iii) S is prima facie justified in forming the belief that p on the basis of episodic memory at t because (i) and (ii) are the case.33 EP and EP* are meant to express essentially the same idea. The only difference is that EP* restricts the scope of the preservativist thesis to propositions that the subject entertained in the past. For that reason, EP entails EP*. The converse does not seem to be the case though. For the view that episodic memories are causally self-​referential suggests that EP is wrong, and yet it is consistent with EP*. After all, those beliefs which, as argued in section 7.7, falsify EP are not beliefs in propositions that we were in a position to entertain in the past. It does seem reasonable to wonder, then, whether the generativist’s target should have really been EP, or whether the generativist should have targeted, more ambitiously, a weaker formulation of preservativism such as EP*.

33 EP* does not require that if S entertained that p in the past, then every time at which they did must have been a time at which they satisfied (i) and (ii). But every time in the past at which S satisfied (i) and (ii) will have been, in fact, a time at which S entertained that p.

204  The Epistemology of Memory Where has the rejection of EP taken us? The rejection of EP suggests that episodic memory provides us with grounds, or evidence, for forming beliefs for which we had no grounds previous to remembering episodically. Thus, the question of whether targeting EP was the correct approach to take seems to come down to the question of whether this outcome is an interesting enough result about memory. I believe it is, for two reasons. Firstly, as noted in section 7.2, if episodic memory provides us with grounds, or evidence, for forming beliefs for which we had no grounds previous to remembering episodically, then it seems that the type of epistemic justification that episodic memory provides us with is not reducible to the types of justification provided to us by other sources. It is not reducible in that the justification that memory provides for forming a belief does not depend on those other types of justification. What the variety of generativism which amounts to a rejection of EP teaches us, then, is that memory is a basic source of justification. Since EP* is entailed by EP, the rejection of EP* would also lead us to that conclusion. However, the fact remains that, for the purposes of establishing that memory is a basic source of justification, it is sufficient to challenge the stronger preservativist thesis, that is, EP. Furthermore, the reason why episodic memory provides us with grounds, or evidence, for forming beliefs for which we had no grounds previous to remembering episodically shows something important about the scope of memory. If memory is epistemically generative because, when we remember facts that we perceived in the past episodically, memory also puts us in cognitive contact with new facts, then it seems that the scope of episodic memory is broader than that of perception. As a matter of fact, if episodic memories are causally self-​referential, then it seems that the scope of episodic memory is broader than the scope of any of the faculties the deliverances of which give rise to our episodic memories. (It is broader than the scope of introspection, the scope of proprioception, and so on.) Thus, the approach that I have pursued here to challenge EP seems to yield some interesting results about memory not only in epistemology, but also in the philosophy of mind. My intention has been to adopt a concessive approach by focusing on EP. One can challenge the EP variety of preservativism and remain neutral on the truth of weaker varieties of preservativism such as EP*. Which means that, at the conclusion of our discussion, it is still open for the preservativist to retreat to one of those weaker varieties of preservativism. This is not a defeat, since it is possible for such varieties of preservativism to retain some substantial interest. EP* seems to be an example of it. For EP* tells us that,

Memory as a Generative Epistemic Source  205 once we are in a position to form a belief on the basis of some source other than memory, whatever justification we have for forming it will not improve as time goes by thanks to our memory. It may be that, by giving us cognitive access to a fact to which we had no access before, memory puts us in a positive epistemic position with respect to that fact which we never enjoyed before. But memory, EP* tells us, cannot improve our epistemic position with respect to those facts to which we have had cognitive access in the past. Surely this is a preservativist thesis which is substantial enough for it to be worth exploring. Ultimately, though, the lesson to draw from our discussion is that, even if it turns out to be correct that memory cannot improve our epistemic position with respect to those facts to which we have had cognitive access in the past, such a result will not threaten the status of memory as a basic source of epistemic justification.

Conclusion At the onset of this investigation, I distinguished four questions about the metaphysics, the intentionality, the phenomenology and the epistemology of memory. We are now in a position to see how the account of memory that we have been building offers an answer to each of these questions. We are also in a position to see which of those answers depend on others, and why. The questions at issue were the following: i. Metaphysics What are the conditions under which an experience qualifies as an episodic memory? ii. Intentionality What are the contents of our episodic memories? iii. Phenomenology What is it like for us to have episodic memories? iv. Epistemology What kind of epistemic justification do our episodic memories afford? With regards to the metaphysics of memory, we have seen that episodic memory involves two different types of mental states. One of them is the state of remembering episodically, which consists in having some experience or other that plays a certain functional role in the cognitive economy of the subject. The other state is the experience that, in fact, plays the functional role in question. Within this picture of episodic memory, memories do not qualify as such in virtue of their content, or in virtue of their phenomenology, or in virtue of the knowledge that they provide us with. They qualify as such in virtue of some of the causal relations into which they normally enter. This has been the answer to question (i) delivered by our enquiry. What is interesting about this outcome is that, by associating the essence of memories with their functional role, and not with some of their intrinsic properties, this picture of episodic memories allows for memories to be malleable; a feature of memories that we should respect. With regards to the intentionality of memory, we have seen that memories are self-​referential in that they represent their own causal origins. They

Conclusion  207 represent themselves as originating in perceptions of objective facts. This has been the answer to question (ii) delivered by our investigation. This answer has been very productive, thanks to an assumption that we made at the start of this project. This is the assumption that the intentional properties of memories are responsible for their phenomenal properties. If this is correct, then we can view some of the phenomenal properties of memories as the ways in which we experience some of the things that our memories represent. This has been our guiding thought while trying to answer question (iii). With regards to the phenomenology of memory, we have distinguished three phenomenal properties of memories; the feeling that remembered facts took place in the past, the feeling that they were experienced by us in a specific way, and the feeling that the relevant memories are our own. The first of these phenomenal features, we have seen, is the way in which we experience the causal origin of our memories, which they represent. The second of these phenomenal features is the way in which we experience our past perceptual experiences of remembered facts, which our memories also represent. And the third of these phenomenal features is the way in which we experience the fact that our memories match the past, which our memories represent by representing our past perceptual experiences as having been veridical. It seems, therefore, that the considerable amount of detail in the content of our episodic memories is there for a reason. Each element in the content of our memories is responsible for some phenomenal feature of those memories. Since the phenomenology of memory is considerably rich, it is not surprising that the content of our memories is quite complex as well. The complex content of memories also helped us to account for certain aspects of the epistemology of memory. We have seen that, since memories represent the past perceptual experiences in which they typically originate, and since those perceptual experiences represent us as their subjects, memories are protected from a particular kind of error. They are immune to error through misidentification, ultimately, because they represent us as the subjects of the perceptual experiences in which they originate. Our answer to question (iv), therefore, is partly that the epistemic justification afforded by our memories is special in, at least, that way. Furthermore, we have seen that the justification at issue is, for certain beliefs (such as beliefs about the causal origins of our memories), produced by memory and not only conserved by it. Thus, our answer to question (iv) has also been that the epistemic justification afforded by our memories is special in that way as well. Both features of the justification that our memory beliefs enjoy have pointed us in the direction

208 Conclusion of more general conclusions about memory and first-​person thought. The fact that our memories represent us indicates that our first-​person conception of ourselves, in part nourished by memory, is that of an object which is extended in time as well as in space. And the fact that our memories produce, and not only conserve, epistemic justification indicates that the faculty of episodic memory is, from an epistemological point of view, basic. So how does it all hang together? Two thoughts are at the heart of the conception of episodic memory proposed in this book. One is the thought that no intrinsic property of a memory makes it qualify as such. Instead, its functional role does. The other one is that a memory is self-​referential in that it represents itself as having a certain causal origin. The rest of outcomes about memory have been built on these two thoughts. The conception of episodic memory being offered here, then, is that of a faculty which delivers mental states which are quite peculiar. Their metaphysics and their intentionality does not need to be in sync. Those mental states may represent themselves as having a causal origin that, as it happens, they lack. But, when all goes well, these are mental states which have the causal origin that they represent themselves as having. They keep an eye, so to speak, on where they are coming from, and they inform us about it. And, in virtue of this fact, they feel like no other mental state that we are capable of having. It is no wonder, then, that we think of memory as such a prominent capacity in our lives.

References Adams, R. M. (1974) “Theories of Actuality.” Noûs 8: 211–​231. Alston, W. P. (1986) “Epistemic Circularity.” Philosophy and Phenomenological Research 47: 1–​30. American Psychiatric Association. (2013) Diagnostic and Statistical Manual of Mental Disorders. Fifth Edition (DSM-​5). Washington, DC: American Psychiatric Association. Arango-​Muñoz, S. (2014) “The Nature of Epistemic Feelings.” Philosophical Psychology 27: 193–​211. Arango-​Muñoz, S., and K. Michaelian (2014) “Epistemic Feelings, Epistemic Emotions: Review and Introduction to the Focus Section.” Philosophical Inquiries 2: 97–​122. Aristotle (1972) “De Memoria.” In: Aristotle on Memory, R. Sorabji (ed. & trans.), 47–​61. London: Duckworth. Audi, R. (1997) “The Place of Testimony in the Fabric of Knowledge and Justification.” American Philosophical Quarterly 34: 405–​422. Audi, R. (2002) “The Sources of Knowledge.” In: The Oxford Handbook of Epistemology, P. K. Moser (ed.), 71–​95. Oxford: Oxford University Press. Barclay, C. R., and De Cooke, P. A. (1988) “Ordinary Everyday Memories: Some of the Things of Which Selves Are Made.” In: Remembering Reconsidered, U. Neisser and E. Winograd (eds.), 91–​125. Cambridge: Cambridge University Press. Barsalou, L. W. (1988) “The Content and Organization of Autobiographical Memories.” In:  Remembering Reconsidered:  Ecological and Traditional Approaches to the Study of Memory, U. Neisser and E. Winograd (eds.), 193–​243. Cambridge:  Cambridge University Press. Bartlett, F. C. (1932) Remembering:  A Study in Experimental and Social Psychology. Cambridge: Cambridge University Press. Bayne, T. (2008) “The Phenomenology of Agency.” Philosophy Compass 3: 1–​21. Bayne, T., and Montague, M. (2011) “Cognitive Phenomenology:  An Introduction.” In: Cognitive Phenomenology, T. Bayne and M. Montague (eds.), 1–​35. Oxford: Oxford University Press. Behan, D. P. (1979) “Locke on Persons and Personal Identity.” Canadian Journal of Philosophy 9: 53–​75. Bergson, H. (1911) Matter and Memory. London: George Allen and Unwin. Bernecker, S. (2010) Memory: A Philosophical Study. Oxford: Oxford University Press. Bernecker, S. (2017) “A Causal Theory of Mnemonic Confabulation.” Front Psychology 8: 1207. Bernecker, S., and T. Grundmann (2017) “Knowledge from Forgetting.” Philosophy and Phenomenological Research. https://​doi.org/​10.1111/​phpr.12469. Block, N. (1978) “Troubles with Functionalism.” Minnesota Studies in the Philosophy of Science 9: 261–​325. Brewer, B. (2007) “How to Account for Illusion.” In: Disjunctivism: Perception, Action, Knowledge, F. Mcpherson and A. Haddock (eds.), 168–​180. Oxford: Oxford University Press.

210 References Broad, C.D. (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. Brown, A. S. (1991) “A Review of the Tip-​of-​the-​Tongue Experience.” Psychological Bulletin 109: 204–​223. Burge, T. (1991) “Vision and Intentional Content.” In:  John Searle and His Critics, E. LePore and R. Van Gulick (eds.), 195–​214. Oxford: Blackwell. Byrne, A. (2010) “Recollection, Perception, Imagination.” Philosophical Studies 148: 15–​26. Campbell, J. (1997) “The Structure of Time in Autobiographical Memory.” European Journal of Philosophy 5: 105–​118. Campbell, J. (2010) “Demonstrative Reference, the Relational View of Experience, and the Proximality Principle.” In:  New Essays on Singular Thought, R. Jeshion (ed.), 193–​212. Oxford: Oxford University Press. Chalmers, D. (2002) “The Components of Content.” In: Philosophy of Mind: Classical and Contemporary Readings, D. Chalmers (ed.), 608–​634. Oxford: Oxford University Press. Chalmers, D. (2004) “The Representational Character of Experience.” In: The Future for Philosophy, B. Leiter (ed.), 153–​181. Oxford: Oxford University Press. Chalmers, D. (2017) “The Virtual and the Real.” Disputatio 46: 309–​352. Coady, C. A. J. (1992) Testimony: A Philosophical Study. Oxford: Oxford University Press. Conway, M. A., and C. W. Pleydell-​Pearce (2000) “The Construction of Autobiographical Memories in the Self Memory System.” Psychological Review 107: 261–​288. De Brigard, F. (2014) “Is Memory for Remembering? Recollection as a Form of Episodic Hypothetical Thinking.” Synthese 191: 155–​185. Debus, D. (2010) “Accounting for Epistemic Relevance. A New Problem for the Causal Theory of Memory.” American Philosophical Quarterly 47: 17–​29. Della Sala, S. (2009) “Anarchic Hand.” In:  The Oxford Companion to Consciousness, T. Bayne, A. Cleeremans, and P. Wilken (eds.), 37–​39. Oxford:  Oxford University Press. Della Sala, S., C. Marchetti, and H. Spinnler (1991) “Right-​Sided Anarchic (Alien) Hand: A Longitudinal Study.” Neuropsychologia 29: 1113–​1127. Dretske, F. (1986) “Misrepresentation.” In: Belief: Form, Content, and Function, R. Bogdan (ed.), 17–​36. Oxford: Oxford University Press. Dretske, F. (1995) Naturalizing the Mind. Cambridge, MA: MIT Press. Dummett, M. (1994) “Testimony and Memory.” In: Knowing From Words, B. Matilal and A. Chakrabarti (eds.), 251–​272. Dordrecht: Kluwer Academic Publishers. Eddinghaus, H. (1885) On Memory. New York: Dover Press. Evans, G. (1982) The Varieties of Reference. New York: Oxford University Press. Feinberg, T. E., R. J. Schindler, N. G. Flanagan, and L. D. Haber (1992) “Two Alien Hand Syndromes.” Neurology 42: 19–​24. Feldman, R., and E. Conee (1985) “Evidentialism.” Philosophical Studies 48: 15–​34. Fernández, J. (2006) “The Intentionality of Memory.” Australasian Journal of Philosophy 84: 39–​57. Fernández, J. (2008) “Memory and Time.” Philosophical Studies 141: 333–​356. Fernández, J. (2010) “Thought Insertion and Self-​ Knowledge.” Mind & Language 25: 66–​88. Fernández, J. (2013) Transparent Minds:  A Study of Self-​Knowledge. Oxford:  Oxford University Press.

References  211 Fernández, J. (2014) “Memory and Immunity to Error through Misidentification.” Review of Philosophy and Psychology 5: 373–​390. Fernández, J. (2015) “Externalism, Self-​Knowledge, and Memory.” In:  Externalism, Self-​Knowledge and Skepticism, S. Goldberg (ed.), 197–​214. Cambridge: Cambridge University Press. Fernández, J. (2016) “Epistemic Generation in Memory.” Philosophy and Phenomenological Research 92: 620–​644. Fernández, J. (2018) “The Functional Character of Memory.” In:  New Directions in the Philosophy of Memory, K. Michaelian, D. Debus and D. Perrin (eds.), 52–​72. London: Routledge. Fernández, J. (Forthcoming) “The Ownership of Memories.” In: The Sense of Mineness, M. Guillot and M. García-​Carpintero (eds.). Oxford: Oxford University Press. Friedman, W.J. (2001) “Memory Processes Underlying Humans” Chronological Sense of the Past.” In:  Time and Memory, C. Hoerl and T. McCormack (eds.), 139–​169. Oxford: Oxford University Press. Funkhouser, E. (2006) “The Determinable-​Determinate Relation.” Noûs 40: 548–​569. Gallagher, S. (2005) How the Body Shapes the Mind. Oxford: Oxford University Press. Gibson, J.J. (1979) The Ecological Approach to Visual Perception. Boston, MA: Houghton Mifflin. Giovannetti, T., L. J. Buxbaum, I. Biran, and A. Chatterjee (2005) “Reduced Endogeneous Control in Alien Hand Syndrome: Evidence from Naturalistic Action.” Neuropsychologia 43: 75–​88. Glenberg, A. M., and N. G. Swanson (1986) “A Temporal Distinctiveness Theory of Recency and Modality Effects.” Journal of Experimental Psychology: Learning, Memory and Cognition 12: 3–​24. Goldberg, G., and K. K. Bloom (1990) “The Alien Hand Sign. Localization, Lateralization and Recovery.” American Journal of Physical Medicine and Rehabilitation 69: 228–​238. Goldberg G., N. H. Mayer and, J. U. Toglia (1981) “Medial Frontal Cortex and the Alien Hand Sign .” Archives of Neurology 38: 683–​686. Goldie, P. (2012) The Mess Inside:  Narrative, Emotion, and the Mind. Oxford:  Oxford University Press. Goldman, A.I. (1979) “What is Justified Belief?.” In:  Justification and Knowledge, G. Pappas (ed.), 1–​25. Boston: D. Reidel. Graham, G., and G.L. Stephens (2000) When Self-​Consciousness Breaks: Alien Voices and Inserted Thoughts. Cambridge, MA: MIT Press. Hazlett, A. (2010) “The Myth of Factive Verbs.” Philosophy and Phenomenological Research 80: 497–​522. Hirstein, W. (2006) Brain Fiction:  Self-​ Deception and the Riddle of Confabulation. Cambridge, MA: MIT Press. Hoerl, C. (2001) “On Thought Insertion.” Philosophy, Psychiatry and Psychology 8: 189–​200. Hume, D. (2000) A Treatise of Human Nature, D.F. Norton and M.J. Norton (eds.). Oxford: Oxford University Press. Husserl, E. (1964) The Phenomenology of Internal Time-​Consciousness. Bloomington: Indiana University Press. Intraub, H., R. S. Bender, and J. A. Mangels (1992) “Looking at Pictures but Remembering Scenes.” Journal of Experimental Psychology:  Learning, Memory, and Cognition 18: 180–​191.

212 References Intraub, H., and M. Richardson (1989) “Wide-​Angle Memories of Close-​up Scenes.” Journal of Experimental Psychology: Learning, Memory and Cognition 15: 179–​187. James, W. (1890) The Principles of Psychology. London: Macmillan. Kim, J. (2003) “Supervenience, Emergence, Realization, Reduction.” In:  The Oxford Handbook of Metaphysics, M.J. Loux and D.W. Zimmerman (eds.), 556–​587. Oxford: Oxford University Press. Klein, S. B. (2012) “The Self and Its Brain.” Social Cognition 30: 474–​518. Klein, S. B. (2013) “Making the Case that Episodic Recollection is Attributable to Operations Occurring at Retrieval rather than to Content Stored in a Dedicated Subsystem of Long-​Term Memory.” Frontiers in Behavioural Neuroscience 7: 1–​14. Klein, S. B. (2015) “What Memory Is.” Wiley Interdisciplinary Reviews: Cognitive Science 6: 1–​38. Klein, S. B., and S. Nichols (2012) “Memory and the Sense of Personal Identity.” Mind 121: 677–​702. Kriegel, U. (2009) Subjective Consciousness:  A Self-​Representational Theory. Oxford: Oxford University Press. Kriegel, U. (2013) Phenomenal Intentionality. Oxford: Oxford University Press. Lackey, J. (2005) “Memory as a Generative Epistemic Source.” Philosophy and Phenomenological Research 70: 636–​658. Lackey, J. (2007) “Why Memory Really Is a Generative Epistemic Source:  A Reply to Senor.” Philosophy and Phenomenological Research 74: 209–​219. Le Poidevin, R. (2017). “Memory and the Metaphysics of Time.” In:  The Routledge Handbook of Philosophy of Memory, S. Bernecker and K. Michaelian (eds.), 219–​228. London: Routledge. Lewis, D. (1976) “The Paradoxes of Time Travel.” American Philosophical Quarterly 13: 145–​152. Lewis, D. (1980) “Mad Pain and Martian Pain.” In: Readings in the Philosophy of Psychology, N. Block (ed.), 216–​222. Cambridge, MA: Harvard University Press, 216–​222. Lhermitte, F. (1983) “Utilization Behavior and Its Relation to Lesions of the Frontal Lobes.” Brain 106: 237–​255. Locke, J. (1975) Essay Concerning Human Understanding, P. Nidditch (ed.). Oxford: Oxford University Press. Loftus, E. F. (2005) “Planting Misinformation in the Human Mind: A 30-​Year Investigation of the Malleability of Memory.” Learning & Memory 12: 361–​366. Loftus, E. F., and K. Ketcham (1994) The Myth of Repressed Memory. New  York: Martin’s Press. Malcolm, N. (1963) Knowledge and Certainty. Englewood Cliffs: Prentice Hall. Martin, C. B., and M. Deutscher (1966) “Remembering.” Philosophical Review 75: 161–​196. Martin, M. G. F. (2004) “The Limits of Self-​Awareness.” Philosophical Studies 120: 37–​89. Matthen, M. (2010) “Is Memory Preservation?” Philosophical Studies 148: 3–​14. McCarroll, C. J. (2018) Remembering from the Outside:  Personal Memory and the Perspectival Mind. Oxford: Oxford University Press. McDowell, J. (1997) “Reductionism and the First Person.” In: Reading Parfit, J. Dancy (ed.), 230–​250. Oxford: Blackwell. McGrath, M. (2007) “Memory and Epistemic Conservatism.” Synthese 157: 1–​24. Mellor, C. S. (1970) “First Rank Symptoms of Schizophrenia.” The British Journal of Psychiatry 117: 15–​23.

References  213 Michaelian, K. (2011) “Generative Memory.” Philosophical Psychology 24: 323–​342. Michaelian, K. (2013) “The Information Effect: Constructive Memory, Testimony, and Epistemic Luck.” Synthese 190: 2429–​2456. Michaelian, K. (2016a) “Confabulating, Misremembering, Relearning:  The Simulation Theory of Memory and Unsuccessful Remembering.” Frontiers in Psychology 7: 1857. Michaelian, K. (2016b) Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past. Cambridge, MA: MIT Press. Nigro, G., and U. Neisser (1983) “Point of View in Personal Memories.” Cognitive Psychology 15: 467–​482. Owens, D. (2000) Reason without Freedom:  The Problem of Epistemic Normativity. London: Routledge. Pacherie, E. (2007) “The Anarchic Hand Syndrome and Utilization Behaviour: A Window onto Agentive Self-​Awareness.” Functional Neurology 22: 211–​217. Parfit, D. (1971) “Personal Identity.” Philosophical Review 80: 3–​27. Perrin, D. (2018) “A Case for Procedural Causality in Episodic Recollection.” In:  New Directions in the Philosophy of Memory, K. Michaelian, D. Debus and D. Perrin (eds.), 33–​52. London: Routledge. Perry, J. (1999) “Indexicals and Demonstratives.” In: A Companion to the Philosophy of Language, B. Hale and C. Wright (eds.), 586–​613. Oxford: Blackwell. Plantinga, A. (1993) Warrant and Proper Function. Oxford: Oxford University Press. Pollock, J. L. (1985), “Plantinga on Possible Worlds.” In: Alvin Plantinga, J. Tomberlin and P. van Inwagen (eds.), 313–​29. Dordrecht: D. Reidel. Porter, S., and A. R. Birt (2001) “Is Traumatic Memory Special? A  Comparison of Traumatic Memory Characteristics with Memory for other Emotional Life Experiences.” Applied Cognitive Psychology 15: 101–​117. Prosser, S., and F. Recanati (eds.) (2012) Immunity to Error through Misidentification: New Essays. Cambridge: Cambridge University Press. Pryor, J. (1999) “Immunity to Error through Misidentification.” Philosophical Topics 26: 271–​304. Recanati, F. (2007) Perspectival Thought: A Plea for (Moderate) Relativism. Oxford: Oxford University Press. Reid, T. (1969) Essays on the Intellectual Powers of Man, B. Brody (ed.). Cambridge, MA: MIT Press. Robins, S. K. (2016) “Misremembering.” Philosophical Psychology 29: 432–​447. Robins, S. K. (2017) “Contiguity and the Causal Theory of Memory.” Canadian Journal of Philosophy 47: 1–​19. Roediger, H. L., and K. A. DeSoto (2015) “Psychology of Reconstructive Memory.” In: International Encyclopedia of the Social and Behavioral Sciences, J. D. Wright (ed.), 50–​55. Amsterdam: Elsevier. Rowlands, M. (2016) Memory and the Self: Phenomenology, Science and Autobiography. Oxford: Oxford University Press. Russell, B. (1921) The Analysis of Mind. London: Allen and Unwin. Saks, E. R. (2007) The Centre Cannot Hold:  A Memoir of my Schizophrenia. London: Virago. Schacter, D. L., and D. R. Addis (2007) “The Cognitive Neuroscience of Constructive Memory: Remembering the Past and Imagining the Future.” Philosophical Transactions of the Royal Society of London. B. 362: 773–​786. Schechtman, M. (1994) “The Truth about Memory.” Philosophical Psychology 7: 3–​18.

214 References Searle, J. R. (1983) Intentionality:  An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Senor, T. D. (1996) “The Prima/​Ultima Facie Justification Distinction in Epistemology.” Philosophy and Phenomenological Research 56: 551–​566. Senor, T. D. (2007) “Preserving Preservativism:  A Reply to Lackey.” Philosophy and Phenomenological Research 74: 199–​208. Senor, T. D. (2017). “Preservation and Generation.” In:  The Routledge Handbook of Philosophy of Memory, S. Bernecker and K. Michaelian (eds.), 323–​335. London: Routledge. Shoemaker, S. (1996) “Self-​Knowledge and ‘Inner Sense.’ Lecture I: The Object Perception Model.” In:  The First-​ Person Perspective and Other Essays, 201–​ 224. Cambridge: Cambridge University Press. Shoemaker, S. (1963) Self-​Knowledge and Self-​Identity. Ithaca: Cornell University Press. Shoemaker, S. (1968) “Self-​Reference and Self-​Awareness.” Journal of Philosophy 65: 555–​567. Shoemaker, S. (1970) “Persons and their Pasts.” American Philosophical Quarterly 7: 269–​285. Shoemaker, S. (1982) “The Inverted Spectrum.” Journal of Philosophy 79: 357–​381. Shoemaker, S. (1984) Identity, Cause and Mind. Cambridge:  Cambridge University Press. Shoemaker, S. (1987) “Introspection and the Self.” Midwest Studies in Philosophy 10: 101–​120. Soteriou, M. (2016) Disjunctivism. New York: Routledge. Stalnaker, R. C. (1976) “Propositions.” In:  Issues in the Philosophy of Language, A. F. MacKay and D. D. Merrill (eds.), 79–​91. New Haven: Yale University Press. Sutton, J. (2010) “Observer Perspective and Acentred Memory: Some Puzzles about Point of View in Personal Memory.” Philosophical Studies 148: 27–​37. Talarico, J. M., K. S. LaBar, and D. C. Rubin (2004) “Emotional Intensity Predicts Autobiographical Memory Experience.” Memory & Cognition 32: 1118–​1132. Talland, G.A. (1961) “Confabulation in the Wernicke-​Korsakoff Syndrome.” The Journal of Nervous and Mental Disease 132: 361–​381. Teroni, F. (2017) “The Phenomenology of Memory.” In:  The Routledge Handbook of Philosophy of Memory, S. Bernecker and K. Michaelian (eds.), 21–​ 34. London: Routledge. Thau, M. (2002) Consciousness and Cognition. Oxford: Oxford University Press. Tulving, E. (1972) “Episodic and Semantic Memory.” In:  Organization of Memory, E. Tulving and W. Donaldson (eds.), 381–​403. New York: Academic Press. Tulving, E. (1983) Elements of Episodic Memory. Oxford: Clarendon Press. Tulving, E. (2001). “Origin of Autonoesis in Episodic Memory.” In:  The Nature of Remembering: Essays in Honor of Robert G. Crowder, H.L. Roediger III, J.S. Nairne, I. Neath, and A.M. Surprenant (eds.), 17–​34. Washington:  American Psychological Association. Tulving, E. (2002a) “Episodic Memory: From Mind to Brain.” Annual Review of Psychology 53: 1–​25. Tulving, E. (2002b) “Chronesthesia:  Conscious Awareness of Subjective Time.” In: Principles of Frontal Lobe Functions, D. T. Stuss and R. C. Knight (eds.), 311–​325. New York: Oxford University Press.

References  215 Tulving, E. (2005). “Episodic Memory and Autonoesis:  Uniquely Human?” In:  The Missing Link in Cognition, H. S. Terrace and J. Metcalfe (eds.), 4–​56. New York: Oxford University Press. Tzeng. O. J.  L. (1976) “A Precedence Effect in the Processing of Verbal Information.” American Journal of Psychology 89: 577–​599. Von Leyden, W. (1961) Remembering: A Philosophical Problem. London: Duckworth. Wittgenstein, L. (1958) The Blue and Brown Books. Oxford: Blackwell. Wollheim, R. (1984) The Thread of Life. Cambridge: Cambridge University Press. Zahavi, D. (2008) Subjectivity and Selfhood:  Investigating the First Person Perspective. Cambridge, MA: MIT Press.

Index For the benefit of digital users, indexed terms that span two pages (e.g., 52–​53) may, on occasion, appear on only one of those pages.

abstraction, 171–​72,  191–​99 action delusions of mader (see delusion) disowned, 113–​14, 129–​30, 131n26,  131–​34 perception and, 162, 165–​66 Adams, Robert M., 59n2 Addis, Donna R., 191–​92 Alston, William, 14n13 anarchic hand syndrome, 132–​34 Arango-​Muñoz, Santiago, 19n19, 19n21 Aristotle, 37n7 attention, 6, 171–​72, 177–​81, 184–​85,  187–​91 Audi, Robert, 171n2, 176n7   Barclay, Craig R., 191n23 Barsalou, Lawrence W., 41n15 Bartlett, Frederic C., 38n10 Bayne, Tim, 112n1 Behan, David P., 8n6 belief as an act, 4–​7, 12, 47–​49, 51–​52, 145 contents of, 37, 179–​81, 188–​91 evidence for, 172–​73, 175–​77, 201–​2,  204 formation, 5–​7, 12, 16–​17, 19–​20, 21–​22, 33, 47, 65–​67, 70–​71, 79–​80, 166, 173–​88,  192–​99 justified, 12–​14, 16–​17, 19–​21, 171–​72, 173–​76, 179–​81, 182–​88, 193–​94, 196–​98, 201–​2,  203–​4 memory, 12–​14, 19–​22, 25–​26, 139n1, 188–​90, 192–​94, 198–​99,  207–​8 preserved,  173–​75 as a state, 6n4, 28, 37, 38–​39, 41–​42, 47–​48, 52–​53, 59–​60, 128n20, 146–​47, 162, 174, 179–​80, 207–​8

Bender, Rachel, 196n31 Bergson, Henri, 87n2 Bernecker, Sven, 21n22, 35n4, 46n24, 171n3, 177n10, 180n14, 182–​86, 193–​94 Birt, Angela R., 150n18 Block, Ned, 48n26 Bloom, Karen K., 132n30 boundary extension, 196–​98 Brewer, Bill, 165n28 Broad, Charlie D., 37n7 Brockmeier, Jens, 41n14 Brown, Alan, 19n20 Burge, Tyler, 164n26 Byrne, Alex, 92n9   Campbell, John, 42n18, 165n28 causal theory of memory, 33–​40, 54–​55n30, 54–​55, 55n31, 62n7 causation, 15, 16–​17, 33, 35–​36, 39–​40, 75 Chalmers, David, 59n5, 162n23, 164n25 Coady, Cecil Anthony J., 176n8 Conee, Earl, 16n16 content mnemonic causally self-​referential, 72–​80, 106–​9, 123–​27, 157–​60, 199–​202 (see also causal theory of memory; causation) conjunctive view of, 57–​58, 67–​72,  79–​80 evidence and, 16–​17, 20–​21 objective view of, 62–​64, 72–​73,  80–​81 phenomenology and, 16–​19, 23–​25, 29–​31,  206–​7 possible worlds and, 4–​5, 60–​62, 64–​65, 68–​69, 74, 107–​8, 124, 159–​60,  199

218 Index content (cont.)

reflexive view of, 23–​24, 25–​26, 78–​82, 85–​86, 88–​89, 94–​96, 98–​99, 106–​8, 109–​11, 124–​25, 136, 140, 157–​59, 160–​61, 167–​68, 171, 199 subjective view of, 57–​58, 64–​67, 70, 71n13, 79–​80,  95–​96 temporal approaches to, 99, 100–​1, 104, 106 perceptual extrinsic view of, 162–​63, 164–​65, 164n27, 166 Fregean,  164–​65 possible worlds and, 163 Russellian, 164 subjective presence in, 186–​87, 190–​91, 199–​200,  204–​5 See also belief; truth conditions Conway, Martin A., 41n15   De Brigard, Felipe, 41n14, 89n4, 195n29 Debus, Dorothea, 39n13 De Cooke, Peggy A., 191n23 Della Sala, Sergio, 132n29, 133n33 delusion,  128–​32 dependence, 17–​18, 20–​21, 29–​31, 39–​40,  46n24 DeSoto, Kurt A., 38n10 determinable-​determinate relation,  30–​31 Dretske, Fred, 15n14, 16n15 Dummett, Michael, 171n2   Eddinghaus, Hermann, 87n3 Evans, Gareth, 143n11, 146–​48, 169 evidence, 12, 16–​17, 18, 20–​21, 78n19, 128n20, 130–​31, 169–​70, 172–​73, 174–​77, 182, 186, 196, 201–​2, 204 experience awareness of previous, 85–​89, 91, 106–​7,  109–​11 of origin, 106, 108 of ownership endorsement model, 113, 119–​24, 127, 131n26, 131n27, 132–​34 identification model, 113, 117–​19,  121–​22 of time, 86–​89, 108–​9

feeling delusions of made (see delusion) epistemic,  19–​20 of mineness, 112–​13, 125–​26 of pastness mentioned, 85–​89, 92–​93, 98–​99, 106–​11, 125–​26,  135–​36 subject-​dependent view and, 104 subject-​independent view and,  100–​2 Feinberg, Todd E., 132n30 Feldman, Richard, 16n16 Flanagan, Natalie G. See Feinberg Friedman, William J., 99n11 functionalism mentioned, 23–​24, 32–​33,  48–​49 realiser,  47–​49 role, 48–​50, 53 functionalist theory of memory, 23–​26, 49–​50, 53–​56,  155–​56 Funkhouser, Eric, 31n26   Gallagher, Shaun, 112n2 generation of epistemic justification, 23–​26, 136, 171–​72, 175–​77, 179–​80, 182, 187,  193–​94 of mnemonic content, 25–​26, 171, 198–​99,  201 generativism, 25–​26, 171–​73, 175, 176n9, 177, 179, 181–​91, 193–​94, 195–​99, 200–​1n32, 201–​2,  204 Gibson, James J., 162–​63n24 Giovannetti, Tania, 132n30 Glenberg, Arthur M., 100n12 Goldberg, Gary, 132n30 Goldie, Peter, 41n14, 43n19 Goldman, Alvin I., 17n17 Graham, George, 131n27 Grundmann, Thomas, 171n3, 182–​86,  193–​94   Hazlett, Allan, 3n1 Hirstein, William, 46n24 Hoerl, Christoph, 131n28 Hume, David, 37n7 Husserl, Edmund, 103n14

Index  219 imagination, 10–​11, 39–​40, 41n14, 54–​55, 105–​6, 120–​21, 123–​24, 126–​27, 191–​92, 194, 201 immunity to error through misidentification memory and, 12–​13, 25–​26, 140–​44, 157–​60,  168–​70 mnemonic content and, 167 observer memory and, 139–​40, 150–​57 quasi-​memory and, 139–​40,  144–​49 impulses. See delusion Intraub, Helene, 196n31 introspection, 13, 33–​34, 140–​41, 142–​43, 169–​70, 176–​77, 194–​96,  204   James, William, 87n2 justification basic/​non-​basic, 14, 16, 172–​73, 188, 195–​96, 204–​5,  207–​8 causation and, 16–​17, 183, 193–​94 evidence and, 16–​17, 18, 20–​21, 174–​75, 182, 186, 201–​2, 204 prima facie/​ultima facie, 174–​75 reliability and, 16, 183, 185–​86, 196–​97 See also belief; generation; immunity to error through misidentification; memory; preservation   Ketcham, Katherine, 191n23 Kim, Jaegwon, 30n25 Klein, Stanley B., 38n11, 78n19, 114, 115n11, 116n12, 116n13, 117, 121n16, 121n17, 122–​23 knowledge, 9–​10, 21–​22, 23–​24, 55–​56, 140–​41, 171n2, 176n7, 178, 206 Kriegel, Uriah, 18n18, 112n2   LaBar, Kevin S., 150n18 Lackey, Jennifer, 171n3, 177–​80 Le Poidevin, Robin, 9n7 Lewis, David, 48n25, 105n17 Von Leyden, Wolfgang, 65n10, 71n12 Lhermitte, François, 133n31 Locke, John, 8n6, 37n7, 76n18 Loftus, Elizabeth F., 12n10, 191–​92   Malcolm, Norman, 37n7 Mangels, Jennifer A., 196n31

Marchetti, Clelia. See Della Sala Martin, Charles B., 33n2, 37–​38n9, 39n12 Martin, Michael G. F., 165n28 Matthen, Mohan, 92n9 Mayer, Nathaniel H. See Goldberg McCarroll, Christopher J., 150n18 McDowell, John, 146–​48 McGrath, Matthew, 174n4 Mellor, Clive S., 128n22, 128n23, 129n24, 131n28 memory autobiographical, 41–​42, 42n16 confabulation and, 45–​46 disowned, 24–​25, 78n19, 113, 114, 116–​17, 123–​24, 125–​28, 129–​30,  134–​35 eidetic, 34 episodic and semantic, 5–​7 innocence of, 65–​67, 70–​71, 80 intentionality of, 9–​11, 14–​16, 17–​19, 20–​21, 22, 23, 25–​26, 27, 29, 30–​31, 63, 65, 72–​73n14, 88, 101, 125–​26, 136, 140, 159–​60, 193–​94, 199, 206–​7,  208 metaphysics of, 8–​9, 14–​16, 17, 18, 20, 21–​22, 23, 26, 27, 31, 32–​33, 139, 206 observer, 43n21, 78n19, 139–​40, 139n1, 150–​55, 156, 157, 159–​60, 161 phenomenology of, 10–​15, 16, 17–​21, 22, 23, 24–​25, 26, 27, 29, 31, 55–​56, 62n7, 72–​73n14, 85–​87, 89–​90, 91, 92–​94, 97n10, 102, 107–​9, 110–​11, 116–​17, 125–​27, 136,  206–​7 quasi, 139–​40, 144–​47, 157, 159–​60 reconstructive conception of, 25–​26, 40–​41, 44, 45n22, 46, 54–​56, 92–​93, 171–​72,  191–​97 recorder model of, 25–​26, 37n7, 171–​91, 198–​201, 200–​1n32 temporal structure of, 42–​43 mental image, 5–​6, 7–​8n5, 8, 9, 28, 34, 36–​40, 42–​43, 45–​46, 49–​56, 57, 125–​26, 150–​51,  154–​56 mental time travel mentioned, 24–​25, 85–​86, 88–​89, 106–​7, 123n18, 125–​26,  135–​36

220 Index mental time travel (cont.) and re-​presentation,  89–​94 and representation, 94–​99 See also representation Michaelian, Kourken, 19n21, 41n14, 46n24, 89n4, 156n20, 171n3,  193–​96 misinformation effect, 12n10, 194n27 Montague, Michelle, 112n1   narrative theory of memory, 40–​46, 53–​54,  192 Neisser, Ulric, 43n21, 78n19, 150n18 Nichols, Shaun, 25n23, 38n11, 78n19, 114, 115n11, 117, 122–​23 Nigro, Georgia, 43n21, 78n19, 150n18   Owens, David, 171n2   Pacherie, Elisabeth, 133n32 Parfit, Derek, 145–​46 perception, 10–​11, 13, 25–​26, 36, 44, 51, 62, 63–​66, 67–​69, 70–​72, 74–​75, 76–​78, 81n20, 94, 103n14, 105n17, 107–​9, 110–​11, 124–​27, 134–​35, 140, 154, 157, 158–​59, 160–​67, 168, 172–​73, 174n4, 176–​77, 179, 180–​81, 190–​91, 195–​97, 198–​99, 200–​2, 204, 207 Perrin, Denis, 72–​73n14 Perry, John, 59n4 personal identity, 8–​9, 114, 117, 145 Plantinga, Alvin, 171n2 Pleydell-​Pearce, Chistopher W., 41n15 Pollock, John L., 59n3 Porter, Stephen, 150n18 preservation of content, 37, 40, 53–​54, 71, 187–​92, 198–​201 of epistemic justification, 171, 203–​5 preservativism, 37n8, 171, 173–​77, 178, 179, 180–​81, 187–​88, 189n20, 193–​96, 199–​200, 201, 202–​3,  204–​5 property affective,  43–​44 epistemic, 15, 20–​22, 25–​26, 139 extrinsic, 162–​63, 164–​67,  168–​70

intentional, 15–​19, 20–​21,  206–​7 intrinsic, 47, 50, 206, 208 mental, 60, 169 normative/​non-​normative,  175–​76 phenomenal, 15–​19, 20–​21, 22, 112n2, 124n19, 139n1, 206–​7 physical,  169–​70 qualitative, 95 temporal, 108–​9,  169–​70 See also state Prosser, Simon, 13n11 Pryor, James, 140–​41n4   rationality, 140–​41, 178 realization, 15, 30–​31 Recanati, François, 13n11, 149–​50n17 reconstruction, 54–​55, 171n1, 171–​72, 191–​94, 195–​96, 198–​99,  201 Reid, Thomas, 37n7, 189–​90 relearning, 35–​36, 36n6 reliability, 14, 54–​55, 171n2, 172–​73, 191–​92, 193–​94, 196, 197–​98,  201–​2 representation, 5, 9–​10, 15n14, 16n15, 26, 33–​36, 40–​42, 91–​92, 95–​99, 103n14, 123n18, 124–​25, 131–​32, 156, 165–​66,  191–​92 representationalism, 16n15 Richardson, Michael, 196n31 Robins, Sarah K., 35n5, 54–​55n30 Roediger, Henry L., 38n10 Rowlands, Mark, 78n19, 159n22 Rubin, David C., 150n18 Russell, Bertrand, 87n2   Saks, Elyn R., 128n21 Schacter, Daniel L., 191–​92 Schechtman, Marya, 41n14, 43n20 Schindler, Rachel J. See Feinberg schizophrenia,  127–​28 Searle, John R., 73–76, 200–​1n32 self, 11–​12, 89–​90, 110–​11, 151–​52,  160–​61 self-​attribution,  112–​13 self-​awareness, 140,  143–​44 self-​identification,  143n11 self-​knowledge, 9–​10,  10n8 Senor, Thomas D., 13n12, 171n2, 175n5, 176n9, 180n14

Index  221 Shoemaker, Sydney, 37n7, 76n18, 78n19,  140–​48 simulation, 89n4 Soteriou, Matthew, 28n24 source monitoring, 194–​96 Spinnler, Hans. See Della Sala Stalnaker, Robert C., 4–​5n2 state first-​order, 48–​49,  55–​56 higher-​order, 48–​49, 50, 55–​56, 155 intentional, 59, 133–​34 perceptual, 165n28 phenomenal, 112, 113, 127–​28,  129–​31 See also property Stephens, G. Lynn, 131n27 Sutton, John, 150n18 Swanson, Naomi G., 100n12   Talarico, Jennifer M., 150n18 Talland, George A., 46n23 Teroni, Fabrice, 81n20 Thau, Michael, 164n27 thought insertion. See delusion

time, 5–​12, 13, 19–​20, 23–​26, 31, 37, 45–​46, 51, 54, 55–​56, 62–​65, 66–​67, 75, 85–​86, 89, 90n6, 91–​94, 99–​109, 110–​11, 115–​16, 118–​19, 122–​23, 125–​26, 135–​36, 141, 151–​52, 153–​56, 171, 173–​77, 178, 179, 180–​84, 185–​87, 189–​91, 194–​98, 201–​2, 204–​5,  207–​8 time travel, 105–​6. See also mental time travel Toglia, Joseph U. See Goldberg truth conditions, 10, 23–​24, 28–​29, 57–​60, 74n17, 141–​42 (see also content) maker, 16–​17, 60 Tulving, Endel, 6n4, 89–​91, 93–​94, 95 Tzeng. Ovid J. L., 100n12   utilization behaviour, 132–​34   Wittgenstein, Ludwig, 140–​41n4 Wollheim, Richard, 43n19   Zahavi, Dan, 112n2