Memory and the Self (2017), authored by Mark Rowlands, is a fascinating book that has all the qualities of good philosophical writing. It deals with a topic, memory, that has not received too much attention in philosophy of mind. It inquires about specific issues of memory that have received no attention at all, and it makes use of ideas from different philosophical traditions. Additionally it appeals to a various range of arguments, including experimental and introspective evidence to justify his claims. What is more, this delicious “combo” for the mind comes in a lucid and elegant prose, extremely clear and fluid, even for non-professional philosophers—and also for non-native English speakers— which at times achieves a literary style characteristic of fiction authors; a style that nowadays has unfortunately become more and more rare in academic philosophical writing.
The main aim of Rowland’s book is to give a better account of the key role played by memories in the constitution of personal identity and the explanation of the unity of a person. Probably the reader is familiar with the psychological-continuity views of personal identity that privileges memory as the essential factor for personhood: as Locke (1690) explained, as long as an individual possesses memories, the one remembering and the one remembered are the same person. Nonetheless, this quite intuitive conceptualization of personal identity presents some problems widely known in philosophical literature, such as the problem of circularity: how can memory explain personal identity if it presupposes personal identity? Besides these more metaphysical questions that go beyond the scope of the book, there are other common sense considerations that cast doubt on the explanatory role of the memory criterion for accounting for personal identity. The anthropologist Jannelle Taylor, writing about his mother who developed dementia, considers that despite the massive loss of memories and “all the changes she has been through, my mother ‘still’ is in many ways the cheerful affectionate person I have always known her to be. Mom still enjoys gentle joking and teasing, as she always has. She still enjoys being around people, still beams radiantly at small children when she sees them, still enjoys the give and take of conversation.” (2008, p. 316). Rowlands is of the same opinion as Taylor: regardless of the Alzheimer of Patsy Hasset, his wife’s grandfather, he felt that Patsy was still there, not simply as a human being or a biological organism, but as a person, as a psychological entity with some defining personality traits. And in fact, this opinion seems to be shared by most of us: according to an empirical study done by Jesse Prinz and Shaun Nichols (2016), people in general consider that the loss of memories does not threaten the identity of a person, in comparison with a change of moral values that is considered to have a devastating impact on it.
So our ordinary understanding of the basis of the continuity and unity of our identity over time gives us two ideas that in principle are contradictory. On the one hand, we think that the loss of memories of past experiences does not undermine personal identity; but on the other hand, we also have the intuition that memories play a certain important role in making us who we are. In Memory and the Self, Rowlands provides a clever, original—and also poetic—response that makes these two ideas compatible: memory makes us who we are even if, like Patsy Hasset, we have lost our memories, because memories of past experiences can persist and continue to shape our personhood when these past experiences have been forgotten, that is, when the content of our memories has disappeared. Rowlands calls “Rilkean memories” these mutated memories that do not have content. The origin of the name is due to the inspiration drawn from a passage of the only novel written by Rainer Maria Rilke that makes reference to memories that have been forgotten and are thus “nameless” but return in a new form: “they have changed into our very blood into glance and gesture, and are nameless, no longer to be distinguished from ourselves” (Rowlands, 2017, p. 53). As we shall see later, Rilkean memories refer to behavioural and bodily dispositions, feelings, moods and sensations, which have arisen from episodic memories but which have lost their contents and have become pure mental acts.
The characterisation of Rilkean memories and the investigation of its role in the construction and continuity of personal identity over time led Rowlands to accomplish another important task: to reconfigure our understanding of the structure of memory. Whereas a traditional analytical philosopher understands a memory as a mental representation with a tripartite structure composed of an act, an object and a mode of presentation, Rowlands proposes a four-constituent model of memory, in which (a) the act of remembering is of fundamental importance to understand the structure of a memory experience; (b) the intentional object, that is, the episode remembered (that exists independently of the act of remembering), is different from the content of a memory; and (c) the act, the content and the mode of presentation are conceptually distinguishable but inseparable: the content of a memory exists when the act of remembering operates certain transformations on the episode remembered and presents it in a certain mode. The mineness is one essential mode in which the episode remembered is presented, and this is what explains the undeniable presence of the self in every memory of our past experiences.
Therefore, a novel explanation of the way that memories make us who we are as well as a novel explanation of the structure of memory are the two major accomplishments that Rowlands intends to achieve in Memory and the Self. It remains to be seen (and evaluated) how the author develops these explanations through his book and how both of them are linked together.
Phenomenology, and the autobiographical self
But first, a remark about Rowlands’ methodology. Rowlands’ writings have been widely influenced by the phenomenological tradition, and this book is not an exception.
On one hand, Rowlands remarks (chapter 1) that whereas analytical philosophy and cognitive science have always privileged the mental content over the mental act to account for cognitive states, the phenomenological tradition has done exactly the contrary: it has privileged the study of mental acts as acts without objectifying them, in order to understand the preconditions of our experiences. Mental content, appearences in phenomenological terms, are only studied to get to the act. And this phenomenological method is exactly what Rowlands adopts: he begins with general intuitive ideas of the type of “memories makes us who we are although their lost does not undermine personal identity”, and “there are some behaviours and moods that connect the person to his past and that are thus relevant to the continuity of identity through time”, ideas that could be understood as appearences, and then he works backwards from them in order to identify the features of the act of remembering in virtue of which memories, behaviours and moods, can appear this way. Rowlands considers essential the recovery and privileged role of the act of remembering in order to understand memory, develop a workable conception of memory content and make sense of the idea that memories make us who we are.
On the other hand, Rowlands makes another use of the phenomenological method to delineate his conceptualization of the notion of self. Whereas in philosophy most concerns about the self are part of a metaphysical project that tries to understand the nature of personhood, its essential properties, its persistence through time, etc., Rowlands proposes to bracket these metaphysical questions and examine the way the self presents itself to us. If someone asks us how we would define our own self, we would probably answer her by describing our beliefs about ourselves, our values, our attitudes, our desires, etc. This description would probably be different if we were asked the same question at a different time. The idea that there are multiple selves and that each of them refers to a particular configuration of our self-knowledge at a particular time is not new. In psychology, this is a common conception of a self. The psychologist Martin Conway, for example, considers that the self refers to conceptual self-structures that are not temporally specified, such as self-schemas, self-scripts, possible selves, self-images, self-with-other units, relational schemas, attitudes, values and other self-beliefs (2005, 2009). These configurations of abstract self-knowledge, that Conway calls “the conceptual self”, are formed and ultimately grounded in episodic memories of specific experiences, and can change through time. This conception of the self constitutes a good workable notion—and a good strategy—that allows any theorist to make use of a notion of self and at the same time to set aside all the metaphysical questions related to the self (that would require an entirely different kind of research). I used it myself for this purpose. I considered that the different selves (past selves, present self, future selves) are just many and different configurations of self-knowledge, different conceptual selves in Conway’s terminology, that constitute parts of the same human being who perdures through time (Trakas, 2014, pp. 131-132). Nonetheless, Rowlands goes beyond this idea and supposes that there is a self that transcends these empirical and multiple configurations of the self. He defines this self, called “autobiographical self”, as the principles of the network of these concrete episodes of self-understanding; their laws of appearance. I have trouble understanding how this notion of autobiographical self, which is in certain way a sort of Kantian self, and thus a transcendental self, can explain the unity of the self and its distinctness from other selves (Rowlands, 2017, p. 84) without being similar to the notion of metaphysical self. I think that the practical solution to avoid a metaphysical inquiry, would be to just state (as I did) that these episodes of self-understanding or configurations of self-knowledge constitute parts of the same human being, and that they are interconnected between them because the physical continuity of the human being assures some degree of psychological continuity. This strategy does not suggest, as Rowlands does, that the principle or structure from which all the different episodes of understanding emerge is itself a self. Rowlands should have said more about the autobiographical self to prevent their readers from thinking that he is actually engaged in a metaphysical explanation of the self (even if he explicitly denies it).
Maybe Rowlands introduced this unitary notion of self in order to account for the unfolding characteristic of memory between a self who remembers what a former self experienced. Rowlands mentions the two “selves” involved in memory while discussing the differences between the notions of autobiographical self and narrative self. According to Rowlands, the autobiographical self is not the same as the narrative self and entirely rescinds from the question of whether the self has a narrative structure (Rowlands, 2017, pp. 85-87). Nonetheless, the autobiographical self is compatible with narrative accounts of self-understanding that conceive that the self who remembers adopts the position of narrator with respect to the self that originally experienced. Rowlands calls them R-self and W-self respectively. Both of them are conceptually distinguishable but not ultimately separable, because both of them—the self that is written and the self who reads what is written—form the autobiographical self. But once again, we do not need to suppose a transcendental self to explain the essential unfolding of the self that characterizes memory. Neither do we need to understand this unfolding of the self in narrative terms. We can forget about narrative and about any transcendental conception of the self, and simply state that a present self, that is nothing more than a particular configuration of self-knowledge at a given time, can have access to previous selves and their experiences because they all belong to the same human being. The numerical continuity and the degree of psychological continuity implied in the fact of belonging to the same human being would guarantee the access (to some extent) to past configurations of self-knowledge and past experiences, and thus the unfolding of the self and the possibility of self-reflection through time that are characteristic of memory.
Rilkean memories (and episodic and autobiographical memories)
The disquisition about the nature of the self implied in the claim that “memories make us who we are” is not presented at the beginning of the book but in chapter 4. After a first chapter that constitutes a condensed summary of all the ideas developed in the book—which deserves a second reading after finishing the book, in order to get a better picture of the whole—, the next two chapters (2 and 3) are focused on the characterization of Rilkean memories.
Rowlands does not intend to directly prove the existence of Rilkean memories: “Rilkean memories are theoretical posits whose existential credentials will be established by the sort of explanatory work they do” (Rowlands, 2017, p. 55). But further on: “if they are to play an explanatory role of certain sort [explaining how memories make us who we are], they must have certain features”. In a certain way, Rowlands forces the readers to accept the existence of Rilkean memories: how will the explanatory work they do establish their existence if their characterization is conceived in a way that they could successfully accomplish this explanatory work? In any case, this tricky argument is not so relevant; readers avid of understanding embodied and affective phenomena neglected in cognitive science and philosophy of mind, will become immediately sympathetic to the idea of Rilkean memories.
Furthermore, there are examples of Rilkean memories in literature and poetry, and it is also easy to think of everyday cases. Embodied Rilkean memories refer to patterns of behavioural as well as bodily dispositions inscribed in the body that originated in the past: a curvature of the spine and a consequent back pain that originated in successive episodes of bad posture while writing as a child, a tendency to talk in a very loud voice during a normal conversation originated in successive episodes of conversation with parents who speak too loudly, are (personal) examples of embodied Rilkean memories. Affective Rilkean memories make reference to sensations, feelings and moods strongly environmentally embedded, which have a very low probability of occurring without the requisite environment. The famous episode of la madeleine de Proust, the nostalgia that arises when walking around our hometown left a long time ago, are cases of Rilkean affective memories. These behaviours, bodily dispositions, moods, feelings and sensations can appear when the initial episodic memories have vanished, can coexist with them, or can exist shortly before the onset of them (like Proust’s madeleine).
Rilkean memories can exclusively arise from memories that are person specific in order to play a role in the constitution of the person and, as Rowlands argues, only episodic memories are sufficiently specific to their subject. The same procedural memories, semantic memories, even semantic autobiographical memories, could be in principle possessed by two different people. So Rilkean memories, Rowlands concludes, can only arise from episodic memories.
While reflecting on the characterization of Rilkean memories, Rowlands introduces a new and original conceptualization of episodic memory. Episodic memories are neither memories of episodes—this will render them indistinguishable from some semantic memories that are also memories of episodes—nor memories of experiences—this will entail the falseness of most of our memories due to the fact that memory’s visual, emotional and evaluative perspectives can and often change over time. Episodic memories cannot either be understood as an adverbial modification of the act of remembering: relocating the experiential qualities of episodic memory to the act of remembering threatens the distinction between episodic memory and semantic memory (I can remember a fact angrily) and cannot explain the contradictory experiential qualities that may exist between the act of remembering and what is remembered (I can remember with joy a sad episode). According to Rowlands, episodic memories are best defined as memories of an episode that is subsumed under a specific mode of presentation: beside the rich experiential-emotional complexes that are characteristic of episodic memories, what is essential to the mode of presentation of episodic memories is that the episode remembered is remembered as one that has formerly witnessed, orchestrated or otherwise encountered by the rememberer, and that this “as” is built into the content of the memory (and not on the act of remembering).
I am quite sympathetic to both ideas: that Rilkean memories arise from episodic memories and that the self-involvement or the presence of the self in the content of memories is what makes memories episodic. Nonetheless, I have some doubts about the effectiveness of Rowlands’ arguments. First of all, he dismisses semantic autobiographical memories as a starting point of Rilkean memories because even if unlikely, it is perfectly conceivable that two different people could possess the same semantic autobiographical memories and have forgotten the other ones that would distinguish one person from the other. So because this situation is possible, semantic autobiographical memories are not considered to be sufficiently specific to the subject. The problem with the use of this kind of hypothetical scenario is that we could easily conceive of a similar scenario about episodic memories and thus come to the conclusion that episodic memories are not sufficiently specific. We could think about identical twins—who in general have a significant amount of experiences in common—who exclusively remember the episodes experienced together. In this hypothetical case, episodic memories would not be sufficiently specific to distinguish the two identical twins. This scenario is as unlikely but as possible as the scenario concerning semantic autobiographical memories, especially when we take into consideration that a lot of semantic autobiographical memories are the result of a process of semantization of episodic memories over time (Piolino, & al., 2009). In the hypothetical episodic memory scenario, what would be sufficiently person specific and would allow us to distinguish the identical twins is not the fact that these episodes are remembered as formerly witnessed, orchestrated or encountered by the rememberer, but the fact that they are remembered as episodes that formerly affected the rememberer in terms of harms, benefits, morality or self-image, and that this affection of the event—which is person specific—is part of the content of the memory (see Trakas, 2014). There is less unlikely that identical twins could only remember the events that both have witnessed, orchestrated and encountered, than they could remember these same events under the same affective tone. And this remark leads me to the second point I wanted to make concerning Rowlands’ conceptualization of episodic memory. Episodic memory is a controversial notion, very much used in psychological research, but not very well defined. Endel Tulving, the “father” of the distinction between episodic and semantic memory systems, has defined episodic memory first in terms of its content, then in terms of its phenomenology (which arise out of its mode; see for example McCormack, 2001), but in certain way the debate has just started, with the growing interest that this notion has aroused in the philosophical community in recent years. The point that Rowlands makes about the specificity of episodic memory indubitably marks a novel way of thinking about the nature of episodic memory that is very promising. But it needs further development. Semantic autobiographical memories that are originated from a process of semantization of episodic memories (very characteristic of older adults), differ from episodic memories at least in the neural substrates and mechanisms and in their phenomenology, but they are also remembered as episodes formerly witnessed, orchestrated or encountered by the rememberer. I previously suggested that in an episodic memory we remember episodes (or people, or places, etc.) as episodes that affected me in a specific way (or that stills affect me), and it is through this affection that the self is present in the content of memory. This affection can explicitly be attended to as the intentional object of my memory, or we can be aware of it in a pre-attentive or pre-reflective way; it can take the form of interoceptive bodily sensations, action tendencies or language, and it can refer to a past affection or to a present and occurrent one. According to my view, it is this affection that makes of memories episodic memories—and that is at the origin of the metacognitive phenomenology that is characteristic of episodic memory—and it is this affection that makes of my episodic memories uniquely mine. More should be explored in this line, because it clearly seems that the presence of the self is an excellent alternative to the current views to characterize the specificity of episodic memory.
In chapter 8, Rowlands argues that the presence of the self is a necessary and sufficient condition for a memory to count as an episodic. I have tried to explain before, through the example of semantic autobiographical memories that are the product of a process of semantization of episodic memories, why the presence of the self characterized as a mode of presentation where the episode is remembered as one that the rememberer has formerly witnessed, orchestrated or otherwise encountered, does not seem sufficient for a memory to qualify as episodic. Nonetheless, the arguments that Rowlands presents to defend the necessity of the presence of the self in an episodic memory are very convincing. First, we could think that the presence of the self is not necessary because non-human animals have episodic memories but neither engage in self-reflective thought nor have a self-concept. Rowlands argues that none of them is necessary for the self to be present in a memory, and that a feeling of familiarity could perfectly account for it. In fact, the thesis that non-human animals have episodic memories is quite controversial, and Rowlands should have mentioned it to reinforce his point. It could have also been argued that the semantic / episodic distinction is also present in non-humans animals, but that its characterization is slightly different from one proper to human animals (and this makes sense considering the importance of the influence of human language in the phylogenetic development of our cognitive capabilities). Second, the case of a patient named RB (mentioned by Klein & Nichols, 2012), who seems to have episodic memories that do not present a sense of ownership, could also be used as a counterexample of the necessity of the presence of the self in episodic memories. But it is not the case: or this is an example of attenuation and not of loss of the sense of ownership, argues Rowlands, or else these memories are not episodic. As he correctly points out, in the absence of the presence of the self in episodic memory, there is nothing to distinguish episodic memories from semantic memories. Therefore, Rowlands gives compelling arguments to assert the necessity of the presence of the self in episodic memories, whereas his arguments for its sufficiency in a certain way fail, because his interpretation of the meaning of the presence of the self in episodic memories is not sufficient to distinguish them from semantic (autobiographical) memories.
Before coming back to the characterization of Rilkean memories, I would like to mention an interesting distinction that Rowlands draws concerning autobiographical memory, which should be considered while theorizing about this notion. Autobiographical memory is another notion very much used in psychological research, but again not very well defined. Broadly understood, it refers to a subsystem that includes some episodic memories and different facts about the self (including semantic memories). Rowlands proposes to distinguish three types of autobiographical memories according to their intentional objects: (a) strongly autobiographical memories: the memory contains the rememberer as the intentional object of the memory, and is thus about something that happened to the rememberer (I remember I travelled to Greece or I remember I was born the 15th February 1983), (b) weakly autobiographical memories: the rememberer is not the intentional object, but is implicated in the mode of presentation of the intentional object of the memory, and is thus about something that she witnessed or encountered (I remember the flight to Greece took off 5 hours later than scheduled); (c) minimally autobiographical memories: these memories, which have no intentional object, are autobiographical because they are the descendant of a memory that is at least weakly autobiographical. While episodic and semantic memories can be strongly autobiographical, only episodic memories can be weakly autobiographical—only episodic memories can include the self in their mode of presentation—and only Rilkean memories can be minimally autobiographical. The common characteristic between all these subtypes of autobiographical memory is that all of them ultimately refer to the rememberer, and it is in this sense that all of them receive the epithet “autobiographical”.
This distinction allows Rowlands to give a minimal definition of embodied and affective Rilkean memories: Rilkean memories are involuntary memories that have no intentional content and are minimally autobiographical because they derive from episodic memories, when their content has been forgotten and only the act of remembering persists. This definition is given in chapter 3, after a series of arguments that (convincingly) show why Rilkean memories cannot be conceived as Freudian memories, nor procedural memories, nor declarative memories, nor semantic memories, nor episodic memories, nor explicit memories, nor implicit memories.
More about episodic memories: their structure
In the next section, I will come back to Rilkean memories, and to their importance for the unity and identity of the self. In this section, I will focus on the characterisation of the structure of episodic memory developed by Rowlands in chapters 8 and 9.
In the introduction, I already anticipated that Rowlands reconfigures the traditional understanding of the structure of memory by proposing a four-constituent model of episodic memory: intentional object, content, mode of presentation and act. In his model, the intentional object is different from the content, and the mode of presentation and act of remembering are conceptually distinguishable but inseparable. These two ideas are the key theses defended by Rowlands in order to change the traditional conceptualization of episodic memory that is characterized by the standard tripartite model of intentionality and the two-model of meaning.
The two-model of meaning (which, according to Rowlands, would be at the origin of Wittgenstein’s rule-following paradox) supposes that items are intrinsically semantically inert and only get meaning and reference by an act of interpretation. This model is useful to account for the semantics of photographs, and because we tend to consider episodic memories as “pictures” of past episodes, we think mistakenly that this model is also useful to understand the structure of episodic memories, when really it is not. Although I have some doubts about the intrinsically semantical inertia of photographs (a specific photograph is ambiguous but cannot “be about anything”), Rowlands makes a good point: photographs exist independently of any act of interpretation whereas episodic memories do not; and photographs need an act of interpretation at least to remove their intrinsic semantic ambiguity, whereas episodic memories do not. The list of differences could be developed (episodic memories can essentially change over time whereas photographs do not, etc.), and this would be an interesting project to finally abandon the photographic model of memories, but this is not Rowlands’ purpose in this book: he only wants to state that, unlike photographs, the contents of our episodic memories are never pure objects, unadulterated by the interpretative activities implicated in my awareness of them. The content of our episodic memories is always presented to us as something, under a mode of presentation, and this mode of presentation is not externally attached to the content, but is essentially built into it. When I remember the face of my father, I remember this face as the face of my father, and not as a visual image of a face whose appearance needs a subsequent act of interpretation to determine that it is a memory and that it is the face of my father. It may be the case that I cannot remember whose face it is, but if I have a memory experience I remember the face at least as a face that belongs to someone I previously saw. For Rowlands, in an episodic memory, meaning and reference are thus not added in a subsequent phase to the presentation of the content to the mind, but are an intrinsic part of it, entangled with it. The meaning and reference includes not only the meaning and reference that is specific to a particular memory content, but also the meaning and reference that is given in every episodic memory: the pastness and the presence of the self who remembers. The meaning and reference is given to the episode remembered, which is not inherently interpreted, when the act of remembering performs on it certain operations of transformation that present the episode remembered under different modes of presentation. These modes of presentation (which are characterized by Rowlands as complex combinations of perception, cognition, emotion and sensation) not only individuate the memory and, more importantly, render the presence of the self a necessary feature of it, but also give rise to memory content. The content of an episodic memory is thus created by the act of remembering.
And this leads us to Rowlands’ four-constituent model of episodic memory and his revision of the standard tripartite model of intentionality. Whereas the standard model considers that the intentional object of an episodic memory is equal to its content, and that this object / content is an episode—defined as a state-of-affairs—that is independent of the act of remembering and propositional in form, Rowlands not only denies the necessarily propositional nature of episodes, but also the identification between the object and the content of a memory. Whereas the intentional object of memory, that is, the episode remembered, is a state-of-affairs independent of the act of remembering, which only plays a passive causal role in the origin of our memories, the memory content is what is available to our consciousness. It is what one can discern and have access to when one remembers, and it is the product of a constructive and active process of remembering.
This later distinction is not new, but has a long tradition—recently recovered but neglected for many years— that goes back at least to the introduction of the notion of intentionality in contemporary philosophy made by Franz Brentano. The distinction between object and content was explicitly formulated by Kazimierz Twardowski (Brentano’s student) in his book On the Content and Object of Presentations (1894) and later endorsed and developed by Alexius Meinong (1899), another one of Brentano’s students. It was also more explicitly applied to the understanding of memory phenomena by Bertrand Russell (1921) and Charlie Broad (1925). All of them, in different ways and with different terminology, defended the existence of a difference between the object of a mental act and its content. I personally got back to this rich tradition and proposed a representationalist account of personal memories based on this distinction (Trakas 2014). I found it a bit disappointing that Rowlands did not mention the origin of this distinction in his book, although I understand that historical references sometimes may cut the argumentative fluidity. Nonetheless, a small footnote would not have done any harm, and it would have been a nice initiative to recognize the often forgotten rich ideas that precede us and still influence us in many ways.
Rowlands justifies the need of this distinction by means of three convincing arguments. If the memory content were identical with the episode remembered:
(a) the idea of mental content should be abandoned (there is nothing “mental” in a state-of-affairs; a state-of-affairs would be mental and non-mental at the same time), or the mentality should exclusively be placed on the act of remembering. The only way to assure the mentality of the content is to distinguish the state-of-affairs from the content and adopt the view that the content is brought into existence by a process of transformation operated by the mental act on the state-of-affairs;
(b) it would be impossible to explain why two states-of-affairs can be identical (such as Oedipus marring Jocasta and Oedipus marrying his mother) whereas the memories of them are not (Oedipus remembers marrying Jocasta but not his mother). States-of-affairs and memory content must be different because their standards of individuation are different: a mental act narrows the standard of individuation of mental content by subsuming one or more constituents of a state-of-affairs (object, property) under different modes of presentation;
(c) the presence of the self would not be essential to the memory, and thus the episode would not appear to the rememberer as one that she formerly experienced. The only way to render necessary the presence of the self and thus episodically remember an episode is to impose on that episode one or more modes of presentation. This process of transformation creates mental content, which is different from the episode.
I have also given some arguments in favour of the distinction between object and content (even if I used different terms), focused on the possible discrepancies between the content and the object of the same personal memory (Trakas, 2014, p. 32-35). The arguments that Rowlands gives are nonetheless persuasive and sufficient by themselves to convince the readers of the need for this conceptual distinction. What is more, his explanation along these two chapters shows the inseparability that is characteristic of the act of remembering, the memory content and the mode of presentation, as well as the key role played by the act of remembering in the construction of our episodic memories: it is finally the act of remembering which is responsible for the mentality, the individuation and the ownership of the remembered content.
Before coming back to Rilkean memories, I would like to make a comment about a remark made by Rowlands. According to our author, his conception of content must not be understood as something that stands between the subject and the episode, but simply as a way or mode of remembering an episode. Because the content is nothing more than the episode transformed in certain ways, Rowlands concludes that while remembering “content” we are in direct contact with the past. Like other authors, Rowlands couples a representationalist conceptualization of memory to a direct realism theory of memory. I profoundly believe not only in the incompatibility of these two conceptions of memory, but also in the impossibility of defending a direct realist view of memory. Direct realist accounts of memory cannot accommodate the existence of memory traces and fail to explain the fallibility and change that characterize our memory representations. They also fail to give a criterion to distinguish between immediate acquaintance in perception and immediate acquaintance in memory (Trakas, 2014, pp. 10-17). Memory researchers would do better to abandon the idea that memory allows us to be in direct and immediate contact with the past and to ask, instead, how a capacity that does not allow us to be in direct contact with the past can nevertheless produce reliable representations of the past.
Forgetting, endemic inaccuracy and a person’s unity and identity (for her and for the others)
In this last section I focus on chapters 5, 6, 7 and 10, chapters where Rowlands develops the role that Rilkean memories—these memories that have no content and are pure act—play in making us who we are.
As I already mentioned, episodic memories are in general considered to give an answer to the metaphysical problem of the self’s unity and identity through time (what makes a person at a time t2 a unified individual identical to a person at a time t1?). Nonetheless, the endemic inaccuracy and the forgetting of episodic memories compromise the identity of the person over time and thus threatens the role played by episodic memory in the explanation of the unity and identity of the metaphysical self. On the contrary, the endemic inaccuracy and the forgetting of memories is not a threat for the autobiographical self, neither from a first person point of view (that is, the self-experience of unity and identity) nor from a third person point of view (the recognition of the unity and identity of another self). Rowlands considers them as self-constructing opportunities that can play a positive role in the constitution of a person.
I will come back in a few lines to Rowlands’ idea of the positive role played by the inaccuracy and the forgetting of episodic memories in the constitution of the autobiographical self. I would now like to make a brief comment about Rowlands’ arguments to state the endemic unreliability of memory. Rowlands asserts the endemic unreliability of memory based on empirical studies on false memories (like studies on flashbulb memories) as well as on memory reconsolidation that, according to our author, would explain why most of our memories are unreliable: every time we access a memory trace, it returns to the unstable and labile state characteristic of short-term memory, and becomes thus sensitive to change. The idea that most of our memories are “false” is not new and has been advocated by psychologists like Elizabeth Loftus: “in essence, all memory is false to some degree” (Bernstein & Loftus, 2009, p. 373). Rowlands rightly recognizes that the notions of accuracy and inaccuracy (conceived as a spectrum) are better suited to characterize memories than the notion of truth and falsity, but he still holds that inaccuracy is endemic to memory. I believe first, that Rowlands misunderstands the notion of change during the process of reconsolidation: “change” does not necessarily mean “distortion” (a term that he explicitly uses), and several times, a change of a memory trace is necessary to render a memory more accurate (for example, when we acquire new information that allows us to better understand a past experience). Secondly, Rowlands—and Loftus—present radical and extremist conceptions of the notions of truth / falsity and accuracy / inaccuracy: all memory representations that are different (even slightly different) from a past representation would be false or inaccurate, and that is why inaccuracy (or falsity) is endemic to episodic memories. This is a surprising conceptualization for someone who proposed to conceive the epistemic values of memories in terms of a spectrum of accuracy versus inaccuracy. Third, I do believe that people who think that memory is endemically unreliable are wrong. Instead of looking at empirical studies on false memories, we would do better to look at our everyday functioning and the way it would be affected if a large number of our memories would be unreliable: not only could we not successfully navigate the physical and social world, but probably we could not even have evolved as we did. Most of our everyday actions are guided by semantic as well as episodic memories, and a human being with an unreliable memory system would be very different from what we are; maybe she will not even be human. Anthropological studies take time and are not often practiced to study psychological phenomena, but they would be of great help to provide empirical data on the reliability of the human memory system(s).
In any case, it remains to be seen how the endemic inaccuracy and the forgetting of memories can be self-constructive for the autobiographical self. Rowlands does not give an explanation of the positive role that endemic inaccuracy plays; he only states that “for an autobiographical project, false memories can be just as self-constructive as real memories” (Rowlands, 2017, p. 115). If confabulations can present some benefits for the confabulator (at least she has a story to tell to herself about who she is), it remains an open question as to whether confabulations are as self-constructive as real memories. The case of forgetting is analysed with more detail, in a specific and interesting chapter about this notion (chapter 5). Passive forgetting (memory decay over time) compromise the memory-based version of the metaphysical explanation of the self and also plays a negative role in the construction of the autobiographical self (by unbalancing the story of who we are, or making us repeat old mistakes). Nonetheless, active forgetting, that is, the conscious and unconscious engagement in a process of forgetting, plays a positive role in the construction of the autobiographical self: it allows us to forget the useless—in order to release cognitive resources—and to forget the pernicious. Furthermore, active projects of forgetting, which can include the explicit manipulation of the environment in order to facilitate or scaffold the process of forgetting (like destroying photographs), say a lot about the person you are. But there is a more pervasive and primitive process of forgetting than active forgetting, which does not require the existence of an autobiographical self who conducts the forgetting, but plays a significant role in the development and preservation of the autobiographical self. This primitive, passive but positive process of forgetting memory content refers to the process that originates in Rilkean memories. Rilkean memories play a positive role in holding the identity and the unity of the autobiographical self through time, in the face of the lost and inaccuracy of episodic memories, and more especially when the self is no longer capable of engaging in remembering (or forgetting), like the cases of Patsy Hasset or Taylor’s mother.
Rowlands compares Rilkean memories to literary style (to understand this analogy, it is worth mentioning that Rilkean memories are pure acts of remembering, without content). If we find a couple of disconnected pages of a book, the style of these pages combined with the remaining content can be sufficient to establish or at least suggest the identity of the author. The same applies to Rilkean memories. Embodied Rilkean memories, that is, the tendency to do things in certain ways in certain circumstances, and affective Rilkean memories, that is, the disposition to have certain moods and feelings in certain environmental circumstances, are part of a person’s existential style. Rilkean memories connect the person to her past and provides a form of continuity between the person who has the Rilkean memories and the person who had their episodic ancestor. Rilkean memories, as part of a person’s existential style, allow an outsider observer to distinguish and recognize individuals on their basis. That is why Rilkean memories play a key role in the recognition of the unity and identity of a person made by a third party. That is why Rowlands is still able to recognize his wife’s grandfather Patsy as the same person he used to be before developing Alzheimer’s disease and thus losing all his episodic memories. This is the right time to remember Taylor’s description of her mother quoted at the beginning of this review. For Taylor, her mother was the same person as before, because she could still recognize her existential style, that is, her particular way of being, acting and feeling: her mother was still a cheerful and affectionate person, who still enjoyed gentle joking and teasing, being around people and having a conversation, and who also still beamed radiantly at small children. Rilkean memories are finally what justify third person recognition judgements.
Rilkean memories solve then the puzzle of the unity and identity of a person from a third point of view, that is, the puzzle of the recognition of another person. But there is a still another puzzle: the problem of explaining the self-experience of unity and identity, that is, the way in which the present self (R-self) experiences a past self (W-self) as a unified individual, identical with herself. According to Rowlands, Rilkean memories are also the key to solve this puzzle, but they do not feature as what they are—Rilkean memories—but as what they were before becoming Rilkean memories: as episodic memories. The necessary presence of the self in episodic memories is the key to first-person recognition: “The person who remembers is, therefore, in her memories even when those memories are not about her. She is in her memories not simply because she has carved or shaped them from the block of the episode. Rather, it is because she had to do this in order to make them something that could be remembered. The content of memory is always infused with the person who remembered and where she is in her life. The content of memory is, in this sense, infused with style. It is infused with, and therefore shaped by, the act of remembering (…) Style and content may eventually go their separate ways—this is what happens when a Rilkean memory is formed. But before this happens, the two are entangled. The style of a person is always there, in the midst of content” (Rowlands, 2017, pp. 194-195). Therefore, because the autobiographical self is present in each and every one of the episodic memories that collectively form the record of her life, the self who remembers (R-self) experiences herself as a unified individual identical with any of her past selves (W-self). This means that Patsy and Taylor’s mother, as well as other people with dementia, could still experience their unity and identity through time if they have at least one episodic memory that remains accessible to their consciousness.
Memory and the Self is an excellent book on memory, with a highly sophisticated dose of philosophical content and literary style. However, I must admit that at the end of the book I was slightly disappointed. The main purpose of the book is to introduce the notion of Rilkean memories and explain the key role they play in maintaining the unity and identity of the (autobiographical) self. Nonetheless, from the first-person recognition perspective, Rilkean memories finally do not play any role; episodic memories do all the work. Saying, as Rowlands does, that Rilkean memories play such an important role because they were episodic memories before becoming Rilkean memories, does not help to assign a real role to Rilkean memories in the self-experience of identity and unity. Although one derives from the other, Rilkean memories and episodic memories are very different. Furthermore, episodic memories do not necessarily become Rilkean memories. The truth is that Rilkean memories do not play any explanatory role in first-person recognition, and that episodic memories are the key to understand how we experience our autobiographical selves as a unified individual, identical to itself through time, despite Rowlands denying this in chapter 6: “these two facts [inaccuracy and forgetting] present a problem for the idea that our episodic memories play a major role in the construction of the autobiographical self” (Rowlands, 2017, p. 122). Moreover, in this section I would have expected more discussion with Stan Klein’s view—an author who is known and mentioned by Rowlands in this book—for whom the unity that we attribute ourselves as persons can be interpreted as a pre-reflective feeling of personal continuity that would permeate all our experiences (for evidence of an amnesic patient who maintains a sense of personal identity despite being unable to retrieve episodic and semantic personal memories, see Klein, 2014).
Rilkean memories do play a key role from the third-person recognition perspective. However, when analysing these cases, we realize that what allows us to recognize someone as the same unified individual identical through time is nothing more than different kinds of habits and character traits. Rilkean memories are finally nothing more than environmental embedded habits and character traits. Rowlands is aware that Rilkean memories may not be a new, non-standard form of memory, but just the product of a process of transformation of episodic memories (Rowlands, 2017, p. 54). This is nonetheless unimportant to him, and maybe it should also be unimportant to the reader in order to get Rowlands’ message: the recognition of these habits and character traits as states that carry in them a trace of the personal past and that allow the personal past to live in the subject in a different way than memories (understood in a familiar sense).
In spite of this small disappointment that other readers may share with me, Memory and the Self is a very pleasant book to read that truly deserves to be read, reread, and discussed by those interested in philosophy of mind and in memory.
Bernstein D. & Loftus E. (2009). How to tell if a particular memory is true or false. Perspectives on Psychological Science, 4(4), 370-374.
Broad, C. D. (1925). The mind and its place in nature. Londres, Kegan Paul, Trench, Trubner & Co.
Conway, M. (2005). Memory and the self. Journal of Memory and Language, 53, 594-628.
Conway, M. (2009). Episodic memories. Neuropsychologia, 47, 2305-2313.
Conway M. A., & Loveday C. (2015). Remembering, Imagining, False Memories and Meaning. Consciousness and Cognition, 33, 574-581.
Klein, S. & Nichols, S. (2012). Memory and the sense of personal identity. Mind, 121, 677-702.
Klein, S. (2014). The two selves: Their metaphysical commitments and functional independence. New York, Oxford University Press.
Locke, J. (1690). An Essay Concerning Human Understanding. [electronic version]. Retrieved from http://www.earlymoderntexts.com/assets/pdfs/locke1690book1.pdf
McCormack, T. (2001). Attributing episodic memory to animals and children. In T. McCormack & C. Hoerl (Ed.), Time and Memory. New York, Oxford University Press, 285-313.
Meinong, A. (1899). On Objects of Higher Order and their Relationship to Internal Perception. Zeitschrift für Psychologie und Physiologie der Sinnesorgane, vol. XXI, 181-271.
Piolino, P., Desgranges, B, & Eustache F. (2009). Episodic autobiographical memories over the course of time: Cognitive, neuropsychological and neuroimagining findings. Neuropsychologia, 47, 2314-2329.
Prinz, J. J., & Nichols, S. (2017). Diachronic Identity and the Moral Self. In J. Kiverstein (Ed.), The Routledge Handbook of Philosophy of the Social Mind. New York, Routledge, 449-464.
Russell, B. (1921). The Analysis of mind [electronic version]. Retrieved from https://www.gutenberg.org/files/2529/2529-h/2529-h.htm
Taylor, J. S. (2008). On Recognition, Caring, and Dementia. Medical Anthropology Quarterly, 22(4), 313-335.
Trakas, M. (2014). Personal Memories. PhD thesis, Ecole des Hautes Etudes en Sciences Sociales & Macquarie University.
Twardowski, K. (1894). On the Content and Object of Presentations. The Hague, Martinus Nijhoff (1977).