Michaelian , Kourken , Mental Time Travel : Episodic Memory and Our Knowledge of the Personal Past

The topic at the heart of this book is the phenomenon of episodic memory, and the book aims to fill a ‘major gap in the philosophical literature on memory’ [xv]: While other philosophers have considered the phenomenon of episodic memory without making much of an ‘attempt to take nonphilosophical work into account’ [ibid.], this book sets out to develop ‘a systematic philosophical view of remembering based on current empirical research’ [ibid.], especially on so-called ‘mental time travel’. Thus, the aim is to offer a philosophical account of episodic memory which is distinctive in virtue of its attention to the relevant empirical literature. This methodological commitment might require philosophical choices. Indeed, some important philosophical work on memory is explicitly excluded from consideration in the book because those ‘philosophical theories are couched in terms that get little empirical traction’ [6]. This might be problematic because a main task of a project that so explicitly tries to bridge the gap between philosophical and empirical work on the same topic might be to show how philosophical work that has no immediate ‘empirical traction’ might, nevertheless, be brought into contact with relevant empirical work. Still, the book offers a comprehensive and knowledgeable survey of the literature with which it does engage, covering historical work (Plato, Aristotle, Locke, and Hume feature) as well as contemporary work, in both philosophy and the empirical sciences. Indeed, substantial parts of the book are descriptive, summarizing the extant literature on memory. For example, Part I offers a chapter (‘Situating Episodic Memory’) that places episodic memory into the wider context of various other memory phenomena in great detail, and another chapter (‘Memory Knowledge’) that summarizes views on epistemological naturalism; similarly, chapter 5 carefully discusses the ‘Causal Theory of Memory’ (which is ultimately put to one side, being said to be incompatible with empirical findings). Each of these (and various other) chapters provides detailed surveys of relevant debates, although it remains somewhat unclear whether, and if so precisely how, these surveys are needed as a basis for the more original contributions that the book aims to make. Maybe the book might have benefited from slightly shorter and more concise surveys and a more detailed development of its original contribution. The book’s main original contribution is the view that we should conceive of episodic memory as a form of ‘imagining or simulating the past’ [57]. According to this ‘simulation theory of memory’, ‘memory does not have a privileged status relative to other forms of imagination: episodic memory is distinguished from other forms of episodic imagination only by its specific temporal orientation’ [ibid.]. We should think of episodic ‘remembering as simulating or imagining episodes from the personal past’ [97]. This is obviously a provocative claim: in everyday contexts, we usually assume that, if it becomes clear that a current mental state or mental event is due to the subject’s AUSTRALASIAN JOURNAL OF PHILOSOPHY, 2017

imagining something, the relevant mental state or event should probably not count as a memory. Thus, in everyday contexts we usually conceive of memory as explicitly not a case of imagination. Michaelian suggests that we should give up this everyday assumption and hold instead that episodic memory is a form of imagination that is 'distinguished from other forms of episodic imagination only by its specific temporal orientation ' [57], because he believes that this view fits best with recent empirical work which shows that 'episodic remembering is a constructive process intimately related to future-oriented mental time travel and to a broader range of forms of episodic imagination' [ibid].
So what, exactly, does the 'simulation theory' say? Michaelian tells us that 'the only factor that distinguishes remembering an episode from merely imagining it is that the relevant representation is produced by a properly functioning episodic construction system … which aims to simulate an episode from the personal past' [97]. Thus [107], the simulation theory says that S remembers an episode e just in case S now has a representation R of e, [and] R is produced by a properly functioning episodic construction system which aims to produce a representation of an episode belonging to S's personal past.
In order to understand this claim better, we need to understand what it takes for an episodic construction system to 'aim to produce a representation of an episode belonging to S's past'. It would seem that, in order for a system to 'aim' at producing a representation of an event that occurred in the past, the system needs to have some means of 'guiding' attempts at producing relevant representations in a way that enables it to reach the goal at which it is aiming-namely, producing a representation of an event that occurred in the past. In order to guide any such attempt, the system needs to 'know' what it is aiming to produce a representation of: that is, somehow the system needs to be in a position to access information about the relevant past event of which it aims to produce a representation. This in turn would imply that there needs to be some relation between relevant past events and the present production of relevant representations which governs the constructive processes that presently occur, which in turn would mean that an episodic construction system aiming to produce a representation of an episode belonging to the subject's past is not quite as unconstrained as it might be when aiming to produce 'merely' imaginary representations.
This, however, is not how Michaelian conceives of the situation. Rather, because he wants to conceive of episodic memory as just another form of imagination, on his view we are not to think of relevant constructive processes as governed by any factors external to the 'episodic construction system' (such as some link with relevant past events)after all, processes of 'mere' imagination are not governed by such external factors either. Instead, Michaelian seems to suggest that it is a basic fact that, in the case of episodic memory, relevant systems produce representations of past events; we just find ourselves equipped with episodic construction systems that work in this well-calibrated fashion. Indeed, in the introduction, Michaelian rather disarmingly states, 'The suggestion is, in effect, that there is a sort of 'preestablished harmony' between the way in which memory tends to simulate episodes and the shape of the episodes that we actually experience' [11].
However, this claim in turn generates serious problems. First, it does seem to entail that, in cases of episodic memory, the episodic construction system does not, strictly speaking, 'aim' at anything at all; rather, the relevant system produces representations, and it just so happens that those representations represent events that subjects witnessed in the past. Thus, if the relation between how memory tends to simulate episodes, and the shape of the episodes that we actually experience, is a relation of 'preestablished harmony', then it seems false to say, as Michaelian does, that what 'distinguishes remembering an episode from merely imagining it is that the relevant representation is produced by a properly functioning episodic construction system … which aims to simulate an episode from the personal past' [97, my emphasis]. Presumably, what we should say is that the difference between episodic memories and mere imaginations, on this account, is that, in the case of episodic memories, some 'preestablished harmony' just happens to obtain between the production of relevant representations at present, and how things were in the personal past, whereas this is not so in cases of 'mere' imagining.
This in turn generates a second problem. Any account of episodic memory should, if possible, explain the relation between a present memory and the event in the subject's personal past of which it is a memory. However, in a context where we are trying to explain a certain relation, reliance on the claim that the relevant relation should be conceived of as some sort of 'preestablished harmony' should be a last resort, because it entails the admission that no proper explanation can be given. And while in some cases where philosophers resort to postulating a relation of 'preestablished harmony' it might indeed be that no further explanations can be given, the case of episodic memory is probably not of this kind.
On the contrary, philosophers who endorse some 'causal theory' of memory can explain how a present memory might be appropriately linked with a relevant past event: they will say that memory representations reliably represent relevant past events accurately because appropriate causal links obtain between relevant past events and present memory representations. Again, Michaelian himself discusses the 'causal theory' of memory in detail, but then rejects (most of) it because he holds that the 'constructive character of remembering ultimately requires us to abandon' it [97]. This, however, might be the move that we should reconsider and ultimately resist. There might be other ways of interpreting the empirical literature that leave room for a modified and tenable causal theory of memory.
Indeed, we might here have reached the point at which 'one person's modus ponens is another person's modus tollens.' On Michaelian's view, we should accept the 'simulation theory' of memory, which entails that the best that we can do to account for the link between an accurate memory representation and the past event that it represents is to describe this relation as one of 'preestablished harmony'. Michaelian concludes that we should therefore hold that such a relation obtains.
Given this rather unappealing conclusion, one might want to insist that any account of memory with that entailment should be relinquished; indeed, the opponent might insist, a satisfactory account of memory should say something explanatory about how a current memory 'links' the remembering subject to the past remembered event, and we should not accept any account of episodic memory entailing that no such explanation can be offered.
Indeed, given that an alternative account that does (at least partly) explain the relation between the remembering subject and the past remembered event is available in the form of the causal theory of memory, readers of this book might conclude that Michaelian's 'simulation account' should be rejected, and that we should instead make use of his own careful discussion of the various modifications that one might have to make to a 'traditional' causal theory of memory in order to accommodate memory's 'constructive character'.
Such a modified causal theory that accommodates the 'constructive character' of memory might well attract all of those for whom Michaelian's modus ponens (concluding that the relation between a present memory and the relevant past event is one of 'pre-established harmony') is a modus tollens; and in chapter 5 Michaelian provides those people with much useful material upon which to reflect further in developing their own accounts.
In summary, the book discusses a wide range of important issues in the philosophy of memory. To some, the book's main original contribution, the 'simulation theory' of memory, might seem too radical, but it certainly deserves close attention; the book also discusses an impressive amount of the relevant extant literature, providing very interesting avenues for further exploration. Anybody interested in the phenomenon of episodic memory should therefore benefit greatly from reading this book.

Dorothea Debus
University of York dorothea.debus@york.ac.uk