Episodic Imagination and Episodic Memory: What's the Difference?

A post by Kourken Michaelian.

Kourken Michaelian is senior lecturer in philosophy at the University of Otago (New Zealand). He is the author of Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past (MIT 2016) and co-editor of Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel (OUP 2016) and The Routledge Handbook of Philosophy of Memory (Routledge 2017).

Kourken Michaelian is senior lecturer in philosophy at the University of Otago (New Zealand). He is the author of Mental Time Travel: Episodic Memory and Our Knowledge of the Personal Past (MIT 2016) and co-editor of Seeing the Future: Theoretical Perspectives on Future-Oriented Mental Time Travel (OUP 2016) and The Routledge Handbook of Philosophy of Memory (Routledge 2017).

What is the difference between episodic imagination and episodic memory? At first glance, imagining events and remembering events would seem to be highly similar processes. Philosophers of memory have, however, usually tried to draw a sharp distinction between them. Indeed, one natural understanding of traditional philosophical theories of remembering treats them precisely as attempts to specify the difference between imagination and memory. It may be, however, that traditional theorists have been barking up the wrong tree—that there is, after all, no deep difference between imagination and memory.

As John Sutton and I point out in a forthcoming entry in the Stanford Encyclopedia of Philosophy, the question of how to distinguish between memory and imagination is ambiguous. On the one hand, we sometimes remember but do so in a way that is in some sense inadequate; in such cases, we naturally say that we are “only imagining”. The question can thus be taken to concern the distinction between cases in which the subject remembers successfully and cases in which he remembers unsuccessfully. On the other hand, we sometimes think about the past in a way that—if traditional theories are right—does not amount to remembering at all; in such cases, too, we naturally say that we are “only imagining”. A theory of remembering must therefore describe both the relationship between successful and unsuccessful remembering and the relationship between remembering, whether successful or unsuccessful, and “mere” imagining.

Most theories of remembering have focused on the latter relationship. Empiricists such as Hume, for example, attempted to distinguish between remembering and imagining by appealing to the latter's greater flexibility or to the former's greater vivacity. The empiricist theory is no longer very popular, but the dominant contemporary theory, the causal theory of memory, continues to draw a sharp distinction between remembering and imagining. According to the causal theory—which was given its most influential statement by C. B. Martin and Max Deutscher—what makes the difference between remembering and imagining is that the former but not the latter essentially involves an appropriate causal connection between the subject's current representation of an event and his earlier experience of the event. What makes a causal connection “appropriate”, in turn,  is that it goes via a memory trace, where a memory trace is understood as stored content deriving from the subject's experience of the event in question. If the causal theory is right, then, one might experience an event and later entertain a representation of it but not count as remembering it even if one's representation is highly accurate—one will not count as remembering it, in particular, when the accuracy of one's representation is due not to the existence of an appropriate causal connection with one's experience but rather to some other factor.

The causal theory remains extremely popular: it is defended in detail in a recent book by Sven Bernecker, and even those who do not explicitly endorse the theory do not normally explicitly reject it. But it is difficult to reconcile the theory with the picture of the nature of remembering that is painted by contemporary psychology. There is overwhelming evidence that remembering is a thoroughly reconstructive process, in the sense that the content of a retrieved memory representation derives not just from the subject's experience of the remembered event but also from a variety of other sources. When we remember an event, we often import content deriving from our experience of other events. We reshape content on the basis of our general semantic knowledge. We rely on various heuristics to simplify or to elaborate on stored content. We allow information available in the environment in which retrieval occurs to shape the content of the retrieved representation. The upshot is that, in all or most cases, a retrieved representation will include content not deriving from the subject's experience of the remembered event. It is clear, in fact, that, in many cases, only a minority of the content derives from the subject's experience of the event, and we have good reason to think that, in some cases, none of the content derives from the subject's experience of the event.

The latter point, if we accept it, undermines the causal theory: if there are cases of remembering in which none of the content of the retrieved representation derives from the subject's experience of the event, it cannot be the existence of an appropriate causal connection with the event that makes the difference between remembering it and imagining it. Thus causal theorists have, naturally, resisted this point, insisting that there is a difference in kind between cases in which none of the content of the representation derives from the subject's experience of the event and cases in which at least some of the content derives from the experience. The motivation for this insistence is, however, entirely intuitive, in the sense that causal theorists do not typically appeal to empirical evidence for a difference in kind between cases in which content is transmitted from the past experience to the present representation and cases in which it is not.

It looks increasingly doubtful that such evidence will be forthcoming. Indeed, episodic memory is now viewed in psychology as a form of mental time travel analogous to episodic future thought, and a multitude of findings demonstrate that remembering the past and imagining the future have similar phenomenologies, involve similar cognitive processes, and rely on similar brain regions. If it were not for a philosophical tradition which has rarely considered episodic memory in connection with episodic future thought, these findings would perhaps come as no surprise—again, memory and imagination seem at first glance to be highly similar processes. In line with the tradition, some philosophers, such as Dorothea Debus and Denis Perrin, have attempted to push back against the mental time travel framework, but the arguments that they have offered end up presupposing the causal theory. Others, such as Felipe De Brigard, have suggested that the framework can be reconciled with the causal theory, but it is not clear how a theory that sees the transmission of content from past experience to present representation as essential to remembering might be reconciled with a framework that sees episodic remembering as being a process of the same kind as episodic future thinking. For obvious reasons, imagining the future does not presuppose the existence of a causal connection between the subject's current representation and his experience of the represented event. If episodic remembering is a process of the same kind as episodic future thinking, then we may have to admit that remembering the past likewise does not presuppose the existence of a causal connection between the subject's current representation and his experience of the represented event.

The mental time travel framework, in other words, suggests that we should consider the possibility that the attempt to distinguish between remembering the past and imagining it is ultimately misguided, that the causal theory (like other traditional theories) has been barking up the wrong tree. I take this possibility seriously in my recent book, where I reject the causal theory in favour of a simulation theory according to which there is nothing that makes the difference between remembering and imagining the past, for remembering just is a matter of imagining the past. The simulation theory is compatible with the claim that remembering usually involves an appropriate causal connection. But—unlike the causal theory—it is also compatible with the claim, suggested by the evidence for memory as a form of reconstructive mental time travel, that remembering does not always involve an appropriate causal connection. Thus, if the simulation theory is right, then a case of the sort that the causal theory classifies as one of mere imagining—a case in which one experiences an event and later entertains an accurate representation of it but in which the accuracy of the representation is not due to the existence of an appropriate causal connection—may be appropriately classified as one of genuine remembering.

It will be appropriately so classified as long as one further condition is met: the simulation theory does not actually say simply that remembering is a matter of imagining the past; instead, it says that remembering is a matter of reliably imagining the past. Recall that a theory of remembering must describe two relationships: the relationship between remembering and “mere” imagining and the relationship between successful and unsuccessful remembering. So far, we have been concerned with the former relationship. Concerning the latter, some, such as Sarah Robins, have worried that, if we give up the causal condition, we will be unable to distinguish between successful remembering and unsuccessful remembering, for example, confabulation. But the causal condition is not required in order to draw this distinction, nor is it the best means of doing so. The difference between successful remembering and unsuccessful remembering is best understood not in terms of causal connectedness (or the lack thereof) but rather in terms of (un)reliability: what goes wrong, in cases of confabulation, is not that no content is transmitted from past experience to present representation but rather that the subject's memory system is malfunctioning—that he is simply “making things up”.

What is the difference between episodic imagination and episodic memory? If the simulation theory is right, there is none. More precisely: there is no difference between successfully episodically remembering and successfully episodically imagining past events. This, at any rate, is how things look to one philosopher of memory. Given the nature of this blog, I close with a question: how do they look to philosophers of imagination?