This week I am writing a series of posts about my new book Memory: A Self-Referential Account (Oxford University Press, 2019). My first post, today, concerns the metaphysics of experiential, or episodic, memory.
In chapter two, I address the question of what it takes for a subject’s mental state to qualify as an episode of remembering. My proposal is that a mental state qualifies as an episode of remembering just in case it has a certain functional role within the subject’s cognitive economy. I motivate this view by highlighting the shortcomings of the main alternative views.
There seem to be two main views about this issue. One of them is popular within philosophy, whereas the other one is popular within psychology. The first view (the so-called ‘causal theory of memory’) is that a subject’s mental state qualifies as an episode of remembering just in case that mental state has a certain causal history. Specifically, if a subject represents a certain fact in the past in virtue of having some mental state, then the subject is remembering the fact in question just in case their mental state causally originates in the subject’s past experience of that fact.
This view has a number of difficulties, but I will concentrate on just one, having to do with the possibility of misremembering. We routinely misremember things that we correctly perceived in the past. I may correctly perceive an apple to be red today, and misremember it as having been green tomorrow. It must be possible, therefore, for our memories to be, so to speak, embellished with details that were not part of our past experience. And yet, this is impossible if the causal theory is correct. The moment an extra detail is added, the relevant mental state no longer counts as an episode of remembering and, hence, it does not count as an episode of remembering falsely.
The second view (we may call it the ‘narrative theory of memory’) is that a subject’s mental state qualifies as an episode of remembering just in case it contributes to a narrative of the subject’s life. More specifically, if a subject represents a certain fact in the past in virtue of having some mental state, then the subject is remembering the fact in question just in case they are using their mental state to construct a smooth and robust story of their life.
This view has a number of difficulties, but I will concentrate on just one, having to do with the demand for a narrative. Sometimes, subjects remember things that took place in their past, even though they are unable to relate the relevant fact to anything else that they remember about their lives. I may remember scratching my knees badly while playing in some playground as a kid, and yet I may be unable to relate that fact, temporally, to any other fact that I remember about my childhood.
My suggestion is that if a subject represents a certain fact in the past in virtue of having some mental state, then the subject is remembering the fact just in case their mental state is of the type which tends to be caused by an experience of that fact, and tends to cause the subject to believe both that they experienced the fact, and that the fact took place in the past. The backward-looking part of this functional role is meant to preserve the main idea in the causal theory, while allowing for the possibility of misremembering. If I visualise the apple as having been green, for instance, I am not occupying a mental state that originates in my past experience. But it is the kind of mental state that I would normally have if I had perceived a green apple, which is why it counts as an episode of remembering (remembering falsely, as it happens).
The forward-looking part of the functional role is meant to capture the main idea in the narrative theory, while allowing for the possibility of remembering facts which are, in our mental lives, isolated from anything else we remember. If I visualise falling in a playground, for instance, I cannot tell a story about what happened before, and after, that episode. But I am inclined to think, in virtue of visualising my fall, that it happened to me, and that it happened in the past. And this is, partly, why it counts as an episode of remembering.
The functionalist proposal, however, is neutral on the kinds of contents that memories have. It is neutral on the kind of phenomenal features that memories have as well. And, finally, it is neutral on the kind of knowledge that memories afford. Those issues require independent investigations which are the subjects of the remaining chapters in the book.
Thank you, Dr. Fernandez, for your post.
A couple thoughts. My take on episodic memory is that it is not strictly causal, because it does directly drive behavior. It is, by contrast, informational, meaning that it can be recalled later, to inform future decision-making. It is only weakly causal, because the information from memory may drive behavior (or not).
Memories are generated (almost) every moment of our waking life. Most memories that are created are probably never recalled, because they are too mundane and non-newsworthy.
I think of episodic memory as a ‘movie’, created by the brain, of current events, so the movie of those events can be played back later, and learned from. We see this particularly in REM sleep and rest, when the hippocampus re-activates to ‘tutor’ the neocortex on new skills and knowledge it recently learned.
I do think that we cannot be neutral about the phenomenal aspects of episodic memory, because those aspects are precisely what allows us to remember. If I witness a robbery, and the police ask me what the robber was wearing, I only have one source of information on that. That source is the visual qualities of my episodic memory. If I am asked to remember the smell of cinnamon, the recall = the phenomenal experience. They are one and the same. The proxy = the qualia.
This is particularly true of animals that do not have language, which is a powerful mnemonic. If a rat encounters a novel berry, its first response is to go into exploratory behavior, observing, smelling, making small tastes, encoding the bitter with the sweet with the sour. During this time, the hippocampus is highly activated, encoding all these subtle, multidimensional qualities for later recall. If the berry turns out to be poisonous (and the rat survives), then the rat needs to recall the various qualia of the berry, and associate them with the subsequent qualia of feeling ill. Later, when the rat encounters a similar berry, and goes into exploratory phase, it will ultimately find enough qualia that match the old memory, and will decide to avoid the second encounter with the poisonous berry. In this way, the qualia of episodic memory serve as the mnemonics, since there is no language to do the job, nor any previous experience with this kind of berry, to train a concept into the brain. Qualia are all the rat has, to match the subsequent encounters with the novel berry.
Hi Matt, thank you for your comments.
On the causality issue: I’m not denying that memory is an important source of information. The question is whether that is what makes a particular mental state a memory. It seems to me that if we do not factor in some kind of requirement about the causal origin of the mental state, then we may end up accepting, as memories, mental states which offer information not significantly different from that offered by memories, even though they are not memories. I suspect this is why we are so bad at telling apart memories from episodes of imagination, for example.
On the phenomenal qualities of memory: Once again, I’m not denying that the phenomenal features of memories are important. But the question is whether those are essential to memories. That is, could a memory lack the phenomenal features that, usually, memories have and still count as a memory? It seems to me that, once more, something like this must be going on when we mistake an episode of imagination for a memory.
Hello Jordi (if I may) — I’ve enjoyed your work on thought insertion. Glad to hear of this new book on memory.
A couple questions:
What’s your reply to the causal theorist who says that he is giving an account of (successful) remembering, whereas you are giving an account of what successful and unsuccessful remembering have in common?
Second, could you say a bit more about how mental states are to be typed when we assess whether some putative remembering is of the same type as a particular previous experience–i.e., in what sense must they must be of the same type for one to be a memory? The question behind the question: could a belief about the past be of the same type as some previous experience of a fact, and be caused by that experience? Any barrier on that? And thus is there any barrier, on your view, on rememberings being occurrent beliefs about what one experienced in the past (as opposed to states apt to cause such)?
thanks,
Peter
Hi Peter, thank you for your comments.
I guess that a causal theorist of memory *has* to say that they are giving an account of remembering as a factive mental state. My concern about this type of approach is analogous to a concern which is sometimes raised against disjunctivism in perception: Let us suppose that there is no such thing as misremembering. There are episodes of remembering, and there are states that we falsely judge to be episodes of remembering. It seems to me that the causal theorist we are entertaining is committed to saying that these are, metaphysically, different types of states even though, from the first-person point of view, they are indistinguishable. Doesn’t the causal theorist owe us an explanation for this?
With regards to this question: “could a belief about the past be of the same type as some previous experience of a fact, and be caused by that experience? Any barrier on that?”
I should have started by saying that the book is an investigation of episodic, and not semantic memory, and the distinction between the two types of memory (which, granted, is not easy to draw) seems to concern this issue precisely. So let me say a bit about this distinction, and then I’ll circle back to your question. When you remember semantically, you are having a belief; a belief which has been preserved from some time in the past, that is, the time at which you originally learnt the remembered fact. By contrast, when you remember episodically, you are having an experience; an experience which is somehow related to a past perceptual experience of the fact. What is important for your question is that, whereas remembering semantically that P requires believing that P and having believed that P in the past (and the former occurring as a result of the latter), remembering episodically that P does not require this. It is independent from belief in that sense.
Now, let’s get to your question: Could a belief, which has been caused by a past experience of a fact, count as a memory of that fact? I am inclined to answer ‘no’. That belief could have two different origins: If the belief is based on a current experience wherein one represents the fact, then the experience at issue is the memory of the fact; not the belief. And one is remembering the fact episodically. (Notice that this could happen even if, in the past, one did not believe that the relevant fact was the case when one experienced it, perhaps because one was not paying attention to it.) If the belief is a belief that one acquired in the past, when one experienced the fact, then one is remembering the fact, but one is remembering it semantically. Notice that this could happen even if, in the present, one has no experience wherein one represents the fact, perhaps because one has forgotten what it was like to experience it. If this is what is happening, then one is only remembering the fact semantically, and therefore this is outside of the scope of the functionalist view offered above.
I hope this helps!
Hi Jordi,
Thanks for your helpful replies. I’m left with a few more questions, in case you have time to continue. In answer to your question of what the subjectively indistinguishable cases of remembering and misremembering an event have in common, it seems the causal theorist could say: “both involve mentally representing (perhaps via imagery) some event as having occurred in the past; but we only call this ‘remembering’ when the representation was caused in the right way by a past experience of the event (by analogy, we’ll only call the belief that p a case of knowing that p when certain other conditions are met).” If this were the right way to look at things, then there is not a deep disagreement between you and the causal theorist. So I suspect it’s not the right way to look at things. I’m just not clear on why yet. But note that this is coming from someone new to the memory literature.
My other question is about the relation you see between occurrent semantic memories and episodic memories. Isn’t it fair to say that episodic memories are “preserved” from the time at which the initial experience occurs, in roughly the same sense in which semantic memories are preserved from the time at which they are learned? (Even if they involve different cognitive systems.) On the supposition that episodic memories are a subclass of beliefs, then remembering episodically that P requires believing that P and that one has believed that P in the past. It seems one would already need to have decided that episodic memories are not beliefs in order to conclude that having them doesn’t require one to have believed their content in the past. Does the issue turn on the idea that episodic remembering is “having an experience”? Is there reason to think that semantic remembering cannot involve “having an experience,” or that having an occurrent belief, in general, cannot involve having an experience?
No worries if you’re too busy with your other postings to reply..
Hi Peter,
on the first issue: I think you are right, that would be a reasonable response. In fact, it would be the response I myself would give to the question of why we cannot easily distinguish episodes of imagination from remembering episodes. So it comes down to whether you think that memory can only be successful. I think your reading of the dialectic here is right.
on the second issue: Yes, I am assuming that there is a distinction between having experiences and beliefs. In perception, for example, I walk around seeing lots of things, but I don’t believe, of each of them, that they are the case. (We don’t have the cognitive capacity for doing this.) The thought is that a similar distinction can be drawn in memory. There are experiences that we have, experiences wherein certain things are presented to us as being in the past, and then there are beliefs that we may or may not form on the basis of those experiences. You are right that this is presupposed, and not argued for. I imagine that, if there is a problem here, it is a broader issue that does not only concern memory.
Thanks — these are much appreciated!
Hi Jordi,
Thanks for this wonderfully clear summary of a complex topic. It is an ideal jumping off place for discussion.
1. I suspect that the causal theory has resources to resist your critique. Is this a correct summary of the causal theory?—M is a memory of E if M originates in experience of E and M represents E. You say that this does not accommodate misremembering.
Can’t the causal theorist evade your critique by saying that M can incorporate many representational elements? Suppose M that represents having dinner with you in mid-summer in Adelaide and suppose that M originated in experience of dining with you in July (which is not mid-summer in Adelaide) Can’t the causal theorist say that M is a memory of dining with you, but not a memory of dining with you in mid-summer (since I never dined with you in mid-summer)?
2. I’m not sure that your view clearly wins over the causal theory. You tie memory to belief about the past. Now, imagine that I have a vivid image I of a traumatic event T. Suppose this leads me to believe that I had a dream about T. Suppose that I did in fact have a dream about T and that I is an image of what happened in the dream. Your theory says that I am remembering the dream. Okay good.
But what if in fact the memory originated in an actual traumatic event T’ that antedated the dream. Suppose that I doesn’t lead me to any belief about T’. Your account doesn’t allow that I is a memory of T’, but the causal theory does. I’m wondering what you make of the difference. I myself am inclined to think that both are right: It’s a memory of both. (This raises a question about transitive memory: if E is a memory of mental event E1, and E1 is a memory of E2, then is E a memory of E2?)
Let me leave it at that. I really enjoyed the post.
Hi Mohan, thank you for your comments.
I think I’m missing something about the causal theorist’s reply:
If M represents your having dinner with me in mid-summer in Adelaide, and M originates in an experience of dining with me in July (not mid-summer down here), then, sure, we can say that M is a memory of dining with me. But the point is that we also want to say that, when you have M, you are misremembering something about that dinner. And this is not something that the causal theorist can say. They can only say that you remember having dinner with me, and you … imagine, I’m guessing, the fact that it happened in mid-summer. But there is nothing that you are misremembering, strictly speaking. This seems counter-intuitive to me, but (see Peter’s point above) I suppose the causal theorist can just say that ‘remembering’ is factive and we should not try to accommodate misremembering in the first place.
With regards to your second case. I am going to assume that, even though my dream of T did not cause I on this occasion, I is the type of image that, normally, I would have when I have a dream of T of that type. You say that I is an image of what happened in the dream so I’m guessing you’ll grant me this assumption. It is an image of what happened in the dream so, normally, the dream would have caused it but, on this occasion, it did not cause it. Now, if the mental image I makes no difference as to whether I believe that T happened or not, (that is, if I am not inclined to believe, when I have I, that T was a part of my life), then I would not count I as a memory of T. I would only count it as a memory of the dream of T.
There is classical case of the painter who believes he is imagining a house, and paints it in his canvas, only to find a remarkable similarity (pointed out by others) between the house he’s painted and a house he used to visit on family holidays as a child. This case is commonly used to motivate the view that, I think, you’re subscribing: Remembering something does not require the inclination to believe that it happened. I, by contrast, count the painter’s mental image, on the basis of which he paints the house in the canvas, as an episode of imagination. And I take this case to simply show that, sometimes, we are able to imagine a particular object, with all the detail that we visualise in it, only because, in the past, we perceived that object. That, in my view, does not make the relevant mental state a memory.
Does that make sense?
Thanks, Jordi.
On point 1, I think we are in general agreement about the case; it’s just that I think the causalist has some resources to meet your criticism.
On point 2, the case of the painter is a little different from that of the dream. In the case of the painter, the house caused the image. In the case of the dream, there are two relevant causes: the dream and the dream’s causal antecedent. Your account seems right regarding the dream: I leads the subject to have certain beliefs about the dream. And the causalist’s account seems right also about the antecedent trauma: the subject has no belief about the latter, but I is caused by it, and plausibly I is about it (as well?).
Hi Mohan,
on point 2, my original reading of your case was that the mental image only had a cause, namely, the dream’s causal antecedent (and not the dream itself). I thought it was a case in which only the forward-looking part of the functional role was satisfied by one candidate (namely, by the dream) and only the backward-looking part of the functional role was satisfied by the other candidate (namely, by the dream’s causal antecedent; the traumatic event). In any case, I think my response should remain the same: The reason why I don’t think it’s a memory of the traumatic event is that, as far as I’m concerned, the traumatic event did not happen in my life. The memory makes no impact on my beliefs about whether that event took place or not. This is why I thought the case of the painter was relevant but, you are right, these are two different cases.