Owens, D. (1996) A Lockean theory of memory experience. Philosophy and Phenomenological Research, 56 (2). pp. 319-332.
A LOCKEAN THEORY OF MEMORY EXPERIENCE
The paper aims to provide an account of the phenomenological differences between perception, recognition and recall. In the first section, recall is distinguished from non-experiential forms of memory. In the second section, it is argued that we can't distinguish perceptual experience from the experience of recall by means of perception's present tense content because it is possible to perceive as well as to recall the past. The Lockean theory of recall as a revival of previous perceptual experience is then introduced, applied and defended against objections. Next, recall is distinguish from memory recognition. Finally, some relevant psychological data is described.
For Locke, memory is a power of the mind "to revive Perceptions, which it has once had, with this additional perception annexed to them, that it has had them before". In my view, this is a correct and complete account of one form of memory: experiential recall. First, it tells us that a recollection counts as veridical only if the experience of the object recalled is an experience the subject has had before. Second, it explains the phenomenological difference between recollection and other experiential states by noting that a recollection must present itself to the subject as an experience he previously enjoyed.
Locke's theory of recall was attacked by Reid who poured scorn on the notion that memory experience is a revival of previous perceptual experience. Reid did agree that "Things remembered must be things formerly perceived or known. I remember the transit of Venus over the sun in the year 1769. I must therefore have perceived it". But this claim is weaker than Locke's because Locke requires that the experience of recall be a revival of the previous perceptual experience. I shall argue that this feature of Locke's account is needed in order to distinguish recall from both perception and recognition.
I shall first delineate the phenomenon I mean to describe and separate it from other forms of memory. Then I shall reject one obvious account of experiential recall in terms of tense. On the positive side, I shall elucidate Locke's theory of recollection, defend it against objections and show how it alone can give us a proper account of the difference between recognition and recall. In this way, we capture the distinctive phenomenology of memory experience in the representational content of that experience, in the information which that experience carries about its own relationship to the object recalled. The conditions which (as Reid and many others agree) must obtain if the experience is to count as veridical thus emerge from the phenomenal character of the experience itself.
Quite often, remembering something is simply a matter of regurgitating knowledge of it which we acquired at some point in the past. So to remember that the Battle of Hastings occurred in 1066 is just to believe that it did occur it 1066 having once learnt that this was so. Such a memory involves no experience of the battle nor even of the occasion on which we learnt of it: the date of the battle is simply dredged up from our current stock of beliefs. Philosophers have called this factual or propositional memory. However, memory frequently does involve experience of the event or thing remembered and this experience can tell us new things about the past in much the same way as perceptual experience tells us things we didn't previously know about the present. It is this experiential memory which is the subject of this paper.
Suppose you attended the Mardi Gras yesterday and I now ask you how many floats there were in the parade. You haven't committed this fact to memory and must reflect on what the parade was like - bringing to mind each of the floats which passed you in the course of the evening. Having done this, you can tell me that there were at least a dozen different floats, something you didn't previously know. Here you learn how many floats there were from your experience of the parade in much the same way as you would have done had someone asked you to count them during the parade. Before recalling the parade, you had no idea whether the proposition 'there were more than twelve floats' was true or false, now you are convinced of it.
Take another example. I am looking at a line but I have no idea how long the line is in centimetres because I have not yet been taught any units of measurement. Later I learn to measure by sight and recalling the length of this particular line, I discover that it was between 10 and 12 centimetres long. My perceptual experience of the line gave me no belief about the line's length in units, so remembering the length can't be a matter of regurgitating some previously acquired belief. Rather my memory confronts me with that length once more and I can learn how long the line was from this experience much as I might learn how long it was by looking at it.
There is a further point of contrast between factual and experiential memory we should note. If I recall the Mardi Gras, this event is experienced as being in the past and to experientially remember the length of the line is to be aware of how long the line once was. This is no mere contingency, for it is hard to imagine how something which presents itself as an experiential memory could be anything other than an experience as of the past. By contrast, I can non-experientially remember all sorts of facts about the past, present and future or facts without any tense at all like the boiling point of water.
Some philosophers have argued that the experiential aspect of memory is dispensable: all memory is factual memory and any accompanying memory image is at most an aide memoire which might help us to recover the knowledge we formerly had but which can never be a source of knowledge. I hope it is now clear that this is no more true of memory experience than it is of perceptual experience. Memory experience gives us reasons to acquire new beliefs just like perception. Nevertheless, there is a clear phenomenological difference between perceiving and recalling. How are we to account for this difference?
Tense and Memory
We enjoy experiences as of the past and experiences as of the present: events are represented as occurring either now or at some point before now and the apparent tense of an event is as integral to our experience of it as any other feature it appears to have. Earlier I noted that experiential memory concerns the past: to recall something is to experience it as having been some way or other. Perception, on the other hand, typically concerns the present. Could this
be the mark that distinguishes memory from perceptual experience?
Mellor denies that we see events as tensed. He claims that the content of sensory experience is tenseless and that tense comes in only when we make a judgement about the temporal relation between our experience and the event experienced. But Mellor's view is implausible. There is a big difference between judging an event to be a past event and seeing it as a past event. For instance, the natural thing to say about the events we see through a powerful telescope is that we believe them to be events in the distant past but they still look to us as if they are occurring in the present. Our judgement that they are past events does not remove the appearance of happening as we watch, it just means that we are no longer fooled by what our eyes tell us.
Given that the content of our experience is tensed, why not (following Reid once more) elucidate the distinction between perceptual and memory experience by reference to the tense content of the experience? "The object of memory, or thing remembered, must be something that is past; as the object of perception and of consciousness must be something which is present. What now is, cannot be the object of memory; neither can that which is past and gone be an object of perception or of consciousness". Reid has given us two criteria for differentiating memory from perception - one in terms of previous awareness, a second in terms of tense. Might these criteria diverge? I think so, for I see no reason to deny that we could have veridical experience as of the past without any previous awareness of it.
It is known that the deliverances of one sense can affect those of another. For instance, ventriloquism relies on the fact that we hear the sound as coming from the moving mouth when we see the mouth move. We are much better at locating the sound with our eyes closed, so the content of our auditory experience here must be influenced by that of our visual experience. Visual input might also affect the tense content of an auditory experience. Suppose an experienced artillery spotter is observing some distant guns through his field glasses. He sees a gun fire and, a second or two later, he hears the report. Might not the tense content of his visual experience influence that of his auditory experience, so that he begins to hear the reports as simultaneous with the flashes he sees through his field glasses? That would be a case of hearing the reports as past events, as events which happen before he hears them. But it would not be a case of recalling the reports: our artillery spotter would have absolutely no inclination to treat this auditory experience as a recollection.
Another example may help. Normal subjects have been equipped with inverting spectacles which ensure that everything in the visual field looks upside down to them. If they are then allowed to interact with their (non-visual) environment, within a few days they adapt so that everything seems the right way up again. Here visual experience adapts to the deliverances of the other senses. We can imagine that the inverting spectacles are replaced by light-delaying goggles which ensure that there is a gap of a second or two between the time light arrives at the lens and the time it enters the eye. Again, the subject is allowed to interact with his non-visual environment. Might he not adapt in due course so that he comes to see events as occurring not now but rather in the very recent past? Such an adaptation would not turn his visual experiences into recollections, it would simply give them a past tense content.
The obvious reply to both of these examples is that we should treat them like the case of the scientist using his telescope who sees the explosion of a distant star as happening in the present but knows that, in fact, it is happening in the past. But it is certainly not like this when we fall victim to the ventriloquist illusion: we actually hear the sound as coming from the wrong direction. Nor do those who have adapted to their inverting spectacles merely judge that the world is the right way up, they actually see it the right way up. And this is because the spatial information delivered by one sense is modified in the light of the spatial information received through another. In this respect, both the artillery spotter and the begoggled subject are in the same boat, for they too are receiving information through more than one sense, the only difference being that in their case it is temporal rather than spatial information. The astronomer, by contrast, has only his theory to contradict the evidence of his eyes, so his visual experience is unlikely to be transformed.
I can spot no incoherence, or even implausibility in my description of these examples. Whether such things occur is a empirical question and surely, if they did occur, the past tense perceptual experiences of the artillery spotter or the subject who has adapted to his goggles would be veridical. Therefore Reid is mistaken: we cannot equate perceptual experience with experience as of the present, nor therefore can we equate memory experience with experience as of the past. So Reid must fall back on his requirement of previous awareness to differentiate perceptual from memory experience. It is because the artillery spotter and begoggled subject have had no previous acquaintance with what they now experience that they are not recalling it.
In the next section, I shall propose that we build Locke's version of the previous awareness condition directly into the representational content of memory experience. Once we do this, we have a simple and, I believe, correct account of the phenomenological difference between perception and recall: in recall the experience itself is familiar, while in perception it is not.
The Previous Awareness Condition
Locke's claim was that memory is a power of the mind "to revive Perceptions, which it has once had, with this additional perception annexed to them, that it has had them before". But what exactly does this assertion amount to?
First, I take Locke to be making a claim about the phenomenology of memory experience: something experienced as a memory presents itself as an experience one has previously enjoyed, it is not merely thought or judged to be an experience previously enjoyed. For example, I cannot get a certain extraordinary tune out of my head and when I hum it to a friend of mine, he assures me that it is the tune I heard on the bagpipes at a Burns Night Dinner we both attended last year. I can recall the occasion but the tune seems totally unfamiliar. Nevertheless, I trust my friend and next time it appears in my auditory imagination, I think of it as the tune I heard at that Burns Night dinner. Here what happens is that I hear the tune while judging this experience to be one I enjoyed during that encounter with the bagpipes. This experience does not present itself as a memory, as a source of information about the past: I do not hear this tune as a tune I have previously heard, I merely believe it to be so. And if my belief turned out to be false, this experience could not be called illusory. Therefore, this is not a memory experience.
On Locke's view, in order to recall a given event I need not remember anything about the context in which I perceived it. I might be totally ignorant of when and where the event occurred - all I need be aware of is that the experience I now have is a revival of some previous perceptual experience of it. This requirement should be read so as to allow for the possibility that my memory experience is, in fact, a revival of many different perceptual experiences. If I determine how many windows there are in my front room by recalling the interior of that room, there may be no one perceptual experience which my memory experience is a revival of. I have seen my front room a million times and very likely no one of these viewings was the decisive determinant of my current memory experience. Nevertheless, there will be some set of perceptual experiences which have been amalgamated or refined in such a way as to generate this memory experience and the memory experience is a revival of these perceptions.
In the last section I claimed that all memory experience is experience as of the past, while perceptual experience may be as of either the past or the present. Does Locke's proposal explain this fact? It certainly allows for perception as of the past where, like the artillery spotter, one enjoys an experience which has a past tensed content but does not present itself as a revival of some previous experience. Furthermore, it implies that memory experience must have a past tensed content (provided there is no perception of the future). If part of what any memory experience tells us is that the experience in question is one that has been had before, then that experience, if it is to be veridical, must be an experience of a past scene. Indeed, it is plausible to conjecture that the mechanism which gives the experience a past tense content does this precisely because it is sensitive to the fact that the experience has been had before.
Reid raises several objections to Locke's revivalist model of memory and it will help us to consider one of them. The objection is in two parts. First, Reid remarks that "When a thing is once annihilated, the same thing cannot be again produced, though another thing similar to it may... From this it follows, that an ability to create new ideas or perceptions, after they have ceased to be, can signify no more but an ability to create new ideas or perceptions similar to those we have had before." But, once this is conceded, a new problem emerges. "I see before me the picture of a friend. I shut my eyes, or turn them another way; and the picture disappears, or is, at it were, laid out of sight. I have a power to turn my eyes again toward the picture, and immediately the perception is revived. But is this memory? no, surely; yet is answers the definition as well as a memory itself can do."
One might try to get round Reid's example by appeal to tense: memory experience is experience as of the past while the experience of my friend's picture is experience as of the present. But if, as I argued, there can be perception as of the past, we should not rely on tense content to differentiate perceiving from remembering the picture. The truth is that Reid has put his finger on an important point here. Locke's previous awareness condition must be construed as the demand that recollection involve an experience previously enjoyed and not merely an experience similar to one previously enjoyed. Locke can respond to Reid's counterexample by saying that when I see the picture of my friend for a second time, I do not count as recalling the earlier picture because my current experience of it is not (and nor does it present itself as) a revival of some previous experience. Rather, I have a new experience, albeit of a type that is familiar, and I recognise my friend's picture. So if we are to have the resources to distinguish a case of recall from a case of recognition of something currently perceived we must make sense of the idea that recall is a faculty for reviving perceptual experiences. But is this so hard?
Say that I video a wedding scene and I subsequently watch the video on more than one occasion. Here there is a sense in which I am watching the same film more than once. It is not just that the film shown on these two occasions represents the same set objects: it is that showings present the same representation of those objects. The showings of the film are, of course, distinct events but the same film is being rerun or re shown on different occasions. Similarly we can distinguish the objects an experience represents, the experience which represents them and the event of having or undergoing that experience. As there can be more than one showing of a film, so there can be more than one occasion on which one enjoys a given experience.
The two showings of our video are produced by a single set of chemical records on the video tape and this is enough to ensure that they are showings of the same film. It would be a quite different matter to show two exactly similar videos, records of similar weddings perhaps. An analogous mechanism, perhaps involving memory traces in the brain, might connect a memory experience with its perceptual original. On this hypothesis, the brain is sensitive to the difference between an experience whose character is determined solely by a neural record and one whose character is determined by new input from the sensory organs and this is what decides whether the experience is presented to the subject as a memory or as a perceptual experience. But what gets into the conscious character of the experience of recall is not such details about the underlying causal structure but simply a sense that the experience has been had before.
A film shown more than once may not look quite the same on the two occasions - it may have deteriorated for instance - but it is still a showing of the same film. In the case of memory, the experience will differ in tense content from the perception it revives but it will still count as a rerun of that experience. And if the memory is a revival of several different experiences, it is no different from a film that is produced by means of several different exposures to the same scene (integrated so that a coherent image results). The result is not exactly similar to any previous experience but it is a rerun of the experiences which contributed to it
Does my account of memory introduce an unwarranted indirectness into the experience of recall? When we discover facts about past events by watching a film, our experience of those past events is at best indirect: we see them by seeing certain non-representational features of the screen before us (the movements of coloured patches on its surface and so forth). Is it like this in recall? Do we recall something by being aware of the non-representational features of an intermediary, the experience of recall? I think not. When we recall, we are aware of features of the event recalled and of the fact that we have had this experience before: nothing else. Recall does this much without making us aware of any further non-representational (e.g. physical) features of the memory experience. In this, having a memory experience differs from seeing a video but the disanalogy is irrelevant. What the film analogy was meant to do was simply to highlight the possibility that the same experience could be enjoyed on more than one occasion and this possibility does not depend on our being aware of the non-representational character of the experience.
Locke observes that "many of those Ideas are produced in us with pain, which afterwards we remember without the least offence. Thus the pain of Heat or Cold, when the Idea of it is revived in our Minds, gives us no disturbance; which when felt was very troublesome, and is again, when actually repeated". This observation does not worry Locke: the experience of heat or cold, he thinks, can be revived without the pleasure or pain which accompanied them. But, someone may ask, what about the pain or pleasure itself? We can recall the pain we felt after being burnt by a fire, exactly how intense it was and where it was felt, without experiencing any discomfort at all. But if this recollection involve a revived experience of pain, the memory experience would provoke severe aversion, since intense pain is necessarily experienced as intensely uncomfortable. Therefore, recollection of pain cannot involve a revival of the experience of pain.
My inclination is simply to face down the objection by denying that the experience of intense pain must be uncomfortable. People who are lobotomised or under morphine analgesia report pains, rank them in terms of intensity but claim not to mind the pain. This is a striking phenomenon but these people are not talking nonsense. Their testimony shows that while aversion is a normal accompaniment of the perception of pain it is inessential to it. Therefore, there is no reason why a pain which was uncomfortable the first time round must still be unpleasant when revived. Equally, there is no reason why it must be a matter of indifference: vividly recalled pain can make you sick, just as perceived pain does.
Recognition and Recall
Recognition, like recall, is a form of memory experience. Yet it is also a form of perception. If I am to recognise something, I must perceive it. But how can it be both, given the clear dichotomy between perception and recall? In this section, I shall answer this query by highlighting the similarities and differences between these two forms of memory experience.
We must first distinguish the memory phenomenon called recognition from other experiences which go under that name. In particular, we must firmly distinguish recognition as mere perceptual classification from memory recognition. Say that I have never seen my new boss before and I must meet her at the airport. I ask her to describe herself over the phone, which she does with unusual skill. As a result, I am able to recognise her when she emerges from the arrivals area. Here we have a perceptual experience - my seeing this woman - and we have my recollection of the content of the description she gave of herself which enables me to recognise her as my boss. But though I recognise her as my boss, I have no memory of her.
Perhaps what is wrong here is that I do not see this woman as my boss at all, rather I see her as a woman with certain features and then work out that she is my boss; in which case it is the absence of a present experience of her as my boss rather than lack of a past experience which matters here. But the case need not be like that: hearing the description may induce an ability to recognise my boss without my consciously applying it. Say I completely forget the description and go to the airport in the hope of locating my boss by some other means but the moment I see her, I realise that she is in fact my boss. Here it may be clear that the description has done its work - I may even be able to recall the description once I see someone who satisfies it - but there is no question of a conscious inference and no case for saying that I fail to see this woman as my boss. Nevertheless, I still do not remember her: there is no memory recognition.
In order to distinguish memory recognition from perceptual recognition, we must put a reference to previous experience into the content of the experience of memory recognition. If I have never seen my boss, I may still see someone as my boss but I cannot see them as someone whom I remember. But how exactly does a reference to the past enter into the content of a recognition experience when the object recognised is patently an object currently perceived? To put it another way, what must the present and the past be like for a memory recognition experience to be veridical?
One plausible proposal is this: to recognise a face is to see it as a face I have seen before. To recognise a face as my bosses' face is to see it as a face I have previously seen as belonging to my boss. Here one need have no particular previous sighting in mind; nevertheless there must have been some such occasion for the experience to be veridical. On this account of memory recognition, there is indeed more to remembering my boss's face than merely knowing what she looks like in such a way as to be able to identify her at a glance, for I might know what she looks like without ever having met her. Equally, recognising her as my boss involves more than merely seeing her face as familiar and then inferring that it looks familiar because it belongs to someone I have seen before, namely my boss: rather I must see this face as a face I have previously seen as my bosses' face.
Nonetheless, recognising my boss does not mean reviving some previous experience of my boss, it does not involve recalling how she once looked. A woman may be recognised as someone that was previously experienced as my boss without the recognition of her involving a revival of that previous experience. This is an instance of quite a general point: something may be mentioned in a correct specification of the content of a given experience without that thing's being an object of that experience. I can have an experience as of X which involves mention of Y without having an experience as of Y. A man may remind me of a wolf, may look similar to a wolf, without prompting me to recall any wolf or wolf-picture, without engendering an experience as of a wolf. This remains true where Y is a previous experience of X.
So, to recognise a face (where recognition is a form a memory) is to perceive it but to perceive it as something one has previously perceived. To recall a face is to have an experience as of a face which presents itself as an experience one has had before. This distinction is meant to be a dichotomy: there could not be a state which was, or even appeared to be, an instance of both recognition and recall. Any state which is apprehended as a revival of a previous perceptual experience is ipso facto felt as non-perceptual. There is no obstacle to feeling your perceptual experiences to be similar to previous experiences but it is impossible to apprehend an experience both as perceptual and as a revival of a previous perceptual experience.
There could certainly be an experience which presented itself as revival of a perceptual experience but was not. This might be a case of an apparent recollection which was in fact the product of the sensory imagination. Or else it might be a perception which presented itself as a case of recollection. Certainly the latter does not often happen and we are rarely in doubt as to whether we are perceiving or recalling but there is no conceptual obstacle to confusing the two, so far as I can see. I could stand before my boss in a such a way that I count as perceiving her and yet sincerely report a vivid memory of how she looked in the recent past and my experience might well be veridical, except in respect of its memory content.
Previous Experience v Prior Registration
I have asserted that an object either recognised or recalled must have been previously experienced. Ayer poses the following question: why require a previous experience of the thing remembered? Say that the content of a certain memory experience has been determined by some past event in such a way as to make that event the object of the experience but without that event's having caused a prior experience of itself. Of course, the past event must have registered with the subject in some way if he is now to remember it. Furthermore, this previous registration must be alluded to in the content of the current memory experience. But, it might be said, this registration need not have involved any perceptual awareness of the event; it might, for instance, have been subliminal, an event which we have become aware of only in retrospect.
Consider the following story: While I am visiting the cinema, a subliminal advertising message is flashed onto the screen in an effort to get me to buy a certain brand of popcorn during the interval. I experience no interruption in the film and watch undisturbed until the interval. During the interval, I go up to the vendor and I recognise the popcorn he is selling. Because the popcorn seems familiar, I buy it and then, whilst eating it, vaguely recall the popcorn's appearance in the film. Is my experience of recognition veridical? Is my experience of recall veridical?
A defender of my account can answer 'yes' to both questions only if he insists that I must have been aware of the popcorn, however fleetingly, when it appeared in the film. But are there any grounds for such an insistence other than an adherence to the previous awareness condition? If, on the other hand, he admits that I was unaware of the popcorn, he must deny that either memory experience is veridical. But why deny that; after all, both experiences were produced by a past event in such a way as to ensure that the memory experience reflects the event's true character?
I think we ought to stick by the requirement of previous awareness because we can provide a plausible account of the phenomenological difference between recognising the popcorn and recalling it only if we insist that recollection requires previous awareness of the event recalled and not just prior registration. On my view to recognise the popcorn is to see it as something you have seen before while to recall the popcorn is to have an experience of it which you are aware of having enjoyed before. The objector must say that to recognise the popcorn is to see it as something you have previously registered while to recall the popcorn is to have an experience which is somehow a revival of a previous registration of it. But what does it mean for an experience to be a revival of a previous registration, rather than merely an experience of something previously registered? No doubt a different causal/physiological story can be told for the two different cases but it is very hard to believe that such stories enter into the content of experience. It is much more plausible to suppose that recall involves finding a certain experience to be familiar (to be an experience one has had before) while recognition involves finding an object familiar (to be something one has experienced before). And if that is correct then one's recollection of an object can be veridical only if one has previously enjoyed an experience of the object recalled.
Although Ayer's question can be answered by a Lockean like myself, it creates severe difficulties for other views. Most philosophers of memory follow Reid in endorsing the requirement of previous awareness while declining Locke's invitation to say that in recollection we are actually aware of the previous experience. But if prior registration is enough to ensure that the past is correctly represented in present experience, why should previous experience also be required, unless, that is, this experience itself is mentioned in the content of the recollection?
Some Psychological Evidence
I have argued that in order to account for the phenomenological difference between perceptual experience on the one hand and memory experience on the other, we must incorporate the previous awareness condition into the content of memory experience. But how psychologically realistic is this? Several philosophers have expressed the worry that this would make the content of the memory experience implausibly complicated and would involve using rather sophisticated notions like 'perceptual experience' to formulate the content of all memory experience, however simple. Is there any psychological evidence that memory experience does involve such an introspective awareness of one's own mental states?
Tulving has recently claimed that the best way to explain various psychological phenomena is precisely to incorporate reference to previous experience into the content of memory experience. Some time ago, he drew a distinction between semantic and episodic memory, one which corresponds roughly to the philosopher's distinction between factual and experiential memory (including both recognition and recall), using behavioural and neurological data to argue that there are distinct memory systems here. More recently, he has also hypothesised that there is a phenomenological distinction between semantic and episodic memory: "When a person remembers a [personally experienced] event, he is aware of the event as a veridical part of his own past existence. It is autonoetic consciousness that confers a special phenomenal flavour on the remembering of past events, a flavour that distinguishes remembering from other kinds of awareness, such as those characterising perceiving, thinking, imagining or dreaming".
What evidence is there to support this claim? The obvious thing to do is to look at those who lack the use of experiential memory and see if this deficit can be accounted for by the absence of the sort of self-consciousness which Tulving thinks essential to episodic memory. If Tulving is right, those who lack the concept of experience will be unable to use experiential memory to make judgements about the past. Is this prediction born out? Tulving notes that there is a consensus among psychologists that young children make no use of episodic memory, though they have a great capacity to retain semantic knowledge and 'know how'. Could this be linked to their lack of the concept of a perceptual experience?
Perner investigated this question. He ran two batteries of tests on children aged between 3 and 6. One battery of tests were memory tests. The children were shown certain pictures and later asked to describe what was in the pictures. In an effort to test their experiential memory, rather than their memory of facts about the picture, they were given no clue as to what was depicted. The other battery was designed to test the extent to which the children were aware of the sources of their beliefs. The children were asked whether they knew something because they had been told it was the case or because they had seen it was the case. They were also asked to differentiate cases where they got something right because they had known the answer from cases in which they had merely guessed correctly. Finally, they were tested for their sensitivity to differences in the information supplied by the different sensory modalities. It might be thought that both batteries of tests were memory tests on the grounds that children could use recall to answer questions about where they had gathered certain information. Perner excluded this possibility by also asking the children which sensory modality they would use to find out certain facts.
Perner found that those children who displayed an awareness of the sources of their knowledge also did well on the memory tests while the others did poorly. Furthermore, the correlation was specific to the tests of free recall. Children who showed little awareness of the sources of their knowledge were still able to state facts about what was in the picture provided they had been encouraged to memorise these at the time (by means of questions like: what fruit is there?) and were prompted with suitable questions later. Thus, he concluded that while factual (or semantic) memory is available to those without any awareness of previous perception, experiential memory is a source of knowledge only for those who possess the concept of experience.
One of the debates currently raging in the philosophy of mind is over whether we can characterise the conscious character of any mental state purely by saying how it represents the word as being, purely by specifying what information it supplies us with about the world around us and our relationship to it. Those who think we can do this must capture the phenomenological differences between mental states by pointing to differences in the informational content of those states. In this paper, I have argued that the distinction between perception, recognition and recall at least can be drawn in these terms.
J. Locke - Essay Concerning Human Understanding Ed. P. Nidditch (Oxford 1975) p. 150. Similar claims are made by W. James - The Principles of Psychology, Volume One ( Dover 1950) pp. 648-50, H. Ebbinghaus - On Memory (Dover ) p. 1 and J. Searle - Intentionality (Cambridge 1983) pp. 96-7.
T. Reid - Essays on the Intellectual Powers of Man Ed. B. Brody (MIT. 1969) p. 326. Reid's claim has been endorsed, in our own century, by many philosophers: C. Broad - Mind and its Place in Nature (Routledge 1937) pp. 239-41, N. Malcolm - 'Three Forms of Memory' in his Knowledge and Certainty (Prentice Hall, 1963) p. 208, Martin and Deutscher 'Remembering' in Philosophical Review 75 (1966) p. 163 and S. Shoemaker - Identity, Cause and Mind (Cambridge 1984), p. 19. In this paper I shall not discuss the issue, raised by Shoemaker, as to whether the previous experience might have been enjoyed by someone other than the subject of the memory experience.
 Philosophers have distinguished 'factual' memory from 'experiential' or 'perceptual' memory while psychologists have distinguished 'semantic' from 'episodic' memory. Different people draw these distinctions differently. The philosopher's distinction goes back at least to H. Bergson. See his Matter and Memory (Zone, 1991), Chapter 2. For subsequent discussion, see B. Russell - The Analysis of Mind (George, Allen and Unwin, 1921) pp. 166-8, Broad - Mind and its Place in Nature op cit. pp. 221-2, , A.J. Ayer - The Problem of Knowledge (Penguin 1956) pp. 134-48, Malcolm - 'Three Forms of Memory' op cit. and Martin and Deutscher 'Remembering' op cit. pp. 161-66. For the psychological distinction, see E. Tulving - Some Elements of Episodic Memory (Oxford 1983).
A. Ayer - The Problem of Knowledge op. cit. pp. 138-142.
H. Mellor - Real Time (Cambridge 1981) p. 26.
T. Reid - Essays op cit p. 325.
Robert Hopkins convinced me of this point and provided me with the goggles example below. David Lewis made me think about what an artillery spotter sees.
We must, of course, carefully distinguish the time at which an experience of a given event occurs from the time at which the experience represents this event as occurring. If the two were always identical, memory experience as of the past would be impossible. Dennett argues that we must also make this distinction in the case of perceptual experience, see his Consciousness Explained (Little Brown 1991) Chapter 6.
Someone might insist that this experience was a memory on the grounds that it was related in the right way to a previous experience and regardless of the fact that it did not present itself as such, thus pushing memory-status out of the phenomenal realm. I am concerned to provide an account of the phenomenology of memory experience, of the difference between what it is like to recall and what it is like to perceive or recognise something. Someone who wants to tie the word 'memory' to facts about causal origin can accept my account in this spirit.
Pace W. James - Principle of Psychology Volume 1 op cit. p. 650.
Reid Essays op cit. p. 369.
Reid Essays op cit. p. 370.
J. Locke - Essay op. cit. p. 633.
B. Russell makes this distinction on pp. 169-70 of The Analysis of Mind op cit.
B. Russell - The Analysis of Mind op cit. p. 170. Compare Searle - Intentionality op cit. pp. 67-8. Robert Hopkins pointed out that one needs to add "and I see it as a face I have seen before because I have seen it before" in order to ensure that the recognition experience is veridical.
C. Peacocke - 'Depiction' Philosophical Review 96 (1987) pp. 384-5.
Ayer raises this objection on p. 145 of The Problem of Knowledge op cit. So far as I know, none of the many defenders of the previous awareness condition have attempted to answer it.
For instance, Russell expresses this worry on p. 179 of The Analysis of Mind op cit.
E. Tulving - 'Memory and Consciousness' Canadian Psychology 26 (1985) p. 3.
J. Perner - "Episodic Memory and Autonoetic Consciousness" forthcoming in Journal of Experimental Child Psychology.
Many thanks to Peter Carruthers, Robert Hopkins, Frank Jackson, Michael Martin and Tom Pink for their comments on earlier versions of this paper.