H. P. Grice and B. S. Benjamin remembered that they met punting by Magdalen Bridge

Nothing is so uniquely personal to a man as his memories. Our inner lives revolve around their contemplation, and in guarding their privacy we seem almost to be protecting the very basiB of our personalities. Most of our remembering is done in private and though we speak of sharing our memories with others there seems to be a sense in which we could no more share a memory than we could share a para , the best that we can do is to try to describe it Yet unlike pains and aches, feelings of anger or of amusement, there are no natural and public signs of memories. We do not have to learn how to keep our recollections private as we have to learn to suppress feelings of amusement, boredom and discomfort. To the prefatory expression “ I remember ” there seems to attach the aura of a voluntary disclosure about oneself. It would seem, then, natural and indeed essential to construe the concept of remembering upon the model of an avowal about one’s state of mind, about one’s inner and in- accessible experiences It is indeed traditional to approach the concept of remembering as though it has this kind of logic ; in this paper I shall argue that such an approach is radically at fault. Although it is not my purpose either to examine particular theories of memory as they bear upon the problem, or to do justice to the literature in the field by subjecting it to detailed criticism, I shall attempt to mention what seem to me to he typical and mistaken moves made in analysing the concept And I E J la ^U 3e Sbi by considering Hume’s analysis as it is presented in the xieatise, because not only does it present in its purest form a thesis which I wish to attack, but it has exerted a powerful influence upon subsequent analyses of remembering. According to Hume, to remember something is to have a special kmd of mental experience in the form of a mental image different from any other kmd of image or idea. In part, of course, he was led to talk of * an idea of the memory ’ hy his .? cu , ^ psychology, hut it is important to notice that although facult 7 psychology are over we still feel compelled to me “ or y 111 a not very dissimilar way. Although anc * neurologists still know practically nothing ..bout the brain mechanism » which enables us to recall past experiences and previously acquired skills, it seems natural tothink of the memory as a unitary function of some sort, and from this it is easy to conclude that our memories, bemg the products of a single process, must m some way be stamped with the sign of then manufacture. This is one feature of the problem which suggests that the task set to any philosophical theory of memory is to detect those characteristics of the mental experience of remembering that will serve to isolate and define it. There is another puzzle which leads one in the same direction How do we know that certain of our images and thoughts are images of, and thoughts about, the past ? How do we know, if we are trying to remember say the look of a town, that none of the images we can summon are right, that none of them are memory images, and then, suddenly, that this one is a memory image ? Hume’s general concept of the mi nd as an entity which can perceive only its own thoughts, inevitably suggested an answer to these questions, which also did justice to the solution suggested by a faculty psychology For if one holds that a second order perception is involved in all thinking, it is natural to apply to this second order perceiving an analysis that is obvious and seems perfectly adequate to account for certain features of first order perception The commonplace that one defines words referring to physical objects by attempting to isolate those features of a physical object that are necessary and unique to it was applied by Hume to the “ objects ” of second order perception The impressions of sense, the ideas of memory and the ideas of imagination differ from each other, he said, with respect to their strength and vivacity, hence perception (first order), remembering and imagining may be defined in terms of relative strength and vivacity. Ryle mentions the absurdities involved in the classical theory of second order perception, but it is worth pointing out here that the difficulties involved m talking of perceiving an impression of sense (or a sense datum) are not by any means so obviously involved m ta lking about perceiving one’s mental image. Most students, when they first read Hume, are struck by the strangeness of his suggesting that the impressions of sense differ from the images of memory only m degree, in a way that they do not think it strange to suggest that the images of memory and those of imagination might so differ This is not surprising, for although no one has quite been able to describe how one should set about obeying a request to attend to one’s impression of sense or sense datum of an object, as distmct from looking at the object itself, everyone knows how to attend to a mental image of an object In ordinary discourse we use the verbs of perception in this second order sense quite naturally, as when we speak of hearing the tune we heard last night or of seeing the accident as clearly as if it were occurring in front of one’s eyes, and only a philosopher would feel obliged to supply inverted commas for the verbs in this use Furthermore, not only can one describe the contents of an image but one can also discriminate between and report upon properties of the image as a whole ; for example, upon its comparative vividness, intensity, blurriness, definition, and so on. I conclude, then, that when Hume suggested that the images of memory and those of imagination difier intrinsically in respect of their relative strength and vivacity, he was not making a suggestion that is difficult to understand and implausible in the way that Ins similar remarks about the impressions of sense are It is generally agreed, however, that the characteristics of strength and vivacity of an image fail to mark off unambiguously our rememberings from our imaginings. For quite often the images of our imaginings and fantasies are very much more vivacious and vivid than are many of our memory images. Usually this standard criticism of Hume’s theory is made with the suggestion, explicit or implied, that if only we were sufficiently attentive and ingenious we could discern what combination of characteristics invariably attend our rememberings and are absent when we are imagining A number have been suggested, eg. that our rememberings are accompanied by a feeling of famibarity, that this feeling makes us apply the concept of pastness to the image, and so on, though I think it is rarely suggested that such characteristics provide the criteria of remembering as Hume maintained his did. The trouble witb tbis criticism is that it fails to expose tbe real nature of Hume’s failures and m fact it simply encourages speculations which must prove equally inadequate Hume’s theory of remembering is the purest example of what I might call the mental datum theory : that to remember is to have a certain sort of mental datum or experience, and to tell others what one remembers is to inform them of the details of tins datum (usually thought to be an image). So far in this paper I nave been chiefly concerned to point out considerations which make this a natural theory to put forward. I wish now to show rt i 8 an “ n P oss ft , le thesis to maintain in any f or m.. I I ^ person statement that asserts an inner experience UJce the possession of an image, a certain sort of feeling or sensa- tion^ can be corrected by a third person. It must be accepted as e or rejected as a he ; it cannot be shown to be mistaken. If I say that I Lave a vivid image of a tree before me, my bearers can, in principle, disbelieve me on the grounds that I am lying, though it is very difficult to imagine what such grounds might be. But it would make no sense to accuse me of being mistaken about the matter. The concept of mistake only applies to cases where it is both theoretically possible to obtain independent evidence on the matter and to explain how the mistake arose. There is no way at present known of obtaining evidence independent of a man’s word as to whether or not he has a mental image of the sort he claims to have. It is sometimes suggested, indeed, that avowals of one’s state of mind or body are m principle incorrigible and hence self-certifymg. But I suggest this is an incorrect way of stating the point. "We are under no logical obhgation to accept the truth of an avowal and avowals of certain types of inner experience are regarded as corrigible by indirect evidence. To mention only one example, claims to be in pam aresometimes rejected on the evidence of medical authorities when there is a recognised correlation between the described pam and a physiologically morbid condition But malingerers are lying, not mistaken. The important logical difference between statements about one’s inner and private experiences and statements about the external and public world does not run along the cleavage line of corrigible and incorrigible assertions, but along the gap separating claims that may be true, mistaken or deceitful, and those that can only be true or deceitful. Sometimes, like George IV, people have entirely delusive memory experiences ; more frequently, they claim to remember something when there is independent evidence to show that they must be mistaken m so thinking. It is this fact that theories of memory like Hume’s cannot account for. It is plain that on Hume’s theory one must have either veridical memories or be lying, for no one can be mistaken as to whether or not he has an image of a certain strength and vivacity. This is the point at which the analogy between first and second order perception ceases to hold. We can explain how we made the mistake of taking an overcoat on the floor to be the body of a man : the light was had, we had lost our spectacles, we were too far away to see properly. But one needs no light by which to see a mental image, no oculist can attend to the defects of the inner eye, and try as we may we can neither approach nor retreat from our images. . We use no organs to detect our images, states of mind and sensations ; they are separated from us by no medium ; there is no mechanism to go wrong ; there is no inference made into which error could creep The absurd necessity that Hume's theory would impose upon us of declaring to be a liar anyone who thought himself to be remembering when he was not, clearly follows equally from any revised version of the theory. There are no special images, accompanying feelings of familiarity, or intense convictions that one is truly remembering and so on, from the experiencing of which it follows conclusively that one is indeed remembering.This follows from the fact that any claim to remember, no matter how confidently it may be based upon the possession of dear and distinct images, feelings of fa mili arity and so on, may in principle he falsified by evidence of a non-subjective kind. For instance, I may he absolutely certain that I can remember meeting a friend in the street yesterday. Yet, if it were proved that the friend whom I thought I remembered meeting had been at that time a hundred miles away, I would have to accept the fact that I could not possibly remember meeting bfm and that I must have imagined the occurrence. It follows from this corrigibility of claims to remember, that no mental datum or combination of mental data can possibly function as sufficient criteria of remem- bering as the Humean type of theory suggests they do. Or, to put the point in the way I have been doing, one can only maintain the enterprise of taking mental data to he sufficient criteria at the expense of rendering it impossible to talk of people making honestly mistaken memory claims. It is well worth noticing thatmental data ofthe type I have been considering do not function as necessary criteria of remembering either. Hume’s doctrine that one remembers if one has an image of a certain kind has only recently been abandoned in the face of a mass of psychological evidence to the contrary. Indeed, some would seem to have abandoned the memory-image theory with the poignant reluctance of the theorist confronted with impossibly indigestible facts. The facts are certainly indigestible, hut the weak points in a conceptual analysis (which eveiy philosophical theory of memory should be) may be detected independently of experimental evidence, and in this case one can certainly show without recourse to experimental evidence that the possession of mental imager}' is not a necessary condition (or criterion) of remembering. For example, if a barrister conducting a cross- examination attempted to tlirow doubt on the reliabibty of a witness s memory by demonstrating, per impossible, that the witness s account was unaccompanied by mental imagerv, no one m the court, would understand lus point at all. Ye” have e s o eci e or to help us decide whether or not people remember an . ome o t icse are tests for the presence or absence of necessary remembering 317 conditions, but tbe possession or lack of possession of images is not one of these tests By substituting for ‘ mental image 5 any other mental datum in the example above, one can see similarly that no specific mental datum need be present before one can be said to remember. To avoid misunderstanding, it should be mentioned that of course these arguments do not apply to any non-introspectible concomitant of remembering, c.g. such as a certain pattern of neural discharge It may veil be discovered that phenomena of this kind are necessary to remembering, but non-introspectible or unconscious mental phenomena in the nature of the case do not, and could not, usefully function as cnteria of remembering, which is the point in question Modem versions of the memory image theory avoid the notorious flaw in Hume’s theory by attempting to account for the fact of honest but mistaken memory claims The core of the modified version is the suggestion that a memory image is a representative image of a past perceptual experience, and when one makes an honest mistake of memory one is faithfully reporting or observing an image that fails to represent the past experience. This is unobjectionable so far as it goes, but if the argument above, showing that images have no necessary role m remembering, is correct, it follows that this theory fails to expose any part of the logical structure of the concept at all, and merely describes a phenomenon that may or may not take place It is, furthermore, obvious that anyone who assigns to the memory-image a central role m the analysis of remembering must explain the connexion or lack of connexion between our rememberings when memory-images naturally are likely to occur, as in our memories of places and faces, and those when they are not, as for instance when we remember how to tie a running bowline or the first four hues of Paradise Lost It is worth considering one such attempted explanation, made by Professor Broad in The JSlind and Its Place in Nature. In a chapter entitled “ Memory ” m that work he says that the word “ memory ” is highly ambiguous m tbe sense of covering “ a number of very diff erent acts ” ; thus we use the verb to remember ” m different senses when we speak of remembering a set of nonsense-syllables, a poem, a proposition in Euclid, how to swim, and people and places, because in each case what we remember differs from tbe others Given this technique of discriminating diff erent senses of the verb to remember , Broad is able to declare the sense m which we remember past perceptual experiences (when imaging is most likely normally to occur), to be different from other cases, most of which, he suggests, can be called instances of remembering only by courtesy. Broad’s contention that ‘ remembering ’ is an ambiguous word which has many senses would be of the first importance if true. There are, however, several reasons for denying its truth, at least in the frm in which he presents it In the first place it is plainly incorrect to assert that, for example, in the claims ‘ I remember her face ‘ I remember her name and telephone number ’, ‘ I remember how to rhumba ’, one could possibly mistake what was being claimed, through an ambiguity of ‘ remember ’, in the way that one might mistake ox misunderstand the claim 4 1 have been to the bank ’. Neither is it apparent that more subtle distinctions of sense are involved asm ‘I feel a penny ’, ' I feel sick ’ and ‘ I feel happy ’. The differences of sense involved in the latter cases can be indicated by the blatant inappropriateness of certain questions asked of one or more of the claims, which axe quite appropriate if asked of others, e g. what is the location of what you feel ? what did you feel it with? and so on But Broad does not elicit dis- tinctions of sense in the verb ‘ to remember ’ in the same straight- forward way. Instead, he first states an implied analysis of the concept of remembering from which distinctions can be seen to follow, by declaring that remembering is an act, and then not unreasonably concludes that the very different activities involved in, e g. remembering how to swim (bodily activity), the lines of a poem (rote activity), someone’s face (the mental activity of having an image), and so on, must be reflected in different senses of the word we use to refer to these differing acts. But such distinctions rest upon the highly contentious assertion that remembering is the name of an act or set of diffe rent acts, and is m no way a straight-forward statement about usage as the previous examples were. Indeed, the implausible conclusion which follows from thistheory, and which Broad draws, that there is no connexion whatever between the senses of remember elicited, might be taken as prima facie evidence of its falsity. The only way to. decide whether remembering and its cognates are radically multivocal, or substantially univocal as I wish to maintain, is to re-analyse the concept to see whether or not a uniform core of meaning is preserved m its use in different con- texts. This I shall attempt to do, and m the course of my argument it will, I think, become quite clear that it is not possible o construe remembering as an act, except of course m the entirely empty sense in which one might say that eveiy active verb mast- denote an act. Before I proceed to this analysis something further is needed to do justice to at least one of the points which Broad wished to make For anyone, irrespectiveof whether or not he takes remembering to be simply an activity, might well be greatly struck by the difference between remembering a past perceptual experience, and remembering a poem or how to swim The difference seems to lie in the fact that whereas it in no way seems necessaiy for one’s remembering the lines of a poem or how to swim to be accompanied by introspectible mental experiences, it is difficult to understand how one could remember a perceptual experience without an experience analogous to the original one taking place For example, what would it be like to remember the very tones of a voice without m some sense hearing the voice again, or to recollect in detail the view up the High from Magdalen Bridge without seeing it in one’s mind’s eye ? Then again, when one goes over the events of the day one usually does so not by telling oneself the story of what occurred but by seeing, hearing and feeling again in memory fragments of one’s perceptual experiences, in short, by a sort of reliving of the day’s events. What I shall call, for want of a better phrase, reliving, is undeniably typical of our remembering of perceptual experiences, and the memory-image theory is an attempt to do justice to this fact, as also is Broad’s assertion that remembering has different senses. Whether the phenomenon of reliving is marked by a distinction m sense of ' remember ’ is a question that may be decided the better when the remaining analysis has been made. It is fairly obvious, however, that the notion of reliving is capable far better than that of a memory image of accommodating the recall of non-visual perceptual experiences and, hence, that the memory-image theory has the trivial defect of over-narrowness. What is less obvious is whether the argument against the memory- image theory presented earlier has equal weight against the broader reliving theory. . It is certainly no more a standard test of remembering^ to enquire whether a person is in some way reliving what he damns to remember, than it is to enquire if he has an image of it -But one might now feel tempted to argue that it does not follow from the fact that this is not a standard test, that reliving is never theless not a necessary ingredient of remembering P as evpen cnees We may investigate this possibility by se • => _ U P e hypothetical case of a man who was perfectly well able to describe his past perceptual experiences and yet denied that he underwent the experience of reliving in any way at all. Some people might wish to argue that he must be reliving the experiences he describes — otherwise how could he describe them? and that either he fails to understand what we mean by the phrases ‘ reliving ‘ seeing or hearing vividly in recollection ! and so on, or that he undergoes his reliving in a very curious way (as, for example, some hold that people with freakish ability to calculate in their heads must perform sums terribly fast and unbeknown to themselves ). 1 It would take us too far off the course this paper must follow to consider the really interesting points this argument involves. Two, however, may be noted. First, the ability to image sensory experiences does vary widely from person to person, so the disagreement may well have a factual basis . 2 Second, it is a genuine puzzle sometimes to know what is to count as an image. For instance, what the writer pi esumes is his memory image of the High from Magdalen Bridge is so fleeting, blurred and thin that, if, as it were, it could be captured for the requisite time, it would undoubtedly prove impossible to draw it ; yet he can describe the view to himself and others. Would this count as an image, a reliving ? One tends to put an end to such a question, I think, by wearily agreeing that it must be an image. The really important point at issue, though, may he decided, whether or not this preliminary question can be. If reliving is a necessary condition of remembering a past experience, then, were we to find someone who could describe to himself and to others his experiences without any reliving of them, it would have to follow that he was not really remembering. I do not think that this point could be sustained. Our subject would give a perfectly adequate memoiy-performance ; he would pass all the standard tests of remembering ; we could not even say that he is not good at remembering, in the usual sense of the phrase. All that we could say of him is that he does his remembering in a curious fashion, and, at the most, that remembering is for us a much richer experience than it is for him, that, perhaps, it means more to us than it does to him. But the phrase * means more 5 in this context has somewhat the same force that it has in the observation that doing addition means more to me than it does to a bank teller ; we both add, but whereas his answers come 3 A phenomenon interesting in tins connexion is reported in a recent note on “ Loss of Visualisation ’’ by Sir Bussell Brain, Proc of the Royal Society of Meihrmc, Apn! 19J4, Vol. 47, No 4 3 An informative mid brief account of such differences is given in “The Measurement of Mental Images ” by P. L Short, Penguin Science Rcics, No. 24. almost automatically, I leach mine by means of agonised and laborious little sums If only because we are taught to add, there is little temptation for us to confuse our personal methods of doing it with the notion of adding itself. It is perhaps largely because remembering is a natural phenomenon, something which we do not have to learn, that we feel ourselves to be authorities on the subject and that our personal methods and techniques will have some necessary connexion with the logical structure of the concept. But a concept used in public discourse could not be so dependent upon the vagaries of private experiences, the nature of which, as we have seen, it is not even easy to describe 3?or it to have a stand- ard meaning its use must be standardised, and reference to the experiences of reliving undergone by individuals, plainly, could hardly be suitable for this In fact, because of our preoccupation with our experiences of remembering, we tend to simply ignore the standard uses to which we put the concept in our discourse It is to these that we must now turn I referred earlier to the fact that a claim to remember is m principle falsifiable or verifiable by observations which are in no way connected with the state of mmd of the person making the claim The following example lEustrates still further points of difference between the logic of statements about one’s sate of mmd and statements containing the verb ‘ to remember ’ or its cognates But this is incidental to the present purpose of the example, which is to provide further material for the study of how the concept actually behaves when used m discourse. Suppose two men, each of whom is thoroughly acquamted with the painting of a certain artist. This artist has painted a view fa mili ar to each, and one man has seen the painting and the other has neither seen it nor had it described Now suppose we ask each man to try and picture the painting to himself and then to tell us what the painting is like The man who has seen the painting before will probably claim to remember it, and we can test his claim by getting him to describe it to us Should his statement be such that it would count as a description of the painting, it must be allowed that he remembers it Should his statement not be a description of the painting, 1 or shouldhe be unable even to begin a description, Tre would have to declare tliat lie did not remember the painting ; in the first case that either he was guessing or confusing the painting with another one ; in the second, unless he suffered from ° r "° fc C ° lmt aB a descn P tl0n 111 thls of context a speech defect or the painting was such that a verbal description presented great difficulty, simply thathe had failed to remember it.We could hardly expect the other man to be able to describe the painting, but it is just possible that, knowing the scene and being familiar with the artist’s ' vision palette and so on, he could tell us sufficiently well what the painting was hbe. But even were his answer to be in substance exactly the same as that of the man who remembered the painting, and even if both were to have similar mental pictures, the second man would not be remembering the painting but guessing what it was like, and even if he were honestly to believe himself to be remembering it, it could not follow that he was. Alternatively, if the first man had quite forgotten that he had seen the picture and had then told us what it. was like, thinking he was guessing, we should be forced to declare that in fact he was remembering the picture although he was unaware that he was domg so. The obvious point which this example illustrates is the in- dependence between on the one hand, the state of mind, mental imagery and so on, of the person who claims to remember or not remember the painting and, on the other, the factors which, on appeal, decide whether the claim is to be accepted or rejected and reformulated as a guess or an imagining. As was mentioned earlier, that memory claims are verifiable m principle by recourse to pubbcly ascertainable facts 1 indicates a marked difference between the logic of these assertions and those reporting a mental experience An even more staking distinction hes in the fact that if a claim to remember is rejected in toto, it is im- mediately reclassified, as for instance * you couldn’t remember such an occurrence because it never took place ; you must have imagined it or guessed it. or made it up ’. Each of us has been subject to such corrections and unquestionably we were not being convicted of linguistic incompetence : neither, as we have seen,could it be for mistaking the nature of our (conscious) mental experience. What, sort of mistake, then, were we committing ? An answer suggests itself most strongly when one notices the way m which rejected memory-claims are corrected : you didn’t remember, you imagined, dreamt, guessed it, and so on. We use these forms of emendation when certain of the conditions hich must he fulfilled before remembering can take place, have This is not aht ays possible, of course, eg. it hen ne remember our dreams feelings and so on But treat our memories of such tilings nx though they arc verifiable independently, m as much as everyone it ould admit that it is possible that his memory of, e g lus dream, is mistaken or faulty. not been fulfilled, for example, when someone claims to remember seeing an occurrence which m fact never took place, or, if it did, winch he could not have been m a position to witness. In such a case, the statement presented as a memory-claim or the unexpressed thought which is taken to be a memory, cannot have come to mind as a result of the memory process, t e as the retention of past experience, for the experience never took place To correct the claim with ‘ you must have imagined, dreamt, guessed it’, and so on, answers the question which the demal that remembering took place poses, namely, How then did he come to t hink of it? by ascribing to him a mental process different to the one originally claimed. I think that it would be true to say that the everyday view of remembering is simply that it is the final stage of a causal process and that the memory is some sort of causal device or mechanism. The fact that in our language remembering is opposed by other process words like imagining, guessing, inventing, making-up, dreaming, and so on, is itself evidence of this belief The view that remembering should he thought of as part of a causal process is, of course, fundamental to psychology and to neurologists attempting to find the brain mechanism responsible for the phenomenon The process view also underlies the extensions made to the concept of memory when, under the influence of the new evolutionary biology, the notion was apphed to races and groups, or m our own day when we apply it to inanimate objects capable of certain involved causal processes, as when we speak of calculating machines having a memory. The fact that it makes some sense (although also different senses) to talk of human memory, race memory, machine memory and so on, indicates how central to the concept of memory is the process-analysis Add to this the point that if we are to explain the phenomenon of memory we undoubtedly have to assume that a process of some sort is responsible, and one is led to enquire whether the analysis of remembering into a process-concept may not solve the problems which the analyses so far mentioned have been unable to meet For one thing, reference to an underlying mental processprovides a common factor m our use of the memory words over a wide range of differing contexts, and would thus appear to explain our usage For another, it would seem to avoid the difficulty raised by honest but mistaken memory- claims, as there is no reason why privileged access should extend to unconscious mental processes There is, however, something very curious entailed by this suggestion On the one hand, even now very little is known about the processes involved m human memoiy . 1 On the other, it is frequently possible to establish with certainty that a person remembers something The consequence of the process analysis would be, then, that we can establish with certainty that on oc- casions the working of an assumed and unknown process has taken place. It also involves the novel information that when we claim to remember, what we are really claiming is that we are undergoing a certain mental process. If this were so it is undoubtedly true that the vast majority of people when they claim to remember something, simply do not know what it is that they are claiming. These conclusions are quite unacceptable But the difficulties involved m construing the concept as a process concept do not necessitate the equally unpalatable course of maintaining that it is not a process concept. The assertion that a person remembers something does mvolve, or is taken to involve, the ascription to that person of a certain mental process undergone. But it also involves far more, and that of a character vastly more important to the purposes of everyday life m which the concept finds employment. In its constant everyday use it finds employment, I suggest, as one of a group of concepts which we use to classify statements according to their truth-value. The group described earlier as process concepts opposed to remembering likewise serve the same purpose. But the properties which, ru the previous context, it was natural to assume were properties belonging to the utterer of a statement — that lie is guessing, imagining, inventing, dreaming, etc — I wish now to point out are properties which, in the first place, belong to the statement itself. A statement is classified as a guess when it is not backed by evidence which would yield it as a conclusion ; an inference when it is so backed ; an invention, story, dream or imagining when it bears no relation, descriptive or evidential, to the facts which it purports to he about, when the statement is presented as a claim to remember. (To preface a statement with, e g. ' this story ’, * my dream etc , is to declare that it does not purport to be about any facts.) These rough summaries, which are not meant to he characterisations let alone definitions, serve to bring out two pomts. First, that to classify a statement under any one of these or similar heads is to label ltwith regard to Its truth-value. 1 There are. of course, highly informed speculations about its nature See, c g The neurophysiological Basis of Mind, by J. C Eceles, Oxford For a discussion of certain special difficulties that accounts of the nature of tlie memory mechanism must face, see, “ In Search of the Engram ’ by K. S Lashlov, S.E B Symposia, vol. IV, Academic Press, HT.X , 1950. Second, that it is, so to speak, to write upon tlie label tbe support or lack of it which the statement possesses. Thus the labelling of an assertion as a guess informs peoplethat the support or grounds hacking the assertion is of a certain kind, and that although the information asserted is unlikely to be true, it just may be so (it may be a shrewd guess). Though the remaining process verbs I mentioned, and the far larger set that I have not mentioned, would each need individual treatment (which I shall shortly give in the case of remembering) m order to demonstrate that they are used to classify assertions, the sample that I have given is suffi- cient to indicate the nature of this linguistic function. It may be observed that it is clearly necessary m the interests of efficient communication that we should apprise one another, where possible, of the logical status of the assertions we make. The enormous difficulties and frustrations reported by people who have experienced living m primitive societies where such distinctions are not made is sufficient witness of this. On the other hand, except in certain rare instances, it is not apparent that the constant exchange of autobiographical prefaces to our remarks would be particularly useful or interesting The man who purveys information solely about himself is a bore Although the fact that it is grammatically correct to refer the actions denoted by active verbs to the actor (I guess, he guesses, etc ) clearly plays a large part m leading us to misconstrue sentences containing verbs of the group which concerns us simply as sentences giving information about the actor, this linguistic fact is not the whole of the matter Even though the logical propriety of employing active verbs at all in these contexts might now seem dubious, it equally seems unavoidable On the one hand, that the making of assertions or the thinking of thoughts of logically different lands necessarily involves a doing or activity seems guaranteed by the “ causal principle ” ; on the other, it is a matter of experience that frequently the production of different sorts of assertions (or thoughts) is preceded by typically di ff erent mental doings, experiences or processes The fact of the matter is that these verb forms play a multiple role m discourse , they are used simultaneously to label the logical status of an assertion or thought, and to refer to the activities or processes whicare causally responsible for the assertion. Neither role can be reduced to the other, but it is important to see that one is logically primary. That the process role is secondary showed itself m the case of remembering, m that no types of mental experiences or thought processes were found to be either necessary or sufficient conditions of remembering. Equally, the primacy of the logical status marking role appears when. it is realised that the conditions which are necessary and sufficient are those relating to the truth value and truth conditions of the statement of what is remembered, e g. my claim to remember seeing Jones hit Smith is correct only if Jones did hit Smith and I witnessed the occurrence. Once "the relationship existing between the two roles is dear it is not difficult to see why the experiences which typically enter into our remembering have no part in the logical structure of the concept. They could never provide conclusive tests of whether remembering has taken place or not because what the tests must he designed to determine is whether the putative memorr is veridical {eg. whether Jones did hit Smith, etc.). If the assertion ' I remember p 5 is true then it follows that one’s memory processes worked correctly, but the truth of the assertion is not logically dependent on the workings of the memory process. This point can be seen more clearly in the following logically analogous case The test of whether an electronic calculating machine is working correctly (and was designed correctly) hes in the correctness of its calculations. The tests of correctness are of course logically related to the rules of the calculus employed, not to an electronic process. The production of a calculation is causally dependent upon the functioning of some process, mechanical or mental, which employs certain mathematical procedures, just as the production of a memory is causally dependent npon some individual’s memory processes. But logical dependence cannot be assimilated to cansal dependence (or vice versa ) as the theories of remembering examined earlier in fact attempt to do. For instance, even if a totally error-free calculating machine were developed, so that it became practicable to say that the correct answer to a complex calculation is the answer given by the machine, the machine’s answer would still in principle he verifiable and the relations between logical and causal dependence remain unchanged Human memories are not perfect, and when- ever anything of importance hangs npon an individual's meznoiy claim we endeavour to verify it. But although in practice we never accept as quite conclusive those experiences whatever they may be, which lead a man to state with honest conviction I know I remember , we do accord them the staus of strongly presumptive signs of remembering. If our memories were uniformly excellent then undoubtedly we would drop our practice of verifying claims to remember, except as a check for mendacity, and treat them as if they were self-certifying. But. even in thi* remote contingency, the primacy of the status labelling role would remain unchanged. What I liave referred to as the ‘ logical status labelling ! role played by tbe mental process verbs must be elucidated further m the case of remembering. To preface one’s remarks with the phrase ‘ I remember . . . ’ (or a cognate) is to indicate to one’s audience that you certify the truth and accuracy of the infor- mation you are about to give or which you claim to be able to give or to the correctness of the performance you will or could under- take The nature of the status-label affixed to p m a statement of the form ‘ I remember p ’, might, m part, be paraphrased * p is true (or this performance is an instance of p) and you have my word for it ’. In this, the role of these expressions is notably parallel to that of ‘ I know ’, and this comes out in the fact that very frequently we can employ either verb with indifference both to the intended and understood sense of the utterance, eg ‘do you know/remember Ins name* ‘ do you know/remember Ohm’s Law* ’ and so on It similarly shows itself m the fact that if one remembers p it entails that one knows p The reverse, of course does not. hold. One may use the verb ‘ to remember ’ only to certify statements or performances relating to the past, whereas we can, by the use of * know ’ certify statements about the past, present or future I propose to mark tlus common linguistic role by describing each concept as a certificatory concept We may now understand the concept of knowledge to be our most general certificatory concept, and the concept of remembering to be specialised m the respect just mentioned It might at first seem that the certificatory role is at once too obvious and too unimportant to be worth the mention For the norm of communication is essentially the exchange of infor- mation supplied m all sincerity. We expect perfectly ordinary statements like ‘ the car is outside ’, * it was raining yesterday ’, I can see him coming ’ to be true, and lienee there is a perfectly good sense in which one can say that the mere utterance of ap- parently informative statements commits the utterer to standing by the truth of lus statements. But this very fact makes it obviously desirable to have expressions which underline our committal to the truth of our assertions, and which can be used to stress the fact that the information is certified. I shall turn shortly to consider the contexts in which we do and do not use remember ’ , here, it may be noticed that a phrase like * body of knowledge ’ can be translated into ‘ body of certified informa- tion and ‘ theory of knowledge ’ into ‘ theory of certification ’ with some gam m illumination A further gam stems from the power to explain certain features of the logical grammar of the concepts which this way of loo kin g at them allows. If one proffers information accompanied by the formula 'I remember (that) p 5 or ‘I know (that) p’, then, should p turn out to be false, to be misinformation, one is forced not merely to admit the falsity of the information, but also that one did not remember, that one did not know. One is forced by the rules of language formally to eat one’s own words. One may attempt to explain the existence of such a rule by pointng out that, at least in the case of the verb to remember and the verbs of perception, which are also subject to this rule, one is dealing with what Ryle has called achievement words. This is both true and useful, for one can talk here of trying or of failing to achieve the desired result, eg oi trying or failing to remember, to see, to hear and so on ; and success may sometimes indeed be ac- companied by a very real feeling of achievement. But one cannot make this point of the verb ‘ to know ’ : it is not clear what could be meant by ‘ trying to know or ' failing to know ’, except in the acquaintance sense of the word where the rule does not apply. The reason for this difference is undoubtedly that the verbs of ’ achievement ’ are also process (or procedure) verbwhereas the verb ‘ to know ’ in its relevant sense is not Processes may or may not work, they need suitable conditions (‘ the light is too bad to see ‘ it happened too long ago for me to remember the details ’, * my ears are full of water ’) ; and sometimes they can be made to work (‘ turn on the light then you’ll see ’, ‘ if you remember what you did before that it might come back ‘ you’d better see a doctor ’) If a process fails to work one must with- draw a claim which implies that it has worked But the fact that this rule of formal withdrawal applies equally to the verb to know, which is not a process verb, suggests that there must be more behind the rule than reference to processes which may or may not have taken place The remaining explanation emerges, I suggest, if one sees the rule also as a device, embodied m the language, to protect the integrity of certificatory expressions Words may be abused, suffer debasement and lose their force, and if it were not for the presence in the language of a formal rule of this nature, certificatory expressions would soon lose the special emphasis which makes them so valuable. It is important to notice that because we use the verb 'to remember ’ as a mark that the information we are giving is true and correct, it does not follow that memory can never play us false, that we cannot make a mistake in our remembering, that one either remembers or one doesn’t. If a man were to make three mistakes in a recitation of the Ode to a Nightingale no one could sensibly accuse him of failing to remember the poem, unless it happened to be an occasion when perfection was required. On the other hand, if he were to get his telephone number wrong, we would say that he simply doesn’t remember it rather than say merely that he made some mistakes "We have no hard and fast rules for what is to count as a correct description, an accurate sum- mary, getting an account right, knowing a street or an argument or a poem Hence, we cannot state hard and fast rules for what is to count as remembering, for we treat remembering as a function of the truth, correctness and accuracy of a statement (or perform- ance). We evaluate the truth or accuracy of a statement by taking mto account the demands of the contest and situation in which it is made, and express this evaluation by choosing terms from, so to speak, a rough scale of expressions For example: ‘ completely accurate ’, * fairly accurate ’, * inaccurate ’, 1 quite true ’, ‘ partly true ’ and so on. Likewise we evaluate and express the truth or accuracy of what is remembered For example, * he remembered it perfectly ‘ he half-remembered it ’, ‘ he didn’t remember it at all ’ and so on. Similarly we may dimini sh the claim to correctness implied by the use of the verbs * to remember ’, ‘ to recollect ’, * to know ’ by prefacing them with, eg. ‘ I seem to . . ’, ‘ I think I . . . ‘ I believe I . . . ’ and so on The object of such qualifications is of course to warn the listener that there is, to a varying degree, some doubt about the correctness of the information which will follow and that, having given due warning, the speaker cannot be blamed if his listener puts too much weight upon his words A noteworthy consequence of the certaficatory role of remem- bering may be seen in the fact that it is only rarely necessary to bring up the notion of remembering m our everyday discourse. Now one could generate a sense of the verb to remember such that from the demonstration that one has not forgotten p, i e that one has produced or performed p, it would follow that one remembers p This sense would fully accord with the requirements of the process concept of memory, as the latter embodies precisely the rule just used to generate the new sense of remember. Hence, the skills and information that may be said to be memory- dependent, i e. which we may forget (perhaps as a result of injury to the brain) may now be said to beremembered when they are aetualised. Thus one could speak of Englishmen conversing or writing in E nglis h as * remembering words in the English language ’, of accountants doing accounts as e remembering how to add ’, and one might murmur as one signs one’s name ‘ I’ve remembered my name agam ’. The absurd inappropriateness of these examples if ‘ remember ’ is understood in its usual sense, illustrates the opposition between the two senses. It is not, of course, an opposition that permits the crude exposure of its existence by denying that in these examples one remembers one’s name or one’s language, for such a denial would for each sense entail that one had forgotten them The inappropnateness would he in bringing up the notion of remem- bering in its usual sense at all in such connexions. Two -very closely related factors determine when it is appro- priate to bang up the notion of remembering m ordinary discourse It would clearly be nearly as pointless to remark the functioning of one’s memory processes each time they work as it would be to remark the fact that one breathes and is m full possession of one’s faculties ; that one has possession of one’s faculties is usually sufficiently evident. The introduction of one’s remarks with the phrase ‘ I remember ’ or the ascription to the remarks or performances of others as remembered has pomt just because we reserve the use of these expressions for occasions when there is some possibility that one may not remember what- ever happens to be m question. Thus the absurd examples given above would be sensible remarks only if made m a context in which there was reason to believe that the person of whom they are made was suffering an impairment of his faculties, or was a child who was actually learning this information or these skills As a result of this restriction of application, the very bringing up of the question of remembering actually implies the possibility of forgetting Compare the rudeness of ‘ do you remember your name* ’ with ‘ what is your name* ’or ‘do you remember what you had for breakfast today* ’ with ‘ do you remember what you had for breakfast three weeks ago* ’ Part of the absurdity of the earlier examples lies m the ineptness of implying that, for instance, a sane and sober accountant might have forgotten how to add Si m ilar considerations determine the occasions on which it would be sensible to use a certificatory expression We never bring up the notion of remembering nnless there are grounds for supposing that an assertion ma)' need checking or that a performance might be faulty, and in such circumstances it is obviously fitting to employ a certifying expression. By the same token it would he otiose to employ an expression with this force when one’s hearers stand m no need of personal assurances about the truth of what is uttered For instance, no one would say either Tremember’ or ‘Iknow that 2 + 2 = 4’ although he certainly does know it and, in the special first sense, he certainly does remember it. Tlie contiast ■which I have drawn between the special sense and the ordinary sense of the verb to remember lies, of course, more properly, between the concept of memory and the concept of remembering The concept of memory is the concept of a storage system, but the concept of remembering is not its natural corollary, that of the wholesale removal from store of the goods and chattels of experience Rather, as I have briefly attempted to show, it is devoted to the removal, not of special articles, but of any articles when conditions are such that the transaction deserves attention. It is now' possible to attempt an answer to the question which we deferred, namely, whether remembering has a unitary sense maintained in all contexts, or whether it is importantly multivocal If the arguments put forward m this paper are correct, it is plain that in their primary linguistic role as certificatory concepts the verb to remember and its cognates preserve a unitary sense in all contexts Things ns various ns assertions, thoughts, images, feelings, bodily performances and so on, may be similar in respect of their logical status as lmvmg been remembered It is the secondary process attributing lole that misleads theorists on the topic and invites speculations about differing senses of the verb. Whet tier there is a single process responsible for the phenomenon of remembering is a question on which no lay opimon would merit consideration, and m the absence of an accepted view about the nature of the process it seems correct to suggest that generally the verb has an utterly vague secondary sense, rather than a cluster of differing ones But it remains quite open to individuals to nominate specific process references should they wish to, and it is apparently the case that sometimes this does occur, as when people, impressed by the characteristic way m which their own memories come to them, take the verb to have reference to these private happenings In such cases there seems no reason why one should not talk of differing secondary senses of the verb, so long as it is clearly understood that such senses are private ones, and do not provide the material upon which an analysis of the concept must be founded.