Our project for the next six chapters or so is a whirlwind tour of some of the main modules of our epistemic establishment: self-knowledge, memory, perception, knowledge of other persons, testimony, a priori knowledge, induction, and probability. I make no claim to completeness; indeed, I claim incompleteness, and that in two different directions. First, I shall stick to modules about whose existence there is fairly wide agreement. As I see it, however, there are other main modules, modules whose existence is a matter of controversy: there is our way of knowing moral truths,¹ for example, as well as our means of perceiving beauty; and there is the sensus divinitatis of which John Calvin speaks, as well as what some theologians refer to as the Internal Testimony of the Holy Spirit.² Second, I shall make no pretense of completeness with respect to the modules I do discuss; such a pretense would in any case be all too easy to see through. An even reasonably complete account of self-knowledge, say, would require a book all by itself; and the same goes for the other main modules. Accordingly I do not aim at a systematic and complete account of these modules and their working: instead, I shall mention and emphasize those features of these modules that illustrate and elucidate the account of warrant I think correct, and those features of these modules about which it has something special to say. We may therefore take the next few chapters as suggestions for research programs.

I. Knowledge of Myself

When we think of self-knowledge, we often turn, in a Cartesian vein, to our knowledge that we think (for example, that we are appeared to a certain way, or that we hold certain beliefs, or that we are in pain [or not]) and that we exist. On many occasions we think we know that we are appeared to in a certain way (at the moment I know that I am appeared to in the way in which one is appeared to when looking at the screen of a computer); we think we know whether we are in pain (at the moment I know that I am suffering a mild eye-she, no doubt induced by an excess of being appeared to in the way just mentioned) and we think we know at least something about what we believe. According to the Cartesian tradition, we have privileged epistemic access to these matters. Some say it is impossible to be mistaken about them; we have incorrigible knowledge of these matters, where S has incorrigible knowledge of a proposition p if and only if it is not possible that p be false and S believe it, and not possible that p be true and S believe -p. Others say, not that we have incorrigible knowledge here, but that we can have knowledge here by reflection alone; all you need to do to know whether you are in pain, for example, is think about it.

Here the tradition in question seems right. It is right, first, in holding that we have knowledge here. Some of our beliefs about how we are appeared to, or what we believe, or whether we are in pain have warrant, and indeed have about as much warrant as any beliefs we ever form. It looks to me as if there is a top on my desk; I currently believe both that there is a cup on my desk and also that it looks to me as if there is. That latter belief, I think, has a great deal of warrant for me: I hold it with near maximal firmness, and (so at any rate I believe) it is formed by my faculties working properly in a congenial epistemic environment, with the triples of the design plan governing its production both aimed at truth and successfully aimed at truth. This tradition is also right in seeing some of this knowledge as incorrigible, or at any rate nearly so,³ although the extent of incorrigible knowledge is meager at best.

The tradition is right, furthermore, in holding that we also know that we do indeed exist (although in order to know that welcome fact we need not infer it from our knowledge that we think). But do we know what sort of thing we are? Will, what sort of thing do we think we are? First, each of us finds himself with the belief that he has existed for quite some time (for most of us, at least several Fears; for some of us more years than we like to think). Second, each of us thinks (more accurately, takes it for granted) that he is conscious—more exactly, intermittently conscious. That is, I, this very being who I am, am now conscious, was earlier unconscious (as in deep sleep), before that was conscious, before that unconscious, and so on. (Here we are contradicted by David Hume, who says “When my perceptions are removed for any time, as by sound sleep, so long am I insensible of myself and may truly be said not to exist.”)⁴ Each of us, furthermore (and perhaps this is what is most remarkable about mind) is and has been aware of himself, as well as of many other things; and it is I, this very being, who is now aware of the sun shining on the lilacs in my backyard, who a moment ago was thinking instead about a Yosemite rock climb, and who yesterday was paying attention to the student who wanted to postpone his examination (pleading mental anguish due to failure to win the Illinois lottery), Awareness (awareness of) is at the root of intentionality, the property of being of or about something, of being directed at or upon or toward something, of intending something. This property is displayed by belief, but also by many other mental states: for example, anger, love, hatred, contempt, admiration, worship, delight, amusement, lust.

What I believe on this head, of course, is that I myself, this very self, this very person, have existed for quite some time and done these things. Here it is hard to match the eloquence of Thomas Reid:

I take it for granted that all the thoughts I am conscious of, or remember, are the thoughts of one and the same thinking principle, which I call myself or my mind. Every man has an immediate and irresistible conviction, not only of his present existence, but of his continued existence and identity as far back as he can remember. If any man should think fit to demand a proof that the thoughts he is successively conscious of, belong to one and the same thinking principle—if he should demand a proof that he is the same person today as he was yesterday or a year ago—I know no proof that can be given him; he must be left to himself, either as a man that is lunatic or as one who denies first principles, and is not to be reasoned with.⁵

He adds that “The conviction which every man has of his Identity, as far back as his memory reaches, needs no aid of philosophy to strengthen it; and no philosophy can weaken it, without first producing some degree of insanity.”⁶ Part of Reid's point, I take it, is this. Any well-formed adult human being who is in an epistemically congenial environment and whose intellectual faculties are in good working order will typically take utterly for granted at least three things: that she has existed for some time (for some years, say), that she has had many thoughts and feelings, and that she herself is not a thought or feeling or congeries of thoughts and feelings. (“Typically”; perhaps a person who comes to reject these commonsense views by way of philosophical reflection need not be subject to cognitive dysfunction.) Surely Reid is right here. Accordingly, if, as seems likely, the modules of the design plan governing the production of such beliefs are successfully aimed at truth, then (on the present account) such beliefs will have warrant; if true and held with sufficient conviction, they will constitute knowledge.

But might it not be, as Derek Parfit suggests, that “We Are Not What We Believe”?⁷ Perhaps people aren't at all what we ordinarily think they are. It is logically possible that I should exist for a very brief time—a microsecond, for example—displaying all the temporally specific properties I do in fact display at the present moment. (Of course I wouldn't then have such properties as being over fifty years old, or being responsible for something that happened ten minutes ago.) So say that a person slice is a person who exists for a micro-second or so. Perhaps there aren't any enduring persons, but only successions of person slices liked by appropriate causal relations and overlapping series of apparent memories. Perhaps a person slice is what (strictly speaking) thinks, believes, feels, and so on; perhaps what I refer to, when I refer to the appropriate thing that thinks, is a person slice, a thing that exists but for a moment. There is I, and all my successors, and all my predecessors; each of us exists just for a moment, the later slices often involving apparent memories: memories apparently of earlier properties of an enduring self, but actually of slices earlier in the series to which the given slice belongs. Or perhaps there aren't even person slices, if a person slice is a thing distinct from thoughts and feelings that has thoughts and feelings; perhaps there are only the thoughts and feelings, linked by relations of causality and resemblance. Couldn't these things be the case? More to the point, couldn't it be both that our experience be as it is and these things be the case?

Let us suppose for the moment that these things are possible, and compatible with our experience being as it is. (For the record, it seems to me quite impossible that there be thoughts without a thinker, though possible that there be person slices.) It is also possible that I am not a member of any such series of person stages but have popped into existence this very instant, complete with a complement of insistent but wholly false memory beliefs; it is also possible that I have existed for a thousand centuries but suffer from a peculiar sort of amnesia with respect to all but the last few years of this time; it is also possible that it is really your body that is mine, and that I am unaccountably deceived about the connections between me and this body, and so on. These things are indeed possible, and (in a way I shall explain) compatible with my experience's being as it is. But if these things are possible, and consistent with my experience's being as in fact it is, then should we not conclude that those commonsense views about ourselves have little warrant? Indeed, shouldn't we give them up? if we wish to be philosophically responsible, shouldn't we say, with Lichtenberg, “Es denkt”⁸ rather than with Descartes (and the rest of us) “Je pense”?

Parfit thinks so. Considering the view that what I am is a persisting Cartesian ego, he says

As Locke and Kant argued, there might be a series of such entities that were psychologically continuous. Memories might be passed from one to the next like a baton in a relay race. So might all other psychological features. Given the resulting psychological continuity, we would not be aware that one of these entities had been replaced by another. We therefore cannot know that such entities continue to exist. (p. 223)

And,

We could not tell, from the content of our experiences, whether we really are aware of the continued existence of a separately existing subject of experiences. The most that we have are states of mind… when we have had a series of thoughts, the most that we are aware of is the psychological continuity of our stream of consciousness. Some claim that we are aware of the continued existence of separate existing subjects of experiences. As Locke and Kant argued … such awareness cannot in fact be distinguished from our awareness of mere psychological continuity. Our experiences give us no reason to believe in the existence of these entities. Unless we have other reasons to believe in their existence, we should reject the belief. (p. 224)

A few lines down he adds, “My claim is merely like the claim that, since we have no reason to believe that water nymphs or unicorns exist, we should reject these beliefs.” (p. 224)

How shall we understand Parfit here? He puts his point variously:

We could not tell, from the content of our experiences, whether we really are aware of the continued existence of a separately existing subject of experiences… when we have had a series of thoughts, the most that we are aware of is the psychological continuity of our stream of consciousness… such awareness [of a persisting subject of experience] cannot in fact be distinguished from our awareness of mere psychological continuity.

Clearly there is sense to these claims; from one perspective they seem no more than the simple truth. I am now being appeared to in that tiger-lily way; and I believe that I am being thus appeared to. But I also believe something else, which perhaps we can approach as follows. Parfit speaks here of experiences; he speaks as if, among the furniture of the world, there are, in addition to (say) persons and material objects, experiences. So, for example, if I am now being appeared to in that tiger-lily way, then there is in this situation a tiger-lily appearance or experience. Now the idea that there are such things as appearances is problematic along several dimensions (dimensions that were extensively explored earlier this century); perhaps the truth of the matter is not that there exists a tiger-lily appearance but simply that I am being appeared to tiger-lilyishly, in a tiger-lily way. But suppose for simplicity we concede Parfit's point. Then I can state a further belief I have: I believe that I am being appeared to tiger-lilyishly, that I am aware of a tiger-lily appearance, but also that there is something—I myself—that is both aware of that tiger-lily appearance and that is distinct from it. (I am not identical with that tiger-lilyish experience.)⁹

Now consider this last belief. The point made by Locke, Hume, Kant, Parfit, and others is that there is a certain way in which this belief is not supported by my experience, or by the appearances I am now aware of, or by the succession of ways in which I am being appeared to. Hume is right: “For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure, I never can catch myself at any time without a perception, never can observe any thing but the perception” (A Treatise of Human Nature, p. 252; Hume's emphasis). Say that I am directly aware of something if I am aware of it and am not aware of it by way of being aware of something else; and again suppose for present purposes that when I am appeared to thus and so, then typically there is something, a thus and so appearance, of which I am aware. Then when I perceive the tiger lilies I am directly aware of a tiger-lilyish appearance; but I am not directly aware of myself.¹⁰ My belief that there is tiger-lily appearance is supported by my being directly aware of a tiger-lily appearance; but my belief that I am distinct from that tiger-lily appearance is not supported by way of my being directly aware of myself. I am also being appeared to in the way that goes with perceiving a squirrel; I also believe that the tiger-lily appearance is distinct from that squirrel appearance; and this belief is supported by my being directly aware of both of these appearances and nothing that one is distinct from the other. But not so for my belief that I myself am distinct from each of these appearances. It is not the case that I am directly aware both of the appearance and of myself and note that one is distinct from the other.

We might perhaps suppose that it is memory by virtue of which I know that I am not identical with the present appearance. I remember being aware a moment ago of a squirrel appearance, and am now aware of a tiger-lily appearance. I reason as follows: if I were identical with the present appearance, then I would have been identical with that past appearance; but in that case the present appearance would be identical with the past experience; but that could be the case only if the squirrel experience had somehow turned into the tiger-lily appearance, which certainly did not seem to happen. So (the suggestion goes), even if I am not directly aware of myself, I am directly aware of some things such that propositions reporting those awarenesses immediately entail that I am not identical with present appearances.

But this reasoning is flawed. What I am directly aware of, at the moment, is the tiger-lily appearance and (at best) my memory belief that I was aware of a squirrel experience; from propositions reporting these things it doesn't relevantly follow¹¹ that I am not identical with the tiger-lily appearance. First, from the fact that I can't be identical both with that squirrel appearance and with this tiger-lily appearance, it just does not relevantly follow that I am not identical with the tiger-lily experience. Second, any appearance of plausibility for the claim that it does follow arises from the fact that one who offers the argument takes it for granted that it is I, this very thing, who am now aware of the one appearance and was then aware of the other appearance. In so doing she is of course doing no more than following nature's promptings; the memory belief (that I was aware of the squirrel appearance) already comes with a subject-experience structure; it comes as the memory that one thing—I—was aware of another thing: that experience. I don't merely believe by way of memory that there was a squirrel appearance: what I remember is that I was aware of such an appearance. But if I can't take my present belief—that I am aware of but distinct from a tiger-lily experience—at face value, how will it help to appeal to a memory belief that has the same structure? The second is no better off than the first. If I agree that what I should say about my present experience is “There is a tiger-lily experience,” then I will find it hard to deny that what I should say about my past experience is “there was a squirrel experience”; but then my memory gives me no independent grounds for supposing that there is a persisting thing which (who) was aware of the squirrel experience and is (now) aware of the tiger-lily experience.

So Parfit is right: “the most that we are aware of is the psychological continuity of our stream of consciousness.” More exactly, that is the most we are directly aware of. And of course we are not at any one time directly aware even of this; for we must rely upon memory for our awareness of past elements of our stream of consciousness and hence for our awareness of the continuity of that stream. But Parfit (along with his illustrious predecessors) draws a fateful inference from this fact. He concludes that we don't know and cannot justifiably believe that there is such a thing as a persisting subject of experience that is distinct from any of those experiences (and any aggregations of them):

Some claim that we are aware of the continued existence of separate existing subjects of experiences. As Locke and Kant argued … such awareness cannot in fact be distinguished from our awareness of mere psychological continuity. Our experiences give us no reason to believe in the existence of these entities. Unless we have other reasons to believe in their existence, we should reject the belief.

Here, however, we obviously need additional premises. You might as well argue that since we are not directly aware of tiger lilies but only tiger-lily appearances, our experience gives us no reason to believe in the existence of tiger lilies; and unless we can come up with some other evidence we should reject the belief. You might as well argue that since a Cartesian demon could arrange things so that I would believe a false proposition with the same fervor that I believe 2+1=3, I don't know that latter proposition; and if I can't find some other evidence, I should try to give it up. You might as well argue that since my experience could be just what it is even if I were the unlucky brain-in-vat victim of an Alpha Centaurian cognitive experimenter, I don't really know any of the things I think I know; so if I have no other evidence (and how could I?) I should reject those beliefs.

All of these inferences are regrettably hasty. Those arcane possibilities are indeed possibilities; they are also consistent with my experience's being as it is; but it doesn't follow that the commonsense beliefs in question have little warrant. It is entirely possible both that what Locke, Hume, Kant, and Parfit say here is true and that the commonsense beliefs in question have much by way of warrant—enough, even, to constitute knowledge. It is entirely possible both that what Parfit and his mentors say is true, and that we nonetheless can “tell, from the content of our experiences, whether we really are aware of the continued existence of a separately existing subject of experiences.” I can tell “from the content of my experience” that there is a squirrel in my backyard, even if my experience could be just what it is and there be no squirrel there. My experience could be as it is and the belief in question false: but that is nowhere near sufficient to show that the belief has little or no warrant for me, or that it does not constitute knowledge for me. Most of my contingent beliefs are such that my experience could be just what it is but the belief false; how is it supposed to follow that these beliefs don't constitute knowledge? How does it follow that (even more alarmingly) I should try to give them up unless I can find some other evidence? We know that there are trees and flowers, even though our experience could be just what it is and there be no trees and flowers; we know that the world has existed for more than five minutes (and even that there was such a land as ancient Greece) despite the fact that our experience could be just what it is and those beliefs false.

On an adequate account of warrant, what counts is not whether my experience somehow guarantees the truth of the belief in question (and how could it do a thing like that?), but whether I hold it with sufficient confidence and whether it is produced in me by cognitive faculties successfully aimed at truth and functioning properly in an appropriate environment. If so, it has warrant; and if it is also true it constitutes knowledge—and this even if what Parfit and his mentors say is true. These conditions certainly seem to be met for such beliefs as that I have existed for some substantial stretch of time, and that I am neither an experience nor an aggregation of experiences. Therefore the question whether we know that we are persisting subjects of experience (distinct from those experiences) reduces to the question whether these beliefs are true. It would be of great interest to go into the question whether there is reason to doubt what we ordinarily believe here (I can't myself see that any of the usual arguments have any force at all); but I shall resist the temptation to make an excursion from the firm dry ground of epistemology into the misty miasmic morasses of metaphysics.

I conclude this section with a brief excursus in a different direction. We are all inclined to think that we are subjects of experience, distinct from those experiences; but we are also inclined to think that we are not identical with our bodies or even our brains. More exactly, what we are inclined to think are things from which it obviously follows that we are not identical with our bodies or brains. Thus we think it possible, in the broadly logical sense, that we should survive a rapid series of transplants in which each of our bodily organs—arms, legs, liver, stomach, lungs—should be replaced by a prosthesis, one made of some more durable and thus more suitable material, perhaps. And the same goes even for our brains; for it is natural to think it possible, in the broadly logical sense, that I should survive the following sort of brain transplant. First, the left half of my brain is very rapidly replaced by another left half that is in the same functional (computational?) state; then the right half is replaced in the same way; and the original right and left halves are destroyed. It is possible, in that broadly logical sense, that I should survive this operation, and indeed that I should remain fully clothed, conscious, and in my right mind (perhaps reading a current text on neurobiology) throughout the entire process. (It is also natural to think that it is logically possible that I survive the sort of operation in which the neurons are rapidly replaced, neuron by neuron, by other neurons in the same functional state, and similarly for rapid molecule-by-molecule and atom-by-atom replacements.) Further, these beliefs of ours—that the preceding scenarios are broadly logically possible—don't arise (so at any rate we ordinarily think) by way of cognitive malfunction or disorder, or because of wishful thinking or other cognitive processes not aimed at truth, or because of trade-offs and compromises. But then they have warrant; and if they do, then so does the consequent belief that I am not identical with my brain or with my body.¹² So when I think, what thinks is not a brain or a body. That is, I think, and I am not either my body or my brain; it could be, for all I can prove, that my brain or body also thinks. What is clear is that I think and (by the preceding argument) am not identical with brain or body, so if either of them also thinks, then there are at least two of us in the neighborhood who think.

Accordingly, I am not a body or a brain or part of a brain. Am I perhaps something more like a computer program? Many seem to take it utterly for granted that we stand to our brains and nervous systems as a computer program does to a computer. Various different computers can realize (run) the same program; and in the same way I could be realized in various different brains or bodies. Of course I would not be identical with any of those brains, any more than the program for my Macintosh SE is identical with that machine; what I am is a program suited to be realized in a variety of appropriate media.

But what, after all, is a computer program? Well, it seems to be a series of some kind, perhaps a series of instructions or, more abstractly, perhaps a series of zeros and ones. Of course a computer program need not be a series of zeros and ones: it could be a series of other shapes; and perhaps it need not be a series of shapes at all: perhaps it could be instead a series of properties or of still other objects. But at any rate a computer program is ordinarily thought to be a series of one sort or another; on the present suggestion, therefore, what each of us really is, is a series. But what is a series? Well, according to current mathematics, a series is best thought of as an ordered set—an n-tuple, in the finite case. There is some choice as to just which sets would be the members of a series (there are various set theoretical constructions available here), but if we are computer programs, then presumably each of us is a set of some sort. In a widely quoted passage, David Lewis once said that he couldn't believe that he and his surroundings were a set of sentences.¹³ I can go him one better: I can't speak for others, but I can't for a moment bring myself to believe I am a set of any kind.¹⁴ People can be aware of trees and flowers; sets cannot. I know that 7 + 5 = 12; no set knows any such thing. A person can decide to go to Chicago; no set can do that. (No doubt our mental life—like the natural members or Greek history—can be modeled¹⁵ in various set theoretical structures; but that is not to say that what I really am is one of those set theoretical structures.) These beliefs, therefore—the beliefs that we are neither our brains for our bodies nor computer programs—have a good deal of warrant for us.¹⁶

II. Memory

A. Phenomenology

“Memory,” said Aristotle, in a flash of inspiration, “is of the past.”¹⁷ Fair enough, but how does it work and how shall we understand it? In one respect memory is wholly familiar, closer to us than hands and feet; in another it is puzzling in excelsis. How does it come that each of us has this window on the past? We will no doubt be told that present experience modifies the appropriate neuronal networks, storing patterns that can later be activated.¹⁸ Indeed, some go so far as to say that nothing of what has happened to me is lost; all could be recalled to memory by the appropriate cerebral stimulation. Well, suppose this is true; how does it relieve the puzzlement or reduce the wonder? So the passage of time leaves traces in my brain; it does the same for the trees in my backyard, but they don't remember. What is the connection between these traces and my having or forming memory beliefs about the past? How do those traces so much as enable me to grasp or apprehend the notion of the past? And by what alchemy do they get transmuted into the beliefs I have about what has happened to me?

The phenomenology that goes with memory is remarkable and puzzling. I now believe that yesterday a friend gave me a box of raspberries. What there is here by way of phenomenology is a few scraps of sensuous mental imagery, something like a rudimentary and fragmentary case of being appeared to in the way in which one is appeared to upon perceiving a box of raspberries, or perceiving one's friend's face. But why should that incline me to form a belief about the past? There is nothing past-directed about imagery of raspberries or friends. Could it be (in a quasi-Humean vein) that it is precisely that scrappiness of the image, its lack of vivacity, wholeness, vividness, that makes it a memory, makes it refer in this way to the past? Surely not; under certain conditions of lighting or in a movie theater I can have that very sort of fragmentary, unvivacious imagery but form no beliefs whatever about the past. I can imagine getting a box of berries tomorrow; and this imagination is accompanied, so far as I can see, by the same sort of fragmentary, scrappy, indistinct sensuous imagery as my memory that I got one yesterday.¹⁹

Let's think a bit further about the sensuous imagery involved in the phenomenology of memory. I remember having seen Paul a year ago in California; there are scraps of imagery here, as of a wall he was standing before, as well as bright sunshine and blue sky. But I can't now clearly think what he looks like, and certainly don't have before my mind anything like a clear image or mental snapshot of him. I certainly don't remember that it was he I saw then on the basis of anything like recognizing Paul in some part of the sensuous imagery, perhaps nothing that the imagery contains a part that looks a lot like an image of Paul. I certainly don't note the phenomenal imagery, and then see the resemblance to Paul, thus forming the belief that it is Paul I saw there in California. I no more do this than look into a given proposition to see that it is Paul that the proposition is about. The memory in question has a sort of intrinsic intentionality or aboutness; it already comes as the memory of Paul; it isn't that it is a phenomenal image that I then somehow identify as being of Paul and of him in California. (Remembering having seen Paul in California isn't phenomenally different from remembering having seen Paul in Arizona.) It is exceedingly hard to focus one's mind on this imagery, fleeting and fitful as it is. Furthermore, it seems to vary widely from person to person; and, as Wittgenstein never tired of insisting, it isn't essential to remembering. That is, there is no phenomenological imagery of that sort such that, necessarily, one remembers seeing Paul in California only if one undergoes that phenomenology. The imagery may be nothing more than irrelevant scraps of color, maybe the color of the orange shirt Paul always wears, or perhaps a half-formed image of his face: obviously that imagery is not necessary to memory. Neither, of course, is it sufficient: I may have the very same imagery when I don't remember seeing him is all, but simply think about him. In some people, indeed, memory seems to work with no sensuous phenomenology at all; it is more like a matter of feeling strongly inclined, upon being asked whether you saw Paul yesterday (or is some other way prompted), to believe a certain proposition: that you did see him.

What is more important than the accompanying sensuous imagery is a sort of sense of pastness; a memory comes as of something past. What is that sense of pastness, and how shall we understand it? Is there something about the memory—some phenomenal or more precisely sensuous feature—that I can note and on the basis of which I can tell that it is indeed a memory, and that it is about the past? I don't think so. A memory comes as about the past, or about something past (a past event, perhaps) in something like the way in which a belief about Sam comes as a belief about Sam. It has a sort of past tinged feel about it. This past tinged feel, however, is not something sensuous. It isn't at all like a mental image; it doesn't fall under the modality of any of the senses. This is of course at best a very unsatisfactory description: but what more can I say? Here I have no answer, nor any suggestion as to how to try to explain our grasp of this mysterious notion of the past in other terms.

Of course, there is more to a memory than a past-tinged belief. Your belief that North Dakota became a state in the late nineteenth century also has a past-tinged feel about it, but it isn't a memory. (It is no doubt right to say that you remember that North Dakota became a state in the late nineteenth century; but you don't remember that event in the way in which you remember that your friend brought you the box of raspberries.²⁰) There is something more in the case of memory, something one can only call a distinctively memorial feel about it. One recognizes it as a memory, not something learned by testimony or in some other way. There is something about the thought or belief in question that identifies it as a memory thought or belief: but is exceedingly hard to describe it or say more than that about it. At any rate I am not succeeding in doing so.

So there is sensuous imagery, a sense of pastness, an aboutness with respect to the subject of the memory, and something like a recognition that it is indeed a memory. These are all part of the phenomenology of memory. But further, when I form a memory belief—that, for example, I had an orange for breakfast—I do not take the belief to be a part of myself; that is, I do not believe that I am a thing somehow composed of such things as this memory belief. I do not believe that I am an aggregate of such things, or a set of them, or a mereological sum of them. Not only do I not believe these things: I believe their denials; I take it utterly for granted that I am one thing and this memory another. Still further, what I remember is that I tasted that orange: that is, what I remember is that I, the very being now aware of the memory, earlier on tasted the orange. Here I suppose I could (in some sense of ‘could’) be wrong. It could be that it wasn't I but one of my predecessors in some series of person slices who tasted that orange; it could be that a mischievous Alpha Centaurian experimenter implanted this ‘memory’ in me. Still (rightly or wrongly), my memory comes as a memory that I did thus and so; pace Lichtenberg, it comes not as, es hat sich daran erinnert but as ich habe mich daran erinnert. I don't remember just that there was tasting, but that I tasted; the memory comes as a memory of something I did, something done by the very person who is the subject of the present thought I did that. Finally, there is still another phenomenological feature of memory beliefs, as of other sorts of beliefs: the sense of rightness or fittingness of that belief on that occasion. There is a kind of attractiveness about the belief, or just possibly a feeling of being inclined, impelled toward it; it has different feel from some other proposition you might try on in the circumstances. I entertain the proposition that it was a box of plums, not raspberries my friend gave me; entertaining this proposition somehow feels different from entertaining the true proposition that it was a box of raspberries, but about all we can say is that it just doesn't seem right. It lacks that sort of seeming to be right or acceptable or true enjoyed by the belief in fact delivered by memory.

I have been speaking of propositional memory, memory as a source of beliefs about the past. Of course, there are other sorts of memory, and these other sorts are connected, in interesting ways, with propositional memory. You remember how to do something—how to ride a bicycle, for example; you don't, even in exercising this memory, need to remember any particular proposition. The smell of sun screen or oranges can trigger memories of mountaineering, although not necessarily propositional memories. Instead you may remember what it looked like at a particular place on a particular climb, or you may instead remember a given mountain—Mount Baker, perhaps—without explicitly remembering any particular proposition about it. You also remember moods, emotions, ambiences. You can remember feeling lost and desolate when you were benighted on the lower saddle with no tent or bivouac sack, the temperature at 15° F. and a cold, ominous wind blowing in from Idaho; to remember that feeling you need not remember any particular proposition. A whiff of perfume can trigger a sudden, sharp, almost painful stab of awareness of someone you haven't seen for 20 years; but again, there need be no particular propositions involving her that you then remember. Naturally there will be many interesting and subtle connections between propositional memory and these nonpropositional varieties: to do the phenomenology of memory properly would require a whole chapter or a whole book; even if I were capable of giving such an account, there is no space for it here.²¹

Finally, there is of course an important connection between memory and my sense of my self as an enduring subject of experience. My thought of myself is of someone who has done and thought many things; and, of course, it is by virtue of memory that I know that I have done and thought these things. Says Reid, “The remembrance of a past event is necessarily accompanied with the conviction of our own existence at the time the event happened. I cannot remember a thing that happened a year ago, without a conviction as strong as memory can give, that I, the same identical person who now remembers that event, did then exist.” It is of course my memory that is the source of my belief that I am a being that thus persists through time. I couldn't have anything like this sense of myself as a persisting object without memory. It is equally clear, however, that I couldn't have the sense of other objects—trees and houses, for example—as objects that endure through time without memory; for without memory I would not have the notion of time and its passage at all.

B. Memory Beliefs as Basic

Clearly my memory beliefs are typically formed in the basic way; that is, I do not reason to them from other propositions, or accept them on the evidential basis of other propositions.²² If I see a blackbird in my backyard, I may form beliefs about its past by virtue of knowing something about how it usually goes with blackbirds: no doubt this one, like most, was hatched from an egg. But that is not how I come to the belief that what I had for breakfast this morning was an orange. I don't form this belief on the basis of such beliefs as that I often or always have an orange for breakfast; instead I just remember it. I do not ordinarily believe that I remember so and so (or that so and so occurred) on the evidential basis of the fact that it seems to me that I remember so and so. (Of course in special circumstances I could do something like that.) But while I do not form my memory beliefs on the evidential basis of beliefs acquired from other sources of belief, I may very well learn to correct my memory, in part on the basis of what I learn from other sources of belief. Perhaps you seem to remember that the fence surrounding the elementary school you attended was at least six feet high; you return years later, noting that the fence is only three fee high; you also learn by way of testimony that the fence has not been replaced; putting these things together with your belief (acquired partly by induction and testimony) that fences do not shrink with the passage of time, you conclude that you remembered that fence as higher than it was. Of course, even in thus correcting my memory I must rely upon memory: the testimony of memory is an essential element of my grounds for the belief that the fence I see before me is the same fence that I remember.

So I don't ordinarily accept memory beliefs on the evidential basis of beliefs solely about the present. This is fortunate, since it is exceedingly hard to see how there could be anything like an even reasonably decent noncircular argument, for me, from present phenomena to the truth of my memory beliefs. Bertrand Russell is right: it is surely possible, in the broadly logical sense, that the world should have popped into existence five minutes ago, complete with all its apparent traces of the past—all its dusty books, decaying buildings, mature oaks, crumbling mountains, and apparent memories. This is possible; more, it is compatible with my present experience's being as in fact it is.

No doubt someone will propose that the best explanation of these present phenomena is that there has been the sort of past I think there has been. Here I am, confronting present phenomena, including my inclinations to form these memory beliefs; and the best explanation, so the claim goes, of these present phenomena is that there has indeed been a past and that these memory beliefs are in fact true. So I have an argument for the conclusion that these memory beliefs are indeed true.

But can we take such explanations and inferences seriously? It will be agreed on all sides that we don't in fact make any such inference; young children apparently form memory beliefs in the ordinary way long before they have any ideas about explanations at all. But more important, the strength of such an argument is at best exceedingly doubtful. First, how do I know, prior to my knowing anything by way of memory, that these present phenomena have an explanation? By way of my knowledge that most phenomena (or most similar phenomena?) do? But so to think is to rely upon memory. Furthermore, one important question here (important with respect to this argument from apparent memory to memory) is whether I am idiosyncratic in these memory beliefs; if there are lots of other people, and only I have these alleged memory beliefs; if there are lots of other people, and only I have these alleged memory beliefs allegedly about the past, a better explanation might be that I suffer from some kind of pathology. But in answering this question about others I will also rely upon my memory: my memory of the claims and behavior of other.

Finally, how can I hold that whole explanation and argument in my mind at once? Descartes observed that for any but the simplest kinds of argument, one must rely upon memory for knowledge of the conclusion. You clearly and distinctly see that the present step follows from earlier steps S₄–S_n; but you have only your memory to testify to such facts as that earlier on you were able to see that S₄–S_n themselves followed from prior steps. This dependence upon memory holds even for such simple arithmetical calculations as that 24 × 32 = 768. (Of course you can record the steps of the argument in your notebook; but then you will rely on memory for the belief that it was you who wrote them down and that you intended them to be an accurate written record of the argument.) And the same goes here for this purported argument to the best explanation. It isn't possible (for most of us, anyway) to hold in mind at one time the relevant phenomena to be explained, the most salient half dozen or so alternative explanations, the reasons (if any) why these explanations aren't as good as the favored candidate, the reasons (if any) for thinking that there must be (or probably is) a good explanation of those phenomena, and finally the inference from the alleged bestness of the explanation in question to the truth of the memory beliefs in question. So an argument of this sort will have at most very little force. Perhaps it isn't necessary to declare it totally without force; perhaps it has a modicum of cogency; what is utterly apparent, however, is that the strength of such an argument is vastly incommensurate with the serene and unruffled confidence with which we hold memory beliefs.

Memory beliefs are therefore typically taken in the basic way, and that is nothing whatever against them. It is not as if they would have more warrant if they were inferred from beliefs about the present. So far, memory beliefs resemble perceptual beliefs (see pp. 93ff.). But there is a crucially important difference between memory and perception here; for perceptual beliefs are typically formed on the basis of the phenomenal imagery, the appearances. There will typically be a detailed mapping from the way in which one is appeared to, to the perceptual beliefs one forms. When I look at my backyard, I am appeared to in a highly detailed and modulated way; and the detail and modulation is reflected in the perceptual beliefs I form. Appeared to in one way, I form the belief that plant's leaves are about six inches long; appeared to in another way I form the belief the lilacs are now long past their prime; and so on. My beliefs are also responsive to changes in the imagery in a variety of complicated and subtle ways. I look in a certain direction, am appeared to in a certain highly detailed and articulated fashion, and form certain perceptual beliefs: I look in another direction, am appeared to in a different highly detailed and articulated fashion, and form other perceptual beliefs. We can go further: there is a similar mapping from the way in which one is appeared to, to the beliefs one would form in the perceptual situation in question, were they occasioned in one way or another. I am appeared to in a rich and variegated way; I may not be paying much attention and may not form much by way of explicit beliefs about what looks like what, which flowers are which colors, and so on; nevertheless there is a detailed mapping from the way in which I am now appeared to, to the beliefs I would form if for some reason I were to pay more attention. (Perhaps we could say that these beliefs or some of them are virtual). But the same, obviously enough, is not true for memory. Many of us, apparently, don't display much phenomenal imagery in connection with memory at all; and in hardly anyone's case is there that detailed mapping from sensuous imagery to belief that goes with perception.

So we don't in fact form memory beliefs on the basis of beliefs about the present—beliefs about how I am being appeared to, for example; and there isn't available much by way of a decent inference from present phenomena to the truth of memory beliefs. But on an adequate view of warrant, this does not so much as slyly suggest that memory beliefs don't have or can't have a high degree of warrant for us. On the contrary: what counts for warrant is whether memory beliefs typically result from the proper function of our cognitive faculties in an appropriate environment, whether the function of memory is to give us true belief about the past, and whether the design plan in this area is a good one. But the fact is (as we all believe) these conditions are all fulfilled. Memory beliefs, therefore, have warrant. Furthermore, they are often held with great firmness. (There are few things I believe more firmly than that I had an orange for breakfast a couple of hours ago.) I conclude that memory beliefs often have warrant and often have a great deal of warrant; many of them, if true, constitute knowledge.