Memory and Knowledge
What is the connection between memory and knowledge? To judge from the standard literature, very little—in a typical treatise on knowledge memory is hardly mentioned. I wish to urge a strong connection: all knowledge is memory knowledge. Memory enters into every instance of knowledge; all knowledge presupposes memory knowledge. That is, all present-tense knowledge depends upon knowledge derived from the past. I don’t just mean that inferential knowledge derives from premises stored in memory—though that is true enough—I mean that even the premises rest upon antecedently possessed knowledge. If I now judge that there is a cup on my desk, this piece of present-tense knowledge requires that I possess knowledge acquired in the past. The reason for this is that in order so to judge I must already know what a cup is: I must have that knowledge stored in my memory. I must have the concept of a cup (as well as the concept of a desk and myself), and having a concept is an epistemic state. My memory interacts with my current experience to deliver the knowledge that there is a cup on my desk. If my memory were blank in this respect, I would not be able to make such a judgment. This is a point about human knowledge not about all logically possible types of knowledge: it is conceivable that a being should acquire knowledge of the same proposition at the same time as acquiring the concept of a cup. But as things are with humans that is not the case: we first have the concept and then use it to formulate the knowledge in question. According to traditional empiricism, the concept derives from perception of cups in the past, traces of which are stored in memory—a concept is an abstraction from perceived objects retained in the mind over time. According to nativism, a concept is an innate item of knowledge that persists in the mind over time. Strictly speaking, it is not stored in memory but in something analogous to memory: an informational storehouse of some sort. At any rate, it exists in the mind as a result of past events—it functions like a memory proper. It exists in the mind unconsciously and occasionally reaches conscious awareness in the form of explicit knowledge. Thus for both theories present knowledge depends upon past knowledge, because it depends on the prior existence of suitable concepts. You can’t know something now unless you already knew something earlier. Knowledge is a cross-temporal phenomenon.
This makes knowledge different from experience. You can have an experience now without having had any prior experience: I can see a cup now without having seen one before, or seen anything before. Present experience doesn’t depend upon past experience. It is like bodily sensation: having a pain now doesn’t depend on having had pains in the past, or having had any sensations. Pain is an at-a-time phenomenon. Accordingly, it is perfectly possible to have an experience of a cup and yet fail to recognize the presented object as a cup. You could lose your memory of what a cup is and still see one. Similarly, you could forget who John is and yet still see him: you no longer know the identity of the presented person yet he can still come within your field of vision. But you can’t know that John is in front of you without knowing who John is—that is, having the individual concept John. Knowledge is made of concepts, and concepts are remnants of the past (whether stored in individual memory or “species memory”). Present knowledge always depends on past knowledge. Note that this is not regressive: the past knowledge is not knowledge-that but knowledge-what. To know that there is a cup on the desk you have to know what a cup is, but this latter knowledge is not a type of knowledge-that. If it were, knowledge would be impossible: for then the knowledge would depend on an antecedent piece of propositional knowledge, which would require another piece of propositional knowledge, and so on ad infinitum. But conceptual knowledge is not propositional knowledge, so there is no regress. The thesis being defended might more cautiously be stated as follows: all propositional knowledge presupposes non-propositional knowledge (which takes the form of memory). First we have conceptual knowledge (knowing-what), and then we have propositional knowledge (knowing-that). We never have a piece of propositional knowledge that is independent of our past epistemic state (this applies equally to self-knowledge). We never know anything about the present that derives wholly from the present.
This position stands opposed to acquaintance models of knowledge—theories that regard knowledge as simply a type of perception. But we can never know a proposition simply by being acquainted with its subject matter at a given time, as if this knowledge were splendidly cut off from the past. Seeing is not the same as knowing. For knowledge requires classification—bringing something under a concept—and concepts pre-date their exercise in acts of propositional knowledge. I can’t even know that I am in pain by pure acquaintance, since I need to apply the concept pain, and that concept stems from my epistemic storehouse. I must already know what pain is in order to judge and know that I am currently in pain. I can feel pain without any antecedent preparation, but I can’t know that I am in pain without already knowing something about pain. Such knowledge is a coordination of past and present not merely a momentary act of acquaintance. That idea is a myth—an empiricist myth—born of modeling knowledge too closely on perceptual experience. There is a deep distinction between knowledge and experience; the former is never a special case of the latter.
How does this conception of knowledge fit the case of the a priori? The answer is: not smoothly, but this shows something important about a priori knowledge. Suppose I judge that the number 3 is prime: to do this I need to know what a prime number is and what the number 3 is. Does my knowledge result from applying this prior knowledge to my current experience of 3? It does not: I don’t have any such experience. I am not presented with the number 3 and then dredge up my concept of that number, along with the concept of a prime. There is no analogue of perceptual experience to combine with antecedently possessed concepts. If I didn’t have the concept three, I couldn’t be confronted with the number 3. Numbers can only come before the mind as so conceptualized. There is no perception of numbers that leaves it open whether the object in question is a number, or the number it is. There is no pre-conceptual apprehension of numbers. Still, in order to make the judgment that 3 is prime I need to bring to bear my prior knowledge of numbers, i.e. my mathematical concepts. So the basic thesis applies to the case of a priori knowledge: propositional a priori knowledge depends on non-propositional conceptual knowledge. Knowing-that depends on knowing-what. It is just that we don’t have the combination of experience and cognition that we have in the empirical case. This is one way that the a priori differs from the a posteriori: such knowledge is not “by experience”, i.e. it requires no triggering perceptual input. Putting the point in terms of memory, propositional mathematical knowledge requires memory knowledge concerning what the objects of interest are—numbers, sets, geometrical forms. Thus mathematical knowledge involves a type of remembering, as empirical knowledge does: both are excavations of the cognitive past. The faculty of memory is being exercised as knowledge is acquired; and this is a deep-seated fact about human knowledge (if not all logically possible knowledge). Knowledge is not separable from memory. As memory takes us back to the past, so knowledge takes us back to the past—it is backwards looking. Remove someone’s memory and you remove his knowledge. New knowledge is inseparable from old knowledge. This is obvious for inductive knowledge, since we need to remember past observations, but it is also true for non-inductive empirical knowledge and for a priori knowledge. In acquiring any item of knowledge the past is always operative in the present. When knowledge is defined as true justified belief (or some such), not only is the justification typically derived from past observation; the very possibility of forming the belief in question derives from a prior epistemic state. Recollection is always in the picture. Both empiricism and nativism tacitly recognize this necessary involvement with the past: empiricism by holding that all knowledge depends on past experience in the form of stored perceptual encounters; nativism by locating the basis of knowledge in what is cognitively present at birth. Knowledge always has a history—in the individual and in the species. Only a kind of misplaced epistemic atomism could suppose otherwise—the idea of knowledge as an isolated quality of a time-slice. Perhaps experience can be conceived that way, but not knowledge. Realistically, knowledge always builds on the past, cumulatively and derivatively; it never descends from the sky and installs itself in the mind ab initio. Memory is what makes knowledge possible.
 The same can be said of other entities of which we have no experience: concepts themselves, platonic forms (the Good), and the self. In these cases we have no pre-conceptual experience of the entity in question, so that there is no presentation of these entities to the mind that leaves open what they are. We are never in doubt about what is before our mind. For example, we can never misidentify the Good as something else—an elephant or the Bad, say.
 As a corollary, skepticism about memory always dogs human knowledge: if our memories are radically in error, our knowledge is doomed. Memory is fallible, so knowledge based on it is always subject to doubt.