Empiricism, Memory, and Knowledge
In pre-Socratic times there was a school of thought known as “memorism” (or so I once dreamt). The principal doctrine of this school was that all knowledge is stored in memory: whenever you know something there was a past event that laid it down in memory, and knowledge is the recall of that something. Past event, storage, recall: these are the necessary and sufficient conditions of knowledge. The memorists opposed the orthodox school of thought (the “revelationists”) which held that all knowledge arises from direct communication with the gods: whenever you know something the gods are conveying it into your mind by speaking directly to you. The revelationists found the memorists impious in their reliance on memory, a human attribute, instead of the divine action of the gods, to whom we owe everything. There is no invocation of the past, no mysterious storing of information, no ecstatic recall experience, just good old-fashioned godly beneficence. Let the gods be praised! The two schools debated the matter at length, never coming to any firm resolution. The revelationists brought forward counterexamples: what about knowledge of the present and future—surely, we don’t know these things by memory? The memorists responded either by denying the existence of such knowledge (eliminative memorism) or by explaining present and future knowledge as special cases of memory knowledge (reductive memorism). Ingeniously, they contended that by the time knowledge is acquired the thing known is past and retained in memory, and that we only know the future by remembering the past (induction etc.). So, it was either memory naturalism or revelation supernaturalism. The memorists were gaining adherents as their anti-supernaturalism spread and flourished; the revelationists seemed mired in superstition and pseudo-explanation. The gods surely had other things to occupy their time, and anyway were not deemed “empirically verifiable”. Memorism seemed to cover the ground nicely, was rooted in everyday experience, and dispensed with ad hoc appeals to divinity. And indisputably, a vast amount of human knowledge simply is stored in memory—knowledge of history, geography, animal husbandry, who your friends and enemies are. The theory looked warranted by the plain facts of human psychology.
But a new school of thought was taking shape at around this time: this school sympathized with the memorist’s anti-supernaturalism but were troubled by apparent counterexamples to the central doctrine. What about our knowledge that everything is self-identical? Is that based on memory? Was there some past event that laid this information down in memory—say, seeing a bunch of self-identical things and making an inductive leap, or hearing it from a trusted teacher? We never seemed to have learned this truth—never made an observation of it or were taught it in school. Yet we knew it. And there is a lot more knowledge like this, as they quickly pointed out: all of geometry and arithmetic, logic, conceptual truths, ethical propositions, maybe even philosophical theories. No past event triggered and justified this type of knowledge; there was no experience of recall in entertaining it; and people never suffered from difficulties of recollection over it (“I know the answer to this, but I can’t quite bring it to mind”). Such knowledge simply doesn’t bear the marks of memory. Has anyone ever said “I just can’t remember whether everything is self-identical or not”? It thus appears that our knowledge exists in two places in our mind: in memory and in some other faculty not itself a type of memory. When asked what this faculty consists in, the anti-memorists grew dark and pensive: for no name suggested itself and the question was obscure. Some declared it an irresoluble mystery, while others (the “eternalists”) boldly asserted that such knowledge exists in the mind eternally (there was no moment of acquisition) and is a primitive fact of human nature. If we want a name for it, we can call it “un-memory” or “pre-memory”—in any case, it isn’t a form of memory in any normal sense. We just have this knowledge; it exists in the deepest recesses of our soul. It was never put there by anyone, divine or mortal, nor was it the result of an interaction with external reality. Remember, these were ancient times and evolution and genetic inheritance were unknown concepts. What this third school (they had no generally accepted name) was sure of was just that not all knowledge is memory knowledge. They opposed the idea that memory knowledge exhausts the whole of human knowledge; their positive theory, however, was still a work in progress. To be sure, much human knowledge is represented in memory, but there remains a substantial core of knowledge that is not so represented. Thus, there are really two types of knowledge; knowledge is not a homogeneous phenomenon. It has two species, two fundamental forms. They resisted the epistemological monism of the memorists.
Does all this ancient intellectual history remind you of anything? Is my dream a reflection of any actual history? Empiricism versus rationalism, of course: memorism is another version of empiricism and anti-memorism is the analogue of rationalism (or nativism). The memorist substitutes memory for experience: instead of saying that all knowledge derives from experience, he says that it is all dependent on memory. He thus sidesteps the standard problems with the concept of experience—whether it is conceptual or non-conceptual, given or interpreted, justificatory or epistemically idle, opaque or transparent—and replaces it with the concept of memory. Interactions with the environment lay down memories, which are later recalled; this is the source of all knowledge worthy of the name. Surely, something like this picture was implicit in traditional empiricism, since experiences had to be retained in memory in order to provide the basis of subsequent knowledge: you see something, remember it, and later recall it in an exercise of knowledge. In short, knowledge is sensory memory, according to empiricism. And rationalism is the denial of that: some knowledge (mathematics, etc.) is not memory knowledge of past interactions with the observable world; it has a different origin and modus operandi (which is hard to specify). Both empiricism and rationalism were opposed to the revelationists of their day; knowledge is not a gift from the gods (or God) but a fact of human natural psychology—an achievement of memory or a product of instinct. My pre-Socratic dream narrative thus mirrors the actual narrative of later philosophy (as Plato’s belief in innate knowledge anticipates Descartes and Leibniz). For some reason, the empiricists didn’t make memory salient, but it was hovering in the background: knowledge is experience remembered. The rationalists, by contrast, thought that not all knowledge consists in experience remembered—remembering past sensory interactions is not a part of knowledge arrived at by pure reason. The question being debated concerns the role of memory in knowledge, not so much the role of experience—whatever that might be exactly. We can even imagine a form of empiricism that eschews the concept of experience altogether, but still insists on the vital role of memory: perhaps there are just physical excitations of the sensory receptors (conscious experiences having been eliminated from the picture), and anyway we want to make room for subliminally acquired empirical knowledge that involves no conscious experience at all. Memory empiricism thus takes precedence over experience empiricism, theoretically speaking.
Putting traditional empiricism aside, the memory formulation affects our view of the distinction between a prioriand a posteriori knowledge. We can now reformulate this distinction in the obvious way: a posteriori knowledge is knowledge based on memory, while a priori knowledge is knowledge not based on memory. This works pretty well: the role of individual and collective memory in the formation of scientific and commonsense knowledge is acknowledged, while its irrelevance to typical instances of a priori knowledge is highlighted (it is not a type of historical knowledge in the broad sense). If anything, this puts a priori knowledge in a better light, because it sounds like pure dogma to insist that all knowledge proceeds from memory—memory is just one way of storing information. The genes store information, in animals and humans, which is then transmitted to offspring (we call the result “instinct”), but such storage is not the faculty normally labeled “memory”. Memory is just one way of possessing information (as well as skills and pre-dispositions); and there is little prima facieplausibility in the thesis that all knowledge (etc.) is contained in acquired memories. Memory is just one method of being informed, equipped, internally configured. And it sounds completely wrong to claim that logical and mathematical knowledge is arrived at by consulting one’s memory, as if it has a basis in historical records; logical and mathematical reasons are not time-bound in that way (“I remember seeing the law of non-contradiction for the first time when I was ten years old, and I have never forgotten it”). You don’t have to ransack your memory to decide if modus ponens is a sound logical rule, nor is there any danger of forgetting it. This kind of knowledge is completely different from knowledge of historical dates, or the route home, or the results of an experiment. Thus, the distinction between a priori and a posteriori knowledge has a firm and clear foundation, which helps establish the sui generis character of the a priori. Strangely enough, the place of memory in relation to the traditional distinction has not been much recognized (if at all).
The empiricist, whether experiential or memorial, puts space and time at the center of knowledge: you can only make pertinent sensory observations at certain times and in certain places. The doctrine might, indeed, be so defined: all knowledge rests on suitable spatiotemporal proximity to the thing known. But the rationalist points to types of knowledge not restricted in this way: we don’t need to be near numbers at a certain time of day in order to know about them (or logical truths or meanings). This type of knowledge is not dependent on spatiotemporal proximity to the thing known—hence the adequacy of the armchair in arriving at such knowledge. Nor is the subject matter naturally conceived as existing in space and time (what is a number such that we could be near it?). Here we find a marked contrast between the two types of knowledge. We really should not expect that a priori knowledge could be subsumed under the a posteriori umbrella. The empiricist is guilty of overgeneralizing from properties of knowledge characteristic of only certain types of knowledge—those dependent on sensory experience or memory. Such knowledge is only so good as the experiences that (allegedly) ground it, or the memory capacities that make it possible; but rational knowledge is free of these kinds of limitations, being neither experiential nor memorial. How it does work, however, is far from clear. All we can say is that considerations of space and time make no difference to the availability of a prioriknowledge.
 I discuss this in Inborn Knowledge (2015), 44-46.
 I realize that I have been writing and thinking about a priori knowledge for over fifty years, and I never tire of it, difficult though it is. It’s one of the things that got me into philosophy in the first place. I think most discussions of it over the last century have been pretty feeble—exercises in problem avoidance and tendentious stipulation. The nature of a priori knowledge is one of the Big Mysteries of philosophy.