"Memory is necessary for all the operations of reason." (Blaise Pascal, Pensées, Krailsheimer, #651)
This seems right. Consider this quick little argument against scientism, the philosophical, not scientific, view that all knowledge is natural-scientific knowledge:
1. I know by reason alone, a priori, and not by any natural-scientific means, that addition has the associative and the commutative properties and that these properties are distinct.
2. If scientism is true, then it is not the case that (1).
3. Scientism is not true.
I grasp (understand) this argument and its validity by reason. To grasp any such argument, it is not sufficient that a succession of conscious states transpire in my mental life. For if the state represented by (1) falls into oblivion by the time I get to (2), and (2) by the time I get to (3), then all I would undergo would be a succession of consciousnesses but not the consciousness of succession. But the consciousness of succession is necessary to 'take in' the argument. And this consciousness of succession itself presupposes a kind of memory. To grasp the conclusion as a conclusion -- and thus as following from the premises -- I have to have retained the premises. There has to be a diachronic unity of consciousness in which there is a sort of synopsis of the premises together with the conclusion with the former entailing the latter.
But of course something similar holds for each proposition in the argument. The meaning of a compound proposition is built up out of the meanings of its propositional parts, and the meaning of a simple proposition is built up out of the meanings of its sub-propositional parts, and these meanings have to be retained as the discursive intellect runs through the propositions. ('Discursive' from the L. currere, to run.) This retention -- a term Husserl uses -- is a necessary condition of the possibility of understanding.
And so while I do not grasp an argument by memory (let alone by sense perception or introspection), memory is involved in rational knowledge.
The Pascalian aphorism bears up well under scrutiny.
Example of associativity of addition: (7 + 5) + 3 = 7 + (5 + 3). Example of commutativity: (7 + 5) + 3 = (5 + 7) + 3. The difference between the two properties springs to the eye (of the mind). Now what must mind be like if it is to be capable of a priori knowledge? Presumably it can't just be a hunk of meat.
But if the below companion post is right, not even sense knowledge is such that its subject could be a hunk of meat. We are of course meatheads. But squeezing meaning out of mere meat -- there's the trick!
We should look past useless memories to present realities in the way we look past the floaters in our visual field. To concentrate on the detritus of memory is only to enliven what ought to be left to slumber.
Passing a lady in the supermarket I catch a whiff of patchouli. Her scent puts me in mind of hippy-trippy Pamela from the summer of '69. An olfactory stimulus in the present causes a memory, also in the present, of an event long past, a tête-à-tête with a certain girl. How ordinary, but how strange! Suddenly I am 'brought back' to the fantastic and far-off summer of '69. Ah yes! What is memory and how does it work? How is it even possible?
Let's start with the 'datanic' as I like to say:
1. There are (veridical) memories through which we gain epistemic access to the actual past, to events that really happened. The above example is a case of episodic personal memory. I remember an event in my personal past. To be precise, I remember my having experienced an event in my personal past. My having been born by Caesarean section is also an episode from my personal past, and I remember that that was my mode of exiting my mother's body; but I don't remember experiencing that transition. So not every autobiographical memory is a personal episodic memory. The latter is the only sort of memory I will be discussing in this post. The sentence in boldface is the nonnegotiable starting point of our investigation.
We now add a couple of more theoretical and less datanic propositions, ones which are not obvious, but are plausible and accepted by many theorists:
2. Memory is a causal notion. A mental image of a past event needn't be a memory of a past event. So what makes a mental image of a past event a memory image? Its causal history. My present memory has a causal history that begins with the event in 1969 as I experienced it.
3. There is no action at a temporal distance. There is no direct causation over a temporal gap. There are no remote causes; every cause is a proximate cause. A necessary ingredient of causation is spatiotemporal contiguity. So while memory is a causal notion, my present memory of the '69 event is not directly caused by that event. For how could an event that no longer exists directly cause, over a decades-long temporal gap, a memory event in the present? That would seem to be something 'spooky,' a kind of magic.
Each of these propositions lays strong claim to our acceptance. But how can they all be true? (1) and (2) taken together appear to entail the negation of (3). How then can we accommodate them all?
Memory trace theories provide a means of accommodation. Suppose there are memory traces or engrams engraved in some medium. For materialists this medium will have to be the brain. One way to think of a memory trace is as a brain modification that was caused at the time of the original experience, and that persists since that time. So the encounter with Pam in '69 induced a change in my brain, left a trace there, a trace which has persisted since then. When I passed the patchouli lady in the supermarket, the olfactory stimulus 'activated' the dormant memory trace. This activation of the memory trace either is or causes the memory experience whose intentional object is the past event. With the help of memory traces we get causation wthout action at a temporal distance.
(Far out, man!)
The theory or theory-schema just outlined seems to allow us to uphold each of the above propositions. In particular, it seems to allow us to explain how a present memory of a past event can be caused by the past event without the past event having to jump the decades-long temporal gap between event remembered and memory. The memory trace laid down in '69 by the original experience exists in the present and is activated in the present by the sensory stimulus. Thus the temporal contiguity requirement is satisfied. And if the medium in which the memory traces are stored is the brain or central nervous system, then the spatial contiguity requirement is also satisfied.
Question: Could memory traces play merely causal roles?
Given (2) and (3), it seems that memory traces must be introduced as causal mediators between past and present. But could they be just that? Or must they also play a representational role? Intuitively, it seems that nothing could be a memory trace unless it somehow represented the event of which it is a trace. If E isthe original experience, and T is E's trace, then it it seems we must say that T is of E in a two-fold sense corresponding to the difference between the subjective and objective genitive. First, T is of E in that T is E's trace, the one that E caused. Second, T is of E in that T represents E.
It seems obvious that a trace must represent. In my example,the sensory stimulus (the whiff of patchouli) is not of or about the '69 event. It merely activates the trace, rendering the dispositional occurrent. But the memory is about the '69 event. So the aboutness must reside in the trace. The trace must represent the event that caused it -- and no other past event. The memory represents because the trace represents. If the trace didn't represent anything, how could the memory -- which is merely the activation of the trace or an immediate causal consequence of the activation of the trace -- represent anything? How a persisting brain modification -- however it is conceived, whether it is static or dynamic, whether localized or nonlocalized -- can represent anything is an important and vexing question but one I will discuss in a later post.
Right now I want to nail down the claim that memory traces cannot play a merely causal role, but must also bear the burden of representation.
Suppose a number of strangers visit me briefly. I want to remember them, but my power of memory is very weak and I know I will not remember them without the aid of some mnemonic device. So I have my visitors leave calling cards. They do so, except that they are all the same, and all blank (white). These blank cards are their traces, one per visitor. The visitors leave, but the cards remain behind as traces of their visit. I store the cards in a drawer. I 'activate' a card by pulling it out of storage and looking at it. I am then reminded (at most) that I had a visitor, but not put in mind of any particular visitor such as Tom. So even if the card in my hand was produced by Tom, that card is useless for the purpose of remembering Tom. Likewise for every other card. Each was produced by someone in particular and only by that person; but none of them 'bring back' any particular person.
Bear in my mind that I don't directly remember any of my visitors. My only memory access to them is via their traces, their calling cards. For the visitors are long gone just like the '69 experience. So the problem is not merely that I don't know which card is from which person; the problem is that I cannot even distinguish the persons.
Had each visitor left a differently colored card, that would not have helped. Nor are matters helped if each visitor leaves a different sort of trace; a bottle cap, a spark plug, a lock of hair, a guitar pick. Even if Tom is a guitar player and leaves a guitar pick, that is unhelpful too since I have no access to Tom except via his trace.
So it doesn't matter whether my ten visitors leave ten tokens of the same type, or ten tokens each of a different type. Either way I won't be able to remember them via the traces they leave behind. Clearly, what I need from each visitor is an item that uniquely represents him or her -- as opposed to an item that is merely caused to be in my house by the visitor. Suppose Tom left a unique guitar pick, the only one of its kind in existence. That wouldn't help either since no inspection of that unique pick could reveal that it was of Tom rather than of Eric or Eric's cat. Ditto if Tom has signed his card or his pick 'Tom Riff.' That might be a phony name, or the name of him and his guitar -- doesn't B. B . King call his guitar 'Lucille'?
If I can remember that it was Tom who left the guitar pick, then of course I don't need the guitar pick to remember Tom by. I simply remember Tom directly without the need for a trace. On the other hand, if I do need a trace in order to remember long gone Tom, then that trace must have representational power: it cannot be merely something that plays a causal role.
Traces theories have to avoid both circularity and vicious infinite regress.
Circularity. To explain the phenomenon of memory, the trace theory posits the existence of memory traces. But if the explanation in terms of traces ends up presupposing memory, then the theory is circular and worthless. If what makes the guitar pick a trace of Tom is that I remember that Tom left it, then the explanation is circular. Now consider the trace T in my brain which, when activated by stimulus S causes a memory M of past experience E. M represents E because T represents E. What makes T represent E? What makes the memory trace caused by the encounter with Pam in '69 represent Pam or my talking with her? The answer cannot be that I remember the memory trace being caused by the encounter with Pam. For that would be blatantly circular. Besides, memory traces in the brain are not accessible to introspection.
Infinite Regress. Our question is: what makes T represent E and nothing else? To avoid circularity one might say this: There is a trace T* which records the fact of E's production of T, and T represents E in virtue of T*. But this leads to a vicious infinite regress. Suppose Sally leaves a photo of herself. How do I know that the photo is of Sally and not of her sister Ally? If you say that I directly remember Sally and thereby know that the photo is unambiguously of her, then you move in a circle. You may as well just say that we remember directly and not via traces. So, to hold onto the trace theory, one might say the following: There is a photo of Sally and her photograph, side by side. Inspection of this photo reveals that that the first photo is of Sally. But this leads to regress: what makes the second photo a photo of the first?
Conclusion: To avoid both circularity and infinite regress, memory traces must possess intrinsic representational power. Their role cannot be merely causal.
A later post will then address the question whether memory traces could have intrinsic representational power. If you are a regular reader of this blog you will be able to guess my answer.
REFERENCE: John Heil, "Traces of Things Past," Philosophy of Science, vol. 45, no. 1 (March 1978), pp. 60-72. My calling card example above is a reworking of Heil's tennis ball example.
It is we who supply the blood that enlivens the spectral vampires that haunt us from our past. A part of mind control is purgation of memory, and without mind control happiness is achieved with difficulty, if at all.