One view might be, the past experience, which is remembered, together with a stimulus which immediately preceded the remembering. But this involves the possibility of causation at a distance, which (it will be said) is very difficult to maintain. Another view might involve a persistent mental trace (the formation of which was caused by the past experience which is remembered) together with the present stimulus. But again, it will be said, the notion of a "mental trace" is a very difficult one. We are left then with the possibility that it is a persistent physical trace, caused by the past experience, in the body of the person who remembers, together with the present stimulus. Since this trace is usually supposed to be in the brain, I shall refer to it as a "brain-trace". For lack of an alternative, then, we must accept the view that the knowing is caused by existence of the brain-trace plus the occurrence of the stimulus.

Now it is possible that the formation of the brain-trace might be caused, not by the past experience, but by, say, an operation by a clever surgeon.

If this is so, it is possible that a brain-trace,

exactly like that which would be produced by a past experience 348

H. P. GRICE :

of such and such a kind, might exist without any such experience having occurred.

It will further be possible that both the brain-

trace might exist and the stimulus might occur, without the past experience having occurred. But if both the brain-trace existed and the stimulus occurred, the memory-knowing would occur.

Therefore the memory-knowing might occur without the remembered experience having occurred. But that is logically impossible.

Therefore unless the argument is unsound one of

the premisses must be rejected; and the easiest premiss to reject is that memory-knowledge occurs.

Now I think the argument is unsound; but in order to show that it is I must distinguish more closely what the argument asserts, for I think there is an ambiguity in it, due to an ambiguity in the word "possible" which may mean either

"logically

possible "

"causally possible

Suppose, first, that

sible " means logically possible. Then the bare bones of ta

argument will be:

1. The existence of a brain-trace of kind A plus the occurrence of a stimulus of kind B is logically compatible with the non-occurrence of any experience of kind E
2. The existence of a brain-trace of kind A plus the occurrence of a stimulus of kind B causally involves the occurrence of a memory-knowing of an experience of kind E.

Therefore the occurrence of a memory-knowing of an experience of kind E is logically compatible with the non-occurrence of any experience of kind E.

But this is absurd; therefore either (1) is false, which is very, very improbable; or (2) is false, and the falsity of (2) will involve the falsity of the proposition that if these memory-knowings occur they are caused by the existence of a brain-trace plus the occurrence of a stimulus; or there are no memory-knowings, which seems the easiest alternative to accept.

But there is a suppressed premiss in the argument which is false. (Perhaps it is rather a principle than a premiss.) The argument should run:

1. The existence of a brain-trace of kind A plus the occurrence of a stimulus of kind B is logically compatible with the non-occurrence of any experience of kind E.
2. The existence of a brain-trace of kind A plus the occurrence of a stimulus of kind B causally involves the occurrence of a memory-knowing of an experience of kind E
3. For any propositions p, q, r, if p is logically compatible with q, and p causally implies r, then r is logically compatible with q. Therefore the occurrence of a memory-knowing of an experience of kind E is logically compatible with the non-occurrence of any experience of kind E.

   But (3) only has to be considered to be seen to be false. Let

   p = it has been raining, 9 = the ground is not wet, y = the

   ground is wet. Then p will be logically compatible with q, for it is logically possible that it should have been raining without the ground being wet; and p will causally imply r, for whenever it rains the ground does get wet; but q is clearly not logically compatible with r; for it cannot be true both that the ground is wet and that it is not wet.

   I conclude then that the argument in this form is unsound ; but before I pass on to the second form the argument might take, I ought to remark that it must not be supposed that I accept the views about the causes of memory-knowledge involved by the argument.

   Suppose now that "possible" means. " causally possible".

   The argument (including the suppressed premiss or principle will now run :

   1. The existence of a brain-trace of kind A plus the occurrence of a stimulus of kind B is causally compatible with the non-occurrence of any experience of kind E.
   2. The existence of a brain-trace of kind A plus the occurrence of a stimulus of kind B causally involves the occurrence of a memory-knowing of an experience of kind E.
   3. For any p, q, r, if p is causally compatible with q, and 1 causally implies r, then r is logically compatible with q-
   4. Therefore the occurrence of a memory-knowing of an experience of kind E is logically compatible with the non-occurrence of any experience of kind E.

   But this is absurd; therefore (as before) we must reject memory-knowledge.

   (3) is now, I think, true; but its gain is (1)'s loss.

   For

   the supporter of the argument is now committed to maintaining not that it is logically possible that a brain-trace of kind A should exist without the occurrence of an experience of kind E, but that it is causally possible that it should so exist. That means, I think, that he has got to maintain that there are conditions given which there would be a brain-trace of kind E without any experience of kind E having occurred; and in order to support this contention he must maintain, for example, that if a surgeon operated in a certain way he would produce the brain-trace, or give some other explanation how the brain-trace could be produced. But to maintain any such thing as this is something, I should have thought, that no reasonable man would be prepared to do. For I cannot see what evidence in favour of it he could possibly have.

   I do not then think that any real doubt has been cast on the occurrence of memory-knowledge; and it seems to me, therefore, that my theory is untouched by objections of the kind I have just discussed.