The argument from the connection of brain-lesions with loss of memory is not so strong as it looks, though it has also, some weight. What we know is that memory, and mnemic phenomena generally, can be disturbed or destroyed by changes in the brain. This certainly proves that the brain plays an essential part in the causation of memory, but does not prove that a certain state of the brain is, by itself, a sufficient condition for the existence of memory. Yet it is this last that has to be proved. The theory of the engram, or any similar theory, has to maintain that, given a body and brain in a suitable state, a man will have a certain memory, without the need of any further conditions. What is known, however, is only that he will not have memories if his body and brain are not in a suitable state. That is to say, the appropriate state of body and brain is proved to be necessary for memory, but not to be sufficient. So far, therefore, as our definite knowledge goes, memory may require for its causation a past occurrence as well as a certain present state of the brain. In order to prove conclusively that mnemic phenomena arise whenever certain physiological conditions are fulfilled, we ought to be able actually to see differences between the brain of a man who speaks English and that of a man who speaks French, between the brain of a man who has seen New York and can recall it, and that of a man who has never seen that city. It may be that the time will come when this will be possible, but at present we are very far removed from it. At present, there is, so far as I am aware, no good evidence that every difference between the knowledge possessed by A and that possessed by B is paralleled by some difference in their brains. We may believe that this is the case, but if we do, our belief is based upon analogies and general scientific maxims, not upon any foundation of detailed observation. I am myself inclined, as a working hypothesis, to adopt the belief in question, and to hold that past experience only affects present behaviour through modifications of physiological structure. But the evidence seems not quite conclusive, so that I do not think we ought to forget the other hypothesis, or to reject entirely the possibility that mnemic causation may be the ultimate explanation of mnemic phenomena. I say this, not because I think it LIKELY that mnemic causation is ultimate, but merely because I think it POSSIBLE, and because it often turns out important to the progress of science to remember hypotheses which have previously seemed improbable. LECTURE V. PSYCHOLOGICAL AND PHYSICAL CAUSAL LAWS