Page:Mind (New Series) Volume 8.djvu/177

 SOME BEMABKS ON MEMORY AND INFERENCE. 163 ing-point and the way and this surely is erroneous. The normal type of inference is surely the unbroken development of an identical subject, which does not leave the mind by the way and which, therefore, cannot possibly be remembered. This is the normal type, and I will add that, so far as this fails to be present, the operation is really not an inference. 1 With this I must pass from the subject of memory's falli- bility. I will add some words on the question which has been raised about Belief. Memory, we saw, takes its ideas of the past as real, while in mere imagination there is no such claim. It is the addition of belief, then, we hear it said, which turns imagination into memory, and our main task is to find in what this addition consists, or at least to set it down as " a final inexplicability ". But the whole question is in this way mis- understood and the issue radically perverted. To take for granted the existence of "mere ideas" as self-evident and as a matter of course, and to treat belief in these as something supervening, or even adventitious, which we have then got to explain, is fundamentally erroneous. It is to make an assump- tion quite false in its principle and in its consequences most misleading. The presence of and the possibility of these " mere ideas " is, on the contrary, the very thing which calls most for explanation. No such ideas, we may say with con- fidence, can possibly exist in an early mind. To entertain an idea in which you do not believe, a suspended idea held in separation from the presented reality, is a late and, when we reflect, is an enormous mental achievement. It implies a disruption of that immediate unity of theory and practice which is at first throughout prevalent and is also necessary. 1 Even in an indirect argument where I divide A into A6 and Ac, and then by disproving Ac prove A 6, I do not in the operation depend upon memory. Certainly at the end of my disproof of Ac I may have forgotten A6, but I then return to the beginning with the knowledge that A is not c, and now with that in my mind reach the conclusion A6 from A. The knowledge that A is not c does not here depend on memory. It might so depend if, e.g., I had merely found in my notes that I had one day proved Ac to be false, and if I used that bare result. But so far that result obviously does not pretend to be itself made in my inference at all. And with direct reasoning it seems clear that, so far as the subject has lapsed from the mind by the way, there is properly no inference. The operation, to become an inference, must in some form be repeated without that lapse. The retention of an identical content before the mind, and the assumption that where I have seen no difference by the way there is no difference, can neither of them be called memory except by an abuse of language. The points raised by Prof. Ladd are certainly well worth raising and discussing, but his treatment of them seems not satisfactory.