Page:Bergson - Matter and Memory (1911).djvu/235

 it recalls to us first an image A′ which is like it, because it is the recollection A′, and not the perception A, which really touches B in memory. However distant, then, we suppose the terms A and B from each other, a relation of contiguity can always be found between them, provided that the intercalated term A′ bears a sufficiently farfetched resemblance to A. This is as much as to say that between any two ideas chosen at random there is always a resemblance, and always, even, contiguity; so that, when we discover a relation of contiguity or of resemblance between two successive ideas, we have in no way explained why the one evokes the other.

What we really need to discover is how a choice is effected among an infinite number of recollections which all resemble in some way the present perception, and why only one of them,—this rather than that,—emerges into the light of consciousness. But this is just what associationism cannot tell us, because it has made ideas and images into independent entities floating, like the atoms of Epicurus, in an inward space, drawing near to each other and catching hold of each other when chance brings them within the sphere of mutual attraction. And if we try to get to the bottom of the doctrine on this point, we find that its error is that it intellectualizes ideas over much: it attributes to them a purely speculative rôle, believes that they exist for themselves and not for us, and overlooks the relation which they