[45] The fashion in which the term ‘feeling’ shifts about in psychology is notorious as a source of confusion. It would be convenient if it could be kept for pleasure-unpleasure, and used no longer as a synonym for ‘emotion’, since emotions can much more easily be regarded as built up from organic sensations. CHAPTER XIV Memory Within the surface of Time’s fleeting river Its wrinkled image lies, as then it lay Immovably unquiet, and for ever It trembles, but it cannot pass away! Shelley, _Ode to Liberty_. So far we have alluded only casually to memory, to that apparent revival of past experience to which the richness and complexity of experience is due. Every stimulus which is ever received leaves behind it, so it is said, an imprint, a trace capable of being revived later and of contributing its quota to consciousness and to behaviour. To these effects of past experience the systematic, the organised character of our behaviour is due; the fact that they intervene is the explanation of our ability to learn by experience. It is a way peculiar to living tissue by which the past influences our present behaviour across, as it might appear, a gulf of time. How we should conceive this influence is perhaps the most puzzling point in psychology. The old theory of a kind of Somerset House of past impressions has given place to an account in terms of facilitations of neural paths, lowered resistances in synapses, and so forth. It was natural that as the broad outlines of neural activity came to be known, psychologists should attempt to make use of them. But on close examination it is clear that their interpretations were far too crude. Fixed ‘paths’, one for every item of experience which has ever taken place, and others for every kind of connection into which the items come, however multitudinous we make them, no longer explain what can be observed in behaviour and experience. As Von Kries and, more recently, Koffka have insisted, the fact, for example, that we _recognise_ things in cases where it is certain that quite different paths must be involved, is fatal to the scheme. And mere multiplication of the entities invoked leads to no solution. Semon even goes so far as to say that when we listen to a song for the hundredth time we hear not only the singer but a chorus of nine and ninety mnemic voices. This corollary by itself is almost a refutation of his theory. We have to escape from the crude assumption that the only way in which what is past can be repeated is by records being kept. The old associationists supposed the records to be writ small inside separate cells. The more modern view was that they were scored large through a deepening of the channels of conduction. Neither view is adequate.