We know a great many things concerning ourselves which we cannot know nearly so directly concerning animals or even other people. We know when we have a toothache, what we are thinking of, what dreams we have when we are asleep, and a host of other occurrences which we only know about others when they tell us of them, or otherwise make them inferable by their behaviour. Thus, so far as knowledge of detached facts is concerned, the advantage is on the side of self-knowledge as against external observation. But when we come to the analysis and scientific understanding of the facts, the advantages on the side of self-knowledge become far less clear. We know, for example, that we have desires and beliefs, but we do not know what constitutes a desire or a belief. The phenomena are so familiar that it is difficult to realize how little we really know about them. We see in animals, and to a lesser extent in plants, behaviour more or less similar to that which, in us, is prompted by desires and beliefs, and we find that, as we descend in the scale of evolution, behaviour becomes simpler, more easily reducible to rule, more scientifically analysable and predictable. And just because we are not misled by familiarity we find it easier to be cautious in interpreting behaviour when we are dealing with phenomena remote from those of our own minds: Moreover, introspection, as psychoanalysis has demonstrated, is extraordinarily fallible even in cases where we feel a high degree of certainty. The net result seems to be that, though self-knowledge has a definite and important contribution to make to psychology, it is exceedingly misleading unless it is constantly checked and controlled by the test of external observation, and by the theories which such observation suggests when applied to animal behaviour. On the whole, therefore, there is probably more to be learnt about human psychology from animals than about animal psychology from human beings; but this conclusion is one of degree, and must not be pressed beyond a point.