Bergson would not think of denying that this intellectual method, in which facts are used as material for abstraction, is of the utmost practical use for explaining facts and so enabling us to control them. He suggests, however, that our preoccupation with these useful abstractions, classes and their relations, misleads us as to the facts themselves. What actually takes place, he thinks, is a kind of substitution of the explanation for the fact which was to be explained, analogous with what happens when a child at a party, or a guest at dinner, is misled about his actual sensations, only this substitution of which Bergson speaks, being habitual, is much harder to see through. Explanation, as we have seen, consists in constructing a plan or map in terms of such abstractions as classes and their relations, or sometimes, when the abstraction has been carried a step further, in terms simply of words or symbols, by means of which we represent the causal relations between such of the actual directly known facts as can be classified. This plan is more comprehensive and complete than the actual facts which we know directly in the ordinary course of things, for which it stands, and it enables us to explain these facts in terms of the classes of causes from which they follow, and the classes of effects which they produce. No explanation, of course, can actually acquaint us directly with the real antecedent or consequent facts themselves: it can only tell us to what classes these facts must belong. The terms of the plan by which we explain the facts, the classes, for instance, daylight and darkness, and their relation of alternation, or the words or symbols which stand for classes and relations are not themselves facts but abstractions. We cannot think in terms of actual facts: the intellectual activity by which we formulate general laws can only work among abstractions, and in order to explain a fact we are obliged to substitute for it either a class or word or other symbol. All description and explanation of facts consists in substitutions of this kind. The explanation applies provided the abstraction is based on fact, that is, provided it is possible to fit the fact to which the explanation is intended to apply into the class employed to explain it: the general law, for instance, about the alternation of the classes daylight and darkness will explain any facts which can be fitted into one or other of these classes, or again general laws about dogs, such as “dogs lick” will apply to whatever fact belongs at once to all the simpler classes, “warm,” “rough,” “of a certain size, and smell,” out of which the class “dog” is constructed. The general law itself, however, does not consist of such facts but of abstractions substituted for the facts themselves. Such substitution is extremely useful and perfectly legitimate so long as we keep firm hold of the fact as well, and are quite clear about what is fact and what only symbol.