It might, then, he supposed that the consciousness here finds a compensation for the role that has been withdrawn from it. If it is not the thing that creates the relation, it will be said, at least it is that which creates its efficacity of suggestion. Many psychologists have supposed that a relation has the power of evocation only when it has been perceived. The perception of resemblance precedes the action of resemblance. It is consequently the consciousness which assembles the ideas and gives them birth by perceiving their relations. This error, for it is one, has long been wide-spread--indeed, it still persists.[28] We have, however, no difficulty in understanding that the perception of a resemblance between two terms supposes them to be known; so long as only one of the terms is present to the consciousness, this perception does not exist; it cannot therefore possess the property of bringing to light the second term. Suggestion is therefore distinct from recognition; it is when suggestion has acted, when the resemblance in fact has brought the two terms together, that the consciousness, taking cognisance of the work accomplished, verifies the existence of a resemblance, and that this resemblance explains the suggestion. Second objection: we are told that the relations between the objects--that is, the principal categories--must be of a mental nature, because they are _a priori_. That they are _a priori_ means that they are at once anterior and superior to the experience. Let us see what this argument is worth. It appears that it is somewhat misused. With regard to many of the categories, we are content to lay down the necessity of an abstract idea in order to explain the comprehension of a concrete one. It is said, for example: how can it be perceived that two sensations are successive, if we do not already possess the idea of time? The argument is not very convincing, because, for every kind of concrete perception it is possible to establish an abstract category. It might be said of colour that it is impossible to perceive it unless it is known beforehand what colour is; and so on for a heap of other things. A more serious argument consists in saying that relations are _a priori_ because they have a character of universality and of necessity which is not explained by experience, this last being always contingent and peculiar. But it is not necessary that a function should be mental for it to be _a priori_. The identification of the _a priori_ with the mental is entirely gratuitous. We should here draw a distinction between the two senses of the _a priori_: anteriority and superiority.