There are only two possible ways in which synthetical representation and its objects can coincide with and relate necessarily to each other, and, as it were, meet together. Either the object alone makes the representation possible, or the representation alone makes the object possible. In the former case, the relation between them is only empirical, and an a priori representation is impossible. And this is the case with phenomena, as regards that in them which is referable to mere sensation. In the latter case- although representation alone (for of its causality, by means of the will, we do not here speak) does not produce the object as to its existence, it must nevertheless be a priori determinative in regard to the object, if it is only by means of the representation that we can cognize anything as an object. Now there are only two conditions of the possibility of a cognition of objects; firstly, intuition, by means of which the object, though only as phenomenon, is given; secondly, conception, by means of which the object which corresponds to this intuition is thought. But it is evident from what has been said on aesthetic that the first condition, under which alone objects can be intuited, must in fact exist, as a formal basis for them, a priori in the mind. With this formal condition of sensibility, therefore, all phenomena necessarily correspond, because it is only through it that they can be phenomena at all; that is, can be empirically intuited and given. Now the question is whether there do not exist, a priori in the mind, conceptions of understanding also, as conditions under which alone something, if not intuited, is yet thought as object. If this question be answered in the affirmative, it follows that all empirical cognition of objects is necessarily conformable to such conceptions, since, if they are not presupposed, it is impossible that anything can be an object of experience. Now all experience contains, besides the intuition of the senses through which an object is given, a conception also of an object that is given in intuition. Accordingly, conceptions of objects in general must lie as a priori conditions at the foundation of all empirical cognition; and consequently, the objective validity of the categories, as a priori conceptions, will rest upon this, that experience (as far as regards the form of thought) is possible only by their means. For in that case they apply necessarily and a priori to objects of experience, because only through them can an object of experience be thought.
The whole aim of the transcendental deduction of all a priori conceptions is to show that these conceptions are a priori conditions of the possibility of all experience. Conceptions which afford us the objective foundation of the possibility of experience are for that very reason necessary. But the analysis of the experiences in which they are met with is not deduction, but only an illustration of them, because from experience they could never derive the attribute of necessity. Without their original applicability and relation to all possible experience, in which all objects of cognition present themselves, the relation of the categories to objects, of whatever nature, would be quite incomprehensible.
The celebrated Locke, for want of due reflection on these points, and because he met with pure conceptions of the understanding in experience, sought also to deduce them from experience, and yet proceeded so inconsequently as to attempt, with their aid, to arrive it cognitions which lie far beyond the limits of all experience. David Hume perceived that, to render this possible, it was necessary that the conceptions should have an a priori origin. But as he could not explain how it was possible that conceptions which are not connected with each other in the understanding must nevertheless be thought as necessarily connected in the object- and it never occurred to him that the understanding itself might, perhaps, by means of these conceptions, be the author of the experience in which its objects were presented to it- he was forced to drive these conceptions from experience, that is, from a subjective necessity arising from repeated association of experiences erroneously considered to be objective- in one word, from habit. But he proceeded with perfect consequence and declared it to be impossible, with such conceptions and the principles arising from them, to overstep the limits of experience. The empirical derivation, however, which both of these philosophers attributed to these conceptions, cannot possibly be reconciled with the fact that we do possess scientific a priori cognitions, namely, those of pure mathematics and general physics.
The former of these two celebrated men opened a wide door to extravagance- (for if reason has once undoubted right on its side, it will not allow itself to be confined to set limits, by vague recommendations of moderation); the latter gave himself up entirely to scepticism- a natural consequence, after having discovered, as he thought, that the faculty of cognition was not trustworthy. We now intend to make a trial whether it be not possible safely to conduct reason between these two rocks, to assign her determinate limits, and yet leave open for her the entire sphere of her legitimate activity.
I shall merely premise an explanation of what the categories are. They are conceptions of an object in general, by means of which its intuition is contemplated as determined in relation to one of the logical functions of judgement. The following will make this plain. The function of the categorical judgement is that of the relation of subject to predicate; for example, in the proposition: "All bodies are divisible." But in regard to the merely logical use of the understanding, it still remains undetermined to which Of these two conceptions belongs the function Of subject and to which that of predicate. For we could also say: "Some divisible is a body." But the category of substance, when the conception of a body is brought under it, determines that; and its empirical intuition in experience must be contemplated always as subject and never as mere predicate. And so with all the other categories.