Kant fails to notice this second difficulty,[7] and up to a certain point avoids it owing to his distinction between the imagination and the understanding. For he thinks of the understanding as the source of general principles of synthesis, viz. the categories, and attributes individual syntheses to the imagination. Hence the individual syntheses, which involve particular principles, are already effected before the understanding comes into play. But to throw the work of effecting individual syntheses upon the imagination is only to evade the difficulty. For in the end, as has been pointed out,[8] the imagination must be the understanding working unreflectively, and, whether this is so or not, some account must be given of the way in which the imagination furnishes the particular principles of synthesis required.
[7] We should have expected Kant to have noticed this difficulty in A. 105, Mah. 199, where he describes what is involved in the relation of representations to an object, for his instance of representations becoming so related is the process of combining elements into a triangle, which plainly requires a synthesis of a very definite kind. For the reasons of his failure to notice the difficulty cf. p. 207.
The third and last main difficulty of the first kind concerns the relation of the elements of the manifold and the kinds of synthesis by which they are combined. This involves the distinction between relating in general and terms to be related. For to perform a synthesis is in general to relate, and the elements to be combined are the terms to be related.[9] Now it is only necessary to take instances to realize that the possibility of relating terms in certain ways involves two presuppositions, which concern respectively the general and the special nature of the terms to be related.
[9] 'To relate' is used rather than 'to recognize as related', in order to conform to Kant's view of knowledge. But if it be desired to take the argument which follows in connexion with knowledge proper (cf. p. 242), it is only necessary to substitute throughout 'to recognize as related' for 'to relate' and to make the other changes consequent thereon.
In the first place, it is clear that the general nature of the terms must correspond with or be adapted to the general nature of the relationship to be effected. Thus if two terms are to be related as more or less loud, they must be sounds, since the relation in question is one in respect of sound and not, e. g., of time or colour or space. Similarly, terms to be related as right and left must be bodies in space, right and left being a spatial relation. Again, only human beings can be related as parent and child. Kant's doctrine, however, does not conform to this presupposition. For the manifold to be related consists solely of sensations, and of individual spaces, and perhaps individual times, as elements of pure perception; and such a manifold is not of the kind required. Possibly individual spaces may be regarded as adequate terms to be related or combined into geometrical figures, e. g. into lines or triangles. But a house as a synthesis of a manifold cannot be a synthesis of spaces, or of times, or of sensations. Its parts are bodies, which, whatever they may be, are neither sensations nor spaces nor times, nor combinations of them. In reality they are substances of a special kind. Again, the relation of cause and effect is not a relation of sensations or spaces or times, but of successive states of physical things or substances, the relation consisting in the necessity of their succession.
