Kant's doctrine of freedom includes practical freedom (which is not, according to his definition, opposed to causality) and transcendental freedom; he seems, however, not always to have kept the distinction between the two clearly in view. His theory of transcendental freedom is grounded upon the doctrine of the pure ideality of time. The only method of saving the doctrine of freedom is, according to Kant, the theory that the law of necessity applies to things as phenomena but not to things in themselves. If phenomena are not to be regarded as things in themselves, but as mere thought-images, they must themselves have reasons which are not phenomena. Such a cause for pure reason[80] is not determined by phenomena, although its effects appear as phenomena. The causal action of reason does not have a beginning in time, but is the constant condition, outside time, of all free action of the will. Kant failed, however, to prove the pure ideality of time, as Riehl has sufficiently shown. Moreover, were the _intelligibile_ character of reason the cause of action as phenomenon, there would be no possibility of moral improvement, since the noumenon is not affected by phenomena,--an inference which Schopenhauer makes in adopting Kant's theory. Moreover, if space has, as Kant also assumes, transcendental ideality, plurality is not conceivable; hence, the moral difference of characters, and the science of Ethics itself, could have no transcendental significance. It is evident that Kant argues from the standpoint of an assumption of a "soul-thing," a constant "substratum" of psychical phenomena,--a standpoint which he himself criticizes. He identifies this thing-in-itself, moreover, with the reason, although he himself declares that the concept of the thing-in-itself is but a concept limiting reason.[81] He makes the reason a thing-in-itself outside time, although it is an activity, a process of consciousness in time. The thought of duty, of the categorical imperative, is a phenomenon, and if the will is determined thereby, it is determined by something in time. Kant takes but little account, moreover, of the fact of birth. Is the _intelligibile_ character born? If so, it is preceded by something in time; if not, it must be eternal, existing before birth as well as after death. And how can he assert, too, that an action might have been other than it was, if it depended upon the constitution of the _intelligibile_ character, and this is as it is, and operates as it is? Schopenhauer's argument for transcendental freedom contains many self-contradictions, and is founded on the fiction of a first free choice of character. Schopenhauer asserts, however, that character is innate. If so, how is it chosen? The theory assumes that one is before he is. An act of choice presupposes a chooser, and, according to his own words, "Every _existentia_ presupposes an _essentia_"; that is, every existence must have a particular being, essence.