Transcendental idealism, the doctrine that things in space are appearances, offers the possibility of avoiding idealism. Dogmatic idealism is undermined because its arguments depend on treating matter and space as things in themselves. Problematic idealism might be avoided if one could find a way of proving the existence of bodies without having to resort to a dubious causal inference. In the Fourth Paralogism, Kant sought to show that bodies are known immediately, just as we know our own mental states immediately through inner intuition.
Our inner intuitions are representations of ourselves as thinking. Kant argued on Cartesian lines that the presence of these representations is known immediately. Our outer intuitions are representations of bodies, and qua representation, they are known immediately as well. But we do not infer from these representations that bodies exist as their cause; rather, the representations are identified with the bodies. Bodies, as appearances, are nothing more than "species of our representations" and hence are known immediately.
Kant exploited the thesis that as species of representations, bodies are on the same footing as our inner intuitions of ourselves, in order to solve the Cartesian mind/body problem. This problem depends once again on transcendental realism. If body and soul are independently existing beings with nothing in common, it is impossible to understand how they could influence each other. One is driven to extreme positions such as the pre-established harmony (Leibniz) or supernatural intervention (Malebranche). But once it is recognized that both have the status of appearance, we can subsume both under causal laws, which after all are only rules for the ordering of appearances. What underlies the appearances is unknown to us, but it is neither body nor thinking thing. (I will return to this point shortly.) To the student of the history of philosophy, this "solution" to the problem of idealism (thopugh not to the mind/body problem) seems to be exactly that proposed by Berkeley, whom Kant characterized as a dogmatic idealist. But in his early works Principles of Human Knowledge and Three Dialogues between Hylas and Philonous, Berkeley had argued that the esse of corporeal things is percipi. This can fairly be translated into Kant's jargon by stating that bodies are a species of representation. Thus, Kant's first critics held that transcendental idealism is a "freshened up" version of Berkeleyan idealism.
Kant was furious with this characterization of his idealism, accusing his critics of deliberate misrepresentation of his position. In the Prolegomena, he tried to dissassociate himself from Berkeley in two ways. First, (in the Appendix) he noted that there is no a priori element in Berkeley's philosophy. Allison has argued that this makes Berkeley a transcendental realist, and hence open to being called a dogmatic idealist, as in the second edition of the Critique. Be this as it may, it does affect Kant's own description of his own idealism.
In Remarks II and III of Part One, Kant confronted the issue directly, and dangerously. Genuine idealists hold that the only real beings are thinking beings, everything else being representations in the thinking beings. Kant, on the other hand, denies this thesis. Appearances are appearances of things in themselves, so that what we call bodies exist not merely as representations (as stressed in the Fourth Paralogism), but as things in themselves. "I grant by all means that there are bodies without us, that is, things which, though quite unknown to us as to what they are in themselves, we yet know by the representations which their influence on our sensibility procures us. These representations we call 'bodies,' a term signifying merely the appearance of the thing which is unknown to us, but not therefore less actual. Can this be termed idealism? It is the very contrary" (Ak. 290). A vital feature of this claim is that the unknown thing is said to "influence" our sensibility, and as such, we can only postulate it as a cause. But then it seems that we are back to the original problem, i.e., the dubiousness of causal inference. Kant might respond that what is dubious in causal inference is not that there is a cause, but what the cause is. And he has admitt3ed that the cause of our representations of bodies is unknown. "Neither the transcendental object which underlies outer appearances nor that which underlies inner intuition, is in itself either matter or a thinking being, but a ground (to us unknown) of the appearances which supply to us the empirical concept of the former as well as of the latter mode of existence" (A379-80). Unfortunately, Kant had also claimed that the concept of causality can be justifiably applied only to objects of experience, as the condition of rule-governed change of the states of empirical objects. Thus his appeal to the unknown cause of our representations falls victim to his limitation of the use of our understanding to experience.
Whether or not Kant was aware of this conflict within his system, he tried yet another approach in the second edition of the Critique. In his "Refutation of Idealism," Kant did not appeal to the transcendental ideality of space and objects in space. I will approach the argument as an attempt to improve the original, by filling a gap in the argument of the Fourth Paralogism.
Even if we regard bodies as dependent on space, and space as a form of intuition, it does not quite follow that bodies are empirically real. What is required is sensation, which is the material element of intuition. Further, the second Postulate of Empirical Thought holds that existence is known through perception and its connection through causal laws. But perception stemming from sensation must be distinguished from imagination, for we can imagine objects in space as well as perceive them. Kant had stated in the Fourth Paralogism that perception is a condition of imagination, but he did not argue for this claim. The Refutation of Idealism provides an argument to the conclusion that the existence of objects in space is a necessary condition for us to be able to describe ourselves as having thoughts in time.
Whereas in the Fourth Paralogism representations of bodies are placed on a par with representations of ourselves, in the Refutation, representation of things in space are in a sense made prior to the representation of ourselves, thus reversing the argument of skeptical idealism. The key to the argument is the claim that we are conscious of ourselves as determined in time. We represent ourselves as having mental states that occur in a definite order, one after another. But, Kant continues, this order cannot be determined by reference to ourselves alone; we require a reference to something enduring, and only spatial objects fit that bill. Thus we see that perception is a necessary condition for any self-representation: "The consciousness of my existence in time is bound up in the way of identity with the cojsciousness of a relation to something outside me, and it is therefore experience, not invention, sense not imagination, which inseparably connects this outside something with my inner sense" (Bxlii).
This argument is puzzling. One must first ask why mere inner intuition is not sufficient for the determination of the order of one's mental states. Is there not awareness of a before and an after in dreams, for example? Another question concerns the correctness of the primary premise, that we are conscious of ourselves as determined in time. If the premise simply means that we are aware of a sequence of perceptions, it is hard to see why a permanent is necessary for such awareness. On the other hand, if it means that we are able to assign a definite place to each of our states of self-consciousness, then it is questionable whether the premise is true. At the very least, it seems that this is an activity which requires some comparison between those states of which we are conscious and permanent objects. But then are we conscious of ourselves as determined in time?