The Third Part deals with the question, "How is metaphysics in general possible?" We have seen how both mathematics and pure natural science are possible, by appealing to our pure intuitions of time and space and the concepts of our faculty of understanding. We use our pure intuitions and our faculty of understanding to make sense of experience, but metaphysics, as its name suggests, deals with matters that are beyond the realm of experience. It either deals with concepts that lie outside of experience (like God) or it deals with the totality of possible experience (like whether the world has a beginning and an end). Intuition and understanding are of no use here. Metaphysics deals with the faculty of pure reason, and the ideas contained therein.
The distinction between the understanding and reason is crucial. Philosophical error frequently arises from a confusion of one for the other. Any concept that can be applied to experience belongs to the faculty of understanding and has nothing to do with metaphysics. Reason is not directed toward experience, and any attempt to apply the ideas of reason to experience is mistaken.
Reason tries to make experience complete. Reason tries to tie all of experience together and to give it meaning. This drive to metaphysics is not in itself problematic; it becomes wrong-headed only when we apply our pure intuitions or pure concepts of the understanding to the pursuit.
Kant distinguishes three different kinds of "ideas of reason"— psychological ideas, cosmological ideas, and the theological idea—that between them contain all of metaphysics. This summary will deal with psychological ideas, while the summary of sections 50–56 will deal with cosmological and theological ideas.
Psychological ideas try to identify some sort of substance or ultimate subject underlying all the predicates we can apply to a subject. For instance, we can describe a cat as "a thing with claws" or "a thing that purrs" and so on, but what is the "thing" itself? What do we have left over when we peel away all the predicates? Kant suggests that this search is futile: the understanding helps us make sense of experience by applying pure concepts to empirical intuitions, and concepts take the form of predicates. The only knowledge we can have comes in the form of predicates attached to subjects.
A possible candidate for ultimate subject comes in the form of the thinking ego, or soul. When describing internal states ("I think," or "I dream," for example), we refer back to an "I" that is fundamental, indivisible, and unique. However, Kant argues, this "I" is not a thing or a concept that we can have knowledge of in itself. That we are capable of experience at all suggests that we have some sort of consciousness, but we refer to this consciousness (or soul) without having any substantial knowledge of it.
Just as appearances in the external world suggest to us that there are things in themselves, so inner sensations suggest to us that we have some sort of soul or ego. But, just like things in themselves, we can know nothing about this soul; we can know only about the appearances that manifest themselves. This conclusion suggests that Descartes is wrong in thinking we can know about the mind better than we can know about external bodies.
What we can say about the soul we can say only in reference to our own experience. Thus, we can't know whether or not the soul is immortal, since that is to ask questions about the soul outside the realm of experience.
In the Prolegomena, Kant divides mental activity into three major faculties. First, there is the faculty of sensibility that uses the pure intuitions of space and time to form our sensations into empirical intuitions. This faculty helps us organize and make sense of what we see, hear, smell, touch, and taste. We can also use the pure intuitions of space and time to reason mathematically.
Second, there is the faculty of understanding that uses pure concepts to form our empirical intuitions into appearances. This faculty allows to make sense of what we see, hear, smell, touch, and taste as according itself with regular, universal laws of nature, and so helps us make general inferences and conclusions. That, effectively, is the business of natural science
Third, Kant introduces us to the faculty of reason in this part of the text. While the faculties of sensibility and understanding help us make sense of experience, reason helps us make sense of purely mental concepts. It does this by means of ideas, which try to fill out and give completeness to the concepts we apply in experience. For instance, psychological ideas take the concept of substance and try to flesh out what it is we mean when we talk about a "thing." What substance underlies a thing and makes it what it is? We will see similarly that the cosmological ideas try to fill out our concept of cause, trying to identify connections that go beyond those we encounter in experience, and that the theological idea tries to fill out our concept of community, trying to identify what unifies everything that exists.
Our faculty of sensibility gives us math, our faculty of understanding gives us science, and our faculty of reason gives us metaphysics. The important conclusion to draw from this discussion of faculties is that metaphysics is the product of pure reason and deals only with ideas in our head; in other words, metaphysics cannot tell us about how things are in themselves. Metaphysics as Kant conceives it is more a matter of untying mental knots than determining whether or not God exists.
The mental knot Kant associates with psychological ideas is that of substance, and particularly that of a thinking substance. Talk of substances was a major preoccupation of the rationalist metaphysics of the 17th and 18th centuries, and Descartes was one of the major philosophers to discuss substances. Descartes is famous for the statement "I think therefore I am": I cannot doubt that I exist, since the act of doubting is an act of thought and I couldn't think unless I existed. I exist: but what can I know about this "I" that I am? While I know that I think, I can doubt that I have a body (I could be a butterfly dreaming I have this body), so I conclude that I am a thinking (as opposed to bodily) substance. I may think I know, or I may guess, any number of things about my body, but while these thoughts or guesses may be mistaken, I cannot doubt that I am thinking or guessing. From this line of reasoning, I conclude that my mind is better known to me than my body.
And so on. In the Meditations, Descartes questions the reliability of the senses, and then attempts to see how much he can know about himself and the world around him using only his intellect.
According to Kant, all I can know about this "I" is that I am. What I sense and think are representations, and these representations have to take place within a subject. For things to be seen and heard there has to be a consciousness that is doing the seeing and hearing. This "I" essentially represents that logical necessity: there must be something doing the seeing and hearing, and I call that something "I."
This "I" is not something I encounter in experience; it is the basis for my experience. As a result, we cannot apply to it the categories we apply to experience. Descartes tries to do essentially that, applying the concept of substance and other concepts of the pure understanding to it. Kant suggests on the contrary that we should think of this "I" the way we think of things in themselves: we can infer that it is, but we cannot infer anything about it. Pure reason, engaging in metaphysics, cannot tell us anything substantial about the way things are.