Because the imperative is something that stems from the rational will, adherence to it is really only adherence to the law that the will itself created. The will, that is, is acting as the will tells itself it has to. Due to the circular nature of the will's imperative to behave a certain way, obedience to the imperative is actually obedience to the self, and obeying moral law requires nothing more or less than complete self-direction. The premise that moral law -- the categorical imperative -- is born out of the rational will is central to Kant's theory of ethics: "each individual agent regards itself as determining, by its decision to act in a certain way, that everyone (including itself) will always act according to the same general rule in the future" (Kemerling 2002). This is why the imerative is categorical, or universal, and at the same time an entirely anonymous creation of each independent rational will. Though Kant's ability to employ logical and rhetorical devices is far beyond my own, I do not fully trust the conclusions he has come to. more precisely, I do not trust his assertion that the rational will is responsible for creating the categorical imperative, or that morality can be a truly autonomous act. Rationality is shaped by experience; it is impossible to know ro even conceive of anything in a vacuum. Determining how one should act is precisely a determination about external events, and rational decisions must be made with information that might be internalized, but was at some point external....

Imagine a pure innocent (with the sudden gift of language) presented with a moral question: is cheating wrong? The innocent needs to know what cheating is, what money is, how poker is played, the importance of money, etc. We can agree that cheating is wrong because we know the effects of being cheated and have a sense of private property, but these were not inborn, autonomous creations of our rationality but rather are experiential.
Kant argues that "a rational will, insofar as it is rational, is a will conforming itself to those laws valid for any rational will" (Johnson 2008, sec. 10). This simplification of his argument reveals another flaw -- Kant assumes that rationality, by virtue of being rational, will necessarily come to the same conclusions. While this might be true if both rational minds had access to precisely the same (no more, no less) information (again, a dependency on externalities), but this is never the case in actual human affairs. Two rational people faced with a purely rational problem can have two very different answers (look at a married couple deciding the fastest route somewhere and see). Thus moral behavior cannot be universal if it is always entirely rational, or vice versa.