In Kant’s ethical system as developed in Goundwork of the Metaphysics of Morals, the apparent paradoxical circularity of freedom—namely, that morality cannot exist without our being free, while our freedom is demonstrated through our being beholden to the moral law—is resolved through the dual nature of the human being as both a phenomenon and a noumenon.
Central to Kant’s theory is the notion of autonomy, or self-rule. In order to rule oneself, though, one must first be capable of ruling, that is, possess freedom. Otherwise, in the absence of freedom, one is subject to heteronomy, or rule by something beside oneself, something external to the will. An animal, because it lacks reason, is surrendered over to its instincts, which operate naturally, without any input from the animal; and this unconscious life according to instinct allows it to survive without having to think. In contrast, humans are rational agents, meaning that we can determine the maxims on which we act. That is, reason allows us not only to formulate maxims, which are based on our instincts, but also to overcome these instincts. What is distinctive about humanity, then, is that by virtue of its autonomy, it can legislate and then compel itself according to its own laws. Specifically, since reason is universal, we humans abide by the moral law, which, being legislated, requires freedom as its ground.
It is important to further distinguish freedom from autonomy by characterizing the former as negative, and the latter positive. The will is free, according to Kant, when it is undetermined by nature, i.e., not influenced exclusively by the instincts, so that freedom is ultimately freedom from; meanwhile, autonomy is specifically determined to the extent that it posits, or establishes, certain rules for itself, in accordance with which it will act—in a word, law-making. As all causation is lawful, and as freedom is efficient causation, it falls under freedom’s law, i.e., morality. Freedom is thus fundamental since, without it, autonomy cannot be implemented; and moreover, freedom is the indispensable basis for a moral system.
The necessity of autonomy for morality, however, creates a seemingly vicious circle: If freedom is that which enables self-legislation, but if self-legislation is that which constitutes true, positive freedom, then how is that freedom, and with it ethics, can be validly demonstrated? Additionally, and far more problematically, freedom would seem to be impossible in a world governed 
by natural, physical laws. Kant himself argues for the a priori validity of causal determinism in his Critique of Pure Reason, one of whose objectives is to provide a grounding for Newtonian science against Humean skepticism. As it is not merely the case that we know the world—which includes us—operates according to a sequence of effects and their antecedent causes, but also that it must be this way, given that causality is a category of the understanding, we must rule out freedom, and subsequently morality. If my body and mind can be reduced to physical mechanisms, both subject to predetermined causal relations, then I have no power over my actions; I become a slave to my instincts, in which case, having no choice in the matter, responsibility cannot be attributed to me: Strictly speaking, “I” do not cause anything to happen intentionally but am rather a medium for other forces—in a word, I am heteronomous.
To solve this quandary, and thus to salvage the possibility of ethical life, Kant emphasizes two distinctions: First, cognition can be either theoretical or practical, and second, beings can be considered under two different aspects, viz., qua appearances/phenomena or qua things-in-themselves. Simply put, whereas theoretical cognition is receptive, practical cognition is spontaneous. A scientist 
like Newton, when he speaks of the universal law of gravitation, for example, does not “create” the phenomenon of gravity; he observes, records, and formulates it as a mathematical statement. The laws of nature operate independent of us, and our scientific knowledge of it is theoretical in that it takes cognizance of it. Kant also defines nature as phenomenal, that is, as consisting of appearances, which in turn are made intelligible by the understanding by means of pure concepts. Importantly, this means that nature as it appears to us, as it is given in sensibility, is necessarily and universally causally determined, but the “as it appears” makes all the difference, since we do not know what nature is like in itself. In circumscribing nature thus, Kant can make room for the possibility—which also means, as we shall see, the practical necessity—of freedom as it pertains to intelligence.
Practical cognition is not passive with regard to its object but actually produces its object, and one such object, Kant claims, is freedom. The will is nothing but pure practical reason, meaning that it is spontaneous. This is what gives rise to the firsthand 
experience of freedom, the sense that whenever I perform an action, like typing this essay, it is I who am behind it: I form the representation, or maxim, of typing a certain sequence of words, then I execute it. Moreover, I experience moral phenomena—foremost, the ought of the moral law, for which we humans have a predisposition. But what of the fact that I am myself a physical thing, which obviously impairs my spontaneous willing? Despite being physical, I am not merely phenomenal, for I also transcend the limits of appearances: In addition to an empirical ego, I am a transcendental self/will. Because the transcendental is beyond appearance, it eludes both the forms of intuition, which belong to sensibility, and the pure concepts of the understanding; hence, my true inner self is neither spatial nor temporal, nor can it be trammelled by the nexus of physical laws to which all empirical objects adhere.
Given that the category of causality is excluded from the transcendental sphere [1], it should follow that autonomy, as a free positing, which involves the idea of an efficient cause, should be jettisoned alongside it. While this is logically true, Kant insists that the imperative of the moral law calls upon us to postulate, for the sake of action, the freedom of the will. Such a defense is plausible because, after all, noumena are theoretically unknowable. Thus, one cannot claim to “know” that one has free will the way one “knows” that 1+1=2 (a priori) or that, when pulled, a door opens (a posteriori). Nonetheless, the unknowability of things-in-themselves does not prevent us from supposing that we are not unfree. The status of freedom, in other words, is wholly indeterminate from a cognitive standpoint. But because the moral law is binding, and because the moral law requires freedom, we are obliged to have faith in favor of freedom.
Otherwise, if we give into a total agnosticism, renouncing both freedom and determinism, then in terms of practice, we have effectively sided with the latter, resulting in a form of fatalism. Our a priori moral nature, however, demands that we act positively, that we order and organize our inclinations through reason, and that we legislate for ourselves the categorical imperative. Unconditioned freedom, freedom unaffected by nature, is what we posit because we ought to; it is a regulative Idea without which moral action is inconceivable. Kant also thinks the nobility of intelligence is further demonstrated by the fact that we humans understand nature, by which we certainly do not transcend it but comprehend it. One can say that in conceptualizing nature through science, tapping into its a priori laws, nature resides within us, in our representation. We may be in nature, but we are not necessarily of it.
[1] “Sphere” seems to be a better alternative than “realm” because it allows for ambiguity, considering Kant’s dualism can be interpreted either epistemologically, between aspects, or ontologically, between spaces.
Neologikon 
Pingback: Sandel On the Primacy of Right (1 of 3) | Neologikon