Beatrice Han has argued that the theories of subjection (determinism: structure) and subjectivation (freedom: agency) are the “the blind spot[s] of Foucault's work.” Furthermore, she continues, as historical and transcendental theories, respectively, Foucault left them in a state of irresolvable conflict. In the Scientific Temptation I have shown that, as a practicing researcher, Foucault encourages us to situate the theories of the subject in the context of his un-thematized search for a metaphysics of realism, the purpose of which was to ground his complementary reach for a possibility of naturalism. In Returning to Kant I now argue that it is this fundamental feature of “Foucault's Foucault” that drives his returns to Kant, the purpose of which was to resolve the conflicting theories of the subject and thereby solve his Giddensian problem of structure and creativity. Locating the returns and their purpose in the context of my own arguments for the recovery of human agency, I argue that Foucault's attempts to solve his Giddensian problem led to two unfortunate solutions. In the first return, his resort to Baudelaire's aesthetic subject is a regression to a pre-noumenal conception of the Kantian subject. With the second return, the reinstatement of the Kantian subject as causally empowered, minus the noumenalism, is nothing more than a reclamation of Kant's conception. I argue that only a reconstruction of Foucault's scientific realism permits us to understand that he could have moved beyond mere reclamation to the actual recovery of human agency.