This paper defends Michel Foucault's notion of archaeology of knowledge against the influential and putatively devastating criticism by Dreyfus and Rabinow that Foucault's archaeological project is based on an incoherent conception of the rules of the discursive practices it purports to study. I argue first that Foucault's considered view of these rules as simultaneously implicit and historically efficacious corresponds to a general requirement for the normative structure of a discursive practice. Then I argue that Foucault is entitled to that view despite the charges to the contrary by Dreyfus and Rabinow. I also explain in detail how the argument by Dreyfus and Rabinow arises from a misunderstanding of Foucault's archaeological project as transcendental inquiry, while archaeology of knowledge is, in fact, a diagnostic project. The result is a novel understanding of the notion of archaeology of knowledge that enables a reassessment of Foucault's philosophical work in connection with current debates regarding the relationship between reflection and practice in the structure of thought.

Several subjects are fully convinced that they are brains in vats whose experiences are hallucinatory. They confront a ‘skeptic’ who raises the possibility that they are not brains in vats who lack and hallucinate hands but ‘brains in skulls’ who have hands and see them. Familiar responses to skepticism are offered in support of the claim that the subjects know they do not have hands. The philosophical significance of this looking-glass approach to skepticism is also discussed. It is suggested that these familiar responses to skepticism do not achieve anything because, if we begin from the assumption that we are brains in vats, we can employ those same responses to skepticism to support the claim that we know we do not have hands. One interlocutor argues that things are not so dismal. The main conversation is framed by a discussion among night watchmen at some sort of skepticism laboratory.