In his Reflective Knowledge, Ernest Sosa offers a theory of knowledge, broadly virtue-theoretic in character, that is meant to transcend simple ways of contrasting "internalist" with "externalist" or "foundationalist" with "coherentist" approaches to knowledge and justification. Getting beyond such simplifications, Sosa thinks, is the key to finding an exit from "the Pyrrhonian Problematic": the ancient and profound skeptical problem concerning the apparent impossibility of validating the reliability of our basic epistemic faculties and procedures in a way that escapes vicious circularity. Central to Sosa's anti-skeptical strategy is the claim that there are two kinds of knowledge. His thought is that animal knowledge, which can be understood in purely reliabilist terms, can ground justified trust in the reliability of our basic cognitive faculties, thus elevating us (without vicious circularity) to the level of reflective knowledge. I offer a sketch of an alternative approach, linking knowledge and justification with epistemic accountability and responsible belief-management, which casts doubt on the idea that "animal" knowledge is knowledge properly so-called. However, it turns out that this approach is (perhaps surprisingly) close in spirit to Sosa's. I suggest that the differences between us may rest on a disagreement over the possibility of providing a direct answer to the Pyrrhonian challenge.