Abstract
In this essay I examine Roche and Sober’s (R&S) thesis that explanation is evidentially irrelevant, clarify the nodal points of disagreement, and defend explanationism. To do this, I utilize William Lycan’s categories of explanationism (2002) and a distinction between per se explanatoriness and particular explanatoriness. These help show that even if there are cases where explanation identification does not raise the probability of a hypothesis, ferocious explanationism is not even in principle challenged by R&S. Further, R&S’ challenge to inference to best explanation proves too much and ultimately fails.