Is there epistemologically irrational knowledge?

Abstract

I present an epistemological puzzle about perceptual knowledge and its relation to the evaluation of probabilities. It involves cases, concerning a given subject S and a proposition P in a determinate context, where apparently: S has perceptual knowledge of P; the epistemic justification S has for believing Not-P is much greater than her epistemic justification for believing P. If those two theses were true, the following very plausible epistemological principle would fail: If S knows P, then the epistemic justification S has for believing Not-P is not greater than her epistemic justification for believing P. I offer a solution to the puzzle, which is compatible with basic intuitions and theses of orthodox Bayesianism.