Bruce Glymour
Kansas State University
Over the last 40 years or so, a consensus has emerged in philosophy of science that essentially chancy events can be explained, but that such explanations are not deductive arguments. The orthodoxy about the nature of such explanations has recently been challenged on two related fronts. I argue here that if we take these challenges seriously, we must deny that such explanations aim to convey understanding. I propose that instead we treat explanations of essentially chancy events as deductive explanations, and I show how to do this with some of the more troubling cases, namely explanations in classical statistical mechanics.