Abstract

It is an open question whether we ever successfully explain earlier states by appealing to later ones, and, further, whether this is even possible. Typically, these two questions are answered in the same way: if we give and accept ‘backwards explanations,’ they must be possible; if they are impossible, we are right to reject them. I argue that backwards explanations are brittle—they fail if the future event does not occur—and this is part of the reason they are not accepted about the actual world. This does not mean, however, that they must be rejected entirely. I argue that backwards explanations are possible for certain systems. This shields unificationism about scientific explanation from some recent criticisms