講演者：Kevin T. Kelly (Professor and Director of the Center for Formal Epistemology, Carnegie Mellon University)
Title : Rhetoric, Reliability, and Inductive Inference
Absutract : Socrates argued to persuade. He also sought the truth. In deductive reasoning, the two go together—valid argumentation le ads from true information to true conclusions, and that fact supports or may even boost the rhetorical force of valid deduct ive arguments. Inductive reasoning, by definition, generates conclusions that go beyond the information provided. Such rea soning can be very persuasive, as when scientists prefer simple, unified, sharply tested explanations over complex, diffuse explanations that rely on multiple coincidences—a bias known popularly as Ockham’s razor. But it is harder to say how suc h a bias conduces to true belief. It is tempting to try to make induction look deductive, by adding metaphysical assumption s, such as Leibniz’ view that God is an engineer who likes elegant universes. But that strategy is both rhetorically and e pistemically self-defeating, since the added assumptions are not subject to scientific investigation at all. Instead, we wi ll present a mathematical argument, based on ideas from learning theory, to the effect that Ockham’s is necessary for stayi ng on the straightest path to the truth, even if that path cannot be perfectly straight. Our argument singles out Ockham’s razor as the right rhetorical principle for inductive inference, but optimal, inductive truth-conduciveness is so weak that the argument may actually undermine our native credence in simple, unified theories. Inductive skeptics assume that weaker standards of truth conduciveness should be paired with weaker credence in the conclusions. We respond that standards are c ontextually relative to ambitions. High or even full belief is justified by weak standards as long as they are the best sta ndards achievable in the problem at hand.