Time: 2019 Oct 25, 13:30-16:00
Venue: 新斋324, Tsinghua Univ.
Speaker: Kevin Kelly (Carnegie Mellon University)
Inductive knowledge is knowledge that goes beyond the information available. Inductive skeptics think there isn’t any such thing. Maybe they are right—standard modal theories of knowledge (tracking, safety) are too strict to allow for even banal inductive inferences. But maybe they are wrong: perhaps failure to achieve those standards is justified because they do not apply. I will sketch a simple semantics for the semantics of inductive knowledge and will illustrate some of its interesting consequences. One of them is a normative argument for the old, quasi-sociological view in the philosophy of science that there is (can be?) no “logic of discovery”. The idea is that if inductive knowability implies learnability, then inductive knowability cannot be closed under deductive consequence, so inductive learnability must not be necessary for inductive knowability. I will also discuss the surprising consequence that you can inductively know your own Moore sentence: “I know that A but A is false”, even if you are deductively cogent—something impossible in standard epistemic logic. The presented semantics is a much-streamlined version of the semantics presented in “A Computational Learning Semantics for Inductive Empirical Knowledge”, Logical/Informational Dynamics, a Festschrift for Johan van Benthem, A. Baltag and S. Smets eds., Springer: 2014.
Professor in Philosophy department, Director of the Center for Formal Epistemology at Carnegie Mellon University, author of “The Logic of Reliable Inquiry” (Oxford, 1996). His recent work concerns reliable belief revision, the solution of methodological regresses, and efficient convergence. His research areas include Ockham’s razor and realism, Qualitative belief, and the lottery paradox, Learning semantics for epistemic logic, Analogies between empirical and formal reasoning, The learning power of belief revision, Learning theory and the philosophy of science, Infinite methodological regresses.