Fri Apr 15, 2011
5101 Tolman, 11 AM–1 PM
|Institute of Cognitive and Brain Sciences
Seth Yalcin (UC Berkeley)
A Likely Story: Some Lessons from Natural Language Probability Talk
What exactly does it mean to say, in ordinary language, that something is probably the case? The question is of interest both from the perspective of the logic, semantics, and pragmatics of natural language, and from the perspective of psychological investigations which rely on the use of such talk by subjects; and its answer may shed light on the larger issue of how to give realistic models of human uncertainty. In this talk I review some work in linguistic semantics and in psychology attacking the question. On the semantics side, we will see, inter alia, that natural language probability talk makes for counterexamples to the inference pattern Modus Tollens, and hence calls for a nonclassical account of consequence. On the psychological side, we will review cases wherein competent English speakers robustly judge that a proposition is likely, despite being given the information that the negation of the proposition is more likely. While a standard response to these data is to chalk it up as yet another failure of rationality, I will suggest that careful attention to the semantics and pragmatics of this fragment of language reveals that subjects are making no mistake. A larger aim of the talk is to illustrate by example some ways in which formal semantic/pragmatic inquiry and psychological inquiry can be mutually illuminating.