Thursday, January 17, 2013
4:00 PM - 6:00 PM
Kresge Hall, 2-301
1880 Campus Drive
Evanston, IL 60208 map it
Audience: Faculty/Staff - Student
Group: NU Lemmings PhLing Talks
Topic: "epistemic : deontic :: additive : intermediate"
Abstract: The gradable epistemic adjective "likely" and its closest deontic counterpart "good" are grammatically and semantically similar in numerous ways, but they differ in their interaction with disjunction: P is as good as Q, P is as good as R, therefore P is as good as (Q or R) is a valid inference, but the same inference is invalid with likely replacing good. Some might think that this presents a challenge for the hypothesis that different modal flavors can be given a unified semantic treatment (Kratzer 1978). Not I: instead I will motivate a general distinction between "additive" and "intermediate" scales, e.g. those with which weight vs. temperature predicates are associated. I will show that the same parametric difference can explain the different ways that epistemic and deontic adjectives interact with disjunction. This has various consequences for the logic of modality. It calls into question the adequacy of Kratzer's treatment of graded modality and perhaps even the validity of the the principle of agglomeration (ought P and ought Q entails ought P and Q). It also suggests a closer connection between the semantics of graded modalities and Bayesian models of uncertain reasoning and decision-making that are widely employed in cognitive science and elsewhere in philosophy.