Probabilistic models of vagueness
Language and Logic (Advanced)
First week, from 17:00 to 18:30
While many-valued logics are familiar to semanticists working on vagueness, the use of probabilities to represent vagueness is lesser known, although they are of common use in psychology and cognitive science more broadly. The goal of this course will be to introduce and compare four classes of probabilistic models that have been used in recent years to model categorization and to deal with vagueness more specifically: (i) Probabilistic models based on a theory of noisy or imperfect discrimination (Gaussian models) (ii) Probabilistic models based on the notion of variable criteria (Item Response Theory models) (iii) Probabilistic models based on conceptual spaces (Conceptual Space models) (iv) Probabilistic models based on Bayesian inference under uncertainty (Bayesian models) The course will explain how such models can be related to more familiar models based on degrees of truth, and show specific applications, in particular to the modeling of borderline cases, of the sorites paradox, and of the informativity of vague sentences.