Conditionals as Random Variables

SPEAKER

Richard Bradley

ABSTRACT

This talk will explore the view that conditionals are proposition-valued random variables, building on earlier papers of Jeffrey and Stalnaker and Bradley in which views of this kind are floated. It will be argued that the difference between indicative and counterfactual conditionals lies in the values taken by the random variable in worlds in which the antecedent is false. In the case of indicative conditional no value is determined, while in the case of counterfactuals the value is partially determined by non-semantic features. In both cases the resultant semantics are non-Boolean. The credibility of both kinds of conditional is given by its expected value relative to a probability measure on the domain of the random variable, where the value of a conditional at a world is the expected truth of the proposition determined by the random variable at that world.