Modeling Human Conditional Reasoning by Probability Logic

SPEAKER

Niki Pfeifer

ABSTRACT

Recently, probabilistic approaches started to gain popularity in the
psychology of reasoning and seem to replace classical logic as the
normative standard of reference. Empirical data from probabilistic truth
table tasks, a standard test paradigm for the participants'
interpretation of conditionals, suggest that the uncertainty of an
indicative conditional is assessed by a corresponding conditional
probability, but not by the probability of a material conditional. I
will present an alternative experimental paradigm for investigating how
people interpret and reason about conditionals. The approach uses
coherence based probability logic (in the sense of De Finetti, Coletti &
Scozzafava, Gilio) as a normative framework. I will discuss general
properties of probabilistic argument forms, like probabilistic
informativeness, and show how probability logic helps the experimenter
to infer how participants interpret indicative conditionals by which
inferences they draw from uncertain premises. I will present a series of
experimental results that corroborate the conditional probability
interpretation of uncertain indicative conditionals.