Justification for a Probabilistic Account of Conditionals

Abstract

In this dissertation I argue that a probabilistic account of conditionals similar to the one proposed by Robert Stalnaker in 1968 is the logical account of conditionals that most aptly models conditional use in natural language. I argue that a probabilistic account of conditionals is best able to account for the most systematic and widespread uses of conditionals in natural language as is evidenced by both its compatibility with the descriptively accurate psychological account, as well as its ability to take into account expert intuitions that diverge from the material conditional interpretation. I provide expert support for Stalnakers account by describing the ways that a probabilistic conditional can avoid the paradoxes of the material conditional. I argue that the predictive accuracy of the alternative mental models account provides support for the claim that Stalnakers logical account of conditionals is descriptively accurate. I conclude that both expert and naive reasoners uses of conditional statements are most accurately modelled by a probabilistic account of conditionals similar to that proposed by Stalnaker.