A new resolution of the Judy Benjamin problem

SPEAKER

Igor Douven

ABSTRACT

Van Fraassen's Judy Benjamin problem has generally been taken to show
that not all rational changes of belief can be modelled in a
probabilistic framework if the available update rules are restricted
to Bayes's rule and Jeffrey's generalization thereof. But alternative
rules based on distance functions between probability assignments that
allegedly can handle the problem seem to have counterintuitive
consequences. Taking our cue from a recent proposal by Bradley, we
argue that Jeffrey's rule can solve the Judy Benjamin problem after
all. Moreover, we show that the specific instance of Jeffrey's rule
that solves the Judy Benjamin problem can be underpinned by a
particular distance function. Finally, we extend the set of distance
functions to ones that take into account the varying degrees to which
propositions may be epistemically entrenched.