Full Program »
A Popperian Doctrine on Probability RevisitedThe paper discusses a well-known argument by Popper to the effect that simple, nested models cannot have larger probabilities than more complex models. Popper's doctrine on how to assign probabilities to nested models leads to a conflict with the practice of Bayesian inference in phylogeography. In order to resolve the problem, I challenge the premise of Popper's argument by rethinking the notion of a statistical model invoked in the philosophical discourse. Doing so will also offer a novel, albeit limited defence of Jeffreys's simplicity postulate.
University of Bristol