Friday, 24 September 2004

Statistics for Bayesian Philosophers (and some others)

Jason Grossman

History & Philosophy of Science, University of Sydney

12:05 pm, 817R Cathedral of Learning

Abstract: For the sake of argument, let us fix the following sets of hypotheses and possible observations, and the probabilities assigned to each observation by each hypothesis:    
                                     

                                       frog            bunny       philosopher
         seminar room        0.041           0.009            0.95
         swamp                  0.949           0.001            0.05

Suppose we observe a philosopher.  I will discuss principles of statistical inference, dividing them into those which say that we should tentatively conclude that we're in a swamp and those which say we should tentatively conclude that we're in a seminar room.  I will touch on the problem of including additional evidence in this judgment.

I will draw conclusions on behalf of various well-defined groups of philosophers, including Bayesians (of either of two sorts) and those who accept the likelihood principle.