Three-Toed Sloth

Down the Stairs and to the Left (This Week at the Philosophy Colloquium)

I'm speaking at the CMU philosophy department's colloquium this week. I do not pretend to fully understand how this happened, but no doubt that by the end of the day I will enjoy a simultaneously higher and more profound level of puzzlement about many matters.

"Praxis and Ideology in Bayesian Statistics"

A substantial school in the philosophy of science identifies Bayesian inference with inductive inference and even rationality as such, and seems to be strengthened by the rise of Bayesian statistics in applications. In this talk, I hope to persuade you that the most successful practices of Bayesian statistics do not actually support that philosophy but rather accord much better with sophisticated forms of hypothetico-deductivism. Drawing on the literature on the consistency of Bayesian updating and also on experience of applied work, I examine the actual role of prior distributions in Bayesian models, and the crucial aspects of model checking and model revision, which fall outside the scope of Bayesian confirmation theory. I argue that good Bayesian practice is very like good frequentist practice; that Bayesian methods are best understood as regularization devices; and that Bayesian inference is no more inductive than frequentist inference, i.e., not very. At best, the inductivist view has encouraged researchers to fit and compare models without checking them; at worst, theorists have actively discouraged practitioners from performing model checking because it does not conform to their ideology.

Based on joint work with Andrew Gelman.

Date: Thursday, 20 January 2011

Time and place: Reception 4:00--4:35 in Doherty Hall, talk 4:45--6:00 in Baker Hall A53.

Self-Centered; Bayes, anti-Bayes; Philosophy