Are There Statistically Impenetrable Contexts?

Abstract

Bayesians claim that reason is conducted according to the axioms of probability. In addition, to this, they claim that Bayesian methodology gives us an account of the scientific method―an account that describes the actual practice of science. These are two claims. Bayesian decision theory falls short of its descriptivist claims due to psychological, contextual, and qualitative shortcomings. In some contexts, there may be problems with prediction and prior probability. Even though there might be an acceptable conclusion to a problem, Bayes’ cannot even describe it. I use the example of a disjunction, wherein we cannot assign prior probabilities to either disjunct due to qualitative differences. Bayes only deals with degrees of personal belief. Here we run into not only the problem of assigning prior probability but also problems of uncertain evidence, and principles of relational acceptance. Not the descriptive aspect in jeopardy, but also the Bayesian conceptualization of the scientific method.