Bayes and Brentano

Stephen Yablo (Massachusetts Institute of Technology)

When evidence E is acquired, pr(E) rises to 1.  What about its truth-makers' probabilities  pr(ε)? These will have to change too, as E's probability is a function of theirs.  Bayes' rule  is strangely uncurious about this; it doesn't ask about the probabilistic underpinnings of E's newfound certainty.  What possible excuse can it have,  when  E's bearing on H  depends  on the relative probabilities of  pro-H truth-makers and anti-H?.  The explanation (it's not an excuse) is that ε for the Bayesian trails along probabilistically after E.  A rule allowing  E to trail along  after its truth-makers is proposed.  All heck breaks loose, e.g., evidential relations  become hyperintensional.