Propensities and conditional probabilities (Q541817)

This article deals with the objection that Paul Humphreys raised in 1985 against the propensity interpretation of probability, going back to Pierce and Popper. ``Humphreys' paradox'' goes basically as follows: Since propensities reflect asymmetric causal tendencies in time, they cannot obey Kolomogorov's probability calculus, since the latter one implies the Bayes' fomula, which allows to invert conditional probabilities in a non-trival sense. In this article the author provides an updated review on existing solutions, and concludes that none of them is completely satisfactory in view of Humphreys' 2004 rejoinder. Eventually she formulates and discusses a original, positive solution.