Kaufmann on the probabilities of conditionals (Q928708)

This discussion responds to \textit{S. Kaufmann}'s claim [J. Philos. Log. 33, No. 6, 583--606 (2004; Zbl 1061.03011)] that in certain cases the probability of a conditional \(A \rightarrow B\) should be different from the associated conditional probability \(\text{Pr}(B| A)\). While the conditional probability results in \(\text{Pr}(A \rightarrow C)=\sum_{X_i \in \text{range}(X)} \text{Pr}(C | A X_i) \text{Pr}(X_i | A)\), Kaufmann suggests that in the specified circumstances \(\text{Pr}(A \rightarrow C)=\sum_{X_i \in \text{range}(X)} \text{Pr}(C | A X_i)\text{Pr}(X_i)\) better accords with our intuition. Douven gives a new description of this sort of case, and shows how Kaufmann's rule leads to two inconsistent probability assignments. He considers the reply that only one such description can be appropriate, and concludes by noting several experimental results that might explain why our intuition goes wrong in these sorts of cases.