Shafer on conditional probability (Q794046)

In this short discussion note of \textit{G. Shafer's} paper (see the foregoing review; Zbl 0539.60004) the present author shows that some kind of paradox may be derived when one of Shafer's assumptions concerning the agent's beliefs is given up. The issue is whether someone can ever believe that he will fully believe (or ''take as evidence'') a proposition that is actually false. Doxastic logic tells us that whoever fully believes that p ipso facto believes that this belief is not mistaken, thus - with respect to someone's present beliefs - the believer himself cannot think it possible that he is mistaken. On the other hand, we all know that humans are not infallible; thus we should perhaps reasonably assume that in the future we will fully believe some proposition p which is (or will turn out) actually (to be) false. The author shows that if indeed the agent assigns a positive probability to his future becoming sure of some false proposition E, then a Dutch book may be construed against him.