Probabilistic belief expansion and conditioning

The paper discusses probabilistic belief change, in particular the type of change occurring when information is added which is consistent with the old beliefs. In the AGM theory of belief revision this type of belief change is known as an expansion, and is usually considered to be easy compared to revision and contraction, the two other main types of belief changes the AGM theory distinguishes. If logical theories are used to represent the belief states, then an explicit deenition of expansion can be given, and, at rst sight, probabilistic conditioning seems a natural generalization of this notion in the context of probabilistic belief states. However, we argue that it may be more correct to view conditioning and expansion as two diierent kinds of belief change, and we discuss the diierences between the two kinds in the context of belief states represented by partially determined probability functions.