A dynamic conditional decision model

We provide a characterization of decision models based on partial information and on choices among conditional acts. Proper representability of choices in such frameworks can be performed locally and not globally, by grasping the genuine nature of conditional models. In particular we will deal with finite frameworks where acts coincide with categorical random quantities and where conditioning dynamically and implicitly depend on the set of available choices. Imprecise models could naturally arise, implying either a ``credal'' set approach or the synthesis through extreme representatives.