作者: Joseph Y. Halpern , David A. Easley , Lawrence E. Blume
DOI:
关键词: Decision field theory 、 Decision rule 、 Von Neumann–Morgenstern utility theorem 、 Computer science 、 Subjective expected utility 、 Decision theory 、 Artificial intelligence 、 Optimal decision 、 Evidential decision theory 、 Decision analysis 、 Mathematical economics
摘要: In almost all current approaches to decision making, it is assumed that a problem described by set of states and outcomes, the maker (DM) has preferences over rather rich acts, which are functions from outcomes. However, most interesting problems do not come with state space an outcome space. Indeed, in complex often far clear what spaces would be. We present alternate foundation for primitive objects choice syntactic programs. A program can be given semantics as function but does necessarily have this way. representation theorem proved spirit standard theorems, showing if DM's preference relation on programs satisfies appropriate axioms, then there exist S states, O way viewing O, probability S, utility such DM prefers b only expected higher than b. Thus, subjective, just like utility; they part description problem. number benefits approach discussed.