Decision theory in expert systems and artificial intelligence
International Journal of Approximate Reasoning
Probabilistic reasoning in intelligent systems: networks of plausible inference
Probabilistic reasoning in intelligent systems: networks of plausible inference
Valuation-based systems for Bayesian decision analysis
Operations Research
Hi-index | 0.00 |
Johann Pfanzagl proposed an additional decision axiom in 1959, complementary to the usual axioms, which would have tightly restricted permissible monetary utility functions. Subsequent debate partially clarified previously unacknowledged normative issues, and ended with orthodox consensus favoring a principled reform of practice. An analytical device contributed by Pfanzagl is extended here to show that the consensus practice sometimes conflicts with a principle often expressed as a continuity axiom, in situations as simple as Allais' famous example. The axioms have from their beginnings been recognized as incomplete, so these new exceptions cannot much threaten the axioms' role in such theories as a popular semantics for subjective probability. However, the paper suggests an enhancement of sensitivity analysis in applied work, sometimes overruling orthodox advice.