🤖 AI Summary
While Gilboa and Schmeidler’s (1989) ambiguity aversion is foundational in decision theory, experimental evidence reveals persistent behavioral paradoxes under weak ambiguity attitudes.
Method: We propose a novel axiom—“minimal ambiguity aversion”—and construct a rational decision framework termed *Prudent Dual-Self Expected Utility*, wherein robust choices emerge from evaluating subjective scenarios via two distinct selves and selecting the worst-case expected utility.
Contribution: This work provides the first axiomatic characterization of minimal ambiguity aversion, yielding three strictly equivalent utility representation theorems. It identifies “dual-self modeling coupled with conservative worst-case selection” as the core mechanism. The framework unifies invariant biseparable preferences, subjective–objective rationality models, and utility representation theory, thereby extending both the axiomatic foundations and behavioral explanatory power of ambiguity-sensitive decision models.
📝 Abstract
Uncertainty aversion introduced by Gilboa and Schmeidler (1989) has played a central role in decision theory, but at the same time, many incompatible behaviors have been observed in the real world. In this paper, we consider an axiom that postulates only a minimal degree of uncertainty aversion, and examine its implications in the preferences with the basic structure, called the invariant biseparable preferences. We provide three representation theorems for these preferences. Our main result shows that a decision maker with such a preference evaluates each act by considering two"dual"scenarios and then adopting the worse one as its evaluation in a cautious manner. The other two representations share a structure similar to the main result, which clarifies the key implication of weak uncertainty aversion. Furthermore, we offer another foundation for the main representation in the objective/subjective rationality model and characterizations of extensions of the main representation.