In other contexts, interpreting people’s behavior as generally expected-utility maximizing requires undue violence to the need for generality in theory construction. We get better prediction using fewer case-specific restrictions if we suppose that subjects are maximizing according to one or (typically) more of several alternatives (which will not be described here because they are not directly about game theory): cumulative prospect theory (Tversky and Kahneman 1992), or alpha-nu utility theo