The authors assume that a decisionmaker has expected utility preferences over a large space which includes some variables not observable by the theorist. These will induce preferences over observable variables, which typically will not have the expected utility form. This paper focuses on...