Abstract

Peter Wakker impugns the rationality of Allais-preferences. He argues implicitly that otherwise perfectly reasonable subjects who have Allais preferences will in some situations choose to bet on propositions before, rather than after, learning of their truth-values. After spelling out Wakker’s argument, and identifying and repairing a weak point, I turn it around to say that aversions to information, and preferring to bet on propositions without knowing their truth-values, can be reasonable on precisely the grounds that can make Allais-preferences reasonable. Lastly, to accommodate reasonable Allais-preferences, the normative principle of Utility Theory is restricted to pairwise preferences for lotteries the basic outcomes of which are ‘loaded up’, and, of course, to preferences for lotteries that are ‘comparable as alternatives’.