Why known unknowns may be better than knowns, and how that matters for the evolution of happiness

Abstract

Rayo and Becker (2007) model happiness as an imperfect measurement tool: It provides a partial ordering of alternative courses of actions. In this note, decisionmakers use their inability to rank two actions, to infer rankings of other pairs of actions. It is demonstrated that coarser happiness information actually increases the power of inference. As a result behavior is maximizing, not merely satisficing, almost independent of how coarse the happiness information is. Moreover, to support inference, evolution selects a happiness function with different properties than the one maximizing direct sensory information.