Decision-making environments in which unboundedly rational decision makers choose to ignore relevant information

This paper advances the claim that ignoring relevant information is sometimes consistent with good decision making. Although that finding is not new, the argument presented here is. In contrast with bounded rationality models, the decision-making model in this paper presupposes no cognitive constraints or costs associated with processing available information. The paper identifies a class of decision-making environments characterised by asymmetric payoffs and probabilities – a property which gives rise to optimal decision rules that ignore relevant information. In other words, optimal decision procedures used by omniscient agents are sometimes independent of variables that objectively predict future outcomes.