Proceedings of the Annual Meeting of the Cognitive Science Society

The concept of Nash equilibrium has played a profound rolein economics, and is widely accepted as a normative stance forhow people should choose their strategies in competitive envi-ronments. However, extensive empirical evidence shows thatpeople often systematically deviate from Nash equilibrium. Inthis work, we present the first resource-rational mechanisticapproach to one-shot, non-cooperative games (ONG), show-ing that a variant of normative expected-utility maximizationacknowledging cognitive limitations can account for impor-tant deviations from the prescriptions of Nash equilibrium inONGs. Concretely, we show that Nobandegani et al.’s (2018)metacognitively-rational model, sample-based expected util-ity, can account for purportedly irrational cooperation rates ob-served in one-shot, non-cooperative Prisoner’s Dilemma, andcan accurately explain how cooperation rate varies dependingon the parameterization of the game. Additionally, our workprovides a resource-rational explanation of why people withhigher general intelligence tend to cooperate less in OPDs, andserves as the first (Bayesian) rational, process-level explana-tion of a well-known violation of the law of total probability inOPDs, documented by Shafir and Tversky (1992), which hasresisted explanation by a model governed by classical proba-bility theory for nearly three decades. Surprisingly, our workdemonstrates that cooperation can arise from purely selfish,expected-utility maximization subject to cognitive limitations.