This paper proposes a new interpretation of non-cooperative games that shows why the unilateralism of best-reply reasoning fails to capture the mutuality of strategic interdependence. Drawing on an intersubjective approach to theorizing individual agency in shared context, including a non-individualistic model of common belief without infinite regress, the paper develops a general model of a 2 × 2 simultaneous one-shot non-cooperative game and applies it to games including Hi-Lo, Stag Hunt, Prisoners’ Dilemma, Chicken, BoS and Matching Pennies. Results include High as the rational choice in Hi-Lo, and Cooperate as a possible rational choice in the Prisoners’ Dilemma.

本文提出了一种对非合作博弈的新解释，该博弈说明了为什么最佳回答推理的单边主义无法抓住战略相互依存的相互关系。运用主体间的方法在共享上下文中对单个代理进行理论化，包括没有无限回归的具有共同信念的非个体模型，本文开发了一个2×2同时单发非合作博弈的通用模型，并将其应用于博弈包括Hi-Lo，Stag Hunt，囚徒困境，Chicken，BoS和Matching Pennies。结果包括：高是Hi-Lo的理性选择，而合作是囚徒困境中的可能理性选择。