Standard game theory’s theory of cooperation is based upon threatened punishment of non-cooperators in a repeated game, which induces a *Nash equilibrium* in which cooperation is observed. Thus, cooperation in games is explained as a *non-cooperative equilibrium*. Behavioral economics, on the other hand, explains cooperative behavior by inserting ‘exotic’ agruments into preferences (altruism, fairness, etc.), and again deducing cooperation as a *Nash equilibrium* in a game with non-standard preferences. In both variants, cooperation is envisaged as achievable as a Nash equilibrium.I believe a more compelling approach is to model individuals as using a *Kantian optimization protocol,* but with standard, non-exotic preferences. The Kantian protocol inserts morality not into preferences, but into the optimization protocol, and these are distinctly different approaches, as I show. We deduce cooperation in one-shot games in Kantian equilibrium. Kantian optimization resolves both tragedies of the commons and the provision of public goods: Kantian equilibria, in both cases, are Pareto efficient, in contrast to Nash equilibrium. Furthermore, I characterize the class of 2 x 2 symmetric games in which Kantian optimizers will drive Nash optimizers to extinction, and conversely; both are non-empty classes.