In a previous post, I discussed a twist on the Prisoner's Dilemma, wherein it is rational for agents to “feel shame”: to modify their utility function to decrease the utility of defecting against cooperators. What's neat is that it is a fully rational micro-foundation for getting cooperation in the PD. Today, I want to look whether this behavior can be learned. Learning to play One line of research in game theory is in determining how game-theoretic agents learn to play the game they are ...
