The folk theorem establishes an abundance of equilibrium payoffs for a non zero sum repeated game with patient payoffs.
However, considering a multi-agent repeated games where the players employ learning techniques - which of these payoffs is obtained by an equilibrium strategy that the agents can learn?
We present results connecting the repeated game literature with recent learning results.

The paper offers an automated method for solving mechanism design problems that involve repeated interactions without money. The method can be used for arbitrary discount factors and relies on dynamic programming and promised utility. I illustrate the approach with a problem of taking turns at performing an indivisible chore. Additionally, I discuss the relationships between different classes of mechanisms and show that promised utility mechanisms are more general than mappings from histories of finite memory.