Zero-Knowledge Cooperation in Dilemma Games

We consider a very simple adaptive rule that induces cooperative behavior in a large class of dilemma games. The rule has a Pavlovian flavor and can be described as win–continue, lose–reverse. It assumes no knowledge about the underlying structure of the environment (the “rules of the game”) and requires very little cognitive effort. Both features make it an appealing candidate for explaining the emergence of cooperative behavior in non-human species.