Robert E. Rowthorn, Ricardo A. Guzmán, Carlos Rodríguez-Sickert,  2009.

ABSTRACT

In this paper we survey the various theories of gratuitous cooperation (in favour of non-relatives and without repeated interaction), and we describe our own effort to integrate these theories into a self-contained framework. Our main conclusions are the follows. First: altruistic punishment, conformism, and gratuitous cooperation coevolve, and group selection is a necessary ingredient for the coevolution to take place. Second: people do not cooperate by mistake, as most theories imply; on the contrary, people knowingly sacrifice themselves for others. Third: in cooperative dilemmas conformism is an expression of preference, not a learning rule. Fourth, group mutations (e.g., the rare emergence of a charismatic leader that brings order to the group) are necessary to sustain cooperation in the long-run.