The major graduate textbooks in microeconomics are all at least 15 years old. This aging leads to a received wisdom that is out of step with the actual state of knowledge in micro theory. In this article in a recent JEBO on the future of experimental economics, Herbert Gintis suggests that many economists are confused about what implies that agents should play subgame perfect equilibria (common knowledge of rationality, not just rationality, is required) and about what epistemic conditions would lead to a Nash equilibria being played. He suggests that Aumann’s correlated equilibrium is a more sensible prediction of behavior, with social norms serving as the correlating device. Some game theorists dislike the addition of norms to behavior, sensibly replying that all norms or social preferences mean is that the theorist/experimenter has misspecified an agent’s utility function. This critique is valid, but given how often (and how useful) u(x)=x is used, it might be more sensible just to use equilibrium concepts that account for such norms using a linear utility function rather than quibble about what utility function agents actually have. I think that foundations of norm and role driven behavior – perhaps on an axiomatic level, and certainly in a manner stronger than “anything can be explained by some sort of social preference” – is going to be an incredibly active frontier of research in the next decade.
“Toward a Renaissance of Economic Theory,” H. Gintis (2010)