Philip Gorski in Public Books:

K10588The social sciences have an ethics problem. No, I am not referring to the recent scandals about flawed and fudged data in psychology and political science.1 I’m talking about the failure of the social sciences to develop a satisfactory theory of ethical life. A theory that could explain why humans are constantly judging and evaluating, and why we care about other people and what they think of us. A theory that could explain something so trivial as the fact that social scientists care about data fudging.

This is not to say that we have no theories. It’s just that they’re bad theories. Consider evolutionary game theory.2 It says that ethical life results from individual rationality. How so? Assume a population of self-interested actors. (A big assumption!) Have them play a one-on-one, zero-sum game with each other, over and over again. (Prisoner’s dilemma, anyone?) The winning strategy will be something called “tit-for-tat.” The rules of TFT are as follows: 1) be nice in the first round; 2) copy your partner on all subsequent rounds. In other words, if they are mean, you should be mean back; if they act nice, you should, too. In the long run, individuals who follow the TFT strategy will be better off than people who follow a mean strategy. Or so the computer simulations tell us.

This theory is morally satisfying. Nice guys don’t finish last after all! But it is not intellectually satisfying. Human evolution didn’t really work this way. Earlyhomo sapiens was not modern homo economicus. Our ancestors were not isolated monads. They lived in small groups. They were social animals. A good theory would start with good assumptions—realistic ones. Ethical life just doesn’t feel like game theory. Often, it’s hot emotion, not cool calculation. It’s filled with anger and sorrow, love and joy, not minimizing and maximizing. Finally, a good theory would have to account for why we have moral emotions in the first place. In particular, it would have to account for “niceness” itself.

More here.