Andrew Gelman writes, in response to my comments in this post:
Comments on comments on "Voting as a rational decision", by Andrew Gelman:After reading our article, "Voting as a rational decision," Mark Thoma asked,
If helping other people makes me happy, why would caring about other people be contrary to my own self-interest? This is essentially a question about the meaning of the term selfish. I [Mark] assume selfishness means maximizing my utility, which may or may not include the happiness of other people as an argument.
The challenge in all such arguments is to avoid circularity. If selfishness means maximizing utility, and we always maximize utility (by definition, otherwise it isn't our utility, right?), then we're always selfish. But then that's like, if everything in the world is the color red, would we have a word for "red" at all? I'm using selfish in the more usual sense of giving instrumental benefits. For example, if I cut in front of someone in line, I'm being selfish. If I don't do it (because I get pleasure from being a nice guy and pain from being a jerk), then that's other-directed. I'm sacrificing something (my own time) in order to help others. Just because something is enjoyable it doesn't have to be selfish, I think.
To put it another way, if "selfish" means utility-maximization, which by definition is always being done (possibly to the extent of being second-order rational by rationally deciding not to spend the time to exactly optimize our utility function), then everything is selfish. Then let's define a new term, "selfish2," to represent behavior that benefits ourselves instrumentally without concern for the happiness of others. Then our point is that rationality is not the same as selfish2.
Also, some of his commenters questioned whether a single vote could be decisive, what with recounts etc. The answer is, yes, it can, because there is ultimately some threshold (even if unobservable) as to whether the recount occurs. And even if this threshold is itself probabilistic, the probabilities can be added. We demonstrate this mathematically in the Appendix to the 2004 Gelman, Katz, and Bafumi article in the British Journal of Political Science; see page 674 here.
My reply to Andrew:
I almost followed up with something similar. My point, which probably wasn't as clear as it could have been, was about rationality not always being the same as selfishness. The utility functions that we use in maximization problems can, and sometimes do, differ from the definition of selfishness as usually defined, e.g. Barro's "Are Bonds Net Wealth" where the parent's utility function has the utility of the children as an argument. Selfishness means, I think, doing what's best for me no matter what (I have no concern for the effects that my actions have on others), and those types of preferences are easy to model. But we also allow for preferences where I care about how my actions affect others so that, in your example, there would be a net benefit from not cutting in line. We don't ask where preferences come from, or why people hold them, whether it's rational in a psychologist's sense to care about others, we just say that if this is what you like, this is how you will act when presented with a particular set of constraints, prices, etc.
In that context, it is not always rational to be selfish. It is rational to maximize utility, and if I am an other-directed person - if helping others makes me happy - then it is perfectly rational to appear to act contrary to my own self-interest (or better, selfish-interest). I suppose it would look particularly irrational to someone with inward-directed preferences, but so long as the choices maximize utility, there wouldn't be anything irrational about it from an economist's perspective.
[I think we are really talking about Adam Smith's Theory of Moral Sentiments, i.e. the way in which we are endowed with emotions like empathy, sympathy, "fellow-feeling," and so on, that act to redirect our self-interest into what's best for all, i.e. into the maximization of the social interest]