You might think (and some people do) that game theory therefore becomes irrelevant to the real world of human social interaction, because people are not rational seekers of maximum utility, as game theory allegedly assumes. But while game theory is often described in that way, it’s not quite the right picture. Game theory actually only tells you what people would do if they were “rationally” maximizing their utility. That makes game theory the ideal instrument for identifying deviations from that notion of rationality, and many game theorists are happy with that.
There is, however, another interpretation of what’s going on. Perhaps people really do maximize their utility—but utility is not really based on dollars and cents, at least not exclusively. And maybe “emotional” and “rational” are not mutually exclusive descriptions of human behavior. Is it really so irrational to behave in a way that makes you feel good, even if it costs you money? After all, the root notion of utility was really based on happiness, which is surely an emotional notion.
Actually, most economists have long recognized that people are emotional. But when your goal is describing economics scientifically—and mathematically—acknowledging emotions poses a real problem, as Colin Camerer explained to me. “One of the things mainstream economists have said is, well, rationality is mathematically precise,” he said. “There’s one way to be rational. But there are a lot of ways to not be rational. So they’ve often used that as an excuse—anything can happen if people aren’t perfectly rational.” And if anything can happen, there’s no hope of finding a mathematical handle on the situation. “Economists have been a little defeatist about this—if you give up rationality, we’ll never be able to have anything precise.”
This argument seems very much like the strategy of looking for lost keys only under the lamp post, because you couldn’t see them if they were anywhere else. If there’s only one sort of behavior (rational) that you can describe with your math, then that’s the behavior you will assume is correct. But Camerer and other