Putting game theory to the test

When three game theorists won the Nobel Prize in economics last October, the award spotlighted a theoretical tool that has quietly spread across several areas of science, gaining power as it goes. Game theory, which untangles complex situations in which the best strategy of one player depends on the actions of another, was originally devised to study poker, chess, and the like. Later, economists adapted it to explain markets and competition, and since the 1970s it has attracted researchers from other areas, including animal behavior. In the last few years, some evolutionary biologists have taken game theory the next step: testing its predictions in the field. These empiricists are turning the tables on the traditional approach: applying game theory to explain Making sense existing data on, say, fightnestmates can 1 ing or cooperative hunting. Instead, these researchers have been making field observations and doing lab experiments specifically to test game-theory modelsand finding good agreement with the calculations in everything from spiders fighting over web sites to naked mole rats obeying a dominant female. The new give-and-take between theory and data, says Lee Dugatkin of the University of Missouri, Columbia, is allowing researchers to sharpen their models and get an increasingly detailed understanding of a wide range of animal behavior. This effort is also reverberating outside biology, says Peter Hammerstein, a theorist at the Max Planck Institute for Physiology of Behavior in Seewiesen, Germany, who collaborates with both animal behaviorists and economists. After all, game theory's predictions have generally been quite difficult to verify for humans. The problem, explains Oxford University zoologist Martin Nowak, lies in knowing what the payoffs are for a particular "game." Is making a large profit, say, more desirable than driving a competitor out of business? With animals, however, the reward for a successful strategy is easy to identify: an advantage, such as more food, a higher rank in a social hierarchy, or less competition for mates, that ends up increasing of turr the animal's reproductive success. As a result, says Hammerstein, "it's now very fashionable for economists to work in evolutionary game theory," and game theorists from other fields have also taken note. Social scientists, for example, hope that un< derstanding how a behavior 0 such as cooperation is mainh tained in animals will give them insights into the origins of similar behavior in human systems. Biologists had left game theory to the economists until 1973,whenJohn Maynard Smith of the University of Sussex, U.K., used the technique to explain when and why some animals, such as stags or fish, fight with each other. Maynard Smith proposed treating a given behavior as a strategy in a game and assuming that strategies strife ... Egret evolve just as physical charnmurderous. acteristics do. Thus, any welladapted population will follow the "best" strategy in this sense: Any mutants practicing a different strategy will reap a lower reproductive payoff and will die out. Maynard Smith named that optimum strategy an evolutionarily stable strategy, or ESS. Over the past 20 years theorists have modeled nearly every imaginable animal behavior as an ESS: aggression, cooperation, foraging, hunting, ... and oppression rivalry, and many more. queen threatens a "It was very exciting that the models seemed to predict all this," Hammerstein says. "But to see that it was more than just a superficial correspondence, they had to be tested." To demonstrate that an animal really does follow an ESS, researchers would have to collect enough data to calculate the exact reproductive payoff for the observed strategy and the alternatives. n. h ba