Making Light Work of Brownian Motion

inspection" is risky for the scouts, but the information can benefit them as well as the rest of the school-if the interloper is not a predator or if it's not hungry, the smaller fish don't need to scatter. A group of scouts approaching a predator, Milinski noted, is playing out a Prisoner's Dilemma: Each has a strong incentive to defect and let the others take all the chances, but if all defect, they learn nothing about the predator. Full cooperation, on the other hand, minimizes the risks because the predator becomes confused if it can't focus on a single target. Because potential predators approach the school again and again, Milinski thought that a tit-for-tat strategy might have evolved among the fish. Milinski and Dugatkin have independently tested the idea-Milinski in sticklebacks and Dugatkin in guppies-and both find that the fish do indeed use a tit-for-tat strategy in predator inspection. Guppies that are paired up in a tank with a predator confined at one end will approach the predator in a sequence of moves, Dugatkin says. "If one of them is trailing, the lead fish will turn around and head back. It will wait for the other to head out, and then it will go by its side." In other words, if one fish defects (holds back), the other will, too, and it then waits for the first one to cooperate (swim forward) before cooperating itself. The guppies even remember from day to day what other guppies did, Dugatkin found. If one of a pair defects in one trial, the other will defect in turn on a second trial the next day. The verification of the tit-for-tat strategy has led to new and more detailed models of the guppies' behavior, Dugatkin says. "After doing that experiment, watching the fish, and thinking about the model, I realized that guppies should prefer to associate with cooperators because it would be in their interest to be near cooperators if a predator appeared." He later found that, given a choice, guppies did indeed spend more time with fish that had cooperated than with defectors. "These models make some new and very interesting predictions about the evolution of cooperation," Dugatkin says, "and we hope they will spur even more empirical work." Researchers from other fields will be watching this work unfold, says Hammerstein. Take economists, who have a hard time explaining how markets end up in Nash equilibrium, in which no competitor can gain an advantage by unilaterally changing strategy. Studies of markets suggest that Nash equilibria-the equivalent of ESSs in animals-do arise, but the theory predicting them assumes that the players act in a perfectly rational fashion, which is impossible. As a result, says Hammerstein, a number of economists are "looking to evolutionary game theory for processes other than rational decision making that could lead to a Nash equilibrium." Perhaps, he says, these stable strategies arise in much the same way as cooperation arises among guppies: People base their behavior not on rational calculation but on experience. Robert Axelrod, a political scientist at the University of Michigan, raises the possibility that evolutionary game theory might even offer insights into the election-year strategies deployed by candidates for the U.S. Congress. Perhaps politicians imitate the strategies of others, or perhaps some other process from evolutionary game theory is at work. One can only hope that negative campaigning does not prove to be an evolutionarily stable strategy. -Robert Pool