News :: story cache

Game theory in the popular press.

Experiments Suggest Animals can Cooperate with Each Other

If Benefits Accumulate and Partners Reciprocate
University of Minnesota News Service
December 13, 2002
text is a cache of

Animals can be a pretty uncooperative lot. While species like lions and prairie dogs cooperate in some cases, scientists seldom know the costs and benefits of cooperative acts in the wild. In particular, scientists have long been interested in situations in which cooperating animals give up something now in order to develop a relationship that pays off in the long run. Experimental studies show, however, that animals don't usually cooperate in these cases; apparently, they are unwilling to pass up an immediate benefit in order to gain more in the long run. Now, experiments with blue jays at the University of Minnesota suggest that animals may be induced to cooperate when their opponent reciprocates by tit-for-tat behavior and rewards accumulate over a sequence of plays. The work, which will be published in the Dec. 13 issue of Science, suggests that these are among the factors guiding evolution of some animals--including humans--toward cooperative behavior.

"Our results suggest that the timing of the benefits of cooperation is really important," said lead investigator David Stephens, an associate professor of ecology, evolution and behavior in the College of Biological Sciences. "For example, suppose you want two toddlers to work together, and you reward good behavior at every turn. They're likely to soon fall back into bad behavior. But if they have to play nicely with each other for 10 trials before receiving a reward, then they are likely to do it."

Animals that behave in reciprocal (tit-for-tat) fashion and reap rewards only after a sequence of interactions may be those most likely to cooperate in nature, said Stephens. An animal that knows it must help out in order to receive help--perhaps by group hunting or defense of young--and that it must interact several times before realizing a reward--waiting minutes or hours before predators withdraw or prey is secured--is the best candidate to exhibit cooperative behavior. By testing pairs of blue jays, Stephens and his colleagues were able to manipulate conditions to see whether reciprocal behavior and/or delayed rewards would lead to a stable pattern of cooperation.

The Stephens group trained the blue jay pairs to hop on one of two perches when a light came on; this led to one or more food pellets being delivered. The birds could "cooperate" by choosing a perch close to the other bird or "defect" by hopping on the perch farther from the other bird. Neither bird could see what the other was doing until they had chosen their perches. "Reciprocation" meant that a bird did whatever the other bird had done on the previous round.

In each experiment, one blue jay--the "stooge"--was forced to either defect all the time or reciprocate all the time. The other bird, which had free choice, was called the "subject." The subject's payoffs were determined by a Prisoner's Dilemma game, a well-known game used in studies of cooperation. When the stooge reciprocated and food pellets were allowed to accumulate in view of the subject for several rounds of play, the subject cooperated.

One implication of this theory is that social (cooperating) animals may have a superior ability to wait for delayed rewards--a hypothesis Stephens said should be tested. As for humans, nature and history have already run tests.

"In the case of humans, I think we are more likely to cooperate when others reciprocate and the benefits accumulate with time," said Stephens.