Evolution of Cooperation Through Coordinated Turn-Taking

 

Auber Bequest Award

 

 

 
Andrew Colman   Lindsay Browning
 

Funding body

 

Background

 

The introduction of genetic algorithms in the 1980s enabled researchers to investigate the natural selection of social behaviour using sophisticated computer simulations. Research has focused almost exclusively on the Prisoner’s Dilemma game, a type of social interaction between two individuals in which both players benefit if both cooperate, but each can do even better, at the expense of the other, by defecting. Prisoner’s Dilemmas occur ubiquitously in naturally occurring human and animal interactions, and the players often cooperate. A typical biological example involves birds that can remove damaging parasites from their bodies but cannot reach the tops of their own heads with their bills. A pair of birds can either groom each other, to their mutual advantage, or defect from this implicit contract. The outcome is better for each if both cooperate than if both defect, but a bird gets the best payoff by defecting while the other cooperates. In that case, the cooperative bird gets the worst payoff, expending time and energy but getting nothing in return.

 

The evolutionary problem is to explain how such social behaviour could have evolved, given that natural selection operates at the level of the individual organism or gene. This problem has been largely solved by the theories of reciprocal altruism and indirect reciprocity, and computer simulations have shown that, after thousands of repetitions, reciprocal strategies such as Tit for Tat (cooperate if and only if the other player cooperated last time) or Pavlov (repeat any action that led to a good outcome last time, otherwise switch) tend to evolve, resulting in widespread joint cooperation.

 

A major shortcoming of this influential research is its focus on games in which cooperation involves both players acting similarly. There are games, no less commonplace than the Prisoner’s Dilemma, in which favourable payoffs are possible only if one player acts one way while the other acts the opposite way. To cooperate successfully, the players have to alternate or take turns, out of phase with each other. A typical example is the game of Hero, illustrated by two predators feeding on prey while being harassed by scavengers. Each can either ignore the scavengers, or abandon the prey temporarily to chase the scavengers. The payoff is best for an animal that ignores the scavengers while the other chases them; second-best for an animal that unilaterally chases the scavengers and thus loses some feeding time; third-best if both ignore the scavengers and lose some of the prey to the scavengers; and worst if both simultaneously abandon the prey to chase the scavengers, thereby running the risk of losing the whole prey. If this interaction is repeated many times, then the animals benefit, in terms of natural selection, by coordinated alternation – by taking turns in chasing away the scavengers – and there is evidence to show that this type of turn-taking occurs quite commonly in nature.

 

Proposed research

 

Browning and Colman (2004) managed to show through agent-based computer simulation how this type of coordinated, alternating cooperation can evolve without any communication between players. Technical details are given in the original article. Using a genetic algorithm incorporating mutation and crossing-over, we showed that coordinated turn-taking can evolves in games with asymmetric Nash equilibria if the players benefit from it. Precisely how coordination evolves without communication is not fully explained, although we have testable hypotheses about it. This project is designed to study the nature, properties and phenomena of coordinated alternating cooperation in a range of games with asymmetric equilibria and to attempt to find a mechanism to explain how it evolves.

 

Research collaborator

 
The Auber Bequest Award was made to Andrew M. Colman. Lindsay Browning, University of Oxford, is a collaborator on this project.
 
Schedule
 
15 September 2005 - 15 December 2005: programming and preparation
15 January 2006 - 15 June 2006: running simulations
15 June 2006 - 14 September 2006: analysis and writing up
 
References

 

Browning, L., & Colman, A. M. (2004). Evolution of coordinated alternating reciprocity in repeated dyadic games. Journal of Theoretical Biology, 229, 549-557.

 

Colman, A. M., & Browning, L. (2009). Evolution of cooperative turn-taking. Evolutionary Ecology Research, 11, 949-963.