Evolution of Cooperation

The New Scientist has an good review of recent research on the evolution of cooperation – Charity begins at Homo sapiens. This research goes beyond ideas about kin selection. Experimental research has shown that people will punish unfairness even when there is nothing to gain. This work suggests that rule enforcement strongly supports the maintenance of cooperation.

Further support for the idea that strong reciprocity is an adaptation in its own right comes from the theoretical studies of economist Herbert Gintis of the University of Massachusetts, Amherst, anthropologist Robert Boyd of the University of California at Los Angeles, and others. They set up a computer model in which groups of individuals interacted, and watched how their behaviour evolved. Individuals were set up in the model to behave initially either as cheats or as cooperators, and in personal interactions the former came off best. When groups competed with one another, however, cooperation came into its own: groups with more cooperators were likely to flourish.

But that was only the start. The individuals, whether initially cooperators or cheats, were also programmed to copy successful behaviour. In simulations with groups ranging from 4 to 256 individuals, the team found that altruism could evolve. The benefits that cooperation conferred on a group outweighed its costs to individuals – but only in groups of less than about 10. Ancestral human hunter-gatherer bands are thought to have numbered 30 or more individuals, so how could cooperative behaviour have evolved and spread in these groups?

The answer lies in the fact that strong reciprocity is not simply a matter of cooperation; it also requires punishment of those who fail to toe the line. When the team added punishment to their models, they found it made a huge difference. In a second round of simulations, they included a new kind of individual: the “punishers”. These punishers were not only willing to cooperate with others but also to punish cheats. By making cheats pay for their antisocial actions, they tipped the balance towards cooperation. This time, competition between groups led to the emergence of cooperation in groups of up to 50 individuals (Proceedings of the National Academy of Sciences, vol 100, p 3531).

altruism graph

Could competition between small groups of our ancestors somehow have turned them into strong reciprocators? Gintis, Boyd and their colleagues believe so. What’s more, subsequent research by Fehr, working with economist Urs Fischbacher of the University of Zurich, suggests that as humans came to live in larger groups, their attitudes towards reciprocity may have become even more hard-line. Using a similar model to Gintis and the others (Nature, vol 425, p 785pdf), they found that cooperation can become the default behaviour in large groups provided punishers are willing to punish not only those who cheat, but also those who fail to punish cheats. “In this case,” Fehr says, “even groups of several hundred individuals can establish cooperation rates of between 70 and 80 per cent.”

This work is important, because understanding how cooperation emerges and persists is vital in understanding and designing human institutions, and in understanding the emergence of complex systems in general.

One thought on “Evolution of Cooperation”

  1. It is a pitty that journalists from popular magazines are ignorant about the literature (I met the author of this New Scientist paper a few years ago and he was largely unaware of the literature not published in Nature, Science and PNAS). The findings described in the paper about are not really novel. The importance of enforecement has been shown in laboratory experiments in the late 1980s/ early 1990s. See for example Ostrom, Gardner and Walker (1994) Rules, Games and Common Pool Resources, Michigan University Press.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>