Researchers have tackled the problem of cooperation (and the related one of the evolution of altruism) for some time now. Initial game theory models suggested that cooperative animals would quickly be supplanted by selfish ones because of a cost to cooperation, as in the the so-called simple prisoners’ dilemma.

In these situations, typically an individual can choose between an action that benefits everyone, but at a cost to oneself, and an action that clearly benefits the individual, but comes at the risk of greater loss if everyone adopts that strategy. It turns out that most of us would rather lose than share, a sad but to many not surprising commentary on the human condition.

However, people eventually realized that cooperation makes the most sense in social groups, where phenomena such as kin and reciprocal altruism can take place. Sure enough, if the prisoners’ dilemma game is played iteratively, instead of in just one round, and if the players are allowed to keep track of what other players are doing (i.e., to build “reputations”) then it turns out that cooperating, even at a certain cost to oneself, is the winning long-term strategy.

Cooperation is therefore possible, but only if it aims at helping our close relatives (kin altruism) or if there is an expectation of reciprocity (I’m going to scratch your back now, but I fully expect you or someone else to return the favor, eventually).

Nature magazine recently published a fairly hard to read but fascinating paper entitled “Social diversity promotes the emergence of cooperation in public good games” (10 July 2008 issue). The authors, Francisco Santos, Marta Santos, and Jorge Pacheco (the first at the University of Brussels, in Belgium, the other two at the University of Lisbon, in Portugal) argue that cooperation in human groups can evolve as a result of social diversity, an outcome that would make concerted efforts at increasing diversity not only ethically good, but practically useful as well.

What Santos and collaborators did was to investigate what happens when another simplifying assumption of game theory models is relaxed: the players are usually considered equivalent in all respects, something that clearly doesn’t happen in the real world. So they simulated groups where there are different “neighborhoods” with individuals characterized by distinct degrees of “connectivity” to other individuals. So not to run into confounding effects from the already well known phenomena of kin and reciprocal cooperation, their model excluded those possibilities, focusing only on the possible effects of diversity.

The results were rather spectacular: Santos et al. observed a significant boost to cooperation in their simulated groups, which they attributed to the fact that the cost of cooperating was no longer fixed, as in the standard models, but depended on the social context of each individual. The more connected you are, the less the cost of cooperation, and as a result your fitness as a cooperator tends to increase. Moreover, the fitness of cheaters tends to diminish, because they breed other cheaters, to the point that their neighborhood is made mostly of cheaters, in a continuous negative feedback loop. If these scenarios remind you of well known human environments, you got the point.

It will be interesting to see what happens when one combines all three known sources of cooperation among individuals: kin selection, reciprocal altruism and increased diversity/connectivity. The point remains, however, that we now have an additional mechanism to explain an otherwise puzzling human (and some other primate’s) trait. Not all news is good from the Nature paper, however: it turns out that connected neighborhoods also create as a byproduct another new phenomenon not observed in standard models: unequal wealth distribution. Oh well, we really don’t live in the best of all possible worlds.