Cooperation through repression

I've been meaning to write about the paper on primate policing by Jessica Flack and colleagues.

Using 'knockout' experiments on a large, captive group of pigtailed macaques (Macaca nemestrina), we show that a policing function, performed infrequently by a small subset of individuals, significantly contributes to maintaining stable resource networks in the face of chronic perturbations that arise through conflict. When policing is absent, social niches destabilize, with group members building smaller, less diverse, and less integrated grooming, play, proximity and contact-sitting networks. Instability is quantified in terms of reduced mean degree, increased clustering, reduced reach, and increased assortativity. Policing not only controls conflict, we find it significantly influences the structure of networks that constitute essential social resources in gregarious primate societies. The structure of such networks plays a critical role in infant survivorship, emergence and spread of cooperative behaviour, social learning and cultural traditions.

The paper cites a review by Steven Frank titled, "Repression of competition and the evolution of cooperation." From the abstract:

Repression of competition within groups joins kin selection as the second major force in the history of life shaping the evolution of cooperation. When opportunities for competition against neighbors are limited within groups, individuals can increase their own success only by enhancing the efficiency and productivity of their group. Thus, characters that repress competition within groups promote cooperation and enhance group success. Leigh first expressed this idea in the context of fair meiosis, in which each chromosome has an equal chance of transmission via gametes. Randomized success means that each part of the genome can increase its own success only by enhancing the total number of progeny and thus increasing the success of the group. Alexander used this insight about repression of competition in fair meiosis to develop his theories for the evolution of human sociality. Alexander argued that human social structures spread when they repress competition within groups and promote successful group-against-group competition.

In this way, Frank explicates the suppression of competition as a classic multilevel selection problem -- comparable to the problem of meiotic drive vs. meiotic fairness, in which chromosomes are selected for their fitness effects at the organismal level because there is no possibility for them to compete at their own level.

The paper opens with a great quote by Richard Alexander:

The function of laws is to regulate ... the reproductive strivings of individuals and subgroups within societies, in the interest of preserving unity in the larger group. ... Presumably, unity in the larger group feeds back beneficial effects to those ... that propose, maintain, adjust, and enforce the laws (Alexander 1979:240).

This raises a question: what are the beneficial effects of policing for the policer that outweigh its costs? Is it true, for instance, that they have more reproductive opportunities? Or that their efforts benefit their likely offspring?

Here is a prediction from Frank:

Repression of competition therefore contributes an immediate benefit to the group, enhancing the reproductive success of those that invest in policing selfish neighbors. Repressing selfish neighbors imposes a direct cost on those that police, thus policing spreads only when the beneficial effects of improved group success flow back to the policing individuals and their kin.
Policing increases from zero when r < 1 - c, where r is the relatedness of an individual to its group and c is the cost to an individual for investing its resources in policing. Note that low relatedness favors policing. When relatedness is high, selection favors self-restraint and cooperation without the need for policing by neighbors. As relatedness declines, selfishness tends to increase, causing a drop in group efficiency and the average success of each group member. Under conditions of poor group efficiency, policing increases because it enhances the quality of life and the success of those that police.

Now this is interesting. Policing can have the same effect as kin selection, in groups where the average relatedness of individuals is low. Its effect will be greater where the cost of policing is low.

But the Nature paper doesn't really discuss the issue of cost for the policer. It merely documents the social disruption that comes from "knocking" the policers out of the group.

In another 2005 paper in American Naturalist, Flack and colleagues discuss the issue of cost more directly. They begin by asserting that primate policing may be costly because it "requires intervention by third parties into ongoing contests." This entails the risk that one of the parties will attack the policer.

Yet, there are no quantitative data on the costs of policing, in terms of aggression received, to policers in primate or other mammalian species. Furthermore, although there are few quantitative data on the incidence of policing in primates, it appears that effective policing occurs in one-male groups, including gorillas (Gorilla gorilla; Watts et al. 2000) and golden monkeys (Rhinopithecus roxellanae; Ren et al. 1991), but, with the exception of chimpanzees (de Waal 1982), is rare in hierarchical multimale, multifemale societies (de Waal 1977; Ehardt and Bernstein 1992; Petit and Thierry 2000) like those of macaques and baboons. It is not understood why this is the case.

They develop the model that the effectiveness of policing is a function of power gradients -- single-male groups concentrate the power in one individual and allow lower-cost policing.

But their observations of pig-tailed macaques show that many of the most frequent policers are females, including one of the most effective ones. That seems to go against the single-male policing idea.

And there is this point:

Mathematical models of policing and of related intervention behaviors such as punishment are able to effectively reduce their number of state variables by assuming that all individuals pay the same cost for engaging in conflict management or repressing competition. Although this is a reasonable simplifying assumption, it makes the evolution of policing appear more difficult or improbable than actually seems to be the case. There is good evidence from the study of dominance relationships in animal societies for individual variation in resource holding potential or vigor (Clutton-Brock and Parker 1995) and, from this study, for individual variation in power. By allowing for variation in state, Frank (1995, 1996) has shown that small differences in individual vigor can lead to large variations in individual contributions to policing when relatedness is low. The claim that variation in individual vigor is related to variation in investment in conflict management requires the additional assumption that the cost of conflict management varies inversely with individual state. One of the findings of this study is that cost is not only a function of the individual intervening but also of the power values of the individuals engaged in the dispute. Thus, variation in individual vigor is not sufficient. A power structure must arise in which individuals also vary in the degree to which group members perceive them capable of successfully using force. It would be interesting to extend the Frank (2003) treatment to include this assumption.

Heterogeneity is what makes all these things complex to model mathematically. It's also why the Santa Fe Institute is interested in this kind of problem in social structure.

The social disruption resulting from lack of policing is interesting in its own right. The Nature study finds that the average level of aggressive interactions increases, the number of affiliative or reconciliatory behaviors (like grooming) decreases. The conclusion is the critical paragraph:

Conflict threatens to destabilize society. Its immediate consequences include injuries and damaged relationships. It has been demonstrated that policing can directly prevent this. We find that policing also has far-reaching indirect consequences, significantly altering construction of social resource networks that make group living advantageous. We demonstrated this by analysing changes to four network properties. We observe that when policing is operational, group members build larger social networks characterized by greater partner diversity and increased potential for socially positive contagion and cooperation. Without policing, high conflict frequency and severity leads to more conservative social interactions and a less integrated society. Mechanisms for buffering frequent conflicts are therefore essential for construction of stable social niches upon which individuals depend for behavioural resources.

What exactly is the problem with society being destabilized? One might speculate that the worst possible outcome is that different cliques of individuals will break apart into smaller groups -- in other words, the presence of policing enables the individuals to live in a larger group than would be possible without it. That might be a suboptimal solution for other reasons. For example, larger groups ought to carry some protection from predation. But many primates live in small groups anyway.

So are the policers an integral element of primate societies, or an emergent phenomenon more or less natural within large groups? In the latter case, policing might make some difference at an unstable boundary between stable, small group size and large groups that might break down without it.

If this were true, there would be no need to see policers as particularly selected for their function in group selection. Instead, there is a much simpler hypothesis: if the effect of policers is to maintain larger groups (at least in some cases), then policers are directly increasing their own mating opportunities (by keeping more females near them).

But that doesn't address why females would police. One idea about it would be that females face costs both ways. There is a cost to policing, but there is also a cost to living in a society without police. For females, this may come down to safety for offspring -- when aggression increases, juvenile mortality may increase with it.

This kind of behavioral variation might well be able to be studied in the wild. Behavioral traditions within groups probably lead to great heterogeneity -- especially if the number of policers in any given group is low. Some groups will have more aggressiveness than others, and juvenile survivorship may track that.


Flack JC, Girvan M, de Waal FBM, Krakauer DC. 2006. Policing stabilizes construction of social niches in primates. Nature 439:426-429. Full text (subscription)

Flack JC, de Waal FBM, Krakauer DC. 2005. Social structure, robustness, and policing cost in a cognitively sophisticated species. Am Nat 165:E126-E139. Full text

Frank SA. Perspective: repression of competition and the evolution of cooperation. Evolution 57:693-705. PubMed