McElreath et al. (2003) Shared norms and ethnic markers

McElreath, R., Boyd, R., & Richerson, P. J. (2003). Shared norms and the evolution of ethnic markers. Current Anthropology, 44, 122-130. 

Background

• Unlike other primates, humans are often subdivided into ethnic groups marked by seemingly arbitrary traits. 
• A recurring intuition is that, since these markers signal group membership and ethnic groups are often loci of cooperation, markers persist because they allow people to selectively direct altruistic behavior toward coethnics. 
• This argument is flawed. Altruism can only evolve if a cue allows altruists to interact with each other preferentially. 
• Another idea is that selection could favor altruists who carried an external marker, allowing them to limit cooperation to those with the marker (greenbeards). 
• This mechanism is unlikely. Nonaltruists with the marker do best. 

This paper argues that markers function to allow individuals to interact with others who share their social norms. They present a simple mathematical model that shows that marked groups can arise and persist if: 

1) Social behavior in groups is regulated by norms, such that interactions between individuals who share norms yield higher payoffs than interactions among individuals who do not;

2) People preferentially interact with people with whom they share easily observable traits (e.g., dress style, dialect); 

3) People imitate successful people, such that behaviors that lead to higher payoffs tend to spread. 

They also show that the preference to interact with people with markers like one's own may be favored by natural selection under plausible conditions. 

The Model

Behaviors: When two individuals meet, they play a coordination game. There is a benefit for both doing the same behavior, of two (0 and 1). The strength of this benefit is \delta. Assume that it is hard to determine what the other player will do. 

Markers: There is also also a marker trait, with two variants (0 and 1). Assume that individuals tend to interact with others of the same marker, and the strength of this propensity is e. When e = 1, individuals interact at random, when e = 0, they always interact with someone with the same trait. 

Evolution: Imitation of individuals is biased for prestige -- individuals with higher payoffs are more likely to be imitated in the next generation. Sometimes, an individual may choose independent targets for imitating marker and behavior traits. Also, they assume two groups, which mix via a migration rate m. 

The model was tested via numerical simulation. 

Results

1. Stable behavioral differences between groups usually become ethnically marked.

Social interaction alone can lead to the evolution of stable differences in behavior between two groups. People with more common behaviors achieve higher payoffs in the coordination game and are more likely to be imitated. Thus if one behavior is initially common in one group and the alternative be- havior is initially common in the other group, payoffs from social behavior coupled with imitation of the successful will cause the groups to become more different. If the diversifying effect of payoff-biased imitation is sufficiently strong compared with the homogenizing effect of migration, the two populations will reach an equilibrium at which behavior 1 is common in group 1 and behavior 2 in group 2. In contrast, if the rate of mixing is too high or if initially the same behavior is common in both populations, only one behavior will be present in both populations at equilibrium. 

2. Spatial structure is needed to generate ethnic markers, but not to maintain them. 

3. Increasing the number of populations increases the range of initial conditions that give rise to ethnic markers.

4. Group differences are strongest at boundaries.

5. A more general model of social interaction leads to similar results.

They got rid of the symmetry of the coordination game and still got ethnic markers.