Human large-scale cooperation as a product of competition between cultural groups

A central question in human evolution is the unusual extent to which humans cooperate with genetically unrelated individuals in transient interactions. This contrasts sharply with cooperation patterns in other vertebrates. While established evolutionary theories of cooperation fail to fully explain this human behavior, Cultural Group Selection (CGS) theory offers a compelling yet controversial alternative. CGS posits that culture's significant role in human societies allows group-level selection to act upon culturally distinct populations. This is because, unlike genetic variation, cultural variation persists despite migration, as migrants adopt the cultural norms of their new groups. Competition between these culturally diverse populations then drives the evolution of social norms that favor the group. These norms would necessitate cooperation among culturally similar individuals, even in one-off interactions. Despite extensive discussion, CGS theory has lacked rigorous testing. Existing evidence is primarily indirect: historical and ethnographic observations congruent with CGS; assertions that the preconditions for CGS exist in humans; and demonstrations of the group-beneficial nature of human social structures. Consequently, the significance of CGS remains debated. This study directly examines a core, untested CGS prediction: the scale of cooperation is directly proportional to the population structure of cultural variation. Humans are embedded within nested social structures (communities, clans, ethnic groups, nations), each offering potential cooperation scales. The strength of CGS at any level will mirror between-group cultural variation at that level. High cultural differentiation between groups at a particular scale should lead to parochial norms promoting cooperation within the group and discouraging it with outsiders. Cultural FST (the proportion of total cultural trait variation between populations) is a crucial metric for assessing the relative strength of group-level selection versus individual-level selection. A negative association between cultural FST and inter-population cooperation is therefore expected. The study's logic is illustrated by comparing two hypothetical norms, X and Y. Norm X, which prohibits livestock theft among adherents, promotes group benefits and flourishes. Norm Y, restricting theft only from relatives, provides less benefit and thus declines through reduced growth, emigration, or adoption of successful norms. The shift toward norms like X, not necessarily altruistic, reflects group-structured cultural selection. Individual adherence to a beneficial norm requires either emigration to a group with that norm or a simultaneous shift by the majority within their own group. This study examines the content of cooperation norms, particularly to whom adherents feel obligated to assist, taking norm enforcement as given and assuming that deviation results in social penalties.

Several existing studies provide indirect support for or suggest the plausibility of Cultural Group Selection (CGS). Bell et al. (2009) demonstrated considerable cultural differentiation between countries using data from the World Value Survey, indicating the potential for CGS. However, this study didn't assess cooperation patterns in relation to cultural differentiation. Other research, such as that by Richerson et al. (2016), provides indirect evidence by highlighting the essential role of CGS in explaining human cooperation. Francois et al. (2018) offer more direct support, showing that increased competition between firms increased intra-firm cooperation, though this doesn't examine cultural variation across different social scales. Various other works have explored the group-beneficial aspects of human social institutions (e.g., Henrich et al., 2006; Fehr & Gachter, 2002), which indirectly supports the CGS mechanism. However, the relative importance of CGS compared to other evolutionary mechanisms for cooperation remains a subject of ongoing debate and this study aims to directly test its key predictions.

This research was conducted among pastoralist communities in northern Kenya, encompassing four ethnolinguistic groups: Borana, Rendille, Samburu, and Turkana. These groups are structured hierarchically by patrilineal clans and moieties, with the Turkana also having territorial subdivisions based on grazing rights. A total of 759 individuals across nine clans and three Turkana territorial sections participated. The sample was fairly balanced in terms of gender, with a slight skew towards female participants. This design allows for the investigation of cooperation at multiple social scales (clan, ethnic group). Researchers used two main instruments: a questionnaire to assess cultural differentiation and vignette scenarios to measure cooperation rates. The cultural differentiation questionnaire included 49 social norms relevant to pastoral livelihoods, categorized into five themes: cooperation, crime and punishment, raiding, cultural markers, and family dynamics. The vignette scenarios presented 16 situations where a character could help or harm another (unknown) character. These scenarios systematically varied the target's social relationship to the main character (same clan, different clan within the same group, different ethnic group). The Rendille also had an additional condition (Samburu) as they maintain unusually friendly relations with Samburu. Cultural FST values were computed for each norm and then averaged across all 49 norms for each pair of groups. Geographic distance between groups was calculated using participant GPS coordinates, also averaged for each pair. Data analysis involved a mixed-effect logistic regression, with the dependent variable indicating whether a subject endorsed the cooperative act. The model included random effects for subject ID, vignette scenario, and the subject's lowest level group membership. Fixed effects were the cultural FST and geographic distance between the actor and target's groups. This model enabled testing the effects of cultural differentiation and geographic distance on cooperation, while controlling for other potential factors.

The study revealed substantial cultural differentiation among the groups, particularly between ethnolinguistic groups, providing scope for CGS. The average FST values for cultural traits ranged from 0.002 between clans to 0.215 between ethnic groups. These findings are comparable to existing studies on cultural FST, with values substantially higher than corresponding genetic FST estimates. This supports the idea that cultural group selection plays a prominent role. Analysis of cooperation rates showed a significant negative association between cultural FST and cooperation. Lower cultural FST values between pairs of groups predicted higher cooperation rates, indicating a strong relationship between cultural similarity and cooperation. Geographic distance did not have a significant effect on cooperation rates. The effect size of cultural FST on cooperation was substantial, demonstrating the impact of cultural differences. This correlation is robust across all subpopulations and nearly all vignette scenarios. The scenarios showing the highest cooperation rates and the least decline with increased FST involved assistance in vulnerable situations (sharing food, helping an injured person). Notably, norms related to warfare, measured by FST values based on raiding norms, had the strongest impact on the scale of cooperation, supporting the CGS hypothesis. The study also revealed variations in cooperation rates across different subpopulations, possibly due to factors such as religious influences (in the Borana group) or strategic alliances (between Rendille and Samburu).

The findings strongly support the hypothesis that cultural group selection has played a significant role in shaping human cooperation. The observed correspondence between cultural differentiation and the scale of cooperation directly confirms a key, previously untested prediction of CGS theory. Competing explanations for human cooperation, such as misfiring of psychology designed for familiar interactions or ease of coordination among culturally similar individuals, are less convincing. The lack of influence of geographic proximity on cooperation rates directly challenges the “misfiring” hypothesis, while the nature of the cooperation scenarios (involving costs and requiring social sanctions) weakens the coordination-based arguments. While within-group evolutionary processes shaping norms are also plausible, they fail to readily explain the observed social scale of cooperation, particularly given the dynamic and overlapping territories of the studied groups and the significant costs associated with inter-group conflict (e.g., high mortality rates in cattle raiding). The results suggest that CGS has potentially shaped the human cooperative psychology throughout evolutionary history, not just contemporary norms. Further research is necessary to determine the extent to which this is due to direct CGS effects versus indirect effects resulting from a CGS-shaped psychology.

This study provides compelling evidence for the influence of cultural group selection on the evolution of human large-scale cooperation. The strong correlation between cultural similarity and cooperation rates directly supports a central prediction of CGS theory. This work significantly advances our understanding of the evolutionary forces shaping human social behavior and its cultural variability. Future research could explore the dynamics of CGS in other cultural contexts, particularly in societies undergoing rapid cultural change, and investigate further its potential role in shaping other aspects of human sociality.

The study is geographically limited to a specific region in Kenya and focuses on pastoralist societies. While the findings are significant, their generalizability to other cultural contexts requires further investigation. The use of vignette scenarios provides valuable insights into cooperation norms, but the scenarios are not real-life situations and may not fully capture the complexity of human interactions. Also, the study's reliance on self-reported data may introduce potential biases. While efforts were made to mitigate these limitations through rigorous methodology and careful data analysis, future research employing diverse methods could enhance understanding further.