Elsevier

Evolution and Human Behavior

Volume 34, Issue 5, September 2013, Pages 342-349
Evolution and Human Behavior

Original Article
Cultural evolution of cooperation: The interplay between forms of social learning and group selection

https://doi.org/10.1016/j.evolhumbehav.2013.06.001Get rights and content

Abstract

The role of cultural group selection in the evolution of human cooperation is hotly debated. It has been argued that group selection is more effective in cultural evolution than in genetic evolution, because some forms of cultural transmission (conformism and/or the tendency to follow a leader) reduce intra-group variation while creating stable cultural variation between groups. This view is supported by some models, while other models lead to contrasting and sometimes opposite conclusions. A consensus view has not yet been achieved, partly because the modelling studies differ in their assumptions on the dynamics of cultural transmission and the mode of group selection. To clarify matters, we created an individual-based model allowing for a systematic comparison of how different social learning rules governing cultural transmission affect the evolution of cooperation in a group-structured population. We consider two modes of group selection (selection by group replacement or by group contagion) and systematically vary the frequency and impact of group-level processes. From our simulations we conclude that the outcome of cultural evolution strongly reflects the interplay of social learning rules and the mode of group selection. For example, conformism hampers or even prevents the evolution of cooperation if group selection acts via contagion; it may facilitate the evolution of cooperation if group selection acts via replacement. In contrast, leader-imitation promotes the evolution of cooperation under a broader range of conditions.

Introduction

The extension, degree and diversity of cooperation among unrelated individuals are keys to the ecological success of humans. The term ‘cooperation’ refers to behaviours by which benefits arise from the interactions between individuals. Hence cooperative behaviour provides benefits at the group level. From the individual perspective, however, the evolutionary emergence and stability of cooperation are often puzzling. In particular, this holds for social dilemmas where performing a cooperative act is costly to the actor, and free-riding individuals can reap the benefits of cooperation without paying the costs.

In evolutionary biology, which is focused on genetic evolution, the evolutionary emergence and stability of cooperation are the subjects of a considerable body of literature (Axelrod, 1985, Lehmann and Keller, 2006, Nowak, 2006). Since the dawn of evolutionary theory, Darwin suggested that the evolution of cooperation might be explained by the differential performance of cooperative and non-cooperative groups in intergroup competition (Darwin, 1859, Darwin, 1871). Ever since then, this idea has been controversial (Leigh, 2010, Maynard Smith, 1964, Queller, 1992, West et al., 2007a). Evolutionary models demonstrate that selection between groups can indeed favour cooperation, but only under a limited range of demographic conditions (Lehmann and Keller, 2006, Lehmann et al., 2006, Leigh, 1983, Maynard Smith, 1964, Traulsen and Nowak, 2006). The problem is that within-group processes are typically faster than between-group processes. The rapid spread of individually favoured strategies (like refraining from cooperation) within groups erodes intergroup variation and, as a consequence, undermines the effectiveness of selection at the group level.

It has been argued that when social strategies are transmitted culturally rather than genetically, group selection can favour the evolution of cooperation under less restrictive conditions. The transmission of cultural traits is mediated by various forms of social learning, some of which play a key role in theories of cultural group selection. In particular, conformism, which is the individual tendency to acquire locally common strategies (Boyd & Richerson, 1985), can retard or prevent the spread of initially rare defective strategies (Henrich & Boyd, 1998). By homogenizing behavioural strategies within groups, conformism changes the distribution of variation within and between groups, rendering cultural group selection a potentially efficient force promoting the evolution of cooperation (Bowles et al., 2003, Boyd et al., 2003, Boyd et al., 2011, Boyd and Richerson, 1985, Boyd and Richerson, 2002, Guzmán et al., 2007, Henrich, 2004, Scheuring, 2009). Recent models, however, cast doubt on the facilitating role of conformism in the evolution of cooperation (Lehmann & Feldman, 2008). These models indicate that conformism can even hamper the evolution of cooperation because it hinders the spread of any new strategy in a group. Some modelling studies suggest that other mechanisms of social learning, such as the tendency to follow a group leader are more efficient in promoting cooperation via cultural group selection (Cavalli-Sforza and Feldman, 1981, Lehmann et al., 2008, Lehmann and Feldman, 2008). The contrasting conclusions of different modelling studies by different schools of thought have led to an on-going debate on the role of cultural group selection. Part of the debate centres on the range of parameter values considered reasonable by different authors. Perhaps more importantly, comparison across models is hampered by the fact that the models differ in their basic assumptions on social learning and group selection (Boyd et al., 2011).

In the theory of genetic evolution, there have been considerable confusion and debate around the meaning of ‘group selection’. Some authors even argue that this term should be abandoned altogether (West, Griffin, & Gardner, 2007b), since group selection is a special case of a more general hierarchical theory of selection (Frank, 1986). Yet the concept of group selection can be useful in the context of a group-structured population, since group traits may be an emergent property of the interaction of group members that cannot easily be reduced to individual-level traits (Okasha, 2006, Simon et al., 2012). When talking about group selection, it is crucial to distinguish between two main modes, each representing different mechanisms that can have different effects on the outcome of evolution (West et al., 2007b). First, group selection can be driven by the differential extinction and colonization of patches, that is, the replacement of less successful groups by more successful ones (Leigh, 1983, Maynard Smith, 1964). This mode of group selection played a prominent role in the earliest discussions on group selection and is currently been referred to as ‘old’ group selection (West et al., 2007b) or multilevel selection 2 (Okasha, 2006). We prefer to use a more descriptive term and will henceforth refer to ‘replacement group selection’ when group selection is driven by the replacement of less successful groups by more successful ones. Second, group selection can be driven by the differential production of individuals migrating to and settling in other groups, that is, the ‘contagion’ of groups by individuals derived from other groups (Rogers, 1990, Wilson, 1975).This mode of group selection has been coined ‘trait-group’ selection (Wilson, 1975), ‘new’ group selection (West et al., 2007b), or multilevel selection 1 (Okasha, 2006). As a more descriptive term, we propose to call this mode ‘contagion group selection’. In the biological literature, the distinction between group selection by replacement and group selection by contagion, or the lack of it, has produced extensive discussion (Okasha, 2004, Okasha, 2006, West et al., 2007a, West et al., 2007b, Wilson, 2007) and confusion (Wilson & Wilson, 2007). Despite this, there have been very few attempts to compare the requirements for each of these processes to work (García and van den Bergh, 2011, Lehmann et al., 2006).

In cultural evolution, the distinction between replacement and contagion group selection is as relevant as in biological evolution (Henrich, 2004). Replacement group selection corresponds to the cultural take-over of whole groups by other more successful groups (Boyd et al., 2003, Guzmán et al., 2007, Traulsen and Nowak, 2006). This may happen as a result of intergroup conflicts, where the winning group imposes their ‘culture’ upon subdued groups (Boyd et al., 2003, Guzmán et al., 2007, Traulsen and Nowak, 2006). It may proceed in a more indirect way, if less successful groups tend to disband and go extinct, while well-performing groups bud off subgroups recolonizing empty patches (Soltis, Boyd, & Richerson, 1995). Contagion group selection is mediated by the more gradual migration of cultural traits from one group to another. This may, for example, happen if the cultural traits observed in well-performing groups are preferentially imitated by the individuals of other groups, leading to the gradual introgression of group-beneficial strategies into less successful groups (Boyd and Richerson, 2002, Henrich and Gil-White, 2001, Lehmann et al., 2008). It is to be expected that, as in genetic evolution, the two modes of group selection have contrasting effects on the course and outcome of cultural evolution. Yet, systematic studies on these effects are currently lacking.

Before continuing it is important to clarify the meaning of the term ‘selection’ in the context of cultural evolution. We use a definition that is analogous to the usage of natural selection in genetic evolution, but somewhat more restricted than the definitions often given in the literature on cultural evolution (e.g. (Boyd and Richerson, 1985, Cavalli-Sforza and Feldman, 1981). In genetic evolution, natural selection refers to the preferential transmission of alleles that enhance the ‘fitness’ of individuals, groups or other entities, that is, the ability of these entities to survive and reproduce. Darwinian fitness is often measured in payoffs that are acquired in interactions with others. Natural selection is an important driver of evolutionary change, but there are many other processes (including genetic drift and mutation pressure) leading to a change in allele frequencies. These processes differ from natural selection in that the evolutionary success of an allele is not related to the effect this allele has on the performance of the individuals, groups or other entities harbouring this gene. Similarly, in cultural evolution, various processes lead to changes in frequencies of cultural variants. In some of these processes, the evolutionary success of a cultural variant is related to the ‘performance’ of individuals, groups or other entities harbouring this variant. Examples include the preferential imitation of high-payoff individuals or the higher rate of cultural take-over from groups with superior organisation or technology. However, as in genetic evolution, the cultural transmission of a trait is not necessarily linked to the ‘performance’ of this trait, or to any of its inherent properties. Examples of forms of social learning that lead to cultural changes independent of payoffs include conformism-based learning (imitating traits that are locally most frequent), and status- or reputation-based learning (e.g. following a leader irrespective of intrinsic qualities of his/her cultural traits). To maintain consistency with genetic evolution, we interpret cultural change that is driven by performance- or payoff-based social learning as analogous to natural selection, while cultural change driven by other forms of transmission (such as conformism) does not fall into this category.

In this paper, we study three forms of social learning: payoff-based learning, conformism and leader imitation. Firstly, humans preferentially copy strategies from successful individuals (Henrich and Gil-White, 2001, Kendal et al., 2009, Lehmann et al., 2008). In the context of a social dilemma, this social learning rule is expected to decrease rates of cooperation within groups since free-riding leads to higher payoffs. Secondly, under conformist learning individuals tend to adopt locally common strategies, thereby further increasing the frequency of those strategies in their groups. Thirdly, individuals may be inclined to follow a leader or a teacher in their group, so that the strategy of one influential individual tends to spread locally (Cavalli-Sforza & Feldman, 1981). Experimental results suggest that more than one of the abovementioned social learning rules can apply at the same time (Efferson et al., 2008, McElreath et al., 2008).

To clarify matters and to help resolving the disagreements in the literature on the role of cultural group selection, we developed an individual-based model of cultural evolution in a group structured population. The model allows to consider various mixtures of social learning rules (payoff-based learning, conformism, leader imitation) in the context of two contrasting modes of cultural group selection (contagion versus replacement). For all combinations of settings, we systematically varied the strength of individual and group selection with the aim of elucidating whether, and under which conditions, cultural group selection favours the evolution of cooperation in a social dilemma.

Section snippets

Overview

We simulate a metapopulation structured in m groups of n individuals each. Individuals face a social dilemma, modelled here as a public goods game. At each point in time, each has a pure strategy, ‘cooperate’ or ‘defect’. Cooperators contribute to the public good of their group at a cost. Defectors contribute nothing. All contributions to the public good yield a benefit that is distributed equally among all group members, irrespective of whether they have contributed to the public good or not.

Results

For each parameter combination, we calculated the frequency of cooperation in the metapopulation at the final time step averaged across the 30 replicate simulations. Fig. 2 gives an overview of the simulations that were initialised with one cooperative group. Each panel summarizes the outcome of cultural evolution as a function of the parameters α (strength of individual selection) and β (strength of group selection). The 24 small panels to the right correspond to the scenarios obtained by

Discussion

The last years have seen a debate about the role of social learning in the evolution of cooperation by cultural group selection. On the one hand, it has been argued that a social learning rule like conformism favours the evolution of cooperation (Boyd et al., 2003, Boyd et al., 2011, Boyd and Richerson, 2002, Choi and Bowles, 2007). This argument is based on the idea that conformism reduces behavioural variation within groups, thereby increasing the effects of variation between groups (Henrich

Acknowledgments

We thank the members of the Theoretical Biology group at the University of Groningen for stimulating discussions. Also we like to thank Rob Boyd and two anonymous reviewers for comments on an earlier version of this paper.

References (47)

  • L. Lehmann et al.

    The co-evolution of culturally inherited altruistic helping and cultural transmission under random group formation

    Theoretical Population Biology

    (2008)
  • R. McElreath et al.

    Applying evolutionary models to the laboratory study of social learning

    Evolution and Human Behavior

    (2005)
  • L. Rendell et al.

    Cognitive culture: Theoretical and empirical insights into social learning strategies

    Trends in Cognitive Sciences

    (2011)
  • I. Scheuring

    Evolution of generous cooperative norms by cultural group selection

    Journal of Theoretical Biology

    (2009)
  • R. Axelrod

    The evolution of cooperation

    (1985)
  • R. Boyd et al.

    The evolution of altruistic punishment

    Proceedings of the National Academy of Sciences

    (2003)
  • R. Boyd et al.

    Rapid cultural adaptation can facilitate the evolution of large-scale cooperation

    Behavioral Ecology and Sociobiology

    (2011)
  • R. Boyd et al.

    Culture and the evolutionary process

    (1985)
  • L.L. Cavalli-Sforza et al.

    Cultural transmission and evolution

    (1981)
  • J.-K. Choi et al.

    The coevolution of parochial altruism and war

    Science

    (2007)
  • C. Darwin

    The origin of species by means of natural selection: The preservation of favored races in the struggle for life

    (1859)
  • C. Darwin

    The descent of man, and selection in relation to sex

    (1871)
  • A.R. Guzmán et al.

    When in Rome, do as the Romans do: The coevolution of altruistic punishment, conformist learning, and cooperation

    Evolution and Human Behavior

    (2007)
  • Cited by (33)

    • Conformity and content-biased cultural transmission in the evolution of altruism

      2022, Theoretical Population Biology
      Citation Excerpt :

      In humans, conformity can increase with larger group sizes (Muthukrishna et al., 2016), and between ages 3 (anti-conformist) to 7 (conformist) (Morgan et al., 2015). Group selection (Boyd and Richerson, 1982, 1985; Lehmann and Feldman, 2008; Molleman et al., 2013) and network (Javarone et al., 2015) models have reached different conclusions as to whether conformity facilitates (Boyd and Richerson, 1982, 1985; Molleman et al., 2013; Javarone et al., 2015) or hinders (Lehmann and Feldman, 2008; Molleman et al., 2013) the evolution of altruism as a cultural trait. In group selection models, differences in the form of group selection and initial frequency of altruists can account for these opposite findings (Molleman et al., 2013).

    • Societal background influences social learning in cooperative decision making

      2018, Evolution and Human Behavior
      Citation Excerpt :

      Two prominent strategies are payoff-based learning (Schlag, 1998), in which people imitate successful peers, and majority-based learning (Boyd & Richerson, 1985), where people tend to conform to locally common behavior. These social learning strategies can have profound consequences for the dynamics of cooperation in groups (Boyd & Richerson, 1985; Burton-Chellew, El Mouden, & West, 2017; Burton-Chellew, Nax, & West, 2015; Lehmann & Feldman, 2008; Lehmann, Feldman, & Foster, 2008; Molleman, Pen, & Weissing, 2013; Molleman, Quiñones, & Weissing, 2013; van den Berg, Molleman, & Weissing, 2015). When individuals in a group mutually benefit from coordinating their behavior, the relative frequencies of payoff-based and majority-based social learning can affect the outcome of social interactions and change the stability of equilibria by altering their basins of attraction (Molleman, Pen, et al., 2013; Pena, Volken, Pestelacci, & Tomassini, 2009; Skyrms, 2005).

    • Human Sociobiology and Group Selection Theory

      2017, On Human Nature: Biology, Psychology, Ethics, Politics, and Religion
    View all citing articles on Scopus

    L.M. was funded by an Ubbo Emmius grant of the University of Groningen. A.E.Q. was funded by the Top Master scholarship of the University of Groningen.

    1

    LM and AEQ contributed equally to the manuscript.

    View full text