Wilson describes a growing consensus concerning the role of culture in human evolution. While not everyone is yet a member (he excepts advocates of memetics and evoked culture), I am heartened by much of what Wilson describes. I readily join this consensus when it holds that cultural inheritance is an important tool that has allowed humans to thrive in wide variety of environments; that the properties of cultural inheritance can better explain human cooperation, including altruism, in larger groups than can be explained by genetic inheritance alone; that group-level selection is a useful way to think about human cultural evolution, especially because many important group-level traits are not easily reducible to individual-level fitness calculations; and that many of these group-level traits can be considered adaptations in that they help groups of humans survive in their environment. However, I would like to sound a couple notes of caution.
First, it can be problematic to describe groups as competing “cultures.” This may seem like a semantic quibble, but in this case terminology has the potential to lead us astray. For example, calling groups “cultures” gives the impression that groups are to be defined primarily by their common cultural traits. However, if our goal is to explain the distribution of cultural traits within and between groups, defining groups based on their distribution of cultural traits introduces endogeneity to our analysis. This issue can be avoided if, as in Sober and Wilson (1999, 92-98), groups are defined by interactions in relation to traits, instead of by the traits themselves.
For example, suppose two villages are involved in separate collective action problems, such as maintaining an irrigation system. In each village, a significant fraction of individuals are contributors and significant fraction are free-riders. In a multilevel cultural selection model, the “groups” should be defined at the scope of the collective action problem, in this case the villages, because this defines the scope of each individuals’ influence on others’ payoffs. The groups would not be defined as the set of cooperators and the set of free-riders because free-riders do not affect the payoffs of free-riders in another village in relation to their free-riding trait. Similarly, if each village was made up of recent immigrants from two different backgrounds, the relevant groups would still be defined by the village-level collective action problem and not by the origins of the village members.
Furthermore, characterizing groups as “cultures,” can also encourage essentialist thinking, i.e., discounting the importance of both within-group cultural variation and between-group cultural similarity. Essentialism has left the ethnographic record mostly bereft of the individual-level data necessary to build empirically-grounded theoretical models (Richerson and Boyd 2005,
246-253). This problem is more easily avoided if groups are not thought of as discrete “cultures,” but as sets of individuals who frequently interact with each other and have distributions of cultural traits that influence those interactions.
A second note of caution concerns what seems, to me, like an overly strong adaptationist emphasis. There are many flavors of adaptationism and it is important to identify which is under consideration. When Wilson states that “human cultures are primarily adaptive at the group level,” this seems to be a statement of what Godfrey-Smith (2001) calls “empirical adaptationism.” Empirical adaptationists posit that selection is the most dominant force governing the distribution of traits. While recognizing that the origins of individual traits must be examined on a case-by-case basis, Wilson seems to imply that selection at the group level is the most dominant force governing the distribution of cultural traits as a whole. This begs the question “most dominant compared to what?”
My sense is that this is still an open question and my concern is that if the consensus is to focus primarily on group-level adaptations, we may miss important group-level cultural traits that are not adaptations and many cultural traits that are more easily understood as a mix of group-level and individual-level selection. Consider intergroup warfare, which at first glance looks like a classic group-level cultural adaptation. Individuals put themselves at great risk and successful groups have the potential to gain wealth, territory and other resources. However, a closer look complicates the view that warfare is a group level cultural adaptation.
Warfare’s status as an adaptation is partially undermined by frequency-dependence between groups. Warfare can spread if war-like groups are more successful than peaceful groups. However, as in the classic Hawk-Dove game (Maynard Smith and Price 1973), the more warlike one’s neighbors, the more costly it is to fight and eventually, it may pay to become peaceful. Frequency dependence and balancing selection diminish the utility of adaptationist explanations. For me, it is much more parsimonious to consider how group structure influences the distribution of traits, than it is to consider whether that distribution, or some subset of that distribution, can be considered a group-level adaptation.
Furthermore, participating in warfare can be quite costly for individuals and they may, with some frequency, shirk fighting. Even in the most warlike groups when shirkers are actively punished, selection at the individual level drives some individuals to not participate (e.g., Mathew and Boyd 2011). If participation in warfare is an adaptation at the group-level, is non-participation in warfare an adaptation at the individual level? Which of these forces is more dominant? If fifty-one percent of individuals shirk, is it useful to conclude that culture is primarily adaptive at the individual level? Again, I think it is a more promising approach to consider all relevant forces, selective and otherwise, at different levels of analysis when thinking about the distribution of cultural traits. A strong focus on group-level adaptations may critically limit the effectiveness of our investigations.
In conclusion, I agree with the spirit of this growing consensus and my comments are merely an attempt to shift the emphasis. I strongly agree that recognizing the importance of group-level cultural selection is essential to understanding the distribution of cultural traits. However, I worry that the larger goal of understanding how the combination of selective and non-selective forces operate on cultural traits at different levels of analysis could be hindered by overreliance on essentialist and adaptationalist thinking.
References
Godfrey-Smith, Peter. 2001. Three Kinds of Adaptationism. In Adaptationism and Optimality, ed. Steven Hecht Orzack and Elliott Sober. Cambridge University Press.
Mathew, Sarah and Robert Boyd. 2011. “Punishment Sustains Large-Scale Cooperation in Prestate Warfare.” Proceedings of the National Academy of Sciences 108(28):1091{6490.
Maynard Smith, John and George R. Price. 1973. “The Logic of Animal Conflict.” Nature 246(5427):15{18.
Richerson, Peter J. and Robert Boyd. 2005. Not by Genes Alone. University of Chicago Press.
Sober, Elliott and David Sloan Wilson. 1999. Unto Others: The Evolution and Psychology of Unselfish Behavior. Harvard University Press.
Matthew R. Zimmerman. Cultural Evolution Lab, UC Davis
The above comment is just wrong as is DSW. As brain research proceeds and accelerates we see that pretty much everything behavioral, including language-behavior, is epiphenomenal to far deeper and instantaneous neural processes.
There is simply no time for any of these vaulted higher order concepts to have any effect, let along influence reproductive success. Yes, religious talk is correlated with more kids but we have no way to claim it is causal.
DSW and the above commentator claim as true what needs to be proven — experimentally, with double blind studies and replicated. Such rhetoric is polemical and ideology – wishful thinking.
Ultimately, higher order concepts like culture, personality, consciousness, emotions, choice, decision making are proving meaningless and carry no information value in predicting much if anything – like the statements of DSW.
Thanks to Zimmerman and the respondent for their comments.
I agree with Zimmerman’s point about describing groups as competing “cultures”. In the example that he gives, two villages contain solid citizens and free riders in different proportions. It is inappropriate to call free riders and solid citizens different cultures, no matter what group they are in. The villages are the salient groups because the villagers are influencing each other’s fitness. This is the basis of the concept of a trait-group, which is defined as “the set of individuals the influence each other’s fitness, with respect to a given trait”. If this seems abstract, consider ANY evolutionary model of social behavior, no matter what it is called. To determine the fitness of an individual, it is necessary to know the phenotype of that individual and the the phenotypes of the other individuals with whom it interacts–which constitutes the group for that individual.
To summarize, Zimmerman puts it well when he states that “This problem is more easily avoided if groups are not thought of as discrete ‘cultures’ but as sets of individuals who frequently interact with each other and have distributions of cultural traits that influence those interactions”.
Zimmerman’s comments on the importance of group-level adaptations are also well taken. The thesis of my essay might seem inconsistent. On one hand I say that all traits must be evaluated on a case-by-case basis. On the other hand I say that all nonhuman species are primarily adapted to their environments and that humans are primarily adapted at the group level. I don’t think that these two points are a contradiction for nonhuman species, and I don’t think that they are a contradiction for humans, but I’m happy to conduct a trait-by-trait evaluation and let the chips fall where they may.
I don’t think that frequency-dependent selection interferes with an adaptationist account. In a hawk-dove game, hawks are favored by within-group selection and doves are favored by between-group selection. If either type evolves to fixation, we call that a case of adaptation by natural selection. If the result is coexistence in a stable equilibrium, this is also a case of adaptation by natural selection.
Group selection is strong in human evolution largely because of mechanisms that suppress myriad forms of within-group selfishness, such as bullying, free-riding, and deception. These behaviors are only suppressed, never eliminated, so in this sense group selection is never complete.
I don’t think that the comment by kevinkindsongs is germane to the discussion. Any trait can be reduced to lower-level processes, but that does not render the higher-level descriptions obsolete. In fact, the opposite can be the case. A sizable literature on reductionism addresses this point.