We agree that cultural group selection (CGS) has been under-theorized and applaud the attempt to develop a more formalized framework. However, there are two areas where the current effort falls short. The first is the lack of engagement with research on the development of complex cooperation and shared intentionality in our evolutionary history. The second area is twofold and involves the lack of explicit connection to many processes in the Extended Evolutionary Synthesis (EES; e.g., Laland et al. Reference Laland, Tobias, Feldman, Sterelny, Muller, Moczek, Jabonka and Odling-Smee2014; Reference Laland, Uller, Feldman, Sterelny, Müller, Moczek, Jablonka and Odling-Smee2015), most of which can assist in developing robust CGS arguments. Richerson et al. also measure and describe cultural variation and culture groups, the key elements in CGS, using radically different, and often non-compatible, scales and definitions, relying on a weak analogy use of the F _st statistic. These flaws undermine the strength of the argument, but are not lethal and can be corrected.

Richerson et al. seek support for their assertions with data from contemporary psychology, ecology, economics, and cultural anthropology, but do so without providing engagement with the evidence for when or how human cooperation began, data that exist in the archaeological and fossil record. The drive toward complex cooperation and the incipient patterns of CGS are significantly older than suggested in the target article. There is evidence of complexity in cooperative behaviors deep in Hominoid (ape) history and complex cooperation between non-kin individuals is part of an adaptive complex that predates the appearance of the genus Homo (Herrmann et al. Reference Herrmann, Call, Hernandez-Lloreda, Hare and Tomasello2007; MacKinnon & Fuentes Reference MacKinnon, Fuentes, Sussman and Cloninger2011; Malone et al. Reference Malone, Fuentes and White2012). In order to most effectively contextualize the emergence and role of CGS in the genus Homo, we must examine the Pleistocene evolutionary and archaeological record, which provides material evidence that can support these claims. The patterns and processes invoked in explaining CGS require that we have a working knowledge of the material, behavior, and ecological features of the expansive human niche across the Pleistocene. Recent work (Andersson et al. Reference Andersson, Törnberg and Törnberg2014; Fuentes Reference Fuentes2015; Grove & Coward Reference Grove and Coward2008; Sterelny & Hiscock Reference Sterelny and Hiscock2014; Stout Reference Stout, Whitten, Hinde, Stringer and Laland2012; Whiten & Erdal Reference Whiten and Erdal2012) demonstrates that symbol-making and socio-cognitive niche construction played a key role in our lineage's success in the terminal Pleistocene and is directly relevant to understanding the evolutionary landscape of interactions within and between human groups. This lack of engagement (there are only a few paragraphs with scant citations) represents a missed opportunity to demonstrate the evolutionary history and processes central in developing the capacity for, and effects of, CGS across the evolution of the genus Homo. Combining the material record, niche construction theory, and paleoanthropological data sets creates a key venue to test the theories presented in the article.

For example, there is archaeological evidence for symbol construction, formation, and use that predates the appearance of Homo sapiens (circa 200,000 years ago). Initial results from our broad-scale assessment of these data indicate support for Richerson et al.'s assertion that CGS was active “as far back as symbolic marking is evident.” While they suggest symbol-making may go back to 164 k.y.a., there are earlier signs such as the engraved bones from Bilzingsleben (Mania & Mania Reference Mania and Mania1988) which may date to 350 k.y.a. Ochre use older than 200 k.y.a has been uncovered at sites in Zambia (Clark & Brown Reference Clark and Brown2001), Sudan (Peer et al. Reference Peer, Rots and Vroomans2004), and the Netherlands (Roebroeks et al. Reference Roebroeks, Sier, Nielsen, De Loecker, Parés, Arps and Mücher2012), suggesting complex behavior and possible CGS at earlier dates. We suggest that key aspects of CGS emerged previous to identifiable symbol-making and then coevolved, via niche construction and selection (alongside drift and gene flow), with that capacity to develop a baseline for the radical leaps in complexity in the last 20,000 years that Richerson et al. focus on.

While we are supportive of the CGS concept, there is a lack of explicit connectivity to the Extended Evolutionary Synthesis (EES). Despite introducing three forms of cultural group selection, Richerson and colleagues retain a reliance on selection-based processes as the key architect of function. This misses the opportunity to engage with niche construction, evo-devo, phenotypic plasticity, and other processes in the EES under which “Organisms are constructed in development, not simply ‘programmed’ to develop by genes. Living things do not evolve to fit into pre-existing environments, but co-construct and coevolve with their environments, in the process changing the structure of ecosystems” (Laland et al. Reference Laland, Tobias, Feldman, Sterelny, Muller, Moczek, Jabonka and Odling-Smee2014, p. 162). The processes in the EES, in addition to selection models, and multiple patterns of evolutionarily relevant inheritance (genetic, epigenetic, behavioral, and symbolic; e.g., Jablonka & Lamb Reference Jablonka and Lamb2005) are central to the development of a more robust CGS approach. The current article does invoke non-genetic inheritance and implicitly uses a form of niche construction but does so without serious connection to the theoretical toolkit available. CGS will be more robust if it were explicitly connected to the EES.

We would be remiss if we did not note that cultural F _st does not function in the same way as a genetic F _st does. F _st measures the variation that exists when a metapopulation lives in discrete groups. However, the assumption that responses to one or two questions on a survey are analogous to “one or two loci for a genetic F _st ” is not correct. Variance used to calculate F _st in genetic or morphological contexts are based on physically definable measurements, but the answers by Saudi and Lebanese citizens to one of two questions about the role of clergy in the political system are not truly quantitative measures. They are influenced by multiple contextual and contingent (and linguistic) cultural variables and cannot be treated as independent material measures analogous to the number of alleles at a locus or inter-parietal breadth on a cranium. It is not clear that a cultural F _st measures what a genetic one does or that such a measure is actually useful for assessing CGS.

CGS is relevant and should be incorporated into our understanding of human evolution, but approaches and analyses need to be rooted in a contemporary evolutionary theory and the archaeological record, be untangled from certain assumptions about selection and fitness, and seek models and assessments appropriate for cultural processes.