Introduction
One of the most intriguing features of hominin evolution is the rapid increase in brain volume, both absolute and relative, which was especially pronounced during the last two million years in the Homo clade. Within this time interval, brain volume increased threefold, from approximately 400-500 cm3 in the ancestral australopiths to 1300-1500 cm3 in the Late Pleistocene species such as Neanderthals and early modern humans (Rightmire, 2004; Schwartz et al., 2004; Roth, Dicke, 2005; Sherwood et al., 2008; Leigh, 2012; Neubauer, Hublin, 2012; Holloway, 2015). This striking pattern of directional and accelerating evolution towards larger brains appears to be quantitatively unique among primates (Miller et al., 2019). Given that a large brain is very expensive metabolically and imposes other costs, e.g., increased load on the cervical region of the spine and difficulties at giving birth to larger-headed offspring (Mink et al., 1981; Leonard, Robertson, 1992; Gavrilets, 2015), its continued expansion during two million years of human evolution implies unusually strong and long-lasting selection pressures in favour of larger-brained individuals (or groups containing such individuals). This, in turn, almost inevitably implies a positive feedback loop in the evolution of the human brain: its expansion must have been promoting further expansion (Holloway, 1967; Crespi, 2004; Miller et al., 2019).
Several kinds of hypothetical feedback mechanisms have been suggested in this context including sexual selection for intelligence accelerated by Fisherian runaway process (Miller, 2000), within-group competition for social status which boosted the evolution of increasingly elaborate ‘Machiavellian intelligence’ (Humphrey N.K., 1976; Byrne, Whiten, 1988), and ever-increasing between-group competition in a higly social, ecologically dominant species which resulted in accelerated evolution of cognitive abilities needed for effective within-group cooperation and group-beneficial behaviours (Alexander, 1989; Gavrilets, 2015).
More recently, another type of hypothetical positive feedback mechanism termed ‘cultural drive’ started to receive increasing attention among researches (Laland, 2017). The term was originally coined by Allan C. Wilson (Wilson, 1985) who hypothesized that enhanced cognitive abilities, especially the abilities for social learning and cultural transmission of adaptive behaviours (Laland, Galef, 2009), can accelerate biological evolution. Smarter animals invent new adaptive behaviours more often and transmit them across generations more effectively; new cultural traditions create new selective environments (‘cultural niche construction’ [Laland et al., 2001; Laland, O’Brien, 2011]) in response to which animals evolve faster (Wilson, 1985). The ‘cultural drive’ hypothesis was further elaborated by Kevin N. Laland (Laland, 2017) and other researches who suggested that the coevolution of social learning, cognitive abilities and culture can be self-sustaining (Whiten, van Schaik, 2007; Heyes, 2012; Laland, Rendell, 2013; Whiten et al., 2017). In its simplest form, the positive feedback mechanism of the ‘cultural drive’ can be described as following: better social learning and cognition → more behavioural innovations become fixed as cultural traditions; richer culture → more skills available to be learned from conspecifics; increased usefulness of learning abilities; more sophisticated and flexible behaviour results in new cognitive challenges → selection for still better social learning and cognition (Fig. 1).
Additional positive feedback loops are conceivable, e.g., via longer lifespan (elaborate culture → enhanced survival → longer lifespan → better prerequisites for intergenerational transfer of knowledge → still more elaborate culture → stronger selection for enhanced social learning) or via better nutrition (elaborate culture, including enhanced food acquisition strategies → enhanced nutrition → relaxed constraints for the evolution of larger brains) (Kaplan, Robson, 2002; Crews, 2003; Caspari, Lee, 2004; Laland, 2017).
The cultural drive hypothesis has received some empirical support (e.g., Kopps et al., 2014; Foote et al., 2016; Whiten, 2017), including recently described positive associations between social learning proclivity, absolute and relative brain volume, longevity, social group size and technical innovation in primates (Navarrete et al., 2016; Street et al., 2017) and cetaceans (Fox et al., 2017).
Importantly, the ‘cultural drive’ hypothesis does not contradict other ideas such as ‘Machiavellian intelligence’, ‘social brain’, ‘cooperative brain’, ‘brain for stone tool production’ or ‘brain for mate attraction’. In fact, the ‘cultural drive’ hypothesis can embrace a wide variety of such ideas, because it does not particularly rely on any specific type of culturally transmitted skills or behaviours. Potentially, any skills will do, given that they are cognitively demanding, require high-fidelity social learning and provide reproductive benefits, thereby enhancing the spread of the good learners’ genes.
This versatility is vividly illustrated by a computer model designed by Gavrilets and Vose (Gavrilets, Vose, 2006). Although their paper is titled “The dynamics of Machiavellian intelligence”, the model actually simulates cultural drive in one of its simplest forms. In fact, it accomodates the theories of Machiavellian intelligence and sexual selection within the framework of cultural drive. A brief description of this model is necessary for understanding the theoretical context of the current study. Simulated males in a polygynous, promiscous population compete for mates. Males sporadically (and very rarely) invent ‘Machiavellian memes’, that is, behaviours that improve their competitive ability. Memes can be acquired by other males via social learning. The chances to successfully learn a meme depend on the meme size (‘complexity’) and the male’s memory capacity and learning ability. Both characters are ‘costly’, that is, they decrease survival (it is assumed that they require larger brains, although brain volume is not modeled expicitly). Genes for memory capacity and learning ability mutate at a specified rate. Meme size positively correlates with its fitness effect, but the correlation is weak. Initially, all males have zero memory capacity and learning ability. The evolution of the simulated population starts with a more or less prolonged ‘dormant phase’ during which both memory capacity and learning ability remain low (as slightly deleterious traits at the mutation-selection equilibrium), and only newly invented memes are present in the meme pool. But sooner or later a self-accelerating process starts which the authors call ‘cognitive explosion’. During this phase cognitive abilities, population’s cultural richness (meme count) and Machiavellian fitness of individuals all increase in a runaway fashion. The cognitive explosion is fuelled by cultural drive (although the authors do not use the term): the more memes there are in the meme pool, the more it pays to have good memory and learning ability (Gavrilets, Vose, 2006).
Here we elaborate on this approach by designing a more complicated model, TribeSim, aimed to explore the impact of different factors on the dynamics of the brain-culture coevolution in a highly social species. The most important of the studied parameters include the extent of within- and between-group competition, different types of memes (individually beneficial, group-beneficial, useless or maladaptive) and their combinations (specialized culture vs. complex culture). We set out to elucidate the possible prerequisites for the most extensive brain expansion observed in the Pleistocene hominins, and we argue that the early representatives of the Homo clade, but not the other apes, probably found themselves in a situation suitable for an unprecedentedly powerful ‘cultural drive’.