Figure 1. The logic of the positive feedback underlying the runaway brain-culture coevolution, as suggested by the ‘cultural drive’ hypothesis. A - the logic of the cultural drive (based primarily on Laland, 2017), B - minimalistic representation of the cultural drive, as implemented in TribeSim program.
Figure 2. Initial stages of brain-culture coevolution under basic parameters (no between-group competition, ‘Machiavellian’ culture, see Supplementary Table 1). See text for further explanations.
Figure 3. Brain-culture coevolution under basic parameters (no between-group competition, ‘Machiavellian’ culture, see Supplementary Table 1). Averages from 10 model runs are shown.
Figure 4. Brain-culture coevolution under three levels of between-group competition: no competition (G= 800), moderate competition (G=40), and strong competition (G=15). The culture is ‘Machiavellian’, all other parameters as in fig. 3 and Supplementary Table 1. Averages from 10 model runs are shown.
Figure 5. ‘Vicious circle’ of meme simplification. The necessary assumption is that learning one meme of the size ntakes less time than learning m memes of the size n/m , which seems likely. Under such circumstances, the saturation of culture with smaller memes results in weaker selection for memory capacity due to less efficient lifetime-wise use of memory.
Figure 6. Simulation results for three different cultures (Machiavellian, Cooperative, Complex) and three levels of between-group competition (15 – very strong, 40 – moderate, 800 – absent). Averages from 10 model runs are shown. For Memory clogging, averages are shown for years 1 – 70000; for other parameters, averages are shown for years 50000 – 70000.
Figure 7. The dynamics of memory loading for two cultures (cooperative and complex) and three levels of BGC (15 – very strong, 40 – moderate, 800 – absent). Averages from 10 model runs are shown. The left top diagram (Cooperative culture, G=800) depicts the situation when the runaway brain-culture coevolution did not start, brain and MC remained very small, and the culture consisted of very few simple memes.
Figure 8. Brain-culture coevolution with evolvable, costly learning efficiency (see text for explanation). All other parameters and designations as in fig. 6. Averages from 10 model runs are shown. For Memory clogging, averages are shown for years 1 – 70000; for other parameters, averages are shown for years 50000 – 70000.
Figure 9. Evolution of social learning (LE, TE). A – genotypic and phenotypic values of LE and TE (phenotypic value = genotypic value + sum of the effects of the corresponding memes). The efficiency of ‘learning culture’ or ‘teaching culture’ is phenotype minus genotype. Note the extensive cultural evolution of teaching efficiency. B – memory loading; note that the TE memes occupy much larger proportion of memory than LE and TrE memes. Only HE memes are more successful, because they are strongly favoured by the cultural group selection. Averages from 10 model runs are shown. Parameters: G=40 (moderate between-group competition); 5 categories of memes are allowed (TrE, HE, Useless, LE and TE); parameters for LE and TE memes: mean efficiency 0.2, standard deviation of efficiency 0.3; C=5; R=2 (see Supplementary Table 1 for further explanations); genes for MC, LE and TE can mutate (for LE and TE genes: mutation rate 0.04, mean mutation effect 0, standard deviation 0.1); brain volume = 20 + MC + 30 x (genotypic value of LE + genotypic value of TE); initial values of LE and TE: 0.05, 0; other parameters were the same as previously described.
Figure 10. The effects of population size, meme invention rate, lifespan and migration rate on brain-culture coevolution. The results are compared with the ‘basic’ situation (complex culture: TrE, HE and Useless memes allowed; costly, evolvable LE; G=40; all other parameters as in fig. 8). Population x 3: the amount of resources available from the environment is multiplied by 3 (R=9000; this results in triple population size), other parameters as in the ‘basic’ situation. Creativity x 3: meme invention rate was 0.0004 (instead of 0.000133) per each meme category per year; other parameters as in the ‘basic’ situation. Lifespan x 2: probability of death per year is Age x 0.0005 (instead of Age x 0.002); this results in average lifespan 52 years instead of 27 years in the ‘basic’ situation. Migration rate 0: no migration between groups. Migration rate x 4: migration chance was 0.004 per individual per year instead of 0.001 in the ‘basic’ situation. Averages for years 50000 – 70000 from 10 model runs are shown.