CONCLUSIONES Y OPINIÓN PERSONAL FUEGO
QUÉ TENEMOS QUE HACER?
The interplay between the fidelity of transmission and population size has been explored by several studies with similar results (Mesoudi, 2011c; Aoki, Wakano and Lehmann, 2012; Kempe, Lycett and Mesoudi, 2014; Nakahashi, 2014; Acerbi, Tennie and Mesoudi, 2016), but it is one specific study that has been most persuasive in definitively establishing population size as a crucial factor for cumulative culture. Henrich developed a model that investigated the conditions under which skill in a population can accumulate and get lost, guided by the case of Tasmanian technology loss. When Tasmania became cut off from the Australian mainland around 10,000 years ago, the population lost a collection of useful complex skills and technologies, like winter clothing and boomerangs, and the author argues that this loss is due to a decrease in effective population size that resulted from the cut-off (Henrich, 2004).
In this model, individuals in a population could accurately identify and attempt to copy the best skilled individual. Learning is error-prone though, and a parameter of the model determines the variation in this individual copy error. In line with how difficult cumulative cultural skills are, most of the time, when attempting to copy, individuals will not manage to reach the best skill level in the population. Nonetheless, through lucky accidents or successful innovations, a small proportion of individuals will acquire a higher skill level, thus improving the overall population skill level. Thus, over time through repeated improvements on the best skill, the population cumulatively advances in skill. The key finding here is that the change in skill depends on the population size. The more individuals there are, the higher the chance that someone will improve the highest skill. The converse of this is that, as the population size decreases, so does the probability of beneficial learning errors, which means that after a certain threshold the population begins losing skill rather than accumulating it, which, Henrich argues, is what explains the Tasmanian scenario.
Henrich’s model relies necessarily on the assumption that individuals can accurately identify and copy the best skilled individual in the population. This assumption, as the discussion of social learning strategies above has shown, is far from trivial. This study was not without its critics (Henrich, 2006; Read, 2006, 2009), but it proved nonetheless influential, and has since been
adopted and modified to investigate questions related to population structure, migration, overlapping generations, and different social learning strategies (Powell, Shennan and Thomas, 2009; Bentley and O’Brien, 2011; Lehmann, Aoki and Feldman, 2011; Mesoudi, 2011c; Vaesen, 2012; Kobayashi, Ohtsuki and Wakano, 2016). Similar models have been fitted to archaeological data, indicating that the appearance and disappearance of complex technologies in the Palaeolithic coincides with demographic changes (Shennan, 2001; Powell, Shennan and Thomas, 2009).
The main criticism of Henrich’s model is that it does not incorporate
population structure. Individuals have cheap unconstrained access to the best model, but once population density or connectedness patterns change, the spread of an innovation will cease to be as straightforward. A model
distinguishing between population size, network size, and connectedness (i.e. in this case, distinguishing between total population size, the number of individuals in a subpopulation that are available for copying, and the number of links between subpopulatons) finds that total population size has little effect on cultural accumulation, but network size and connectedness do (Baldini, 2015). This works emphasises the importance of population structure in the diffusion of information, which, in turn, affects the accumulation of skill. The effect of population connectedness on cultural accumulation is probably more salient than the effect of population size, and indeed this observation has been confirmed empirically (Derex and Boyd, 2016).
The importance of population connectivity for human culture has also been emphasised by Muthukrishna and Henrich (2016), who suggest that useful innovations do not arise as a result of the exceptional cognitive capabilities of isolated geniuses, but rather as an interaction between ordinary human psychology and population connectedness. According to the authors, connected populations produce collective brains, and the bigger and more connected the population, the higher the rate of innovations. This is confirmed in urban areas – urban density predicts the rate of innovation (Carlino, Chatterjee and Hunt, 2007), and so does the population of cities, when innovation is measured in patent numbers (Bettencourt et al., 2007). What is more, too much connectivity can lead to decrease variance, which would lead to a lower rate of useful recombination, thus suggesting there is an optimal amount of interconnectivity that leads to the highest innovation rate. Although not many studies have investigated this relationship using data from real populations, the ones that do show mixed results. Some studies support a positive relationship between complexity and population size (Kline and Boyd, 2010; Collard, Ruttle, et al., 2013), while others find no evidence (Collard, Kemery and Banks, 2005; Collard, Buchanan, et al., 2013). For example, island population size predicts the size of the fishing toolkit in the Pacific (Kline and Boyd, 2010), but there is no link between population size and technological richness in hunter-gatherers from Western North America (Collard, Buchanan, et al., 2013). Empirical evidence from the experimental
laboratory also points to the fact that larger group size promotes higher improvement in skill or better preservation of skill at the micro-evolutionary level (Caldwell and Millen, 2010; Derex et al., 2013; Muthukrishna et al., 2013; Kempe and Mesoudi, 2014), but it is unclear whether the same
processes that support better task performance in the laboratory also explain macro-evolutionary processes in real world populations.
There is mixed evidence that links population size to measures of language complexity (Nettle, 2012). Population size is also related to the word rate of change – languages with larger speaker populations gain words faster and lose words slower in the basic vocabulary (Borenstein, Feldman and Aoki, 2008). Nonetheless, while the evidence points to a positive relationship between speaker population size and phoneme inventory size, the relationship is negative with morphological complexity. Language complexity is an intricate issue though, as languages vary on several dimensions in the way they encode information, with trade-offs between these dimensions, which might
confound any relationship with population size.
Similarly, in a study investigating the relationship between population size and the complexity in a non-technological domain, folktales, the authors found mixed results (Acerbi, Kendal and Tehrani, 2017). There was a significant positive relationship between population size and complexity measured as the overall number of folktale types, but a negative relationship with complexity measured as the number of tale motifs. This mixed result could be attributed to the fact that different levels of cultural complexity depend to varying degrees on population size because they are subject to different pressures, just like language is shaped by both expressivity and learnability. Alternatively, the authors suggest that the relationship between population size and complexity could be domain dependent. For a functional domain like technology we expect a strong relationship, yet a domain like folktales, which is not subject to functional pressures, but in which instead cultural traits can be easily individually reproduced without a need for strong replicative transmission, should show a weaker dependence on population size.