Article for December: Population size does not explain past changes in cultural complexity (?)

Dear friends of the Mint’s Journal Club,

For the last month of 2016 we decided to read the paper from Krist Vaesen et al. Population size does not explain past changes in cultural complexity. Published in April’s edition of PNAS online, the paper criticizes two population dynamic models, one developed by Joseph Henrich (2004, JSTOR, alternative) to explain a decline in cultural complexity of Tasmania from beginning of Holocene until contact with Europeans, and the other developed by Adam Powell and colleagues (2009, Science) to interpret the regional variation in the start of Upper Paleolithic transition.


Vaesen and colleagues criticize the theory and the predictions of two models, and argue that population size is not the driving force behind the changes in cultural complexity, but do not provide an alternative explanation. Henrich, Powell and others wrote a response to this article (PNAS), in which they claim Vaesen and colleagues do not understand cumulative cultural evolution.

What do you think of this debate? And do you have thoughts on what drives cultural changes? Write your comments below.


The Mint wishes you all happy and pleasant holidays, and hopes to continue reading and discussing with you in the new year! 🙂


  • comment-avatar
    Alberto Acerbi 15 December 2016 (14:42)

    I apologise for the shameless self-promotion, but it is in fact closely related to the topic, and I thought it might be of some interest to ICCI readers. Together with Jamie Tehrani and Jeremy Kendal, we have a manuscript in which we discuss how the relationship between demography – population size in particular – and cultural complexity may be domain-dependent. Some cultural domains (technology perhaps) may show it, while others (we study folktales) not.

    We speculatively link this to how cultural stability is obtained. In some domains (again, think technology) cultural stability needs to be supported by hi-fi copying, errors in transmission are random and, on average, deleterious: these domains look like Henrich 2004’s model, and we expect here that small populations would not be able to maintain complex culture.

    In other domains (e.g. folktales) cultural representations are composed by easy-to-remember, attention-catching, etc. features that tend to be reconstructed each time by individuals: these domains *do not* look like Henrich 2004’s model, and we expect here complexity and population size not be correlated.

    There is a preprint of the manuscript with data, codes used for analysis, etc. at

  • comment-avatar
    Piers Kelly 15 December 2016 (16:15)

    I’m not one to wade into the waters of large quantitative models of culture change so I wanted to comment on one slightly tangential aspect that just so happens to be pertinent to Alberto’s remark: How is cultural complexity fairly evaluated and compared within and across cultural repertoires?
    Vaesen et al.’s paper, and the papers it critiques, seem to be most concerned with technological complexity, namely how difficult it is to produce a given artefact and how efficient that artefact is at doing what it’s designed for. And yet in Vaesen et al, there is a certain amount of conflation between technological and symbolic complexity that I’m not at ease with. They write, for example, that “The skills involved in the production of bone points would also have been less complex than the skills involved in a number of the economic and ritual activities that Tasmanians engaged in after 4 kya. These activities include the mining, alteration, and distribution of ochre; the creation of necklaces from human bones and pierced shell beads; body scarification; and funerary rituals.”
    What seems to be under discussion is the *technological complexity* involved in the process of manufacturing as opposed to the *symbolic complexity* of the artefacts themselves.
    Symbolic complexity would need to be derived from an analysis of the painting in and of itself as opposed to the mining and refinement of the ochre used in its composition. Similarly, we ought to assume that beadwork and scarification have a meaning, eg, of status, role etc or perhaps even evocative of an entire cosmology. Judging the complexity of a pierced bead necklace solely by the sophistication of the methods used to produce it seems a little like judging a novel from the refinement of the paper and ink.

    Of course, symbolic complexity in all its dimensions cannot easily be recovered from the archeological record, nor does it lend itself easily to quantitative comparison. However, if the Upper Paleolithic transition is hypothesised to be driven by increased communicative sophistication (or at the very least, coincides with it) then it is worth trying to make some attempt to differentiate symbolic activity from straightforward resource-extraction activity. A way into this might be to look at the relationship between language change and demographics as per Bromham et al 2015 for the Pacific, and search for any meaningful correlations with material symbolic culture as revealed in the archeological record. This might then be extrapolated backwards into the Upper Paleolithic (minus the language part.)

  • comment-avatar
    Hal Morris 19 December 2016 (18:22)

    Responding to Piers Kelly “What seems to be under discussion is the *technological complexity* involved in the process of manufacturing as opposed to the *symbolic complexity* of the artefacts themselves.
    Symbolic complexity would need to be derived from an analysis of the painting in and of itself”.

    Perhaps “cultural complexity” is not the ideal term. To me what it seemed to signify was a a measure of technical skills and understanding and aggregate measure of all all the techniques and understanding that were transmitted from one generation to another. It would also seem that in the case of anything decorative, the the artifact itself represents the thing intended, whereas a pierced bone sewing needle is incomplete in itself, but represents the arts of making it and the arts of using it. One it pointless without the other, and it most likely calls for two separate kinds of artisans, hence there is more complexity than is shown by its mere structure

  • comment-avatar
    Olivier Morin 10 January 2017 (13:28)

    This kind of blanket criticism of an entire body of work is a difficult thing to form an opinion about. On some points I come on Henrich et al.’s side, while on others I see the relevance of some of Vaesen et al.’s critique; but the multiplicity of angles and the acrimony of the debates makes it hard to get a bird’s eye view of the issues at stake.

    Like Henrich et al., I am not impressed by Vaesen et al.’s critique of payoff-biased transmission. The view that cultural transmission (of technology) takes efficiency into account isn’t just experimentally grounded, it has considerable evolutionary and psychological plausibility. (Ethnographic studies may not be so consistent in their results, but I wouldn’t necessarily take all informants at their word when they say that they hold such or such a practice from their father—which could be true in ways that do not imply direct cultural transmission.) Likewise, I didn’t recognise Vaesen et al.’s target in the assumptions of “Complexity maximisation” and “Complexity regression.”

    One point that I am grateful for Vaesen et al.’s paper to have made relates to model comparison: I share their impression that much of the literature on cultural complexity/population size consists in attempts to confirm one hypothesis rather than compare it to others. The view that specialisation and division of labour could be one key driver of cultural complexity makes a great deal of economic sense. For once, Henrich et al. do not deny this—but their response is a sort of retreat into holism: Task specialisation should be understood in a broader context that includes demography. I find this position slightly unconvincing, because the earlier models were quite explicitly focused on a single cause, and because Henrich (at least) never stopped insisting on the paramount importance of transmission fidelity for cultural accumulation.

  • comment-avatar
    Jean-Baptiste André 13 January 2017 (17:34)

    I’m not really going to comment on Vaesen’s paper, but rather discuss a side issue related to the relationship between demography and cultural evolution.

    First of all, as a general point, I think that more links should be made between the literature on cultural complexity / accumulation and the theory of endogenous growth (from economics). Both deal with the accumulation of knowledge. Take for instance the classic Arrow paper « The economic implications of learning-by-doing »(1962). Like Henrich, Arrow suggests that larger populations should innovate faster (« more people means more ideas »), but he also suggests that, ceteris paribus, more useful stuffs should improve faster (because people practice them more), which is an interesting addition (more on this later).

    More generally, what I take from endogenous growth theory that is relevant for the present debate is the idea that the invention of complex techniques, and also the fact to learn these techniques once they have been invented, should be seen as an investment people make (this point is not in Arrow as far as I understand, but definitely in later models of growth such as Aghion and Howitt’s). Of course economists see that as a rational calculation that economic agents/firms make (shall I invest into R&D or shall I not?). But we do not need to commit to this assumption about psychology. As evolutionary psychologists, we also expect that people should be fairly good at allocating their time and effort only to profitable activities, and we sympathize with the idea that cognitive calculations (here, learning and/or improving a technique) is an effort that people make only when it’s expected to pay-off: i.e. when they expect that mastering the technique will reimburse the cost put into learning it (the cognitive effort, the opportunities lost while doing so, etc.).

    Hence, beyond the simple (and, I believe, hard to deny) effect of the sheer number of brains at work, another important dimension must be considered that likely affects cultural complexity: the benefits that people get by inventing and learning complex techniques.

    There are many reasons why these benefits could vary. For instance, the institution of patent shall trigger more investment into innovation because firms foresee that they will gain more from their discoveries (that’s one of the key ideas in Aghion and howitt’s classic paper). But that’s just one effect. There could also be more « biological » effects. For instance, people invest more into inventing/learning complex techniques when they can afford to do so, i.e. when their basic biological needs are already fulfilled (an idea we have been discussing with my colleague Nicolas Baumard lately, and might explore further in the future).

    Another one of these effects is specially related to the current debate: the size of the market effect. The key reason why investing into a complex technique is useful is to use that technique to produce goods that one can then exchange for other goods (either in a genuine economic interaction, or in an informal exchange based on reputation). In fact, and more strongly, because there are economies of scale, many techniques become useful only when the size of the market is large enough. If there was not a large set of potential customers to buy them, no one would be willing to invest into mastering the art of bakery, clothes tailoring, or teaching anthropology, let alone building cars, or creating smartphone applications. And there is even a continuum. The larger the market size, the more complex (and costly) the techniques can be, because they are eventually repaid by economies of scale. That’s why international trade creates growth.

    Note that, even though this effect is not usually discussed in the field of cultural evolution, it’s an absolute classic in economics. It is classic in models of growth (e.g. in Aghion and Howitt, but see also the excellent « Technological progress and regress in pre-industrial times » by Aiyar et al. 2008), but it is also actually Adam Smith’s key idea in The Wealth of Nations! It makes it all the more interesting to reflect on the consequences of this principle for cultural evolution.

    Some of these consequences are similar to Henrich’s copying-error effect, but for a totally different reason. Like Henrich’s model, the market-size effect predicts that smaller and less connected populations should carry less complex technologies. But that’s not because of copying-error. It’s because people in these populations are less interested in investing into complex technologies, because of smaller market and larger transaction costs (hence the opportunities for mutually profitable exchanges are fewer). So the two effects (Henrich’s and market-size) go in the same direction in this case, and might be difficult to disentangle. Interestingly, however, there are also situations in which the two effects go in opposite directions. Here is a short, non-exhaustive, list:

    According to the market effect (but not Henrich’s effect):

    – In a given population, all technological domains should not have the same complexity. More useful domains should be more complex, because people are more motivated to invest cognitive effort into them. This could explain why Tasmanians have kept canoes but not bone points (beyond Alberto Acerbi’s interesting point about non-technological culture).

    – In general, all the usual factors that make exchanges more efficient, beyond population size and connectedness, shall influence cultural complexity. High trust societies for instance should innovate more because they exchange more.

    – A small group of individuals can invent and maintain very complex cultural items if they have access to a large market or if, more generally, they have a lot to gain from it. That’s typically the case in small teams of highly innovating people (e.g. the approximately 60.000 Google employees innovate more than some much larger groups; not to mention the extraordinary achievements of very small teams, or even solitary, scholars devoting all their life to innovation).

    – A large population with a disorganized economic system (due to an external choc) can rapidly lose many complex techniques independently of its demography. This could be tested, perhaps, in historical cases of cultural regression (e.g. the collapse of the Roman Empire, see Aiyar et al. 2008).

    Overall, and to conclude, contra Henrich and colleagues, (i) innovating and imitating are not so different things, (ii) both involve a cognitive effort, none is passive, and (iii) people should make this effort when it’s worth it. Taking this into account should not remove the sheer effect of demography on cultural complexity (more people does mean more ideas, for sure), but it also leads to other, perhaps more interesting effects.