“Cultural Evolution in the Digital Age” Book Club General Discussion

10 Comments

• Dan Sperber 1 July 2020 (01:53)

  How Darwinian is Web cultural evolution ?
  I greatly enjoyed reading Alberto’s book and précis, the commentaries, and Alberto’s response. While the commentators are (I imagine) about to post their rejoinders to Alberto, and to encourage our readers to join the general discussion, let me raise a naïve question of my own.

  Of the three conditions for Darwinian selection — variation, differences in reproductive success, and heritability — the most problematic, in the case of cultural evolution, is heritability. Of course, imitation and other form of copying play a major role in cultural transmission. Still, many types of cultural items cannot be copied at all (compare, in this respect word sounds that tend to be faithfully copied by first-language learner and word meanings that do not lend themselves to any form or copying and that have to be inferentially reconstructed) and when copying is possible, copying errors and other types of departures from the model are so common as to raise a major problem for the view that Darwinian selection is the main force of cultural evolution. I am not mentioning all this to re-open an old debate, but to point out that, while we were debating these issues at the end of the last century, the very-high fidelity secured in biological evolution by DNA replication has, in the case of cultural evolution, been matched and in some respects superseded by electronic replication on the Web. Online sharing, as Alberto points out in his book, is a major vector of cultural transmission, and so, if not digital culture as a whole (for reasons Alberto explains very well), at least Web culture based on online sharing seems to obviously fulfil the three Darwinian conditions: variation, heritability, and difference in reproductive success (or am I missing something?). So, (treating the rest of culture as part of the environment) shouldn’t Web culture provide the perfect example of Darwinian cultural evolution? But does it? Hum… And if not, why not?

• Alberto Acerbi 1 July 2020 (16:05)

  A few thoughts
  Hi Dan,

  Thank you very much for your comment. Yes, I argue that digital online transmission potentially increases fidelity - propensity fidelity, to use Mathieu's terminology. This is not a necessary outcome: I spend a few pages showing how internet memes, intuitively a paradigmatic example of "cultural replication", are in fact result of modifications and adaptations.

  I think this is a very interesting fact with powerful consequences, but does this make for the perfect example of Darwinian cultural evolution? I do not know. I wonder whether it is insightful to think how different supports move cultural transmission in a Godfrey-Smith's 3-d "Darwinian Space", and whether digital online media move it towards more hereditability. It seems quite an attractive position to take, but I am not sure any more on what exactly adds to our understanding, and indeed I avoided to explore this aspect in the book. I'd be very curious to hear what others think about it.

• Hugo Mercier 3 July 2020 (15:50)

  Wary learners also copy blindly sometimes
  In his comment, Alex Mesoudi contrasts wary learners and blind copiers. He suggests that people can be either, depending on their developmental stage, their culture, the situation, etc. As more of a 'wary learner' kind of person, I'd like to push back a bit, not on the data, but on the conclusion drawn from them.

  People (like Alberto, Dan, Olivier, myself and many others) who suggest that humans are wary learners don't mean that they are simply resistant to social influence (incl. communication). Instead, they claim that they use a number of cues to figure out whose influence to follow, and when. As a result, even if people were to sometimes copy blindly, they might still be wary learners who've decided that it makes sense, in this context, to copy blindly.

  For instance, when a child overimitates an adult, they typically are in a situation in which (i) they have low priors (e.g. they face a new device), (ii) the source is benevolent (as far as they can tell), (iii) the source is deemed competent (e.g. because they are in the same place as the object, a heuristic used even by infants, and they are an adult), (iv) the source communicates to them that they're going to show them something (there committing and making the behavior more credible). It seems to make sense, then to copy the adult.

  To the extent that children are less likely to be blindly influenced by an experimenter when these conditions aren't present (which is abundantly shown by the trust in testimony literature), this means that even when the children appear to be blindly copying, they are in fact being wary learners, but wary learners who've decided (unconsciously) that copying was a good idea.

  To show that people really are blind copiers, I think it would have to be shown not only that they copy blindly in some contexts, but also that they keep doing so even when relevant parameters, which make it less rational to copy blindly, change. My reading of the literature is that no such demonstration exists, but I might have missed or misread something.

• Dan Sperber 6 July 2020 (18:35)

  More thoughts, in response to Alberto
  Thanks Alberto for your response. Indeed, in your book, you document directly or indirectly the many ways in which the fact that digital transmission provides for the possibility of reliably exact copying doesn’t, in the end, explain that much. And, in my comment, my goal was merely to reflect how this is remarkable and should be worth at least mentioning in passing. The memetic view and other approaches to cultural evolution have insisted that the three standard conditions of Darwinian selection are met not only in the biological but also in the cultural case, and in the latter thanks to the role of imitation and other copying mechanisms in securing heritability. I have not been the only one to challenge this assumption, which, by now, has been at least seriously qualified if not plainly rejected by most people working in the field. In the historical past, human culture did evolve not only in the few domains where the three conditions were met strictly enough to permit Darwinian selection (e.g., arguably, word phonology), but also in all kinds of domain where they were not (e.g. word meanings). But now comes digital culture, which, because of digital replication, should epitomize the memetic view of culture. As your book shows so well, even if implicitly, it does not, not even approximatively. I thought this was worth making explicit. Don’t you agree?
  This should be a challenge not just to people who take a narrowly Darwinian view of cultural evolution (i.e. one where the main force is Darwinian selection), but to all of us. Why is it that, when the conditions for Darwinian selection apparently are met as they are digital culture, Darwinian selection still may not be the main force of evolution?
  Let me sketch a suggestion based on an analogy with the evolution of infectious diseases (more specifically bacterial diseases). In the evolution of bacteria, Darwinian selection is indeed the main force, and this is true whether or not bacteria live in another organism, and, if they do so, whether they live in host organism as mutualists, pathogens, of mere commensals. If you look at the production of harmful effects on their hosts by pathogenic bacteria, some (e.g. destruction of white blood cells) are beneficial to, and adaptation of the bacteria while the production of other harmful effects (e.g. tooth decay) is just a side-effect from the perspective of the bacteria (as well-explained in Nesse and Williams’ classic Why we get sick). The evolution not of the pathogenic bacteria but of their effects on the host organism (and even more at on populations of hosts organisms), that is the epidemiological evolution of disease, is not the evolution of one species or of one biological adaptation but of an ecological system involving, in this case, at least two types of biological organisms, bacteria and humans or other animals (each of which is likely to evolve in a standard Darwinian way). The evolution of such ecological system and of traits of the system such as diseases in the host populations calls for models that are not provided by standard population genetics even if, obviously, population genetics is crucially relevant to understanding that evolution.
  I would suggest that culture, in general, is an ecological system involving a variety of populations, that of the hosts, and those of a variety of cultural types (mental representations, artifacts, behaviours). For instance, in digital culture, there are (inter alia) populations of human organisms and populations of inputs and output of digital systems. Even if these two populations evolved each in a standard Darwinian way, the ecological system they constitute together would not be explained by the evolution of each of these population taken separately, nor by any simple articulation of the two evolutionary processes. Even if it turns out that the population of outputs of digital networks evolves mostly by Darwinian selection, as it might, it still would not follow that digital culture (with involves not only these digital outputs but also their causal effects on human organisms and conversely) evolves in the same way.

• Alberto Acerbi 7 July 2020 (13:02)

  Culture, (sometimes) selectionist, but not Darwinian?
  Thank you again for your reply Dan, it really makes me think. In the past years, we discussed whether the absence of selection was the main reason why cultural evolution could not be considered a proper Darwinian process. Together with Alex Mesoudi, we argued that this depended on domains, as well as on the granularity of analysis, so that in some domains (and for some levels of analysis), one could indeed consider cultural evolution as a selectionist process. Perhaps we agree on the fact that some domains - specifically, digital online cultural evolution - could be considered as such.

  It seems to me that you are proposing now a different (at least for me) and compelling argument: even if the material items involved (for example) in digital online transmission behave in a proper Darwinian way, what we call culture is more than those ("an ecological system involving a variety of populations [and] a variety of cultural types") so that the full system can not be considered as evolving according to standard Darwinian principles. I hope this characterisation is correct.

  This makes much sense to me and in a way it accounts, post-hoc, for the fact that, even if in the book I explored the consequences of cheap and hi-fi transmission on the web, I avoided to link those directly to the "more Darwinian" motif.

  Out of curiosity: while clearly consistent with your general framework, I wonder whether there is somewhere an explicit discussion of this position, I'd be interested to know more.

• Dan Sperber 7 July 2020 (19:39)

  Epidemiological models are a category of ecological models
  Thanks Alberto for your reply. I am glad that we seem to agree on the idea that

  “even if the material items involved (for example) in digital online transmission behave in a proper Darwinian way, what we call culture is more than those (“an ecological system involving a variety of populations [and] a variety of cultural types”) so that the full system cannot be considered as evolving according to standard Darwinian principles.”

  You ask, “Out of curiosity: while clearly consistent with your general framework, I wonder whether there is somewhere an explicit discussion of this position, I’d be interested to know more.”
  Here is the answer. I was always quite aware that epidemiological phenomena are ecological phenomena and that epidemiological models are a special category of ecological models. Here is what I wrote in my 1985 article, “Anthropology and Psychology: Towards an Epidemiology of Representations”:

  “Epidemiology … is the ecological study of pathological phenomena. It is as eclectic in its ontology as ecology is. It has no more ontological autonomy than ecology has. It does not reduce to pathology, yet it cannot be defined or developed independently of pathology. Of course, one could have an epidemiology of good health or of any other condition; or, as I am suggesting, one can have an epidemiology of representations. But whatever' epidemiology' one is considering, it has to be defined in relationship to some sister discipline. What I want to suggest with the epidemiological analogy is that psychology is necessary but not sufficient for the characterization and explanation of cultural phenomena. Cultural phenomena are ecological patterns of psychological phenomena.”

  I still agree with this except that, for quite sometimes, I have spoken of an epidemiology of representations, artifacts and practices, rather than mentioning just representations.

  Sometimes after this article was published, in the 80s, Jean-Pierre Changeux (the French biologist) suggested I should talk of ecology rather than epidemiology. I didn’t follow his advice because “epidemiology” is more specific and precise, but in terms of getting the message across, maybe I should have. Part of the reason I didn’t is that I was not at the time much concerned with formal models. That began when I collaborated with Nicolas Claidière whose PhD (in 2009) was entitled “Théories darwiniennes de l'évolution Culturelle: Modèles et mécanismes”. Aiming at being more explicit on the ecological character of cultural evolution has only recently become a major goal and a part of the conversation (with Mathieu Charbonneau, Christophe Heintz, Helena Miton, and Thom Scott-Phillips lately). Your help in this respect, Alberto, might make a big difference.

• Alberto Acerbi 7 July 2020 (21:33)

  Food for thought
  Hi Dan,

  Thank you for your clarification.

  I am familiar with the work you refer to and I had indeed guessed that ecological/epidemiological were used similarly. However, I never myself made the explicit connection that this was one of the reasons to be doubtful of the Darwinian characterisation, as I had focused on the "absence of selection" bit (and the related importance of preservative vs reconstructive transmission). Much to think about for me - If I am, hopefully, getting it right!

  ps: thank you also for the curiosity about the ecological/epidemiological choice!

• Alberto Acerbi 8 July 2020 (14:33)

  The 2016 clown craze
  Hi all,

  I am trying to find an explicit example of online cultural evolution where, even if the population of outputs of digital networks evolves by Darwinian selection, we need to keep in consideration the full ecological system to provide a full characterisation of cultural evolution online. (I guess I am a slow and "dialogic" thinker, bear with me.)

  In "Cultural evolution in the Digital Age", after discussing how internet memes are often the product of transformations, active modifications, and similar, I mention the case of the 2016 "clown craze". Let me cite the whole bit, with some amendments:

  "In the autumn of 2016, a clown craze spread for a couple of months, first in the US and then in the rest of the world. Sightings of “evil clowns” in the real world were accompanied, not surprisingly, by circulation of suspicious pictures and news in social media, which in turn produced more sightings of real world clowns. Several accidents, including a murder of a teenager in Pennsylvania were connected to killer clowns, but all claims were deemed unsubstantiated. [...] Memes, such as the clown craze, have their real (or allegedly real) counterpart in the world offline, and real (or allegedly real) events that happen online, such as clown sightings, become digital memes. [...] It went viral because digital media provided a channel for quasi-universal distribution and because enough people were susceptible to it, and enough were willing to contribute with their own material. It went viral because it has plenty of the ingredients that we described in the previous chapter: negative emotions, threat-related information, minimally counterintuitive concepts: clowns are like us, but they also behave in an unpredictable and out-of-the-ordinary manner. Clowns are even super-stimuli to our natural disposition to find faces attention-catching and memorable, with their make-up that exaggerates facial features."

  In sum, while the clown internet memes are selected and copied with hi-fidelity, to understand their cultural evolution we need to take into consideration, among others, the mental representations of the individuals involved in the sharing, other cultural transmission chains that happened offline that may not be "Darwinian", etc. Is that the logic? Do you have better examples related to online culture?

• Mathieu Charbonneau 8 July 2020 (16:58)

  Being more Darwinian than Natural Selection
  To my mind, one of the main reasons for the failure of the memetics model of cultural evolution is that it was set in a nearly vacuous ecology. Traditions struggling for survival had to deal with their capacity to be memorized and further disseminated, but this was about it in terms of ecological interactions. All reduced to the ‘cultural fitness’ of the meme, and the idea that they would compete for brain space, attention, and communication channels. While these factors point to competitive interactions between the memes—natural selection naturally follows in the memetics model—the environment idealized away and replaced with the simple notion of carrying capacity. Adopting this idealization, it shouldn’t be surprising that everything had to be variation + retention + selection: the very understanding of the ecology of cultural evolution—population check—commanded it.

  So how would a Selectionist approach fare in the digital medium? (I say Selectionist instead of Darwinian because Darwin clearly did not reduced the environment in terms of carrying capacity). As I pointed out in my commentary to Alberto’s book, in the digital world, nothing really dies. We have a case of nearly infinite carrying capacity, with some behaviors spreading more than other, for sure, but also with a very low death rate in general (as Alberto points out in the book, astronaut.io allows you to explore part of this world beyond the spot light). In fact, we have something like Lewontin’s famous thought experiment of bacterial strains in an environment of infinite nutrients. Certainly, the one multiplying faster increases in proportion (higher reproductive fitness), but the strain with a lower fitness never goes extinct. But more importantly, no adaptation follows: there is simply no ecological design problem to be solved.

  Adopting a richer ecological perspective asks us to put back units of culture into context, and examine not just how well they manage to spread, but also how they depend on their environment of production (ecological variation regulation), how they link to other units and often form important ecological relations, and what sorts of ecological role they play (which niche they occupy and which niche they provide for other units). This can open up a more serious Selectionist approach to digital age cultural evolution (and I would argue cultural evolution tout court, but that is for another time), but it would also lead to study processes and interactions beyond the faithful transmission of memes, such as ecological interactions between different types of cultural units. To borrow a nomenclature devised by Nicolas Claidière, adopting a richer ecological view allows not only to study homo-impacts (replication/transmission) but also hetero-impacts: how cultural items of some type in a population affect the presence of items of a different type in that same population. We have forms of mutualism: smartphones promote doing TikTok videos, wanting to do TikTok videos increases the sell (and thus use) of smartphones. We certainly have predation: rules and policies of some platforms hunt down and eliminate promotions of terrorism, pornography, harassment, etc. Others are more complex: Facebook’s advertising system promotes advertisements of certain kinds, but it is unclear how these will be affecting the Facebook platform (parasitism?).

  To use again the example of “Camera-eats-first”, the behavior spreads certainly because other individuals do take those pictures and the behavior is made public (transmission). But to be successful cultural items, “Camera-eats-first” also relies on a rich ecosystem made of many other things—smartphones with cameras, internet, and public photo platforms such as Instagram. And not just any ecosystem. One could take a film camera or draw an aquarelle of the lunch and share the resulting pictures by distributing copies of it in the street. The “Camera-eats-first” behavior spreads online not just because it is psychologically attractive, or because we are exposed to more people, or because transmission fidelity has increased (homo-impacts), but because there is a digital ecosystem that favors and promotes its spread (as I argued in the commentary): the participatory Web 2.0 (motivating the creation of content), cheap access to production and sharing tools (both material and digital), etc.

• Hugo Mercier 8 July 2020 (18:38)

  The evolution of a meme
  https://twitter.com/egg_dog/status/1280666587792650240