Alberto Acerbi’s response: There is much work to do
It was a great pleasure having a book club dedicated to Cultural evolution in the digital age. Writing a book feels like a long and solitary experience and it is comforting that, when done, it may result in such productive exchanges. Thus, first of all, I want to thank the organisers, Tiffany Morisseau and Dan Sperber, and all the participants for their commentaries: kind, sometimes even flattering, but always perceptive and stimulating. I organised my reply around three macro-themes that emerged in the book club: the role of producers of cultural traits, the features of specifically digital cultural transmission, and a discussion on some more foundational issues in cultural evolution, namely the importance of faithful transmission for cumulative culture and our reliance on social information.
In my précis, I quickly mentioned that one of the topics I would have liked to elaborate more in the book was the role of producers of cultural traits. It seems I should have indeed, as it picked the interest of several commentaries. Sacha noted that information spreads online not only, or mostly, because of its cognitive attractiveness (the suite of factors to which I paid most attention to, taking the perspective of the consumers of cultural traits), but also because of its shareability. Sex-related information may be attention-grabbing, but it is ill-suited to share. On the contrary, to use again his examples, science or technology news is hardly attractive, but signals competence, and phatic information (“I love you mum”, see Berriche & Altay, 2020) may be highly shared not because catching or interesting, but because it provides a way to keep contacts with friends and family.
In a similar fashion, Tiffany considers the importance of truthfulness in our everyday sharing activity. Whereas sometimes it is important if what we share is true or not, in some occasions this is not the case, and we are less concerned about the accuracy of the content than about, say, its function as an endorsement, or as a joke. This helps to correct the overblown claims about people being fooled by fake news: sharing is not believing. In Cultural evolution in the digital age I report the top-performing fake news in Facebook in 2017, according to Buzzfeed: “Babysitter transported to hospital after inserting a baby in her vagina”, with more than 1.2 million engagements. (In case you want to know, the 2018 champion is “Lottery winner arrested for dumping $200,000 of manure on ex-boss’ lawn”.)
In his commentary, Hugo develops these ideas. He argues that there are two main motivations for sharing information: informing others and managing our reputation. An interesting observation is that the latter motivation may be prominent in social media, but not in other channels of transmission. I rush to share my last preprint on Twitter, possibly with an eye-catching thread (include pictures!), but I would find odd, and frankly embarrassing, to do the same in an email to multiple recipients, perhaps my colleagues. Hugo argues that this is due to the fact that informative communication is better suited for targeted channels of transmission, whereas reputation-enhancing communication can be less discriminating, so more suitable for social media. This asymmetry, in turn, may explain various features of social media sharing, for example a preference for novel information, that enhances our reputation as competent sources, or for information “positively associated” with a coalition, where the informational content is less important than signalling our support or membership.
Finally, Pascal expands these considerations by sketching a more general picture of producers’ motivations. He uses a framework inspired by social evolution theory to propose a possible way to investigate the biological fitness advantages implicated in producing and sharing information. While some production can be purely altruistic (sharing information with relatives, especially offspring) and some selfish (a cult leader sharing ideas to extract labour from cult members), most dissemination of information is, according to Pascal, potentially mutualistic, that is, favouring both the sender and the receiver. In this context, from an evolutionary perspective, good sources of information are valued and rewarded, so that being a good source of information is in itself fitness-enhancing. As in Hugo’s commentary, this allows for some predictions on the content that we are most likely to share online. Threat-related information, for example, as I discussed in the book, is less likely to be verified than information on positive outcomes (it is risky to check whether the information “there are snakes under a rock” is true, but you want to check if “there are diamonds under a rock” is) so that sharing information on threats is less likely to impact on a source’s credibility.
The producers’ perspective is overlooked in cultural evolutionary studies. Cultural evolution theory, for good reasons, has been developed with the implicit goal of explaining how the process of transmission could preserve and diffuse cultural traits and how we choose which ones to acquire, more than with an interest on how these cultural traits arise, or on how we decide which ones to share. The commentaries make a strong point that we should start to widen our focus (see also André et al., 2020). The role of Cultural evolution in the digital age in this shift is undoubtedly minimal, but I like to think that the role of what we observe every day in social media and on the internet is not. For me, at least, it has been a surprising realisation: we are willing to share for uncertain immediate gains an enormous amount of information, sometimes at low cost – a tweet, or correcting a typo in Wikipedia – but often at high cost – a blog post or a YouTube tutorial. Even more surprising, this information is mostly of good quality. Reviews, for example, are a particularly interesting case that is mentioned in the book: they are, no matter the platform, generally useful and comprehensive. Incidentally, and against the common trope that “everybody is mean on the internet”, the overwhelming majority of reviews are also positive, so that the problem becomes how to identify the objective ones against a cheerful background. Besides this general and surprising from a cultural evolution point of view (remember the scroungers/producers asymmetry) willingness to share information, the details are to be filled, and these commentaries open new and unexplored venues of research which I hope to see developed in the next years.
A second issue that emerged in the commentaries is that the digital medium and, even more in details, different platforms or services, impact the cultural evolutionary process in specific ways. A recurring theme in Cultural evolution in the digital age is that it is important to focus on the interaction between the peculiar features that makes digital media special and rather general psychological constraints that act both in our online and offline lives or, as I wrote in the book, it is important taking “a long view”. I attempted a rough taxonomy of the features that make digital cultural evolution special, characterising them as availability, reach, opacity, explicitness, and fidelity.
But, of course, “digital” is a quite broad category. In his commentary, Sacha notes that different social media are used in different ways: “I love you mum” posts are successful in Facebook, but they would hardly be hits in LinkedIn. Twitter forces users to write relatively short posts, while Instagram is only about pictures. “Likes”, as Sacha interestingly reports, are often used as bookmarks in Twitter, as a signal to the algorithm in YouTube (“I want more of this”), and, as mentioned, they often have a phatic function in Facebook.
At the end of the chapter Transmitting and Sharing, I used an apparent ambiguity to discuss exactly this theme: whereas I found negative information being predominant in “fake news”, the opposite was true for the New York Times most successful articles. The reason, however, besides the social and demographic differences, was possibly because the successful articles were defined as the most emailed, and an email can be seen as more personal than a social media sharing. The digital tool one is using does change the cultural dynamics that are generated. At the same time, it is worth to be able to zoom in and out, from digital-general to platform-specific features, according to what we are interested in. Again, I see this as an excellent venue for future works.
Mathieu’s commentary concerns instead a more general feature of the digital medium, which links directly with what we said about the importance of producers of cultural traits. Not only online and digital media provide fidelity amplifiers, as I called them in the book, but they also provide, in Mathieu’s words, generative amplifiers. While I focused on the fact that online and digital tools make easy and cheap sharing a picture, a song, or a video, Mathieu rightly points out that they also make easy and cheap producing a picture, a song, or a video. I am old enough to remember using analogue cameras and thinking twice before taking a picture, but now our iPhone storages are clogged with thousands of shots, often not very memorable. Generative amplifiers not only produce a net increase in the information recorded, but they are likely to change the nature of this information. New genres – Mathieu mentions “Camera eats first” – completely depends on this possibility: another fascinating way in which digital media impact cultural evolution.
Finally, Oliver and Alex’s commentaries focus on broad cultural evolutionary issues that go beyond the digital sphere. Olivier scrutinises my hypothesis of a connection between preservative transmission and cumulative improvement, indeed one of the issues that I highlighted in my précis as “more open to debate”. And the debate is very welcome, especially when the criticism is good enough to make me rethink my assumptions!
In the book, I schematised a distinction between smooth search spaces where individual problem-solving suffices to find solutions and mostly flat, with rare and narrow peaks, search spaces, where it is unlikely it does and faithful transmission is important (see also Acerbi et al., 2011). As Olivier notes, the shape of problem spaces is not given once and for all: transmitted information and individual problem-solving interact and enhance each other. A bit of transmitted information can transform a flat, narrow-peaked space in a smoother one, and then individual problem-solving can do the rest. I agree, and I would say that the distinction of two shapes of search spaces should be treated as the distinction between pure reconstructive and pure preservative transmission. They are abstract ideals, real cultural traits – the canoes and the Greek fires – are always somewhere in between.
The question becomes: is that a useful idealisation? I still think so. There are cases where high-fidelity transmission is likely to do most of the work: as it happened, differential calculus can be invented independently, but the majority of people learn it drawing heavily on social information. This decreases the probability that a trait will disappear as well as increasing the number of possible further innovations. In this sense, I believe that supports that make easier transmitting successfully social information (you know what I have in mind) can as a consequence enhance cultural cumulation or, if you prefer, progress.
In his comment, Alex wonders whether my characterisation of “wary learners” is not simply the other side of the coin of the “blind copiers” characterisation: both are an extreme caricature of a more nuanced reality that stands in between. I hope that from the full reading of Cultural evolution in the digital age is clear that I am not claiming that social information is not important (as I said in the book, that would be a very strange take in a volume about culture), but indeed I claim that we may use it differently with respect to what seems to be implied in classic cultural evolutionary accounts.
Alex is right that I partly used this characterisation as a rhetorical device to counteract the countless worried descriptions of the “dangers of digital influence” and some of their unrealistic assumptions. On the other hand, I do think that there is something more here. While in the book I sketched only a few ideas, the underuse of social information is a robust pattern, for which we do not have yet a clear explanation (Morin et al., submitted). Given the central stage that the concept of social learning has in cultural evolution theory, it seems to me we have some interesting work to do. I fully agree with Alex that various factors influence our usage of social information, painting a complex picture. We have indeed elsewhere explored the importance of different domains (Acerbi & Mesoudi, 2015), and in Cultural evolution in the digital age I additionally discuss the possible role of experimental design (chapter Wary Learners) or transmission supports (chapter Transmitting and sharing).
What concerns me most is not how much we use social information, but how the causal role of concepts as “social learning” or “social learning strategies” is often used to explain cultural success, as I discussed in my précis. I do not know whether “flexible” is better than “wary” (but I may change idea at some point!), but I would be pleased if I had given a small contribution to a literature that makes an effort to keep into consideration the full factors that impact on to the usage of social information, coming both from cultural evolution (the references cited in Alex’s commentary, plus, for example, the recent Miu et al., 2020) and epistemic vigilance (Mercier, 2020).
Overall, the take-home message I got from this book club is that there are several open questions and that Cultural Evolution in the Digital Age only scratched the surface of most of them. This is a positive take-home message: the book club confirms that evolutionary, or cultural evolutionary, approaches are particularly suitable for understanding and possibly influencing the changes and the challenges we face in the digital age: there is much work to do!
Acerbi A & Mesoudi A (2015). If we are all cultural Darwinians what’s the fuss about? Clarifying recent disagreements in the field of cultural evolution. Biology & Philosophy, 30.
Acerbi A, Tennie C & Nunn CL (2011). Modeling imitation and emulation in constrained search spaces. Learning & Behavior, 39 (2).
André JB, Baumard N & Boyer P (2020). The Mystery of Symbolic Culture: What fitness costs? What fitness benefits? OSF Preprints.
Berriche M & Altay S (2020). Internet Users Engage More With Phatic Posts Than With Health Misinformation On Facebook. Palgrave Communications, 6, 71.
Mercier H (2020). Not Born Yesterday, Princeton University Press.
Miu E, Gulley N, Laland KN & Rendell L (2020). Flexible learning, rather than inveterate innovation or copying, drives cumulative knowledge gain. Sciences Advances, 6, 23.
Morin O, Jacquet P, Vaesen K & Acerbi A (submitted). Social information use and social information waste.