The purpose of this article is to explain how cultural evolution and cognitive science can inform each other. The specific way this is done is to show how cognitive science can help address the third way question faced by cultural evolution. This question is, “are human thought and behavior made adaptive…not only by genetic selection and intelligence but, at least sometimes, by cultural selection?” (3). The California school, which holds to dual-inheritance theory, answers yes. The Paris school, which holds to cultural epidemiology, answers no. They argue that cultural items are not replicated. They are reconstructed, and this means that a crucial feature of evolution--replication--cannot be met. Heyes argues that the system 1 and system 2 distinction (see section 4) from cognitive science can bring clarity to the debate. The article ends with a few bonuses. First, she elaborates on how metacognitive social rules can also inform the discussion. Second, cultural evolution can help explain the adaptivity of cognitive mechanisms.
2. Third-way Cultural Selection
The third-way question asks whether, besides natural selection acting on genetic variation and human intelligence, there is a third way (cultural selection) whereby we can explain cognitive or behavioral characteristics. The California school, composed of individuals such as Boyd, Richerson, Cavalli-Sforza, and Henrich, seems to answer yes. They build social learning on genetic adaptations. They focus on gene-culture coevolution. And they use mathematical methods that study genetic evolution to study socially learned characteristics. However, philosophers of biology and evolution, Lewens and Sterelny, have suggested otherwise. They notice that members of the california school rarely do not give robust causal explanations, such as why a cultural trait is fit, who benefits, and what the benefit is. Heyes acknowledges the point but thinks that cultural evolution is meant to be selectionist. The reasons are that cultural evolutionary theory is inspired by Donald Campbell, who was “evangelically selectionist.” Moreover, the California school emphasizes selection, Darwinian characteristics, cultural adaptations, the logic of natural selection, and cultural transmission. Is all this language merely non-literal? The California school needs to be able to positively answer the third-way question in order to calm the reservations of their commentators.
3. Replication and Reconstruction
The paris school, championed by individuals such as Sperber and Morin, argue that the california school does attempt to give an evolutionary explanation. However, the Paris school thinks they are mistaken because cultural entities (ideas, beliefs, artifacts and the like) are reconstructed, not replicated. The importance (they say) is that replication, but not reconstruction, allows for high-fidelity transmission, and if cultural entities are not passed on with sufficiently high fidelity, then selection is less likely to occur. It is less likely to occur because new adaptations cannot accumulate. The problem with the debate, Heyes tells us, is that the replication-reconstruction distinction is put in terms of a product--high fidelity--instead of a psychological process, and this is the very thing under investigation. We should be asking which psychological processes lead to high fidelity. When we know this, then we can see whether the third way question can be answered. (Personal note: I think there is a further issue here: when does high fidelity matter? In On Modeling Cognition and Culture, Henrich and Boyd (2002) argue that low fidelity can at times lead to adaptive results).
4. A Reconstruction of Replication and Reconstruction
Heyes redefines replication and reconstruction so that they are associated with system 1 and system 2, respectively. System 1 is “...fast, involuntary, parallel, unavailable to conscious awareness…” (4). It is based on information from genetics and associative learning. System 2 is “...slow, effortful, serial, available to conscious awareness…” (4). System 2 uses system 1 information and its own information. That being said, we know we engage in replication often: we unintentionally imitate others’ gestures--we are often unaware that we are copying. We tend to copy the preferences and aversions of others (fears or disgust) via observational conditioning, which is a type of associative learning. By contrast, using system 2 to imitate would not be of high-fidelity. Reading about how to copy a pro-tennis player won’t be as helpful. This tells us that the paris school is right about two things and wrong about a third. First, they are right that replication is higher fidelity than reconstruction. In other words, system 1 is better at copying than system two. Second, the Paris school is right that replication depends on psychological mechanisms based on genetics adapted for culture. Here, Heyes cites some research supporting the idea that imitation mechanisms are learned. Thirdly, the Paris school is wrong to think that replication is rare, i.e., we use system 1 all the time.
(I can think of one critical question about this section. The question is whether Sperber, a member of the paris school, would endorse the system 1 and system 2 distinction. Heyes suggests that Sperber would since it is analogous with his distinction of replication-reconstruction as stimulus driven vs inference driven, respectively. However, the citation is to work from 1996-2000. Sperber and Mercier, in the Enigma of Reason (2017), deny the system 1 and system 2 distinction. It is worth asking how they would view this section).
5. Cultural Selection Requires more than replication
The California school faces a hurdle. When selection occurs, “...a change in the frequency in the types in a population constitutes improvement or adaptation” (5) and this improvement cannot merely be due to smart choices by agents. Smart choices would be instances of intelligence or design and by definition would be the second, not the third, way. For example, increasing the frequency by which we shop with reusable shopping bags compared to plastic ones is an improvement, but if this is done by foreseeing the environmental benefits, then the improvement is not selectionist (not third way). It is done by design. The hurdle faced by the California school is how to explain high-fidelity transmission without appeal to smart processes. Compounded to this is the further requirement of recurrent fidelity. The cultural entities don’t just need to be transmitted with high fidelity. They also need to be maintained long enough to improve and then transmit. In the next sections, Heyes will try to suggest how the hurdles can be passed.
6. Metacognitive Social Learning Strategies
The challenge is whether the learning strategies proposed by the California school are sufficiently “dumb” to qualify as a third way mechanism. This would be clearly the case if the strategies were those of associative learning, present in many species and early in development. In section 4, Heyes wrote that replication, via system 1 processes, is partly associative. So to a certain extent, the learning strategies can qualify as a third mechanism. However, the learning strategies also seem to be part of system 2. Here, we can distinguish two kinds of strategies: direct or indirect. The direct strategy bases a rule on features of the trait to be copied, e.g., how fast does his boat go in water? The indirect strategy bases a rule on features of the person who possesses a trait to be copied, e.g., how many boats does he have? Heyes calls both of these strategies “metacognitive” because, unlike the associative rules, they are represented in the organism acting on them. While Pavlov’s dog does not think about salivating at the presentation of a bell, a human could think “I will copy the person with the most hunts.”
So we have to ask whether these latter (meta-cognitive) rules are sufficiently dumb to meet the third way. Heyes thinks that indirect learning strategies can be “dumb” in the sense that “...they bias an agent towards copying better variants without the agent being smart enough to know which variants are better and which are worse” (6). So, while the agent might follow the strategy for a reason (lots of boats indicate what he wants), the agent need not know why what he copies works. (Personal note: I think this discussion is insightful, but I worry that Henrich wouldn’t be satisfied. Henrich writes that, often, the rule is followed as a bias and without the agent knowing that it does anything. Perhaps in these situations, a spectrum of the above factors interact). In closing, Heyes then proposes that these metacognitive rules can help meet the recurrent fidelity challenge by biasing us to attend to cultural models.
7. Cultural Evolution of Cognitive Processes
Humans have a unique set of cognitive abilities. Where do they come from? The evolutionary psychologist's answer appeals to genetic evolution. In our genes, we have very specific programmes for the development of our cognitive abilities, which are triggered in development. This view has fallen out of favor among scientists. Heyes proposes that we apply cultural evolution not just to the product of thought, say behavior, practice, and tools, but also to the mechanisms of thought itself--a view she calls “cognitive gadgets.” On this view, humans have a genetic basis that gets shaped in development and via society to have domain specific modules. Cognitive mechanisms are learned and culturally inherited.
In this view, different metacognitive strategies are handed down culturally. They will have differential success, and hence some cultural groups will be more successful than others. Heyes observes that, as cultural evolutionists tell us, many of these metacognitive strategies are indirect. However, various animals display these strategies, in a sense. For example, female vervet monkeys prefer to copy female vervets. So, indirect biases are not enough. Nevertheless, given cultural evolution, metacognitive strategies can accumulate the experience of many and be passed down.
Heyes, C (2018). Enquire within: cultural evolution and cognitive science. Philosophical Transactions. https://doi.org/10.1098/rstb.2017.0051