Towards an epistemology of cultural learning

The secret of our species’ success resides not in the power of our individual minds, but in the collective brains of our communities. Our collective brains arise from the synthesis of our cultural and social natures—from the fact that we readily learn from others (are cultural) and can, with the right norms, live in large and widely interconnected groups (are social). The striking technologies that characterize our species . . . emerge not from singular geniuses but from the flow and recombination of ideas, practices, lucky errors, and chance insights among interconnected minds and across generations. (Henrich, 2016, pp. 5–6) On August 20th, 1860, an expedition to explore the Australian interior set off from Melbourne led by a former military officer and police inspector named Robert Burke. After several months of travel, Burke and a small lead group reached the Gulf of Carpentaria on the continent's northern shore. On their return trip they were unable to rendezvous with the remainder of their party and eventually ran out of food and supplies while still 600 km from Melbourne. Fortunately, a local Aboriginal group, the Yandruwandha, provided them with fish and some cakes made from the ground seeds of an aquatic fern called nardoo. Burke and his group were eventually able to discover the source of the seeds and set to work grinding them into cakes in an attempt to emulate the Yandruwandha. Nevertheless, despite procuring and eating plenty of nardoo, the men continued to weaken. Nardoo contains a slow-acting toxin that the Yandruwandha were able to remove by, among other things, rinsing the seeds in copious amounts of water. Burke didn't know this, and so he and his men were slowly poisoning themselves. Burke eventually died, as did the rest of his group, save for one man who was spared only because he was taken in and cared for by another Yandruwandha group until a rescue party arrived several months later. A Yandruwandha “Natural History Handbook” would have run to hundreds of pages with sections on the habits of game, efficient hunting techniques, how to find water, how to process toxic ferns, yams, and cycads, and so on. Australian Aborigines are famous among archeologists for the simplicity of their technology. Nonetheless, an “Instruction Manual for Technology” would have to cover the manufacture and proper use of nets, baskets, houses, boomerangs, fire drills, spears and spear-throwers, poisons, adhesives, shields, bark boats, ground stone tools, and much more. . . To become a competent Yandruwandha, you would also have needed to master “Social Policies and Procedures,” “Grammar and Dictionary,” and “Beliefs, Stories, and Songs,” volumes of comparable length. (Boyd, 2018, pp. 18–19) All of the knowledge contained in these and other volumes was not the possession of any individual Yandruwandha on his or her own. It was the possession of the collective, of the cultural group, but it was thereby freely available for individuals to exploit when they were in need. Yandruwandha individuals were able to survive and thrive in their environment by being members of a collective body that over generations had accumulated a body of knowledge that was continuously circulated, nourished, and passed on over time. This capacity to collectively generate, accumulate, build upon, and transmit knowledge over time has enabled our species to expand from our humble origins in Africa to now inhabit nearly every environment on the planet, from the Sahara to the Himalayas, the Amazon to the Arctic. Over the last several decades an expanding body of work in the empirical sciences has converged on the idea that the key to what makes us human—the focal point around which explanations of our development as a species revolve—is culture. Human beings are creatures that have evolved to have distinctive social and cultural capacities that have fed back on the evolutionary process itself, creating “a new kind of animal” previously unseen in our planet's history (Boyd, 2018; Henrich, 2016). A key component of this major evolutionary transition in our human lineage is the emergence of distinctive social learning capacities that make possible cumulative cultural evolution, a process by which environmental and technological information and know-how is aggregated, built upon, and transmitted within and between groups and across generations so that cultural knowledge accumulates over time, with increasing complexity and efficiency. The secret of our success, as Henrich puts it, resides not in our individual intelligence but in the collective intelligence of our communities, in our abilities to learn from and cooperate with one another in ways and to a degree previously unseen in the animal kingdom. While the significance of cultural learning is now widely acknowledged in the empirical human sciences, this empirical literature has received little attention from epistemologists. This paper advocates for the development of a social epistemology of cultural learning, one that draws on the insights of this growing empirical literature. The topic of cultural learning does not fit smoothly into the landscape of contemporary social epistemology. Cultural learning is not simply a matter of how individual epistemic agents deal with social evidence, nor is it a matter of the epistemic properties of groups or collective agents. Instead, cultural learning concerns the processes by which individual epistemic agents “put their heads together” as members of communities so as to transmit knowledge between members and across generations with the kind of efficiency and scope required for cultural knowledge to accumulate. Many of these social learning processes are unique to our species, and they are far more various and complex than the “social evidence” provided by the beliefs and testimony of others. Human children are adept and prodigious social learners from very early in ontogeny, soaking up information from their cultural environment via processes of imitation, socially scaffolded practice, and direct instruction via demonstration, explanation, argument, and explicit testimony. Group-level social norms and practices combine these processes to produce socially structured learning environments that facilitate the efficiency of cultural transmission by encouraging various forms of hybrid-learning, such as the apprentice learning characteristic of the transmission of many craft skills and techniques (Sterelny, 2012). Individual and group-level dynamics thus work in tandem, feeding back on one another in such a way as to facilitate cultural transmission that is not only accurate and reliable, but also flexible, broad in scope, and of sufficient volume to allow for the preservation and accumulation of knowledge over time. Understanding these cultural learning dynamics is thus crucial for understanding how it is that knowledge gets around between actual human minds and within actual human communities. Section II introduces some of the recent empirical literature concerning the significance of cultural transmission for human evolution and development. Section III describes some of the varieties of cultural learning that support human cumulative culture. Section IV then outlines some of the epistemological concerns that arise out of this literature. These include distinguishing between various cultural transmission channels, identifying the adaptively relevant aims of transmission, and explaining how various cognitive and social dynamics contribute to the achievement of these aims. As this is a work of advocacy, I do not here present anything like a detailed epistemology of cultural learning. I merely hope to facilitate an appreciation of the potential benefits of adopting a cultural learning perspective. Three and a half million years ago, our Australopithecene ancestors were an interesting but relatively unobtrusive feature of the African landscape. By one and a half million years ago, early members of our Homo lineage had expanded out of Africa and into Eurasia, eastward as far as China and south to the Indonesian island of Java. Our own species, Homo sapiens, evolved across continental Africa roughly 300kya and quickly expanded out of Africa in multiple dispersals, eventually reaching Australia by 60kya and North and South America by 15 to 20kya. Today we occupy nearly every habitat on the planet, the most obtrusive species the Earth has ever seen, our industry and infrastructure reshaping and destroying ecosystems everywhere. We have even left the planet, walking on the moon and sending our technology to explore our solar system and beyond. What explains this rapid and unprecedented evolutionary trajectory? Perhaps surprisingly, there is now a well-established and widely accepted answer to this question. The answer is culture (Tomasello, 2019, p. 3). At some point in our history, our lineage developed unprecedented capacities for social coordination and the social transmission of knowledge. The human way of being in the world became increasingly dependent on socially learned and cooperative behavioral traditions concerning such things as tool use and manufacture and foraging and food processing techniques. Increased ecological dependence on such socially transmitted cultural traditions selected for individual cognitive and motivational capacities that facilitated interpersonal and collective cooperation and transmission. Human social learning became more and more efficient, far exceeding that present elsewhere in the animal kingdom. We developed abilities to pass on skills and information more quickly and reliably and also more flexibly, with respect to a wider range of topics and activities. Gradually, beginning in fits and starts but solidifying over hundreds of thousands of years, cultural transmission became so efficient that skills and information transmitted between individuals and across generations could be modified, refined, and built upon by others and by subsequent generations. Knowledge and skill began to accumulate and develop over time. Culture became cumulative, and cumulative culture forever altered the processes of human evolution. Biological and cultural inheritance began to feed back on one another, culture shaping our biology as much as our biology shaped our culture. We became increasingly biologically adapted for cultural coordination and transmission, which increased our biological dependence on cultural traditions, which selected for further adaptations for culture. Human physiology and life history became increasingly adapted to culture, and culture became more and more central to the way in which we inhabited our environment. We gradually became a cultural species, and the flexibility and innovativeness of our cultural nature allowed us to expand into and thrive in nearly every environment on Earth.1 This phylogenetic story concerning our development as a cultural species can be supplemented and supported by an ontogenetic story concerning the development of human individuals. Recent work in developmental and comparative psychology suggests that the most distinctive features of human child development concern the emergence early in ontogeny of sophisticated social and cultural capacities that have cascading effects on the shape of human ontogenetic development. The general capacity to soak up and inhabit culture, the development of which is central to our phylogenetic story, is just as central to our ontogenetic story. Michael Tomasello (2019) has called the skills, capacities, and motivations required for participating in cultural coordination and transmission “the shared intentionality infrastructure,” and he argues that this infrastructure develops in two basic steps: first, the emergence of joint intentionality, including capacities for joint attention, perspective taking, and the building of common ground, at around 9 months of age, and second, the emergence of collective intentionality, including capacities for coordinating perspectives, conventional communication, and normative self-regulation, at around 3 years of age (p. 8). By 3 years of age, human children have developed a suite of skills, capacities, and motivations that collectively amount to a shared mental infrastructure that supports our distinctively cultural way of being in the world and radically distinguishes the human mind from that of our closest evolutionary cousins. Comparative studies of the cognitive capacities of great apes and human children bear this out. Hermann et al. (2007) tested 2 ½-year-old children and mostly adult chimpanzees and orangutans in a wide battery of cognitive tasks spanning both the physical and social domains. Tasks in the physical domain measured the understanding of quantities, spatial relations, tool use, and causality, while tasks in the social domain measured intention reading, communication, and social learning. Aggregating the results within these two domains, the study showed that the 2 ½-year-olds performed on par with the chimpanzees and orangutans in the physical domain but far out-performed the apes in the social domain, particularly in the tasks measuring social learning. The results support what the authors call “the cultural intelligence hypothesis,” that human beings’ especially powerful skills of socio-cultural cognition early in ontogeny serve as a bootstrap for the development of cognition generally. Selection operating on the pace and patterning of development can have profound and cascading phenotypic effects. In a subsequent study Wobber et al. (2013) investigated the developmental trajectory of cognition in both the physical and social domains of human and chimpanzee infants from the ages of 2 to 4. Like Hermann et al. (2007), they found that human 2-year-olds’ performance in the physical domain was on par with 2-year-old chimpanzees. It was not until the age of 3 that the human children started to show a statistically significant difference in performance, and this difference was not pronounced until the age of 4. In the social domain, on the other hand, the human children's difference in performance was already pronounced at 2 years old, and this difference increased steeply with age. The chimpanzee infants showed no increase in performance between 2 and 4 years old. Enhanced capacities for social cognition were thus shown to emerge earlier in human ontogeny and develop more quickly than capacities for physical cognition, these early-emerging capacities for social learning and communication likely playing a role in scaffolding the subsequent development of other aspects of cognition. Human beings have evolved to have very different life histories than chimpanzees and, presumably, our last common ancestor, differences that are particularly pronounced in childhood (Nowell, 2021). All primates exhibit a prolonged juvenile period relative to other mammals (Charnov & Berrigan, 1993), but the human juvenile period is longer than would be expected even in comparison to other primates (Bogin & Smith, 1996; Walker et al., 2006) and has steadily lengthened over our evolutionary history, providing more opportunity for the acquisition of cultural skills and for the cultural shaping of cognitive development.2 Furthermore, the comparative developmental data demonstrates that, despite the ultra-prolonged human juvenile period, key cognitive developmental trajectories relevant to cultural coordination and transmission are pushed significantly forward in human ontogeny. The shape of human ontogenetic development is thus markedly different than that of our closest evolutionary cousins (Rosati et al., 2014). Key capacities for joint attention and shared intentionality emerge early in human development and provide the foundation for much that comes after, including the development of conventional language. Human beings thus appear to have been biologically built over our evolutionary history to soak up culture in ways that non-human animals simply have not.3 In virtue of human beings’ evolved and early-emerging capacities for social cognition, human epistemic activity is shared from very early in ontogeny. While social learning is common in the animal kingdom, humans have evolved species-unique social learning capacities that have greatly increased the fidelity and bandwidth of human knowledge transmission, facilitating the preservation and accumulation of cultural knowledge over time. Many animals—from insects to fish to birds to primates—are capable of learning by observing and copying the behavior of others (Laland, 2017). In a way, this should be no surprise. In order to survive, an animal must be capable of extracting information from its environment. Others of the same species are a significant feature of the environment of many animals and are potent reservoirs of information. The ability to extract information from these others can thus be an important tool for survival. In the most rudimentary form, such information can be gleaned from simply observing the behavior of others in their environment. A raven might learn that a food source is safe to approach by observing others approach it, or it might learn that a piece of food can be acquired through incrementally pulling up the string on which it hangs by observing another raven perform this procedure (Heinrich, 1999). In such instances, the behavior of others draws the animal's attention to information it might not have acquired otherwise, and many animals have evolved learning strategies that home-in on the wealth of relevant information revealed by the behavior of others.4 Chimpanzees are particularly adept at this kind of social learning, acquiring an array of foraging techniques, including nut-cracking, termite-fishing, and ant-dipping, by observing the behavior of other chimpanzees. These techniques are population specific, showing significant variation across different chimpanzee groups, and they are largely passed down from mothers to offspring. Chimpanzee infants are in nearly constant contact with their mothers, and as such they have ample opportunity to observe their mothers’ foraging behaviors and experiment with their tools. Simply paying attention to their mothers and being attuned to their mothers’ behavior facilitates chimpanzee infants in acquiring important information about their environment. This ability to learn how to manipulate one's environment by observing it being manipulated by others is often termed “emulation learning” (Tomasello et al., 1993). Chimpanzees’ capacity for emulation is greatly enhanced by their capacity for what Tomasello (2019) calls “individual intentionality,” for understanding others’ goals and intentions and appreciating successes and failures in their intentional activity. By observing others trying, succeeding, and failing to manipulate their environment, chimps are adept at learning about causal features of the world around them, for instance, that nuts can be smashed open by a rock or that termites will cling to an object thrust into their nest. In cultural learning, learners do not just direct their attention to the location of another individual's activity; rather, they actually attempt to see the situation the way the other sees it—from inside the other's perspective, as it were. In this case, learning is social in a way that individual learning enabled or supported by the social environment is not. It is learning in which the learner is attempting to learn not from another, but through another. (Tomasello et al., 1993, p. 496) On Tomasello et al's construal, cultural learning requires not just the ability to understand others as intentional agents with a perspective on the world, something apes can do, but the ability to actively “take” or “inhabit” the perspective of another and to learn from within that perspective. In learning from within another's perspective, one is not merely extracting information from the other's activity, but instead, in a way, participating in that activity, entering into an intersubjectively or collectively shared body of information and know-how. Tomasello et al's definition of cultural learning raises intriguing questions about the metaphysics of human mindedness, but for my purposes here these can be largely left to one side. More important is their detailed documentation of the varieties of learning behavior present in human children but largely absent elsewhere in the animal kingdom. Tomasello et al identify three general types of cultural learning: imitation, instructed learning, and collaborative learning. Imitation learning is distinguished from emulation in that it involves learning and reproducing the specific behavioral strategies used by others to produce results, not just learning about the results themselves. Instructed learning is learning through active teaching, where the learner internalizes something of the instructions of the teacher. Collaborative learning involves peers collaborating to learn together something novel, something that neither knew beforehand. I here focus on the first two transmissive forms of cultural learning, leaving collaborative learning for another occasion. Revealingly, what epistemologists would recognize as the topic of testimony appears nowhere in Tomasello et al's original 1993 paper. Part of the reason for this is that empirical investigations of social learning understandably tend to focus on skill transmission, on knowledge-how rather than knowledge-that, where the activity of telling and being believed has a reduced (though not non-existent) role. More significant for epistemology however, is the fact that so much of the cultural knowledge transmission crucial to our evolutionary story, from food processing and foraging techniques to tool use and manufacture, appears to be supported by forms of learning that are not obviously testimonial and that do not appear in standard epistemological accounts of sources of knowledge or ways of knowing, even within social epistemology. Since the publication of Tomasello et al's 1993 paper, there has been an explosion of work in developmental psychology on children's testimonial learning,6 and as a result, the chapter on cultural learning in Tomasello's recent book, Becoming Human: A Theory of Ontogeny (2019), does include a section on testimony. However, this section consists of only two paragraphs. While I think that testimony is more significant and deserves more attention than a mere two paragraphs would suggest, the relatively minor position that testimony occupies within the cultural learning literature raises the question of whether there is a whole lot about knowledge transmission that epistemologists have been missing.7 To get a sense for what epistemologists might be missing, it will be worth discussing in more detail some of the varieties of cultural learning distinguished within the empirical literature. Imitation is clearly one of the most important forms of cultural learning from an evolutionary and developmental perspective. Psychologists have distinguished several different kinds of imitation, some of which might be present elsewhere in the animal kingdom, some of which appear to be exclusively human. Imitation is distinguished from emulation by the fact that, when imitating, an agent is aiming to reproduce not just a result but the specific behavioral procedure used by another to produce the result. When a chimpanzee observes another chimpanzee open a jar by twisting off the lid, for example, the observer might learn that the lid can be removed and might set about removing the lids of similar jars by whatever means at its disposal, likely by the application of brute strength. This is emulation learning. The observer has learned something about the world (that the lids of jars can be removed) but not a specific procedure (twisting the lid) for producing this result. Like great apes, human children are quite adept at emulation learning. They understand others goals and intentions and pay attention to outcomes, and as a result they are able to learn about causal features of the world by observing the results of others’ activity. However, human children pay much more attention to others’ actions themselves than do great apes, and they are much more disposed to reproduce those actions. 2-year-olds will conform to a specific procedure used by a group even when they already know a different procedure for achieving the same end (Haun et al., 2014). Apes rarely do this. Once they have discovered a successful procedure, they tend to stick with it, regardless of what others are doing (van Leeuwen et al., 2013). Moreover, human children are prone to what is often called, somewhat unfortunately, “over-imitation,” to reproducing elements of a procedure that are unnecessary and irrelevant to achieving the desired result. Chimps show little tendency to do this (Horner & Whiten, 2005). Chimps focus on what it takes to get what they want, ignoring everything else. Whereas children are disposed to do what others do because others do it, apes are disposed to do what others do because of what it reveals about how they can get what they want. Human children are also adept at what is called “role-reversal imitation.” Human children as young as eighteen-months are capable of understanding and reversing roles in a joint procedure (Carpenter et al., 2005). When a child and an adult play different roles in a procedure, such as the adult holding a block steady so that the child can place another block on top, the child learns both roles, later having the ability to switch and play the role previously played by the adult. This requires not only the ability to understand others as intentional agents with a perspective on the world—something apes can do—but also the ability to take the perspective of another, something required for understanding one's role in a practice and, I would argue, for being guided by social norms. Chimpanzees do not do this—they do not reverse roles played by others—and this is connected to a general difficulty with engaging in joint attention and joint activity (Tomasello & Carpenter, 2005). The distinctive forms of imitation in which human children engage at a very young age help to make possible the transmission of much practical knowledge required for cumulative cultural evolution, and they do so by drawing on our species-unique capacities for shared intentionality. Human children are motivated to do what others do, simply because others do it, and they are capable of taking the perspective of another and learning about the world from within the other's perspective. While there is some evidence that great apes are capable of imitating actions, most of this evidence concerns “enculturated” chimpanzees, chimpanzees raised in human families. For instance, Buttleman et al. (2007) found that enculturated adult chimpanzees are capable of behavior comparable to 1-year-old human infants in the “rational imitation” paradigm of Gergely et al. (2002). As Henrich and Tennie (2017) point out, however, even though some human-raised chimps have been shown to be capable of some human-like imitation, their levels of imitation fall far below that of human toddlers and far below the threshold of what would be required for cumulative cultural evolution.8 Empirical studies aimed at measuring imitation must asses whether a subject actually reproduces a demonstrated behavioral procedure. Since a behavioral procedure can be reproduced more or less faithfully, this means that imitative learning can be assessed as being more or less accurate, or as exhibiting a greater or lesser degree of fidelity. The question, then, is not just whether a procedure has been copied or reproduced but how well, and while it has been shown that apes and some other non-human animals might be capable of some imitation, the fidelity of non-human animal imitation is quite low. As Kevin Laland and colleagues have demonstrated using mathematical modeling, the fidelity of transmission is of crucial importance for cumulative culture. If the fidelity of transmission in a population is too low, then no matter how much novel invention or modification occurs, culture cannot accumulate (Lewis & Laland, 2012). The increasingly faithful imitation that we see in human children thus appears to be an important part of our evolutionary story. Human beings evolved the ability to imitate more fluently and accurately than our closest evolutionary cousins, and this was a key factor in what allowed our cultural traditions to become cumulative. Advanced imitative capacities increase not only the fidelity of human cultural transmission, but also its bandwidth, including both the scope of information available for transmission and the sheer volume of information conveyed. Imitation allows for the learning of skills that are causally opaque and thus that cannot be acquired through emulation alone. Learning a skill via emulation requires that the causal affordances of objects be made transparent through the behavior of others. If these causal affordances are not transparent, then emulation will be insufficient for skill transmission, as was sadly the case for the Burke expedition. Their attempt to emulate the Yandruwandha skill of producing Nardoo cakes was a failure because key detoxifying components of the Yandruwandha skill were not transparent from the seeds and cakes themselves. Burke and his men needed to imitate the Yandruwandha's actions themselves, even though the underlying causal basis for those actions was opaque. The underlying casual basis for a particular cultural practice is often opaque even to skilled practitioners themselves. This is frequently the case for the elaborate food detoxification procedures employed by small-scale foraging societies. No one in the community might know why exactly bitter manioc must be processed through a multi-step procedur