1Ethological research on sociability and culture in the animal world has boomed in the last half century. Nevertheless, this research is still relatively unknown in the social sciences. This was not always the case. In their early years, the social sciences developed in close association with naturalists’ reflections on animal societies, as demonstrated by the works of Auguste Comte, Herbert Spencer, Alfred Espinas, and Émile Durkheim (Stark 1962; Guillo 2003). At the turn of the twentieth century, however, the social sciences began to distance themselves, slowly but surely, from that research topic and from the branches of the life sciences in which it is now studied.  Indeed, one could even say without exaggeration that the research conducted in the life sciences on animal society and culture has been either largely ignored—often in the name of the epistemological autonomy of the social sciences—or vehemently rejected as representing a fallacious naturalist reductionism.  However, understandings of animal society and culture have undergone a conceptual revolution since the 1960s, with studies producing numerous empirical discoveries.
2This article aims to provide an overview of this research by describing, as faithfully as possible, the information it has brought to light, the way in which it approaches the question of animal sociability and culture, and its specific conceptual features, all while bearing in mind how sociocultural phenomena have traditionally been understood in the social sciences. By way of introduction, we will attempt to show that the category of “animal-machine,” used freely in certain strands of the social sciences to disparage ethological models, is far too vague and unstable to provide an accurate representation of ethology’s characteristic perspectives. Next, we will summarize the two main topics—relatively autonomous despite connections between the two—upon which ethological research into animal sociability, in the broadest sense, has focused in the last half century: social behavior sensu stricto, and cultural phenomena. Finally, an overview of how ethology understands animal society, culture, and agency will enable us to outline the conditions under which a critical debate that engages with the social sciences could be productive.
Does ethology view animals as “machines”?
3Current studies aimed at rehabilitating animal agency in the social sciences or philosophy commonly depict ethology as a discipline based on a conception of animals as “machines.”  The conditions for the recognition of animal subjectivity supposedly imposed by the “animal-machine” model and its associated epistemology are so strict that researchers trying to meet them would be forced to abandon the search for subjectivity or the attempt to shed any light on its content. Nevertheless, unless it can specify precisely how ethology expresses and deploys this purportedly “mechanical” understanding of living things, this characterization may turn out to be extremely misleading. The concept of “machine” is, essentially, too vague and polysemous to provide a sufficiently rigorous account of an entire discipline’s concerns, let alone the twists and turns of its historical development.
4We make no attempt, here, to present an analysis of the epistemology that underlies ethology, even for the recent period. Our aim is simply to move beyond the “animal-machine” characterization in order to point to several distinctions that are important for the purposes of dialogue with the social sciences. Namely, we must distinguish clearly between at least five stances, on the one hand, and five lines of questioning, on the other. Any one of these stances could be taken as evidence that the ethologists who adopt it adhere to some extent to a form of “mechanism.” But they all involve different problems and may be adopted, or rejected, independently of each other, a fact that is inevitably obscured by the use of a single category like “animal-machine.” These stances are: the refusal to use a mentalist vocabulary; the requirement to always choose the explanation that presents animal psychological abilities in the least complex way (a principle generally known as “Morgan’s Canon” [Morgan 1903]); the use of the experimental method; the explanation of behaviors using the stimulus-response model; the explanation of behavior on the basis of innate patterns. The five lines of questioning to be distinguished relate to the conventional basic typology of ethology : the description of studied behavior; the determination of the immediate causes that trigger it in an animal; the determination of its function; its ontogenesis; and finally its phylogenesis.
5These distinctions are extremely important. They allow us to show how the “animal-machine” characterization creates conceptual confusion that obscures the differences between ethological paradigms, and, more broadly, the different ways the various strands of the discipline understand animal agency. In fact, these paradigms can all be characterized simultaneously as “mechanistic” with regard to some of these questions or stances, and “non-mechanistic” with regard to others.
6Behaviorism, for example, is firmly opposed to the use of mentalist vocabulary in the description of behaviors. For the experts who adhere to this school of thought, such vocabulary refers to entities—mind, instinct, thought, will, etc.—that are non-observable and cannot, therefore, be subjected to scientific analysis. Nevertheless, they treat the description of human behavior in exactly the same way. In behaviorism, the “mechanistic approach” is not reserved for animals. The paradigm does not, therefore, propose a “Great Divide” between humans and non-humans in this respect. In behaviorism, moreover, this stance on description is widely associated with Pavlovian views according to which conditioning can explain the proximate causes of behavior. Behavior is understood here as a response that is proportionate to a typical stimulus, whereby the whole pattern is the product of a learning process based on conditioning—in other words, acquisition during ontogenesis. In Pavlov’s classic example, dogs can learn to salivate in response to an aural or visual signal if the signal is repeatedly associated with the presentation of food. The animal does not, in this scenario, seem to be responding to patterns inscribed “mechanically” in its innate instincts.
7Classical ethology, which emerged in the 1930s with the works of, most notably, Nikolaas Tinbergen and Konrad Lorenz, broke radically with many of these principles. Lorenz, in particular, expanded the list of examples that countered the idea of behavior as a “set of conditioned reflexes.” He asserted the existence of “coordinations of innate and highly specialized movements.” Regarding the ontogenesis of complex behaviors, he maintained that these are formed by a complex layering of sequences, some of which are innate, while others contain gaps, “representing, rather than an instinctive act, an ‘ability to acquire’” (Lorenz 1970 , 20).  In a famous study, he demonstrated that during a short and clearly defined period of their development, graylag goslings tend to bond with a being in their environment that has certain characteristics that their mothers have, although the goslings can be deceived up to a certain point. This being serves as a default image that, once the goslings reach maturity, will enable them to recognize members of their own species and so trigger innate sexual behaviors.  In short, for Lorenz, the functionally homogenous behaviors of higher animals are not simply acquired through repeated stimulation, as behaviorists claim, but are innately programmed to be acquired. The rehabilitation of the innate basis of behaviors has prompted this strand of ethology to tackle the question of their function. The task is to explain why such instinctive behaviors became established in the species. In this approach, function is considered from a phylogenetic perspective: it is defined in terms of a behavior’s usefulness for the survival of the species. Together, the rehabilitation of instinct on the one hand, and the interest in the function of behaviors on the other, led classical ethology to significantly relax its demands for strict physicalist description and (as attested by Lorenz’s use of notions like instinct and ritual) to partially lift the ban that, in behaviorism, prohibited the use of a thick mentalist or sociological vocabulary. Moreover, as this approach is no longer focused on the quantitative measurement of the relationships between stimulus and response, researchers have gone back to collecting data in the natural environment, thus emphasizing the limits of the experimental method.
8In contemporary ethology, non-physicalist—i.e., mentalist or social—vocabulary is largely accepted, or at least tolerated, for describing behaviors in terms of their function when the latter is deemed by the observer to be self-evident. For example, as the author of a methodological reference work writes, we could agree to describe a bird’s flight when approached by a predator as “escape” behavior rather than as “the rapid alternate contraction and relaxation of the pectoralis muscle” (Lehner 1996, 82). Similarly, the use of slightly anthropomorphic sociological categories like “cooperation,” “dominance,” “altruism,” or “reconciliation” (de Waal 1990) is now seen as acceptable by many ethologists as long as they are used in a clearly functional sense and/or to explain the evolutionary forces that influenced the selection of the behaviors they designate. As the ethologist Adam Miklósi points out:
[…] animals living in groups might have similar problems to solve (dominance, cooperation, etc.) or similar evolutionary forces have selected them for living in a group in the first place.
10This is why, for some ethologists, a form of “functional anthropomorphism” is acceptable (Miklósi 2007, 15).
11In contrast, the overwhelming majority of ethologists nowadays are very—in some cases extremely—reluctant to use this sort of vocabulary when it comes to accounting for the proximate causes of a given behavior, or in other words the organico-psychic mechanisms that underlie it. They point out that behaviors described in identical functional terms—altruism, reconciliation, cooperation, for example—may be the result of very different organico-psychic mechanisms in different species. Similarly, a function like “escape” may be fulfilled by organs that are completely morphologically and phylogenetically different (the wings of flies or birds, for example). Thus, a dog’s behavior toward the members of the human family with which it lives is very similar, in Western societies, to that of the family’s children: play, relative submissiveness, etc.  But the mental states of the dog, according to these ethologists, have nothing in common with those of the children. On this question (of the proximate mechanisms involved in the triggering and performance of a behavior) and on the question of the ontogenesis of behaviors, most contemporary ethologists follow Morgan’s Canon, although some, such as the cognitive ethologists, apply the rule less strictly or radically than others.
12As should be clear from the above, we will be unable to gain any insight into how contemporary ethology conceptualizes animal behavior until we move beyond the idea that it treats animals as “machines.” At the very least, it is essential to clarify in exactly what sense, and with respect to which specific questions, it does rely on such a “model.”
13We will now focus more narrowly on ethology’s approach to social behavior. As we will see, this question has long been an enigma, even within classical ethology. A revolutionary new paradigmatic framework began to unravel the puzzle in the 1960s, with important conceptual implications for a whole strand of ethology, which from then on would remain firmly tied to the neo-Darwinian theory of evolution.
The explanation of animal social behavior: The success of neo-Darwinian evolutionary models
The kin selection model
14It would not be an overstatement to say that classical ethologists did not see the theoretical challenges posed by social behavior as profoundly different from those associated with other animal behaviors. As discussed above, most of them focused their attention on the description and study of the proximate mechanisms involved in the structure, triggering, and performance of a given behavior, as well as on its ontogenesis. Phylogenetic explanation was not completely absent, but it was applied with reference to a behavior’s “usefulness for the survival of the species,” which more often than not was deemed sufficiently obvious as to render further investigation unnecessary.
15When it comes to social behaviors, however, this sort of explanation could hardly satisfy supporters of the “evolutionary synthesis” that began to dominate biology in the 1930s (see Mayr 2001). This theory accorded the mechanism of natural selection a crucial explanatory role. In this context, natural selection is defined as differential reproduction among individuals in the same population who all carry variants of the same trait, when such variants are heritable and do not all have the same “adaptive value.” In other words, the different variants do not give the individuals who carry them the same probability of producing offspring. For example, a slight heritable difference in skin color between two soles living in the same geographic niche gives each sole a different probability of being eaten by a predator, and therefore a different probability of producing offspring. Over many generations of descendants of the two individuals, this reproductive differential can bring about a process of natural selection in the population within which they live.
16Yet, as Darwin himself emphasized (Darwin 1998 , 192–3), several behaviors usually classified as social in the natural world present serious problems for this explanation. How is it possible for natural selection to perpetuate behaviors that reduce, for the benefit of others, an individual’s own chances of reproducing—for example, worker ants and bees who work or are willing to defend their society to the death against intruders? Let us imagine what the ancestors of such insects were like before this complex division of labor and the social behaviors associated with it began to spread through the populations formed by their descendants. Let us then imagine that one of the traits upon which this social life now depends arose in one of the ancestors through mutation—for example, an inclination to bring back food for a sister, or to defend the hive to the death against intruders, or a primitive form of sting that kills its possessor when used. This “altruistic”  individual would, by definition, be less likely to have descendants than others in the same population, not just because this trait would reduce its own chances of survival, but also because it would improve the survival rate of its peers without the trait, who would benefit from the help of the “altruist.” This reproductive differential should destine such traits to disappear from the population almost as soon as they appear, starting in the subsequent generation, and to an even greater extent after several generations.
17This argument has long been used by defenders of neo-Darwinism to counter theories based on the principle of “usefulness for the species”; simply showing how a trait might be useful for a population or a species does not explain why and how it was selected. To circumvent this difficulty, certain researchers in the 1960s proposed a more elaborate version of the principle of usefulness for the species: group selection. In a now classic work, the Scottish naturalist Vero Copner Wynne-Edwards discussed the example of Scottish ptarmigans. In times of overpopulation, some male individuals seem to resign themselves to leaving their territory and so renouncing the chance to procreate (Wynne-Edwards 1962). Wynne-Edwards claimed that this sort of “altruistic,” “social” behavior—in that the individual adopts a behavior that is manifestly beneficial to the rest of the population but reproductively detrimental to itself—had been able to establish itself in the species because selection operated here not at the level of genes or individuals, but at that of groups or populations within the species. Groups in which such behavior took place fared better in inter-group competition than others, which died out.
18Against this model, a group of researchers including William Hamilton, John Maynard Smith, and George Christopher Williams, argued that, despite appearances, the principle of “group selection” was unable to solve the enigma described above. No matter how useful such traits are for the group, there was still no explanation for how they could be passed down through successive generations once they have appeared, given that they reduce the reproductive chances of individuals who possess them, while increasing the chances of individuals who do not. Building on this criticism, these researchers went on to construct an evolutionary model that was compatible with the principle of natural selection as it is understood in neo-Darwinism. Their theory would revolutionize how biology approached social behavior in the natural world. It was expressed concisely by Hamilton at the beginning of the 1960s in an article on Hymenoptera, an insect order comprising mainly social species, including bees and ants (Hamilton 1964). 
19To solve the enigma, Hamilton extended the classic idea of the fitness (the adaptive or reproductive value) of a trait, or more accurately of the gene that governs it, to include not just the usual definition of that trait’s effect on its possessor’s ability to produce direct descendants, but also its effect on the number of descendants that its possessor’s close kin could have—siblings, father, mother, for example. Incorporating this “inclusive fitness” in the calculation makes it possible to accommodate a key fact; some genes can be passed on by other means than directly through the reproduction of the individual that carries them. Genes can spread through populations by virtue of their positive effect on the reproduction rate of individuals with a high chance of possessing the same gene—namely, the kin of the individual with the gene. To put it another way, there are two ways genes can permeate through populations over the course of successive generations: first, the classic, direct route whereby a gene increases the reproductive chances of the individuals that carry it; second, the indirect route whereby a gene increases the reproductive chances of the close kin of the individuals that carry it.
20The second mechanism is associated with a specific type of natural selection called “kin selection” (Hamilton 1964). As Hamilton points out, the “kin selection” model provides a solid Darwinian evolutionary explanation for many social behaviors that he calls “altruistic.”  A gene that governs an altruistic behavior can spread through a population over the course of generations if the reproductive benefits that the behavior accords individuals with a high chance of carrying the gene (i.e., kin) outweigh its costs—the reduced probability that the individual who carries the gene will reproduce, up to and including sterility, as among worker bees.
21Hamilton formalized this argument in a “rule” that now carries his name. Let C be the cost of the altruistic behavior in terms of the fitness of the individual that carries the gene; let B be the behavior’s benefit in terms of the fitness of the individuals that benefit from it; and let r be the coefficient of genetic relatedness between the first individual and the others. On that basis, an altruistic behavior can permeate through a population over the course of generations if:
23To give an example, this rule means that in an animal population where siblings have a genetic relatedness coefficient of ½—i.e., where they are likely to share half their genes, as among mammals—the kin selection mechanism will select for an altruistic behavior if, all other things being equal, it allows the sibling of a completely altruistic individual to increase its total number of descendants by more than double the number of direct descendants the altruistic individual would have had if it had not been altruistic, in other words if it had devoted all its energy to its own reproduction. In short, if a behavior reduces the number of direct descendants an individual can expect to have by one, it must allow that individual to have at least two extra nephews or nieces in order for the gene that governs it to be selected by this mechanism.
24For the same reason, “Hamilton’s rule” predicts that the closer the genetic relationship between individuals of a species, the greater the chance that “altruistic” behaviors will be spread through that species by the kin selection mechanism. This prediction was convincingly confirmed by Hamilton himself with his study of Hymenoptera—an order of insects that includes bees, wasps, ants, and bumblebees. Sociality has been observed thirteen times in insects, twelve of which were among the Hymenoptera. In these species, reproduction follows “haplodiploid” stages of development, giving the coefficient of genetic relatedness between individuals some rather unusual properties: sisters share, on average, 75% of their genes (in species with diploid development, like mammals, they share only 50%); meanwhile males share 100% of their genes, compared to 50% in species with diploid development. This significantly reduces the amount of reproductive benefit that altruistic behavior needs to confer indirectly to kin (by increasing their reproductive chances) in order to outweigh the cost of partial or total reproductive abstinence in the altruistic individual. Therefore, it makes the emergence of sociality much more likely in these species. This feature of the social Hymenoptera provided striking confirmation for the “kin selection” theory, which has since become a cornerstone of explanations of the emergence of sociality in the animal world.
Other routes to social life
25Starting in the 1960s, and particularly in the 1970s, other neo-Darwinian models were added to the kin selection mechanism in recognition of the diversity of evolutionary routes to social life in the natural world. First, the “reciprocal altruism” model put forward by Robert Ludlow Trivers (Trivers 1971) demonstrated that, under certain conditions, a gene governing altruistic behaviors can have a positive selective value—in other words, be perpetuated by natural selection—even when those behaviors benefit non-kin. Trivers states that this is possible as long as the beneficiary of the altruistic behavior is willing to return the favor and help those individuals that previously helped it. In such cases, the individual cost of cooperation does not constitute an evolutionary barrier to the benefits that cooperation can provide. However, as Hamilton points out, this mechanism only works in very specific circumstances. In order for it to function, reciprocity must actually exist in practice: if an animal helps ungrateful individuals, the genes that cause it to behave altruistically will rapidly disappear from the population, for the reasons discussed above. To put it another way, this form of altruism can only really develop in species where individuals are capable of recognizing who has helped them and whom they have helped, in order to avoid free-riders. To date, support for this hypothesis has only been found in a few mammal species, including the olive baboons of Tanzania (Packer 1977), vampire bats (which share blood) (Wilkinson 1984), and above all—to a very high degree according to biologists—humans.
26In the 1970s, researchers, led by Maynard Smith,  also began to draw on game theory to develop neo-Darwinian theories for the evolution of animal social behaviors. Maynard Smith attempted to model the evolution of social behaviors by treating them as strategies in a game where victory is measured in terms of the probability of an individual’s survival, and so ultimately how many descendants it has. In this formal structure, the frequency with which each strategy is used varies with each new turn—each reproductive cycle—as the game is repeated. Thus, we can measure the evolution (in the Darwinian sense) through successive generations of the respective frequencies of two or more competing strategies—two variants, or “alleles,” of the same gene—for example, cooperating or displaying hostility when meeting a conspecific. This allows us to identify the conditions under which a gene governing such behaviors is able to spread through a population. This “evolutionary game theory” has generated important and, in some cases, counter-intuitive theoretical results. In particular, it has demonstrated that, far from always leading to the disappearance of one strategy in favor of another, natural selection can in fact produce situations of equilibrium in which the two competing strategies co-exist, although with very different frequencies, one being much more common than the other. 
27In the 1970s, these and other similar  models formed the backbone of a paradigm described by its originator, the entomologist Edward Osborne Wilson, as a “synthesis” (Wilson, 2000 ) of new ethological perspectives on social behavior: sociobiology. In this synthesis, Wilson suggested extending sociobiological principles directly to humans and commented that, thanks to these models, the life sciences would soon “phagocytize” the social sciences. These summary extensions to mankind provoked fierce and highly ideological controversy that obscured the real objects of study, the models and concepts discussed above, which were thenceforth indiscriminately lumped together under Wilson’s synthesis.
28This controversy largely petered out in the 1980s. Neo-Darwinian explanatory models have flourished since then. They have enabled the collection and interpretation of a large amount of data on social behaviors in the animal world. Nevertheless, when the paradigm was extended to ethology, it was more often than not referred to as something other than sociobiology—namely, “behavioral ecology”—even though the theoretical perspectives denoted by the various labels were the same. Moreover, it was mainly preserved in its original form and dealt with species or topics that were relatively far removed from the thorny and controversial questions raised by human sociability. Explicitly neo-Darwinian studies of sociability in humans and species seen as close to humans—in particular, non-human primates, or even dogs—underwent a gradual change of theoretical focus that once again radically altered how ethology approached these issues.
Research on species close to humans: The growing influence of cognitive science
29While sociobiology—as its name suggests—generally drew a direct line between genes and sociocultural traits, at the end of the 1980s a number of studies on species close to humans started to look more at psychological questions.  This reorientation was due in large part to the remarkable rise of cognitive science around the same time—a rise that has continued unabated since then. Many of the questions explored in this context revolved around social cognition, and more broadly the cognitive abilities and mechanisms that make human culture possible (on this topic, see section three of this article). First among them was “theory of mind,” or the capacity to interpret the behavior of others by imagining their mental states—especially beliefs and desires—rather than relying on the cognitive schema of ordinary physical causality, as if such behavior were just like any other natural phenomena (Wellman 1990; Premack and Woodruff 1978; Tomasello 1999).
30Comparison between the abilities of monkeys, children, and dogs  was central to this approach. From around three to four years old, according to ethologists and psychologists, children begin to interpret other people’s behavior in terms of desires and beliefs. Other animals (including monkeys) seem not to do this, or at least not to the same degree of complexity.  Some currently very influential researchers hold that theory of mind was the prerequisite ability for the acquisition of language and the development of culture, in the full sense of the term (Tomasello 1999). This ability would feature prominently in the evolutionary explanation of what has been seen by many ethologists in the last twenty years (notwithstanding current in-depth debate) as a central difference between humans and other primates. Namely, humans are deemed to be unique among primates in that they tend to be spontaneously “cooperative” with members of their own species, while other primates are more “competitive,” except with their close kin (Hare and Tomasello 2004).
31Since the 1990s, this research has developed in close association with more traditional studies in primatology—in particular, numerous studies of dominance, hierarchy, and intraspecific aggression—and in developmental psychology. In the process, it has brought about a radical shake-up of the questions and debates that preoccupy those disciplines and shifted the conceptual center of the most theoretical strands of ethology. Thanks to the close ties between these ethologists and cognitive science, as well as interest in the phylogenetic (evolutionary) function of social behaviors, the long-standing ban on the use of mentalist vocabulary in behaviorism has been significantly eased. Nevertheless, although this vocabulary now seems tolerated—albeit with strong reservations among the most traditional ethologists—when talking about phylogenesis and the function of social behaviors, its use is still fiercely debated when talking about the immediate causes of such behaviors and the cognitive mechanisms that underlie them. Some ethologists, like Frans de Waal among the primatologists (de Waal 1990), interpret Morgan’s Canon much more radically than others. More generally, many ethologists remain detached from studies influenced by cognitive science, confining the majority of their work to the classic descriptive task of drawing up an ethogram of social behaviors for each species.
The question of culture in the animal world
32In parallel with this research into social behaviors sensu stricto—but in constant interaction with it—a vast field of theoretical considerations and empirical studies has developed in ethology around the topic of “animal cultures.” In contrast to the term “social,” which per se is rarely defined or debated in ethology (the “social” question is usually approached via other concepts, such as altruism, reciprocity, cooperation, or hierarchy), the notion of “culture” is, as we will see, the central concept of a field of research devoted to the study of behaviors that spread through populations of living organisms through a process of “social learning.”
Two groundbreaking studies: Tits and macaques
33Two studies are now seen as the pioneering works in this field. The first looked at the behavior of populations of great tits observed in the United Kingdom from the 1920s to the 1940s (Fisher and Hinde 1949). Over the course of the study, the researchers noticed that, in certain areas, more and more great tits were drinking milk from the bottles left outside houses in the mornings, which they accessed by flipping open or piercing the bottle tops. James Fisher and Robert Hinde showed that this behavior appeared independently among (was “invented” by) tits in several different parts of the United Kingdom, and that it had then been disseminated at a steady pace through numerous tit populations around the country. The trait spread far too quickly to have been caused by a genetic mutation making the birds adopt the behavior. Moreover, it spread in gradually increasing concentric circles centered on nodes of “invention.” In other words, its diffusion followed a pattern similar to that of an epidemic. For Fisher and Hinde, everything pointed towards the behavior being passed from individual to individual by a mechanism of tit-to-tit transmission.
34The second study widely seen as one of the founding works of the field investigated the Macaca fuscata macaques of the island of Koshima, Japan (Itani and Nishimura 1973). In 1952, researchers observing a troop of macaques noticed that they regularly ate sweet potatoes prepared using a technique, invented by a female that the researchers named Immo, that involved washing the vegetables in the sea to clean off the sand that covered them. This technique—or, if one prefers, this behavioral trait—gradually spread through the individuals in the troop until, five years after Immo’s innovation, it had been adopted by 80 percent of them. Here, again, the relatively rapid rate of diffusion ruled out the possibility that the trait could be associated with specific genes, and instead suggested an individual-to-individual transmission mechanism.
Phenomena found in multiple animal species
35Studies of these sorts of “cultural” traits or “animal cultures”—as researchers felt justified in calling them from then on—proliferated in the following decades. Numerous behaviors of this type were discovered among chimpanzees. In a wide-ranging synthesis, renowned primatologists combined their work to compile a list of thirty-nine traits that had spread through a similar mechanism among chimpanzee (Pan troglodytes) populations (Whiten et al. 1999). The list includes, for example, using modified pieces of wood to extract termites from their nests, catching termites using the central vein of a leaf, cracking nuts between two pieces of wood or by resting them on a stone and hitting them with a piece of wood, using leaves as bandages for wounds, etc.
36As the example of the tits shows, this type of trait is found in an extremely wide range of species, and not just those seen as close to humans. Rats, for example, seem to trust other rats’ choices when it comes to foraging and food preferences (Laland and Plotkin 1993). Captive dolphins have proven to be highly capable of imitating an action performed in front of them by a researcher (Kuczaj and Yeater 2006), as have dogs (Kubinyi, Pongracz, and Miklósi 2009). Numerous “cultural” variations in birdsong (Slater 2003) or in food-finding methods (Lefebvre 1986) have also been documented in certain bird species. In some fish species, preferences for specific migratory routes seem to be transmitted from individual to individual (Helfman and Schultz 1984).  Certain insects also display this capacity for transmission: for example, bumblebees in search of food abandon low-yield flowers more quickly when accompanied by an experienced individual (Leadbeater and Chittka 2007).
Culture as a set of traits transmitted through social learning
37These various studies now form a large and dynamic field of study based on a widely accepted definition of culture—although the finer details are still being debated (Laland and Galef 2009). A trait is classified as cultural, in the broad sense, if its presence in a population of a species is the result of individual-to-individual transmission. In other words, such traits
result from social learning rather than from genetic differences between populations [of a species] or differences in the way diverse ecologies shape behavioral development of individuals.
39To put it another way, for a trait to be classified as “cultural” it is not enough for it to be present in one population of a species and absent in another population of the same species. Its uneven distribution may be due, firstly, to genetic differences between populations. It may also be the result of variation in the resources available in the ecological niches occupied by the two populations. In that case, the trait’s presence in a population is due to the fact that the individuals in the population each acquired it individually (“individual learning”) by interacting with an environment that contained a resource with the potential to trigger acquisition of the trait—for example, a particular type of prey, or twigs that are particularly suitable for extracting food. The trait is absent in another population simply because that population’s niche does not contain the same resource. Here, the observable differences in behavior between the two populations are due to differences in their ecological niches rather than to the social transmission of an innovation that appeared in one of the populations. In summary, the behavioral trait is acquired in this case through individual learning rather than social learning. 
40As the central position accorded to social learning suggests, these studies were not unconnected—indeed, they were sometimes closely connected—to research into the social (in the strict sense) behaviors discussed above (see section two of this article). Nevertheless, given the definition of culture adopted by the two fields of study, the theoretical and empirical questions they explore are, in many ways, markedly different. In studies of sociality, many of the traits investigated are seen as typical features, embedded in genes, of the ethogram of the species in question, as is the case, for example, with “altruistic” behaviors among the Hymenoptera, or with the intraspecific aggression associated with social dominance among mammals like wolves or non-human primates. Moreover, it is precisely because such traits are embedded in genes that their occurrence in a species can be explained phylogenetically as the result of natural selection. However, there is no a priori reason that behaviors cannot be seen as “social” even if they are not embedded in genes. This is the crucial point: the origin of a trait—genetic or acquired (through individual learning or from a conspecific)—is irrelevant when it comes to determining whether or not the trait is social, although the social traits being investigated are, most of the time, thought to have a fairly direct genetic basis. Essentially, a behavior is classed as social or not depending on its effects on both the individual that performs it and on other individuals (see sections two and four of this article).
41In contrast, the question of a trait’s origin—or, if one prefers, the factors that explain its occurrence in an individual—is by definition of central importance in studies of culture. As discussed above, a trait is considered to be cultural if it is transmitted through social learning, which straightaway excludes traits that are—primarily—governed by genes. In this context, the “cultural” part of the natural world is universally defined as whatever is not “genetic.” More precisely, biological traits are all seen as the physical, physiological, or behavioral expression of information. This information itself is either contained in genes (the shape of a hoof, the color of the integument, the sterility of Hymenoptera, their “altruistic” behaviors), individually acquired (the color of the hydrangea flower—which varies depending on the soil where it grows—, the type of prey hunted by a predator species), or acquired through social transmission.  The third category by definition encompasses all “cultural” (in the broad sense) phenomena.
Variety and uneven complexity in social learning mechanisms
42Given this definition, the problems facing ethologists who specialize in the study of “cultural” phenomena are inherently different to those, discussed above, that form the basis of the study of sociality sensu stricto. Research and debate on this topic have largely focused on social learning. The first studies in the field showed that a behavioral trait can emerge in an animal due to contact with a conspecific. But in doing so, they opened up a vast field of study: what learning mechanisms and what cognitive abilities are involved in passing information from one individual to another in different species?
43Right from the earliest studies, many ethologists adopted a cautious stance and cast doubt on the idea that social learning in animals is based on a process of imitation similar to that seen in humans (see Hopper and Whiten 2012, 455).  Continuing research has revealed a diverse range of mechanisms and cognitive abilities, with highly varying degrees of complexity, in the different species where transmission of a trait from one individual to another has been observed.
44On that basis, certain ethologists have been able to show that many of the phenomena grouped under this label—in particular, the example of tits opening milk bottles—involve mechanisms termed “local enhancement” or “stimulus enhancement,”  neither of which require more than rudimentary cognitive or social abilities. In such cases, the individual’s attention is directed towards an object or a place by a conspecific. The first individual explores the place or object—the milk bottle, for example—and of its own accord ends up treating the object in the same way as the conspecific; by pecking at the bottle, the tit eventually pierces the top and discovers that the bottle contains a food source. Here, the transmission of information hardly requires any social inclination—some form of herd behavior will suffice. Moreover, the animals involved do not actually need to possess the cognitive ability to extract and replicate the information that drives the behavior they observe in another individual—which would be an extremely complex ability.
45Another relatively simple mechanism discussed in this field is “response facilitation.” It occurs when a behavior that all members of a species are capable of performing is triggered when an individual observes a conspecific performing that behavior (Hoppitt, Blackburn, and Laland 2007). For example, individual A takes flight because it sees individual B take flight (individual B’s flight is in response to a predator).
46The term emulation denotes a mechanism that relies on more complex cognitive abilities, in that it refers to a situation where the information required for performing a behavior is not initially in the repertoire of the individual that acquires it through contact with a conspecific. In social transmission through emulation, the observer is motivated by the result or goal of its conspecific’s action, with no concern for the methods used to achieve it. Its acquisition of a new behavior—for example, feeding on a particular type of prey—is solely driven by that motivation, and “[it does] not copy the behavioral strategies demonstrated to reach it” (Hopper and Whiten 2012, 454). 
47Finally, imitation in the strict sense refers to a process whereby an individual acquires a new behavior by copying and adding to its repertoire an action it observes another individual performing (Whiten and Ham 1992; Tomasello 1999). This mechanism presupposes more complex cognitive abilities than the previous ones, especially if one considers, as many psychologists and ethologists currently do, that it would be impossible without some kind of understanding of others’ intentions, in other words, “theory of mind.” 
48As this graduated scale of social learning mechanisms culminating in “imitation” shows, the question of what makes human sociality and culture unique is often present in the background of ethological research on this topic or on social behaviors.
Current consensus and debates in ethology: The background question of what makes humans unique
49There seems to have been a fairly broad consensus in ethology for the last twenty years regarding the evolutionary explanation of social behaviors in the animal world. Similarly, the vast majority of ethologists now acknowledge that numerous species can display traits that are disseminated through certain populations by social learning. Most ethologists, however, hold that there are fundamental differences between such traits and human cultural traits (see Laland and Galef 2009). First, each species only displays a limited number of them—birdsong variation, for example, is often the only trait of this type found in the bird species in which it occurs. Above all, they are not cumulative. In contrast to animals, according to most ethologists, humans tend to build on and transform pre-existing cultural traits, sometimes making them more complex. Human social transmission immerses cultures in a cumulative dynamic, characterized by the “ratchet effect” (Tomasello 1999), that increases the number, range, and complexity of such traits until they encompass every aspect of the life of Homo sapiens sapiens: technology, language, and more broadly the countless institutions that frame human life. From an evolutionary perspective, therefore, humans possess a very special trait: culture. Produced in a constant stream by humans, culture becomes their own ecological “niche” and thereby significantly alters the action of natural selection, which no longer operates directly but is mediated by coevolutionary processes (Odling-Smee, Laland, and Feldman 2003). Most ethologists also subscribe to the idea that the characteristics of human cultures arise from the specific cognitive abilities used by humans during social transmission. First, the human species has an active tendency to teach conspecifics (“active teaching”), and not just offspring (Whiten 2009). Second, and above all, from a very young age humans are both able and highly inclined to imitate (in the fullest sense of the word) other humans. According to ethologists, this ability is itself enabled and enhanced by “theory of mind.” 
50More broadly, as we pointed out above, many ethologists now think it appropriate to use vocabulary normally used with reference to humans when talking about animals—society, hierarchy, dominance, culture, tradition, etc. But, again, for the vast majority of them it is vitally important that the use of this kind of vocabulary is not taken to imply that the mental mechanisms behind these behaviors—the proximate causes—are the same in animals as in humans. On this question, most ethologists fully subscribe to the principle of parsimony expressed in Morgan’s Canon. Nevertheless, as we alluded to above, not all ethologists apply the principle with the same rigor. Although many think that the majority of sociocultural phenomena in the animal world are rooted in basic learning mechanisms of the behaviorist type, others credit animal cognitive abilities with markedly more depth and substance. 
51Beyond these general commonalities, there are lively ongoing debates in all these fields of research. First, the role of genes continues to provoke significant controversy. Some researchers, following Wilson, champion a highly reductionist interpretation of social behaviors and culture, both among animals and among humans. Others, in contrast, hold that social learning mechanisms, which are by definition not subject to genetic determinism, belong to a unique class of phenomenon that is widely distributed in the animal world and even more so among humans. These researchers are relatively numerous in the field of research devoted to “animal cultures” (see Laland and Galef 2009). Still others reject the distinction—and the theoretical choice—between behaviors that are totally and unambiguously governed by genes on the one hand, and behaviors that are purely subject to social transmission with no genetic influence at all on the other. 
52More specifically, a number of ethologists believe that the differences in social learning phenomena observed in humans and animals are too great to warrant the use of the term “culture” when referring to animals. They prefer to talk about animal “traditions,” reserving “culture” solely for humans (see Laland and Galef 2009). For others, conversely, empirical studies suggest that the distinction between the two is not actually that clear-cut; in their eyes, the difference between human and animal culture is one of degree rather than kind. For example, they claim that a form of cumulativity has been observed in the tool use of certain wild chimpanzees, or that teaching behaviors are found among various mammal species, such as mongooses or cetaceans. 
53Finally, theory of mind is increasingly the subject of acrimonious discussions. Although a large majority of ethologists are nowadays in agreement regarding the extent of human uniqueness in this respect, some nonetheless feel that, as in the previous case, the border between humans and animals is not as clear-cut as we might think. According to these researchers, some animals—especially certain non-human primates, but also dogs—are not limited to a “physicalist” understanding of others’ actions, in other words one based solely on cause and effect. Rather, they are able, up to a certain point and under certain circumstances, to interpret others’ actions in “mentalist” terms, or in other words as the result of intentions.  For some researchers, like Frans de Waal, there are even grounds for suggesting that the great apes use a form of morality in their dealings with each other (de Waal 1990).
Ethology and the social sciences: What potential points of contact are there regarding the social and cultural agency of animals?
54It would certainly not be excessive to say, as we did in the introduction, that this ethological research into sociocultural phenomena in the animal world has been largely ignored—if not vehemently rejected—by the social sciences for a little more than a century. There are, admittedly, clear references to the discipline in major authors like George Herbert Mead, Claude Lévi-Strauss, or Erving Goffman. But in their works, ethology is more a tool reinterpreted in a new context than a set of facts and hypotheses that deserve to be foregrounded and articulated with socio-anthropological thought.
55Without going into the details of complex arguments and debates, we can say here that this rejection—or indifference—more often than not rests on a widespread belief that there is a qualitative difference between the human phenomena studied in the social sciences and the animal phenomena studied by ethologists. According to this view, humans and animals are separated by an epistemological, if not ontological, divide. This gap is thought to be so significant that the social sciences have no need for ethology, while the concepts of the social sciences can, at best, only be used in a rough, metaphorical way for explaining animal behavior—with all the risks of slippage that metaphors entail. 
56This belief in a “divide” has appeared under numerous guises: a distinction between nature and culture; a self-proclaimed Durkheimian holism; an insistence—also inspired by Durkheim, and more broadly by Kant—on the normativity of human social rules and their irreducibility to the actual behavior of individuals; an insistence, in culturalism or symbolic interactionism, on the existence of a symbolic faculty that transforms all spheres of human activity; or more recently, in pragmatic sociology or ethnomethodology, a redeployment in the social sciences of Wittgensteinian ideas like the distinction between reasons and causes or the irreducibility of action to cognition (Coulter 1989; Quéré 2011). 
57The idea that there is a clear distinction between nature and culture is currently being contested in the social sciences by a wave of studies that aim to rehabilitate animal agency and that draw on the works of Bruno Latour (Latour 1993 ), Philippe Descola (Descola 2014 ), or Tim Ingold (Ingold 2007), as discussed above. The researchers developing these perspectives often invoke the need to open the social sciences up to “ethology.” Nevertheless, the few academic ethologists whom these authors cite are mainly those whose model of the animal mind is the richest and most complex, or alternatively unorthodox researchers at the margins of or outside the academic field. In reality, the ethological works mentioned above are more often than not ignored, and they rarely appear in the bibliographies of this new strand of socio-anthropology. They are often rejected, outright and implicitly, in very general theoretical arguments directed against the “animal-machine” model, which is seen as the epistemological foundation of mainstream ethology. 
58There is no scope in this article for even a brief discussion of the validity of these highly complex arguments, which have provoked lively debate, particularly on the question of “naturalism.” For now, we will simply offer a few comments intended to define the contours of the epistemology that characterizes this branch of ethology, and of the concept of social and cultural agency to which it gives rise, in order to highlight areas that could provide points of common interest, connections, or critical debate with the social sciences.
59First, as discussed above, in these ethological works the word “social” is rarely elaborated or discussed in depth. When it is explicitly defined, it is usually reduced, without much clarification, to “cooperation”  or “altruism” (see above, Hamilton 1964). And although, as discussed previously, there is some debate about the definition of culture, it fades in comparison to that raised by the same concept in the social sciences. In general, most ethological research adheres to the same clear-cut and widely accepted definition of the term (as discussed above).
60For the purpose of our study, we must clarify the possible implications of and justifications for this epistemological stance in order to be able to pin down the meaning of what ethologists actually say—and what they are capable of saying—about sociability and culture. According to this epistemological stance, which is explicitly adopted by a number of researchers, these preliminary, clear-cut, and simple definitions are absolutely not—or at least ought not to be—intended to convey the philosophical meaning a concept should have, to match the meaning they can have in ordinary language, or to account for the mental states associated with the behaviors they allow us to describe. Definitions are understood here as no more than starting points whose essentially methodological aim is to delimit the field of study clearly. In this context, a good definition of “social” is simply a definition precise enough to enable the establishment of theoretical models or empirical data collection protocols that may be useful for tackling the cluster of complex and interrelated questions associated, at first sight, with the broad topic of social and cultural phenomena among animals. As the ethologist Richard Dawkins says:
I am not concerned here with the psychology of motives. I am not going to argue about whether people who behave altruistically are “really” doing it for secret or subconscious selfish motives. […] My definition is concerned only with whether the effect of an act is to lower or raise the survival prospects of the presumed altruist and the survival prospects of the presumed beneficiary.
62As discussed above, this sort of definition has unquestionably made it possible to provide a solid evolutionary explanation for collective phenomena among the Hymenoptera. Moreover, this descriptive language has allowed researchers to identify and empirically measure the extremely numerous and diverse behaviors used by individuals in the animal world to devote a portion of their resources to increasing their conspecifics’ chances of survival. For all these reasons, it hardly seems appropriate to reject such studies outright simply on the basis of the inadequate, cursory nature of the definitions they assign to the traditional concepts of the humanities or to common-sense categories.
63On the other hand, it is undeniable that some ethologists do not adhere strictly to this approach or to this interpretation of their data. Some of them—particularly the sociobiologists, or more recently the evolutionary psychologists—, ignoring the methodological status and limited scope of their definitions, are gradually slipping towards an ambition to account for all phenomena such terms might refer to, and thereby to resolve the philosophical, sociological, and anthropological questions that underlie them. There is no question that altruism as defined above cannot alone directly account for a phenomenon like gift-giving as it is understood in the social sciences. The behaviors covered by the term in the social sciences—as indeed according to common sense—form a set that only partially overlaps with altruism as defined in ethology: the gift, as described in anthropology, where it involves a system of obligation imposed on the recipient, does not directly—at least not with any certainty—“reduce the survival chances” of the altruist who gives it. Similarly, war or crime cannot be explained solely on the basis of aggression: as Gabriel Tarde pointed out, many criminal activities do not involve aggression, while some acts that do involve aggression are not criminal (Tarde 1890). 
64Current research into sociocultural phenomena in the animal world cannot, therefore, directly explain the origins, and even less the nature, of the set of phenomena to which the same concepts refer in humans. But it would also be wrong to say, as social scientists sometimes do, that ethological research into altruism tells us nothing about the constellation of behaviors described and grouped into these categories according to common sense or in the humanities. The real question when it comes to this sensitive topic is now, surely, not whether or not these studies fully explain human or animal sociability, but rather what exactly they say about it and where their limits lie. In other words, how can they relate to the knowledge acquired in the social sciences using methods specific to the social sciences (see, for example, Conein 2001)? For all these reasons, it is regrettable that the social sciences tend to largely ignore or outright dismiss ethological studies that focus solely on animal species, and in doing so abandon the lines of investigation raised by these points of contact to a crude naturalism. Moreover, by taking a close interest in this research, the social sciences could themselves play a prominent role in ethology, enriching it with their own ideas and methods. 
This research was conducted as part of and with the support of the ANR “LICORNES” (ANR-12-CULT-0002) and “ASCE” (ANR-13-PDOC-0004) programs.
Translator’s note: Unless otherwise stated, all translations of cited foreign language material in this article are our own.
Although some major works, such as those of George Herbert Mead or Erving Goffman, do contain references to ethology, overall the discipline is very rarely cited in the social sciences.
This was particularly the case in the 1970s and 1980s, with the controversy caused by sociobiology. We will return to this point in the final sections of this article.
In the last thirty years, a number of socio-anthropological studies have tried to rehabilitate the idea of animal agency by critiquing the idea of a sharp distinction between nature and culture, which is seen as a specifically Western ontology or perspective. In this view, this “Great Divide” structures every aspect of the West’s understanding (particularly its scientific understanding) of the beings that inhabit the world. As a result, animals are essentially treated as belonging to the same category as things, especially in the discipline that studies their behavior—ethology—which, therefore, tends to view animals as machines. On this school of thought in sociology and anthropology, and on the authors that inspired it (notably Latour 1991; Descola 2005; Ingold 2007), see the articles by Vanessa Manceron and Catherine Rémy in this issue; see also Guillo 2015.
The last four have constituted one of the discipline’s epistemological cornerstones since the work of Nikolaas Tinbergen (Tinbergen 1963).
Lorenz illustrated this point with his well-known theory of imprinting (Lorenz 1970 , 20).
In the same way, and with a similar theoretical model, Tinbergen demonstrated the existence of strictly ordered sequences of instinctive behaviors in the reproductive behavior of the stickleback (Tinbergen 1952).
On the “babymorphism” of such dogs, see the studies cited in Miklósi 2007.
We will return later on to this concept, which is extremely important in these models.
See also Maynard Smith 1964.
Since the first models, “altruistic” behavior has been defined in the ethological literature as “acts that are costly to one individual and directly benefit another individual” (Boesch 2012; Wilson 2000 ) (back-translated from the French). We will return later to the meaning of this concept and, more broadly, the meaning of the term “social” as used in the ethological literature.
Game theory would also in turn be profoundly enriched with new models and concepts—like that of evolutionarily stable strategies—thanks to this new field of application (Maynard Smith 1982).
There has also been significant debate regarding the conditions that enable the emergence and diffusion of “cooperation,” or, if one prefers, “altruism” (see especially Axelrod and Hamilton 1981).
See, for example, the “parental manipulation” model (Alexander and Sherman 1977). It attempts to explain intra-family sociality in certain species from an evolutionary point of view, based on the conflict of interest between parents and children.
Particularly in evolutionary psychology: see Tooby and Cosmides 1992.
Numerous studies of dogs in the last twenty years have shown that their ability to communicate with humans is often more developed and complex than that of monkeys. In particular, dogs seem to understand pointing much more easily and spontaneously than monkeys—they look in the direction indicated by the finger, rather than at the finger itself (Miklósi 2007).
Nevertheless, there is lively debate around this question, which is now seen as central in these disciplines. On the different perspectives regarding humans, see Enfield and Levinson 2006. See also Clément et al. 2011.
For a review of ethological research and findings on this subject, see the overview edited by Laland and Galef 2009.
On social learning in this strand of ethology, and the potential links that could be established with the social sciences, see the article by Nicolas Claidière and Dominique Guillo in this issue.
With, of course, a whole array of intermediary cases. On this point, see also the article by Nicolas Claidière and Dominique Guillo in this issue.
For the ethologists and psychologists involved in this debate, the term “imitation” itself covers a whole range of heterogeneous and complex phenomena among humans (see, for example, Shea 2009).
On the difference between these two mechanisms, see Hopper and Whiten 2012, 454.
For other mechanisms—either special cases or slightly different versions of those already mentioned—see Hopper and Whiten 2012.
For a review of debates on this question, see particularly Huang, Heyes, and Charman 2006, and Tomasello 1999.
There is, however, some debate—particularly in psychology—about how important theory of mind is for uniquely human characteristics. On this point see Enfield and Levinson 2006.
Such as certain ethologists who currently draw heavily on the cognitive sciences: on dogs, for example, see Miklósi 2007.
See especially Claidière and Sperber 2010.
On these points, see the review in Hopper and Whiten 2012, 466–7.
On this perspective, see for example, Kaminski, Pitsch, and Tomasello 2013, or the overview in Miklósi 2007.
A rejection of this ethology—seen as influenced by neo-Darwinism and “sociobiology”—is found even among social scientists who call for a form of interdisciplinary collaboration on the subject of animals. For a French example, see, for example, Digard 2012.
On these points see Guillo 2012.
On this new trend in socio-anthropology, see the other articles in this issue, especially those by Vanessa Manceron, Catherine Rémy, and Chloé Mondémé. For a critique of its attitude towards ethology, and more broadly towards the historical theory of the “grand partage” on which that attitude rests, see also Guillo 2015.
A recent handbook containing contributions by specialists on these subjects includes only a few definitions, always short and rarely discussed, like the following: “the basis of human sociality, like the sociality of any species, is cooperation” (Mesoudi and Jensen 2012, 421).
This argument is developed in more detail in Guillo 2012.
It is possible, for example, to refine the model of culture conventionally used in this research by combining it with arguments drawn from interactionism. The resulting model would also make it possible to account for communication, interaction, and cultures that combine individuals from different species, which ethology has scarcely contemplated to date. On this point, see the article by Nicolas Claidière and Dominique Guillo in this issue.