Collaboration and human social evolution: review of Michael Tomasello’s why we cooperate (MIT Press, 2009)
- First Online:
- Cite this article as:
- McLoone, B. Biol Philos (2012) 27: 137. doi:10.1007/s10539-010-9227-1
- 268 Views
Michael Tomasello’s new book Why We Cooperate explores the ontogeny and evolution of human altruism and human cooperation, paying particular attention to how such behaviors allow humans to create social institutions.
KeywordsAltruismCooperationStag hunt gameSocial institutionsCulture and cognition
Michael Tomasello’s brief, trenchant book Why We Cooperate (WWC) begins with the following observation: If we understand “culture” to refer to socially transmitted behavioral practices and artifacts, there are at least two aspects of human culture that are unique. The first is that it exhibits cumulative cultural evolution. Humans not only transmit behavioral practices and artifacts between generations, we also often modify these behaviors and artifacts and pass these modifications along, too. The second is that human culture involves social institutions—abstract phenomena like money, marriage, and governments—whose ontology is based upon, and constituted by, group-level agreement on particular norms, rules, and beliefs. This feature too, Tomasello claims, is unique to us (Tomasello 2009, x–xi).1
While Tomasello’s The Cultural Origins of Human Cognition (1999) dealt at length with cumulative cultural evolution and the effects it had on human cognitive evolution, WWC explores the ontogeny and evolution of human altruism, human collaboration, and social institutions. Indeed, to a rough approximation, WWC consists of three inter-connected theses. The first is that human altruism and collaboration are unlike anything we see among other primates. The second is that we can understand the evolution of humans’ unique pro-sociality in large part through reference to the Stag Hunt game. And the third is that we can understand human social institutions as extensions of collaborative activities that ancestral hominins engaged in, such as collective foraging and collective hunting.
WWC presents an incisive argument. Tomasello takes bits of research from a number of fields—child development, psychology, primatology, behavioral economics and philosophy—and presents a compelling perspective on human sociality. (The book also includes supplementary chapters by Carol Dweck, Joan Silk, Brian Skyrms, and Elizabeth Spelke.) Here, I’ll review the predominant arguments and themes in WWC, but will pay particular attention to what I take to be two central, underlying topics. The first concerns conceptual issues that arise when utilizing the Stag Hunt game in an analysis of human evolution. The second concerns the extent to which Tomasello’s arguments fit within a nativist approach to human cognition and pro-sociality generally.
Three varieties of altruism: helping, informing, and sharing
In the first chapter of WWC, Tomasello discusses the ontogeny of human altruism and the extent to which we see altruism in chimpanzees. He begins by identifying two mutually exclusive conceptions of human social behavior: On the one hand, we could believe, as Hobbes did, that humans are born selfish, and that we learn through the strong arm of society to behave and act communally. Or we could follow Rousseau: Humans are born pro-social, but society corrupts us. Tomasello says Rousseau’s conception is closer to the truth, so long as we recognize certain “critical complexities” (p. 3). Namely, while it appears human children are indiscriminately altruistic at around age 1—before we can meaningfully say society has “taught” them how to behave—as they grow older, this behavior becomes “mediated by such influences as their judgments of likely reciprocity and their concern for how others in the group judge them…” (p. 4).
It is worthwhile to dwell on how Tomasello uses the words altruism and mutualism, for they are the core topics of WWC. Following the more-or-less standard usage in evolutionary biology, Tomasello defines altruism as “one individual sacrificing in some way for another” (p. xvii). And though he does not use this terminology explicitly, he seems to adopt a “multi-level interpretation of altruism” (in the sense of Kerr et al. 2004), in which an altruist can decrease its fitness within a group, but enhance its fitness across groups, since altruistic groups might showcase greater reproductive output than less altruistic groups (see pp. 84–85, for instance). Alternatively, Tomasello uses the word mutualism to refer to intra-specific cooperation between two or more individuals in cases where joint coordination yields the highest return (p. 52). If two individuals team up to kill a stag, for instance, and no one individual could have killed the stag alone, this would be an instance of mutualism, according to Tomasello.
While the explicit focus of Chapter 1 is human altruism, Tomasello claims at the outset that it is actually a mistake to speak of altruism as though it were a single behavioral phenomenon. Rather, there are types of altruism. Specifically, a behavior qualifies as an instance of helping if it is an altruistic act that provides an instrumental service (say, opening a door); an act is an instance of informing if it involves the altruistic transmission of information (say, telling an otherwise ignorant person where an object is); and an act is an instance of sharing if it involves the altruistic transmission of a physical good. He makes these distinctions because, while many animals showcase altruistic behavior, how they are altruistic varies considerably. In the particular case of humans, for example, helping, informing, and sharing have different ontogenetic trajectories and rely upon different cognitive systems (pp. 5–6; Warneken and Tomasello 2009).
Does dividing altruism in this way make sense? It does, but since altruism is often treated as some single thing, we should focus on what Tomasello is actually doing here. A biologist can address theoretical questions about the evolution of altruism in what we might call a “substrate-neutral” way. For instance, she can concern herself with how altruism as a functional behavior evolved (or could evolve) in a given population, without paying attention to how the organisms in that population actually implement (or could implement) altruistic behavior physiologically (see, for instance, Sober and Wilson 1998).
But a biologist could also address questions about the evolution of altruism in a more fine-grained manner. She could explore how token acts of altruistic behavior are physiologically manifested in the organisms in that population, asking why the organisms evolved to manifest altruistic behavior in some way(s)—for instance, they always “share”—but not others—for instance, they never “inform.” It is in this latter sort of analysis that categorizing altruism into types seems warranted, because to do so is simply to categorize different sorts of behavior—albeit, behavior with the same supervening label. And as we’re about to see, dividing altruism into types allows Tomasello to tease apart important differences in human and chimpanzee altruistic behavior, differences that would otherwise go unnoticed.
With his three different types of altruism in the background, Tomasello argues that, while children are just slightly more helpful than chimpanzees, they are far more informative and sharing. As an example, Tomasello describes an experiment in which an experimenter would drop a small object in front of, variously, an 18-month-old child and a chimpanzee. In both cases, the experimenter would pretend that the object was just out of her reach. The 18-month-old children and the chimpanzees consistently would retrieve the object and bring it to the experimenter, even when no rewards were involved. But in a similar condition, in which an experimenter needed a door to a cabinet opened for her, so she could place some objects inside of it, the children consistently opened the door, while the chimpanzees did not (pp. 6–7).
There are in fact far more salient differences between humans’ and chimpanzees’ informative and sharing behavior. Tomasello recounts an experiment in which 12- and 18-month-old children were placed in a room with an experimenter, who handled a variety of small objects. When the experimenter would drop an object and pretend that she was unable to find it, 88% of 12-month-olds and 93% of 18-month-olds pointed to the object at least once to indicate where it was. Also, when the experimenter retrieved the object, the children would stop drawing attention to it—showing that the children did not want the objects for themselves (pp. 14–15; see also Liszkowski et al. 2006). In contrast, chimpanzees never point for each other in the wild, and when they point for humans (say, in a laboratory experiment like that above), it is typically “imperative”—for instance, to say “get me that food” (pp. 15–17). Furthermore, while children often share some resource they have access to, Tomasello claims that chimpanzees never do (pp. 24–25).
Given that there are these differences in human and chimpanzee altruism, how are we supposed to understand why these differences exist? Are children simply exposed to more altruistic behavior (say, by their parents), and so, as a result, they themselves manifest altruistic behavior? Tomasello wants to avoid this sort of constructivist account. He singles out helping in particular as a “naturally emerging human behavior,” and goes on to detail five reasons to believe enculturation could play only a minor role in explaining why we see helping behavior among children. On the one hand, it emerges far too early ontogenetically to be the result of learning. Also, children’s helping behavior seems to be unaffected by parental encouragement; indeed, if anything, it seems to be hindered by positive re-enforcement, according to Tomasello (pp. 8–10). Also, chimpanzees help, suggesting culture learning is at least not necessary for helping to emerge. Fourth, we see helping among children from cultures where adults don’t intervene as actively in development. And finally, Tomasello claims it is an extension of human empathy, which he takes to be naturally emerging, too (though he gives no justification for treating empathy as naturally emerging).
I am willing to grant Tomasello these points, for I don’t think we need to enter into an debate about “innate” versus “acquired” to accept the basic thrust of Tomasello’s argument. Namely, on a development spectrum, we can treat some traits as clearly the result of social learning (say, riding a bicycle), while some develop within a wide range of environmental contexts (say, developing ten fingers) and hence are not traits we “learn”; they just happen. Tomasello endorses the idea that human altruism is closer to the finger development side of the spectrum than the bicycle riding side, but he is not dogmatic about the topic. As we will see, though, these issues get more complicated for Tomasello in Chapters II and III, for it is unclear whether Tomasello wants to a endorse a similar nativist account of humans’ ability to engage with each other in token acts of mutualistic cooperation. But we will deal with that later.
While Tomasello claims altruism is naturally emerging, he claims that to whom we are altruistic, and in what contexts, becomes mediated by cultural norms as a child develops. Indeed, he says this must be the case, for humans consistently pursue behavioral strategies that deviate from rational decision making in economic games, while chimpanzees seldom do. For instance, consider an un-iterated Ultimatum game. Say some agent, Smith, offers another agent, Jones, some portion of a resource—for example, $2 from a $10 pot. Jones can either accept that offer, in which case he will get $2 and Smith will get $8, or reject that offer, in which case they will both get none. Often, humans will reject low offers, presumably because they treat it as a social affront. But, from the standpoint of rational choice, in a one-time game, an agent should accept any offer greater than $0. Chimpanzees, on the other hand, are shameless; they will pursue optimal behavior in this sort of game (pp. 32–33). Tomasello argues that the difference is that humans have adopted sets of cultural norms that are over-riding our phylogenetically older instinct to be fitness-maximizers.
Man the collaborator: stag hunts and the evolution of social institutions
In Chapter I of WWC, Tomasello was concerned with human altruism and the ways in which it differs from what we see among chimpanzees. We can say Tomasello pursues three topics between Chapters II and III. First, he discusses human collaboration as compared to chimpanzee collaboration; second, he puts forward a story about how humans’ collaborative abilities evolved; and third, he discusses the ways human collaborative activities underpin social institutions. While his thoughts on these three topics can stand alone, Tomasello attempts, rather successfully, to weave them into a coherent narrative. Here, I will march through what Tomasello says about them all.
“The participants have a joint goal in the sense that we (in mutual knowledge) do X together”
“The participants coordinate their roles—their plans and sub-plans of action, including helping the other in her role as needed—which are interdependent” (p. 61).
Tomasello argues that while a human child can engage in SCAs before her second birthday, chimpanzees never can.
How does Tomasello reach this conclusion? He focuses on one set of experiments in particular, in which 14–24-month-old children and mature chimpanzees were tasked to engage in social games and cooperative problem-solving activities. In one condition, children had to team up with an experimenter to access a prize that was encased in a tricky apparatus. One of the two had to push up a lift, which put the prize in reaching distance of the other. But the person who pushed up the lift could not also reach around to grab the prize, and the person who could grab the prize could not also push up the lift. Crucially, when the experimenter stopped participating, the children would often attempt to re-engage her in the activity. In a parallel experiment with chimpanzees, on the other hand, while the chimpanzees could engage in the shared task, they never attempted to re-engage the experimenter when she stopped participating. Indeed, they would sometimes attempt to perform their own sub-tasks with greater intensity, suggesting they simply could not conceptualize the structure of the collaborative activity (pp. 63–65; see also Tomasello et al. 2005; Warneken et al. 2006; Warneken and Tomasello 2007).
The driver initiates the hunt by slowly pushing the arboreal prey in a constant direction, blockers climb tress to prevent the prey from dispersing in different directions, the chasers, may climb under the prey and by rapidly running after them try a capture, and the ambusher may silently climb in front of the escape movement of the prey to block their flight and close a trap around the prey (Boesch 2005, p. 692, emphasis in original, see also Boesch 2001)
This would indeed seem to satisfy features (1) and (2) of an SCA. But Tomasello disagrees. He posits that the “vocabulary” of the above description is suspect, and in fact the hunt simply begins “when one male chimpanzee begins chasing a monkey through the trees, with the understanding that fellow chimpanzees, who are necessary for success, are in the area.” These other chimpanzees then take “the most opportune spatial position still available at any given moment in the emerging hunt,” such that “each participant is attempting to maximize its own chances of catching the prey, without any prior joint goal or plan or assignment of roles” (p. 62). But a flaw in Tomasello’s interpretation is that chimpanzees distribute meat from the dead collobus with respect to which chimpanzees had the most difficult hunting roles, suggesting they are indeed aware of the inter-connectedness of the collaboration (Boesch 2001).
To a large extent, whether chimpanzees can engage in SCAs is a side issue. No one denies that humans engage in SCAs earlier (ontogenetically), more skillfully, and more often. The main goal for Tomasello in Chapter II is to construct an evolutionary story that accounts for how humans came to possess this ability to collaborate with others. To do so, he uses the Stag Hunt game, which describes something like the following scenario: We might imagine that Smith and Jones are each able to hunt and capture two hares over the course of the day. But, Smith and Jones can also team up to hunt larger game, like stag. The important point is that a stag is not worth four hares (the sum of the two individual hare hunts); it is worth far more than that. Thus, assuming the stag is distributed somewhat fairly, both Smith and Jones stand to benefit from cooperation (Skyrms 2004). In short, Tomasello argues that humans evolved the ability to exploit this emergent behavioral potential of joint action. Specifically, he argues that this likely occurred in the context of hominins who were foraging and hunting together, writing that the “collaborative activities of the type we see today in young children are mostly representative of the earliest collaborative activities in human evolution” such as “the collaborative hunting of large game or the collaborative gathering of fruit” (pp. 74–75).
A more abstract analysis of the role of the Stag Hunt game in hominin evolutionary history might have the following form. There is a supervenient behavioral property of coordinated agents, such that in particular contexts—when large items can be hunted, for example—what a collective is able to accomplish can exceed the sum of what each individual within that collective could do, were these individuals to work alone. We see chimpanzees and humans converging toward Stag Hunt behavior because it is simply a structural feature of our world that the group-level behavioral potential of coordinated agents can exceed the sum of the individual-level behavioral-potential of each member within that group. Thus, if the yield of the collective activity is distributed in the right way, individual members of such a group stand to benefit from collaborative action. Humans have evolved an ability to exploit this general structural feature of our world.
Unfortunately, Tomasello spends no time discussing paleontological or archaeological data relating to the emergence of collective foraging and hunting, and this is a serious flaw in WWC—perhaps its biggest. Because of this, we are left with an abstract argument that the Stag Hunt game played an important role in hominin evolutionary history, but no real sense of when Stag Hunts began, or, for that matter, why the ecological context in which ancestral hominins found themselves selected for this ability to participate in SCAs, while the ecological context of ancestral chimpanzees did not.
To his credit, though, Tomasello does dwell on some of the salient theoretical issues that emerge when dealing with the evolution of cooperation in Stag Hunt games. With Stag Hunts, three issues in particular present themselves. First, who is to ensure that the stag will be distributed equitably after the hunt is done, such that it is worthwhile for everyone to collaborate? This is the problem of resource expropriation. In a collaborative experiment with chimpanzees, for instance, when the chimps were presented with a clumped resource—food in a single pile—the dominant would expropriate the prize, such that later collaboration between the chimpanzees was less effective (pp. 24–25; Melis et al. 2006). Second, how can Smith be sure Jones will follow through on his commitment to hunt stag instead of hare? After all, the Machiavellian maneuver for Jones might be to get Smith to commit to some joint project, but then to opt out at the last minute and leave Smith hanging. This is the issue of obligations and trust. And finally, supposing Smith and Jones really do cooperate, how will they be able to coordinate their behaviors effectively enough to accomplish their goal, such that cooperation is actually worthwhile? This is the issue of “generating benefit” (Calcott 2008).
Tomasello deals with all three of these issues by positing that hominins first had to evolve more basic pro-social attitudes before they could collaborate in Stag Hunts. These pro-social attitudes motivate individuals to collaborate in the first place, allow them to collaborate effectively when they do, and underpin the fair division of the stag after the hunt. As he puts it, “there had to be some initial emergence of tolerance and trust…to put a population of our ancestors in a position where selection for sophisticated collaborative skills was viable” (p. 77). He posits that we might imagine that this emerged because the more generous collective foragers were at an advantage in the context of other generous foragers, that groups excluded individuals who were non-egalitarian, or that cooperative childcare was important to establishing within-group altruism (pp. 83–84).
Regardless of the details, however, Tomasello argues that such an evolutionary process happened after we split with the species that led to chimpanzees, for chimpanzees are bad at resource distribution, on the one hand, and seem to have at best only rudimentary abilities to form joint commitments, on the other. Thus there must have been “some initial step in human evolution away from great apes, involving the emotional and motivational side of experience, that propelled humans in a new adaptive space in which complex skills and motivations for collaborative activities and shared intentionality could be selected” (p. 85). Without these pro-social attitudes as the “motivational and emotional” background to collaboration, Stag Hunts might very easily fall apart, given the game theoretic issues mentioned above.
From collaboration to social institutions
The third major piece to Chapters II and III is Tomasello’s attempt to relate the sorts of topics discussed above—namely, the ontogeny and evolution of human collaboration—to the (predominantly) philosophical work on social institutions and institutional reality. To start, Tomasello puts forward the following definition of social institutions: They are “sets of behavioral practices governed by various kinds of mutually recognized norms and rules” (p. xi). Let’s consider money, the paradigmatic example. Bits of paper have specific purchasing power—they count as token examples of money—because individuals believe they count as money, and they believe that others believe this, too. Without these mutual agreements about particular types of paper, money does not exist (p. xii; Searle 1995).
So an evolutionary account of social institutions will have to explain how we get from chimpanzee-like social behavior—which has no such institutions—to human social behavior, which is filled with them. Tomasello sets up this problem brilliantly by contrasting foraging to shopping. When humans (and apes) forage, they operate in an observer-independent world: trees, nuts, fruits, other foragers, and so on. But when we go shopping, we occupy not only an observer-independent world—the physical structure of the grocery store, the food within it, the check-out counter—but also an observer-dependent, institutional world, with items like “money” and “norms” and “store credit”. These institutional items do not have the same ontology as observer-independent items like the store itself and the food and check-out counter within it. If you wipe all of the humans off the planet, there will still be the store itself, the food, and the check-out counter. But there will not be any money, norms, or store credit. The existence of items in this second class depends upon the presence of minds to think of them (pp. 56–57; Searle 1995).2
While there does appear to be a non-trivial chasm between foraging and shopping, Tomasello argues that we can bridge it if we understand the “We-mode” within which humans can collaborate. A human has the ability to engage in a shared task in which she can conceptualize not only the shared task itself, but also her sub-task within it, others’ sub-tasks within it, and how these sub-tasks relate to each other. (Humans are not operating exclusively in the alleged “I-mode” of chimpanzees, who just go after some treat largely ignorant of any collective action.) The idea is that, if I team up with you, I can understand that my sub-task is X, yours is Y, and both X and Y are necessary to accomplish our joint task Z—for instance, downing a stag. But participation in the collaborative activity involves not only these expectations about a joint action plan, but also normative rules about the collaborative activity itself. For instance, that everyone will participate, that they will participate in a certain way, and that the yield will be dealt with in a particular manner afterward. This coupling of collaborative action and norms and rules is, at bottom, the basic structure of social institutions, Tomasello argues.
What is needed in addition to these elements to get from foraging to shopping is the ability to assign status functions through conventionalized symbols. We see this emerge in children as they engage in pretend play, where two children might decide, for example, that a spoon is in fact a magic wand, or that a banana is a gun. Such status functions, Tomasello writes, are “precursors ontogenetically and perhaps phylogenetically to collective agreement that this piece of paper is money, or that that person is president, with all of the rights and obligations those agreements entail” (p. 97). The basic structure is the same, what varies is the scope. In short, Tomasello’s claim is that the ability to conceptualize collaborative activity, to do so in a way that involves rules and obligations about behavior, and to do so while utilizing mutually agreed-upon status functions, are the core behavioral ingredients of social institutions. Breaking down social institutions into these components allows us to understand how such institutions evolved out of more basic collaboration among ancestral hominins.
What lies beneath?
Tomasello never details what role he believes social learning plays in a child’s development of the ability to collaborate with others. To be sure, we know he believes humans have a “naturally emerging” tendency to be altruistic, and that culture seems to only mediate such behavior. But it is unclear whether Tomasello wants to suggest the human mind has much of a “naturally emerging” flavor to it. Early on, he writes that “As they grow, children are equipped to participate in this cooperative groupthink through a special kind of cultural intelligence” (p. xvi), and he later adds that there are “psychological processes underlying” the uniqueness of human culture (p. 59). But these statements are of course just trivially true, and there is never any careful discussion of what these “psychological processes” amount to, either developmentally or cognitively.
I mention this, in particular, because Tomasello has been a sedulous critic of nativist approaches to human language learning (Tomasello 1999, 2004), and so I expected him to offer a model of the psychology of SCAs that stressed the role that social learning and explicit instruction play in the development of a child’s ability to engage in those SCAs. In short, I was expecting something like the constructivist, social learning model of cognition we see in Sterelny (2003, 2007, forthcoming). For instance: A child grows up in a highly engineered social, physical, and epistemic niche. She is exposed to collaborative action by simply being a member of the community, and she will learn to engage in SCAs through exposure to a variety of activities that necessitate such behavior, and through tolerant peers and elders who will correct and reinforce her behavior as needed. But Tomasello does not offer this sort of account, nor does he offer a more nativist alternative.
Whatever his stance in fact is, this is not an issue that should be omitted, for there is much evidence (some of it from Tomasello’s own work) that suggests children’s ability to engage and collaborate with others follows a predictable ontogenetic pathway (Brownell et al. 2006; Warneken et al. 2006; Warneken and Tomasello 2007), and there is some incomplete evidence that this ontogenetic pathway manifests itself in quite varied cultural contexts (Eckerman and Whitehead 1999). If Tomasello wants to endorse an exposure-based account of children’s abilities to collaborate, as he does with a child’s knowledge of language, he needs to account for (1) why children collaborate as early as they do, (2) why their abilities to collaborate follow such a regimented ontogenetic trajectory, perhaps cross-culturally, and (3) why mature chimpanzees struggle markedly with forms of collaboration that 2-year-old children can perform easily. Each of these points might be entirely compatible with a social learning model of collaboration, but we need to know how.
Also, Tomasello is inconsistent about what role he believes altruism has played in the evolution of collaboration. As we have seen, at times he (rightly) claims that altruism must come before collaboration, given the problems of resource expropriation, obligations and trust, and the generation of benefit mentioned above. He writes that “there had to be some initial emergence of tolerance and trust…to put a population of our ancestors in a position where selection for sophisticated collaborative skills was viable” (p. 77); later, he says that for “humans to have evolved complex skills and motivations for collaborative activities…there had to have been an initial step that broke us out the great-ape pattern of strong competition for food, lower tolerance for food sharing, and no food offering at all (p. 83); and so on. But elsewhere he says “I argue that mutualistic collaborative activities were the original source of human altruism…” (p. 47), that “mutualism might also be the birthplace of human altruism” (p. 53) or that collaborative activities “are the birthplace of human altruistic acts” (p. 99).
This is an unresolved tension in WWC. Tomasello, on the one hand, wants to acknowledge that cultural artifacts and institutional life are the core features of human social behavior, but he seems unwilling to acknowledge that—central those these phenomena are—it is not clear that they have much impact on the early stages of human behavioral or cognitive development. Rather, young children are altruistic largely regardless of social instruction, and their ability to engage in SCAs seems to emerge in much the same way that, say, their ability to perceive object-permanence emerges: quickly, effortlessly, and with no explicit instruction. Tomasello wants to highlight the centrality of social institutions in human life, but his argument is strained by the empirical work that shows social institutions appear to play only a minor role in instilling altruistic behavior in developing children and in teaching them how to collaborate with others.
From here on, page numbers without a preceding year will refer to pages in the 2009 hardcover version of Tomasello’s Why We Cooperate.
This is Searle’s (1995) distinction between ontologically objective and ontologically subjective parts of the world. It is worth noting that these ontological distinctions can only go so far, at least in the present context. Humans bring to the process of foraging an immense collection of acquired knowledge, along with a conceptual apparatus that categorizes the biological world. This knowledge and these categories are not “out there” in some observer-independent space.
This research has been supported by a grant from the National Academies Keck Futures Initiative (grant number: NAKFI HS11). I would also like to thank Patrick Forber for helpful and thorough comments.