Solving the Puzzle of Human Cooperation, Apuntes de Diseño

¡Descarga Solving the Puzzle of Human Cooperation y más Apuntes en PDF de Diseño solo en Docsity! Solving the Puzzle of Human Cooperation Robert Boyd Department of Anthropology University of California Los Angeles, CA 90095 rboyd@anthro.ucla.edu Peter J. Richerson Department of Environmental Science and Policy University of California Davis, CA 95616 pjricherson@ucdavis.edu Draft 1.2: Do not cite in any context without permission of authors. 1 No consensus on cooperation Is society an organic whole with each of its many components working together like the organs in a body? Like organisms, societies are composed of many parts which seem to work together enhance their survival. Different people fulfill different, necessary role—subsistence, reproduction, coordination, and defense. Regular exchange of matter and energy guarantees that each component has the resources it needs. Norms, laws and customs regulate virtually every aspect of social interaction, who may marry who, how disputes are resolved, and how verbs should be conjugated. Ritual and religion provide comfort to the sick and fearful, maintain a feeling of solidarity and belonging, and serve to preserve and transmit knowledge through time. Even the simplest human societies seem like complex machines designed for growth and survival. People have long been divided about whether this metaphor is useful or misleading. Many believe that the appearance of design is real. Functionalism, an old and still influential school in anthropology and sociology, holds that beliefs, behaviors and institutions exist because they promote the healthy functioning of social groups. (Spencer, 1891; Radcliffe-Brown, 1952; Malinowski, 1922; Aberle et al., 1950). Such functionalists believe that the existence of some observed behavior or institution is explained if it can be shown how the behavior or institution contributes to the health or welfare of the social group. The conviction that people are selfish drives others to argue that the appearance of design is an illusion—the complex structure merely reflects a standoff in a struggle among selfish individuals. Such rational individualists, mainly economists, political scientists and philosophers, hold that human choices must be explained in terms of individual benefits; any group benefits are an accidental side effect of selfish individual choices. This conflict remains unresolved because the competing protagonists espouse irreconcilable views about the causes of human action. Functionalists view people as being shaped by their society. People acquire a belief in the rightness in the norms and customs of their culture as a result of growing up in that society. Believers in the rational individualism see people as choosing how to behave based on their own interest. People are bound by custom only to the extent that it serves these interests. Let us now consider in more detail why this difference in belief about human nature leads to different views about the function of society. When functionalists do provide a mechanism for the generation or maintenance of group-level adaptations, it is usually in terms of selection among social groups (Turner and Maryanski, 1979). Rappaport (1984) provides an exceptionally clear statement of the idea: such conceptions as honor, morality, altruism, honesty, valor, righteousness, prestige, gods, heaven, and hell, [make] group selection important among humans. By group selection I mean the selection for and perpetuation of conventions enhancing the persistence of groups, even though these conventions 4 example, the societies of highland New Guinea. Here, patrilineally organized groups number from a few hundred to several thousand. These groups have religious, political, and economic specialists, they engage in trade and elaborate ritual exchange with distant groups, and they are able to regularly organize parties numbering several hundred to make war on their neighbors. Even contemporary hunter-gathers who are limited to the least productive parts of the globe have extensive exchange networks and regularly share food and other important goods outside the family. Other animals do none of these things. Thus we have an evolutionary puzzle. Our Miocene primate ancestors presumably cooperated only in small groups mainly made up of relatives like contemporary non- human primates. Such social behavior was consistent with our understanding of how natural selection shapes behavior. Over the next 5 to 10 million years something happened that caused humans to cooperate in large groups. The puzzle is: What caused this radical divergence from the behavior of other social mammals? Did some unusual evolutionary circumstance cause humans to be less selfish than other creatures? Or, do humans have some unique feature that allows them to better organize complex cooperation among selfish nepotists. Solutions to the puzzle. People have proposed five different kinds of solutions to this puzzle: 1. The “heart on your sleeve” hypothesis holds that humans are cooperative because they can truthfully signal cooperative intentions. 2. “Big mistake” hypotheses propose that contemporary human cooperation results from psychological predispositions that were adaptive when humans live in small groups of relatives. 3. Manipulation hypotheses hold that people either tricked or coerced into cooperating in the interests of others. 4. Moralistic reciprocity hypotheses hold that greater human cognitive abilities and human language allow humans to manage larger networks of reciprocity which account for the extent of human cooperation. 5. Cultural group selection hypotheses argue that the importance of culture in determining human behavior causes selection among groups to be more important for humans than for other animals. These five are not mutually exclusive, and, in fact we believe that the most likely explanation is some combination of the last two hypotheses. The Heart on Your Sleeve Hypothesis In his book, Passions within Reason, the economist Robert Frank (1988) argues that humans cooperate with non-relatives because people can reliably detect the true 5 intentions of others. Frank thinks that people have innate dispositions that cause them to be more or less cooperative and that they can signal their disposition to others by their appearance and demeanor. Truthful signals allow cooperators to preferentially interact with other cooperators and this makes possible the evolution of cooperative behavior among non-relatives. Because thinking and feeling are complex physiological processes, Frank argues, it is more costly for a defector to maintain the appearance of a nice guy than it is for a cooperator. The theory of signaling developed in economics and borrowed by biologists (see Gintis 2000 for a discussion) shows that, with this assumption, natural selection may favor honest signals of intent. Thus, Frank's idea is that we wear our hearts, be they good or bad, on our sleeves. This argument cannot explain why humans are more cooperative than other animals because it applies with equal force to other animals. If humans must wear their hearts on their sleeves, then why not chimps or baboons? True, humans seem smarter than other creatures (although this may mainly be a matter of perspective), but this fact cuts both ways. Being clever may allow humans to better penetrate deceptions, but it may allow them to better perpetrate deceptions as well. In fact, we would use the evidence from other animals to argue against the importance of Frank's mechanism among humans. If the evidence about intentions that one could get by the mechanisms that Frank invokes was as reliable as the information conveyed by kinship, then we should see much cooperation among non-relatives in nature. Since we see very little cooperation among non-relatives, we conclude that signaling provides much poorer information than kinship. Perhaps the reason is that the costs of a defector deceptively signaling that they are altruists is really not very high. If deceptive signalers of altruism can easily reap most of the benefits provided by true altruists without paying the costs of providing altruism, the signal will become useless. The Big Mistake Hypothesis Many people have argued that contemporary cooperation results from dispositions that evolved when humans lived in small groups of close relatives (e.g. Alexander 1974, 1987, Hamilton 1975, Tooby and Cosmides 1989). The genus Homo is about two million years old, and the species Homo sapiens at least 90,000 years old. Until the spread of agriculture beginning about 11,500 years ago humans probably lived in relatively small groups (although how small is a matter of debate). In such a world, it is argued, selection could favor psychological mechanisms that led to unselective altruism towards fellow group members because all the potential recipients were relatives. With the advent of agriculture and later urban life, the size of human groups increased dramatically, and these same predispositions led to altruism toward non- relatives. For example, most people feel sympathy for others. Few can fail to be moved by a photograph of a starving Somali child, its eyes rimmed with flies. Such feelings cause many people to act, to send money, to organize relief, or to choose a less than lucrative career in international medicine or development. Sympathy of this kind is genetically maladaptive in the contemporary world because it leads indiscriminate altruism, but it would have been fine, the argument runs, when all of the potential recipients were relatives. We call this the “big mistake hypothesis” because almost everything in modern life, trade, religion, government, and science is a mistake from the genes' point of view. 6 The big mistake hypothesis can be extended to explain inter-group enmity (van den Berghe 1981). In many species of non-human primates, kinship ties within groups are stronger than kinship ties among groups and, as a result, neighboring groups compete for territory or other resources. In such an environment, selection might favor a generalized enmity towards non-group members. Today neighboring groups of hunter are typically physically indistinguishable, but often have different dialects, customs, and styles of manufacturing items like arrow points. Thus, it is argued, selection could have favored the rule: “Be nice to people who talk like you, dress like you, and act like you. Be nasty to everyone else.” The big mistake hypothesis is a cogent story, but we think the evidence from the study of non-human primates tells against it. Many contemporary non-human primates live in small kin based groups very much like those posited for early humans. However, life in such groups does not evidence much non-selective altruism. In all primate species, the members of at least one sex leave their natal group and join another group where their only relatives will be their own children. In most species, it is males who leave, but there are some species such as chimpanzees in which females emigrate. Members of the other sex, usually females, remain in their natal group and live among female relatives. However, when groups are of any size, some of these relatives will be quite close while others will be quite distant. A great deal of evidence suggests that primates are very sensitive to differences in relatedness, directing costly helping behavior mainly toward relatives. Consider baboons for example. A typical baboon group might consist of about 60 animals, about half of them infants and juveniles, 4 or 5 adult male immigrants, and the rest adult females born in the group. Violent conflict among females occurs hourly, and these conflicts sometimes result in serious injuries. Many conflicts involve coalitions, usually made up of matrilineal kin. As a result, female baboons are organized into a dominance hierarchy in which all members of a matriline occupy adjacent positions. Violent conflict among males is also a common occurrence, but coalitions are much rarer (even absent in the chacma baboon), and more ephemeral. When the dominant male confronts a subordinate, the subordinate will often grab an infant to use as a hostage to defend itself. When a dominant male is displaced by a new immigrant, the new dominant will sometimes kill infants sired in the group before he arrived. Such infanticides increase the new dominant's reproductive success because lactation suppresses ovulation, and killing infants causes their mothers to resume ovulation. Other species would tell a similar story—living primates are very good at discriminating between relatives and non-relatives and behave very differently toward each. It is hard to see why early hominids should have been less discriminating in their behavior. Evidence from human behavior also tells against the big mistake hypothesis. Humans do generally know who their their real relatives are. No doubt propensities to cooperate with kin are deeply ingrained in human psychology. One excellent body of evidence comes from the seemingly tangential literature on incest avoidance. Westermarck (1894) suggested that there is an innate avoidance of inbreeding. If so, humans must have an innate kin recognition system. The operation of this device is nicely illustrated by the rarity of marriage among Israeli kibbutz age-mates (Shepher 1983) and the poor success of Taiwanese minor marriages (Wolf 1970). In these 9 that makes manipulation possible is reciprocity, and in its simplest formulation, people, in essence, manipulate themselves in their own group interest. Moralistic Reciprocity Hypotheses A number of authors have suggested that human cooperation is based on reciprocity (e.g. Trivers 1971, Wilson 1975, Alexander 1987, Binmore 1994), and that our more sophisticated mental skills allow us to manage larger social networks than other creatures. Two kinds of evidence support this hypothesis. First, reciprocity clearly does play an important role in contemporary human societies all over the world. Second, some measures of brain size are correlated with social complexity—animal species which have small social networks tend to have smaller brains (corrected for body size) than do animal species with large social networks (Dunbar, 1992). The fact that humans have very large brains for their body size suggests that humans can maintain reciprocal relationships in larger groups than other animals. Field and laboratory experiments suggest that monkeys are much smarter about social problems than non-social problems. For example, vervet monkeys do not seem to know that python tracks (which are obvious and unmistakable) predict the presence of pythons, but they do know that their aggression toward another vervet predicts aggression by that individual's relatives toward them (Cheney and Seyfarth 1990) which suggests that solving social problems is an important for brain evolution. The defining feature of reciprocity is that ongoing interactions allow people to monitor each other's behavior and thereby reward cooperators and punish noncooperators. Beyond this property, there is little agreement among biologists or anthropologists about the details of how reciprocity works. In the simplest models punishment takes the form of withdrawal of further cooperation (for example, Axelrod and Hamilton 1980): I will keep helping you as long as you keep helping me, but if you cheat, I won't help you any more. We will refer to such strategies as “simple reciprocity”. Other authors (e.g. Binmore 1994) argue that punishment takes other forms—non-cooperators are punished by various forms of social ostracism, reduced status, fewer friends, and fewer mating opportunities. Following Trivers (1971) we will call this “moralistic reciprocity.” While these different types of reciprocity are often lumped together, they have very different evolutionary properties. It is very unlikely that large scale human cooperation is supported by simple reciprocity. There is strong theoretical support for the idea that lengthy interactions between pairs of individuals are likely to lead to the evolution of this kind of reciprocating strategy (See Axelrod and Dion 1989, Nowak and Sigmund 1993 for review), but recent work suggests that simple reciprocity cannot support cooperation in larger groups (Boyd and Richerson 1988, 1989). Increasing group size places simple reciprocating strategies on the horns of a dilemma. Strategies which tolerate a substantial number of defectors in the group allow defectors to go unpunished and therefore cannot persist when common because such defectors get the benefits of long term cooperation without paying the cost. Thus, reciprocators must be provoked to defect by the presence of even a few defectors. However, such intolerant strategies cannot increase when rare unless there is a substantial chance that the groups made up mainly of cooperators will form when cooperators are rare and they are extremely sensitive to the existence of errors or uncertainty. This dilemma is not serious when 10 pairs of individuals interact; very minor perturbations allow reciprocating strategies to increase when rare. As groups become larger, however, both of these requirements become impossible to satisfy. This conclusion makes intuitive sense. We know from everyday experience that reciprocity plays an important role in friendship, marriage, and other dyadic relationships. We will stop inviting friends over to dinner if they never reciprocate, we become annoyed at our spouse if he does not take his turn watching the children, and refuse to return to the auto repair shop when they do a bad job. However, it is not plausible that each one of a thousand union members stay out on strike because they are afraid that their defection will break the strike. Nor does each member of a Mae Enga war party maintain his position in the line of battle because he fears that his desertion will precipitate wholesale retreat. Moralistic reciprocity provides a much more plausible mechanism for the maintenance of large scale cooperation. Reciprocators can punish non-cooperators in many ways besides withholding their own cooperation.. Strike breakers can be physically attacked or their property can be vandalized. Even more plausibly they can be socially ostracized—scabs lose status in their community and with it many important benefits of social life. Much the same goes for cowards and deserters who may be attacked by their erstwhile compatriots and shunned by their society, made the targets of gossip, or denied access to territories or mates. Moralistic reciprocity enforced by such of punishment is more effective in supporting large scale cooperation than simple reciprocity for two reasons: When a simple reciprocator stops cooperating in order to punish defectors, he or she induces other reciprocators to stop cooperating. These defections induce still more defections. Innocent cooperators are in effect punished as much as guilty defectors when the only recourse to defection is to stop cooperating. In contrast, other forms of retribution can be targeted so that only defectors are affected. This means that defectors can be penalized without generating a cascade of defection. Second, with simple reciprocity, the severity of the sanction is limited by an individual's effect on the whole group, which becomes diluted as group size increases. Moralistic sanctions can be much more costly to defectors, and therefore, allow rare cooperators to induce others to cooperate in large groups. However, there is also a problem with moralistic reciprocity that remains to be explained: why should individuals punish? Remember the problem with the manipulation hypothesis: If punishing is costly and the benefits of cooperation flow to the group as a whole, administering punishment is a costly group beneficial act, and therefore, selfish individuals will cooperate but do not punish. The elder Maasai who punishes a younger man, suffers a cost to himself and provides a benefit to his fellow oldsters. Similarly, the striker who attacks a scab may be injured himself. The striker who shuns a scab may forgo a satisfying friendship, a beneficial business relationship, or even a desirable marriage partner. Thus, as long as the effect of the punishment administered by a single individual will have little effect on the success of the strike, selfish individuals will not punish. 11 This problem is solved if moralistic reciprocators also punish people who do not punish when they should. This means that moralistic strategies punish non- cooperators, individuals who do not punish non-cooperators, and individuals who do not punish non-punishers. When such strategies are common, rare non-cooperators suffer because they are punished. Individuals who cooperate but do not punish suffer against because they are also punished when they don’t punish. In this way, it can pay to punish, even though the cooperation that results is not sufficient to compensate individual punishers for the cost of punishing. The Maasai gray beard who arranges his son to marry will lose status in the community, the striker who fails to shun the scab will herself be shunned, and the Enga warrior who fails to participate in punishing a coward may become a “garbage man” himself. If moralists are common, the cost of punishing rare noncooperators may be small, and cost of being punished may be large, so that even quite costly group beneficial behaviors can be maintained by this mechanism. (See Boyd and Richerson 1992 for a more formal version of this argument. There is also a closely analogous result in economics. See Hirshleifer and Rasmusen 1989, or Binmore 1994.) There is also a big problem with the moralistic reciprocity hypothesis: It explains how costly group beneficial behavior can persist, but it provides no explanation for why group oriented behavior is more common than any arbitrary behavior. Moralistic punishment can stabilize any arbitrary behavior—wearing a tie, being kind to animals, or eating the brains of dead relatives. It does not matter whether or not the behavior produces group benefits. All that matters is that when moralistic punishers are common it is more costly to be punished than to perform the sanctioned behavior, whatever it might be. When any behavior can persist at a stable equilibrium, then the fact that reciprocity is a stable equilibrium does not tell us whether it is a likely outcome or not. What we need to know is: Does the fact that reciprocity leads to beneficial outcomes for everyone make it more likely that moralistic punishment will support reciprocity than, say, eating the brains of dead relatives. If the answer to this question is no, then moralistic reciprocity is not a complete answer to the puzzle of human cooperation. To answer this question we need to know how a population changes through time. Our approach is to model behavioral change as a process of cultural evolution by biased transmission. People differ in their belief and values about what behaviors should be subject to moralistic punishment. Then depending on the composition of the population, some beliefs will yield higher payoffs than others, and therefore people with those beliefs will be imitated, and they will spread. This process will then go on until the population reaches a stable equilibrium. We have analyzed a model of the evolution of moralistic punishment that incorporates these ideas, and it suggests that the existence of a group benefit does not increase the ability of a strategy to spread when it is rare (Boyd and Richerson, 1992). Cultural evolution by biased transmission is equally likely to reach an equilibrium at which people are punished for not eating the brains of dead relatives as an equilibrium at which they are punished for not defending their group. One answer to this objection (Binmore 1994) is that people don't have to slowly evolve toward an beneficial state through a myopic process of cultural evolution. They can deliberately choose what behaviors should be punished either by getting 14 to know, under what conditions will group selection cause the group beneficial behavior to come to dominate the population? The answer is: Hardly ever. The model we have just sketched is closely analogous to “interdemic group selection” models studied by population geneticists. The only real difference is that biased transmission rather than natural selection leads to the spread of selfish behavior within subpopulations. Extensive analysis of such models by population geneticists indicates that populations will eventually become composed almost completely of noncooperators unless subpopulations are extremely small, there is little mixing among groups, and it is very hard for individuals to discern the costs and benefits of alternative beliefs. Selection among groups fails because biased transmission and migration combine to create a powerful force reducing variation among groups. Migration among groups insures that noncooperators are always present in every group, and biased transmission causes their frequency to increase. In contrast, only force that generates differences among groups—random fluctuations due to small population size and small numbers of individuals colonizing empty habitats—is weak unless groups are very small. Thus there will be little variation among groups even if group selection is very strong, and as group selection can have little effect. The existence of cultural transmission is, by itself, no guarantee that there will be enough variation among groups to allow cooperation will evolve. The impatient anthropologist might say: “Who cares. We anthropologists know that there is heritable variation among societies, so the group selection will work. If sociobiologists and economists and like-minded folks need an account based on individual decisions, let them worry about it. We should get on with the business of determining the function of cultural practices.” Obviously we think that it is important to reconcile anthropological and evolutionary explanations, but many anthropologists do not. Even so, it is a serious mistake to ignore the problem of the maintenance of variation for cooperation between societies. All societies have internal conflict and the balance between cooperation and selfishness varies considerable from one society to the next. The basic empirical facts point to a dynamic balance between selfish and prosocial motives and much complex structure regarding what behavior is appropriate toward co-members of different groups (e.g., co-members of one’s family, extended family, ethnic group, religion, and nation). Humans are not unconditional cooperators so we cannot assume that any given social institution is group function even if many do have such elements. The cases are quite evidently diverse and the devil is in the details of the evolutionary processes in operation (for a nice comparative study see Knauft, 1993). In the remainder of this chapter we will outline two mechanisms that maintain variation in costly-group beneficial behavior. Each mechanism is consistent with human behavior having been shaped by natural selection. However, each mechanism also has empirical requirements that need to be tested with anthropological data, each mechanism makes additional predictions, and each mechanism points to interesting new questions about the logic of the group selection process. 15 How group selection can work. For group selection to be an important process that can generate adaptations at the group level there must be some mechanism that can maintain variation among groups, and not just for some behaviors, but for any arbitrary behaviors. We think that there are at least two such mechanisms: Moralistic reciprocity and frequency dependent bias. Lets see how they work. Variation is maintained by moralistic reciprocity. Group selection can be an important process generating group beneficial behavior when group beneficial behavior is enforced by moralistic punishment. Moralistic punishment allows wide variety of individually costly memes to persist once they become common, and as a result moralistic punishment provides a mechanism for preserving variation among groups, variation that then can be subjected to selection at the group level. By itself, moralistic punishment does not provide and explanation human cooperation, but the combination of moralistic punishment and group selection does. To show how moralistic reciprocity allows group selection to work, we modify the group selection model described above: Once again there is a population subdivided into a fixed number of groups linked by migration and there are two memes. However, now each meme specifies a different set of behaviors that are to be enforced by moralistic punishment. One specifies that a person's primary loyalty is to his or her kin and people who violate this norm are punished; the second, specifies a person's primary loyalty is to his or her group, and, again, violators are punished. We will call these the “nepotistic” and the “nationalistic” memes, respectively. All other things being equal, people are biased in favor of nepotism, but when the nationalistic group loyalty meme is sufficiently common the effects of punishment overcome this bias and people tend to choose the group loyalty meme. Until recently everybody in the population was nepotistic, but then most people in one group adopted the nationalistic meme. We will return the question of why this should occur, but for the moment we simply regard it as a random event, the group level analog of a mutation. The mutant group has a competitive advantage because nationalistic beliefs allow the organization of larger, and more inclusive corporate activities, and as a result it is more likely to survive as a group, and more likely to grow in size. When a group goes extinct, its territory is occupied by others. The question is: Under what circumstances will this group level advantage cause nationalistic beliefs common in the mutant group to spread throughout the population? First, biased transmission must be strong enough to keep the nationalism meme common in the mutant group. To see why, suppose this meme is rarely expressed, and rarely punished so it is difficult for immigrants too learn that nepotism will get them in trouble, and therefore biased transmission is weak. As a result, only a few immigrants will have switched to nationalism by the time that more immigrants arrive. This reduces the penalty for being nepotistic, and thus even fewer of these immigrants switch during the next time period, and eventually nepotism becomes advantageous, and the initially nationalistic group becomes nepotistic like the population as a whole. Unless biased transmission is strong enough to resist the 16 homogenizing effects of migration, nationalism cannot spread by group selection because it cannot persist long enough for selection among groups to occur. In contrast, suppose people with nepotistic beliefs are frequently punished in the nationalistic group so that so that new immigrants whose beliefs differ from the majority rapidly learn that such beliefs get them in trouble and adopt the prevailing norm. Then when yet more nepotistic immigrants arrive they find themselves to be in the minority, quickly learn the local norms, and so maintain the nationalistic meme, that originally arose merely by chance, at high frequency long enough for group selection to have some effect. Second, group reproduction must preserve cultural differences. Suppose that one of groups in which nepotistic memes predominate goes extinct. If their territory is occupied by individuals drawn from all other groups, the majority of the colonists will be nepotistic, and as a result the new group will be nepotistic. Even if the group in which the nationalistic meme is common never goes extinction, nationalism will not spread because the nationalistic group cannot reproduce itself. In contrast, suppose that new groups are formed by the fissioning of existing groups. Then, most of the time, when a nepotistic group fails it will be replaced by a new nepotistic group, but occasionally it will be nationalistic group that fissions, and when this occurs the nationalistic meme spreads. If nationalistic groups are more likely to survive or more likely to grow and fission the result will be that nationalism will spread. The fact that this mechanism only works when biased transmission is strong explains why this it is likely to be more important for cultural evolution than for genetic evolution, and, therefore, provides a potential explanation for why humans are more cooperative than other animals. Evolutionary biologists normally think of selection as being weak, and although there are many exceptions to this rule, it is a useful generalization. So, for example, if one genotype had a 5% selection advantage over the alternative genotype this would be thought to be strong selection. So, suppose that a novel, group beneficial genotype has arisen, and that it has become common in one local group where it has a 5% advantage over the genotype that predominates in the population as a whole. For group selection to be important, the novel type must remain common long enough to spread by group selection which is possible only if the migration rate per generation is less than 5% (see Boyd and Richerson, 1990 for details.) But this is not very much migration. The migration rate between neighboring primate groups is on the order of 50% per generation. [Find a survey on migration rates] While migration rates are notoriously difficult to measure, it seems likely that migration rates are typically high among small local groups that suffer frequent extinction . Migration rates between larger, deme sized groups is much lower, but so to will the extinction rate. If one assumes that migration rates are of the same order of magnitude for cultural evolution, then it seems likely that biased transmission can easily maintain intergroup variation. Variation is maintained by conformist social learning Frequency dependent bias can also maintain variation among groups. We have shown (Boyd and Richerson 1985, Henrich and Boyd 1999) that natural selection can favor a psychological propensity to imitate the common type. When it is difficult, but not impossible, to determine what is the best way to behave in local environment, you 19 the choice may nonetheless have profound effect on people's behavior, including behavior driven by rational choice. Is the model empirically plausible? Knowing that cultural group selection is logically possible is useful. Some authors have claimed that social and cultural institutions cannot be understood as benefiting the group because this is inconsistent with a theory built up from the actions of selfish individuals. Given a showing of logical possibility the interesting question is the empirical one.Have human cultures actually been shaped by cultural group selection? The models suggest that cultural group selection can lead to the evolution of group beneficial traits if: 1. group disruption and dispersal is common, 2. new groups are usually formed by the fissioning of existing groups, and 3. there is variation among groups, and that this variation affects the ability of a group to survive and produce daughter groups. If these conditions held true for substantial stretches of human history then it is quite plausible that during such periods that human cultures were shaped by cultural group selection. To address these questions we (in collaboration with our colleague, Joseph Soltis) have collected the relevant data from the ethnographic literature of Irian Jaya and Papua New Guinea. Until recently, most people in New Guinea lived in small, unstratified societies numbering a few hundred to a few thousand people. Such societies are characteristic of more of human history than the complex, stratified societies that dominate the contemporary world. The highlands of New Guinea were not explored by Europeans until the 1930's and not subjugated by Europeans until after the second World War. As a result, we have high quality ethnographic data describing peoples whose societies had not been drastically altered by a colonial administration. We read as many ethnographies as possible looking for those with accounts of group extinction, new group formation, and variation among local groups (Soltis et al. 1995) The data from New Guinea indicate that group level extinctions are quite common. You need three numbers to estimate extinction rates: The number of groups in the population, the number of extinctions, and the time period during which the extinctions occurred. We were able to find these data for 5 cultural areas in New Guinea, and all 5 reveal high rates of group extinction. The percent of groups estimated to suffer extinction in each generation (25 years) ranges from about 8% among the Maring to about 31% among the Fore (See the table below). Moreover, more than half of the twenty-eight ethnographies surveyed mention the occurrence of 20 at least one extinction, and it is likely that this underestimates the actual number because the failure to mention an extinction does not necessarily mean extinctions never occurred. Region Number of groups Number of extinctions Number of years % groups extinct every 25 years source Mae Enga 14 5 50 17.9% Meggitt, 1977 Maring 13 1 25 7.7% Vayda, 1971 Mendi 9 3 50 16.6% Ryan, 1959 Fore/Usurufa 8--24 1 10 31.2%--10.4% Berndt, 1962 Tor 26 4 40 9.6% Oosterwal, 1961 Second, the data indicate that new groups are formed by the fissioning of existing groups. Detailed accounts for two cultures, the Mae Enga and the Mendi, report that new clans are formed when sub-clans grow to the size that allows them to take on the social functions usually performed by clans. Anecdotal evidence from other ethnographies indicates that the same general processes operate in other parts of New Guinea and Irian Jaya as well. We could not find any account in which colonists of new land are drawn from multiple groups, so the group reproduction process would favor the preservation of between-group variation. Finally, the data also support the existence of cultural variation among local groups, although there is little evidence connecting this variation to local group survival. We have had difficulty finding good ethnographic descriptions of cultural variation among local groups. But for three regions, there is documented local variation. The Ok communities in the fringe highlands vary in terms of ritual and social organization. Tor tribes vary in language and kinship terminology. Abelam groups differentially accepted and rejected non-traditional religions. Furthermore, Barth (1971) argued that group selection has occurred among Ok communities. Ritual variation produced differing levels of centralized organization, and the more centralized groups were able to spread at the expense of the less centralized. At the very least, we could not falsify the hypothesis that cultural variation exists among local groups within larger cultural regions. More generally, if ethnographers are accurate reporters, consequential socio- political differences between different small-scale societies are ubiquitous. (Your experimental data not yet citable?) Rates of change The New Guinea data on extinction rates allow us to estimate the maximum rate of cultural change that can result from cultural group selection. For a given group extinction rate, the rate of cultural change depends on the fraction of group extinctions that are the result of heritable cultural differences among groups. If most extinctions are due to non-heritable environmental differences (e.g. some groups have poor land) or bad luck (e.g. some groups are decimated by natural disasters), then group selection will lead to relatively slow change. If most extinctions are due to heritable differences (e.g. some groups have a more effective system of resolving internal disputes), then cultural change will be relatively rapid. The rate of cultural change will also depend 21 on the number of different, independent cultural characteristics affecting group extinction rates. The more different attributes, the more slowly will any single attribute respond to selection among groups. By assuming that all extinctions result from a single heritable cultural difference between groups, we can calculate the maximum rate of cultural change. Such an estimate suggests that group selection is unlikely to lead to significant cultural change in less than about 1000 years. The length of time it takes a rare cultural attribute to replace a common cultural attribute is one useful measure of the rate of cultural change. Suppose that initially a favorable trait is common in a fraction q0 of the groups in a region. Then the number of generations t necessary for it to become common in a fraction qt of the groups can be estimated using a simple formula derived in Soltis et al. (1995) The time necessary for different parameters is given in following table. If one takes the extinction rate of the Mae Enga as typical, these results suggest that group selection could cause the replacement of one cultural variant by a second, more favorable variant in no less than about 20 generations, or 500 years. Given that not all extinctions result from heritable cultural differences, 1000 years is a reasonable upper bound for the rate of evolution of a single character with a strong influence on group survival. Extinction rateinitial fraction favorable trait final fraction favorable trait 1.6% 10.4% 17.9% 31% 0.1 0.9 192 40.0 22.3 11.8 0.01 .99 570 83.7 46.6 24.8 This rate of adaptation is not nearly fast enough to explain the many cases of cultural change known to have occurred on much shorter time scales. For example, the introduction of the horse to the Great Plains of North America in the 1500's led to the evolution of the culture complex of the Plains Indians in less than 300 years. If the rates of group extinction estimated for New Guinea are representative of small scale societies, cultures like those of the Great Plains cannot be explained in group- functional terms. There has not been enough time for group selection to drive a single cultural attribute to fixation, even if that attribute had a strong effect on group survival. This result also suggest that group selection cannot directly justify the practice of interpreting many different aspects of a culture as group beneficial. While both empirical and theoretical uncertainties prevent us from making a quantitative estimate of the rate of evolution of many different cultural attributes, it is clear that group selection will shape many traits more slowly than a single trait. If group selection can cause the substitution of a single trait on 1000 year time scales, the rate for many traits will be substantially longer. We know from linguistic and archaeological evidence that related cultural groups, such as the Nuer and the Dinka, which differ in many cultural attributes often have diverged from a common ancestral group in the last few thousand years. Thus, there has not been enough time for group selection to produce the many group beneficial attributes that distinguish one culture from another. The result does provide justification for interpreting some aspects of contemporary cultures in terms of their benefit to the group. The model demonstrates that under the 24 Boyd, R. and P.J. Richerson, 1988. The Evolution Reciprocity in Sizable Groups. Journal of Theoretical Biology, 132: 337!356. Boyd, R. and P.J. Richerson, 1989. The Evolution of Indirect Reciprocity. Social Networks, 11:213!236. Boyd, R. and P. J. Richerson, 1990.Group Selection Among Alternative Evolutionarily Stable Strategies. Journal of Theoretical Biology, 145: 331! 342. Boyd, R. and P. J. Richerson. 1992. Punishment allows the evolution of cooperation (or anything else) in sizable groups. Ethology and Sociobiology, 13: 171!195. Becker, G. 1981. A treatise on the family, Cambridge, Mass. : Harvard University Press, Binmore, K. G. 1994. Game theory and the social contract, Cambridge, Mass: MIT Press. Cavalli-Sforza, L. L. and M. W. Feldman 1981. Cultural Transmission and Evolution. Princeton: Princeton University Press Cheney D. L. & R. M. Seyfarth, 1990. How monkeys see the world : inside the mind of another species, Chicago: University of Chicago Press. Coleman, J. 1990. Foundations of Social Theory. Cambridge, MA, Havard University Press. Dunbar, R.I.M. 1992. Neocortical size as a constraint on group size in primates. Journal of Human Evolution. 22: 469–93. Frank, R. 1988. Passions within Reason New York: Norton Hallpike, C.R.1986. The Principles of Social Evolution. Oxford: Clarendon Press. Hamilton, W.D. 1964. Genetical evolution of social behavior I, II. Journal of theoretical Biology, 7: 1–52. Hamilton, W. D. 1975. Innate social aptitudes in man: An approach from evolutionary genetics, in Biosocial anthropology. Edited by R. Fox, pp. 133–55. London: Malaby. Harpending, H. Rogers, A., & Draper, P. 1987. Human Sociobiology, Yearbook Of Physical Anthropology, 30: 127-150 Henrich, J. and R. Boyd. 1998.The Evolution of Conformist Transmission and the Emergence of Between-Group Differences. Evolution and Human Behavior 19: 215–242. 25 Hirshleifer, R. and Rasmusen, E. 1989, Cooperation in a Repeated Prisoner’s Dilemma with Ostracism, Journal of Economic Behavior and Organization, 12: 87–106. Kelly, Raymond C. 1985. The Nuer conquest : the structure and development of an expansionist system, Ann Arbor : University of Michigan Press,. Klein, R. G. 1999.The human career : human biological and cultural origins, 2nd ed. Chicago: University of Chicago Press. Knauft, B.M. 1993. South Coast New Guinea Cultures: History, Comparison, Dialectic. Cambridge: Cambridge University Press. Lande, R.1985. Expected time for random genetic drift of a population between stable phenotypic states. Proceedings of the National Academy of Sciences (U.S.A.) 82: 7641-7645. Malinowski, B. 1922. Argonauts of the western Pacific. Prospect Heights, IL. Waveland Press (Reissue 1984). Nowak, M. & Sigmund, K. 1993. A strategy of win-stay, lose-shift that outperforms tit-for-tat in the prisoner's dilemma game. Nature.364: 56-58. Putnam, R.D. 1993. Making democracy word: Civic traditions in modern Italy. Princeton: Princeton University Press. Radcliffe-Brown, A.R. 1952. Structure and function in primitive society. London: Cohen & West. Rappaport, R. A. 1984. Pigs for the Ancestors.Yale University Press. Richerson, P.J. and R. Boyd. 2000. The biology of commitment to groups: the tribal social instincts hypothesis, in The biology of Commitment, edited by R.M. Nesse. Russell Sage Foundation. In press. Sahlins, M. 1976. Culture and practical reason, Chicago: University of Chicago Press Sepher, J. 1983. Incest, the biosocial view. New York: Academic Press. Sober, E., and D.S. Wilson. 1998. Unto Others: The Evolution and Psychology of Unselfish Behavior. Cambridge: Harvard University Press. Soltis, J., R. Boyd, and P.J. Richerson, 1995. Can group functional behaviors evolve by cultural group selection? An empirical test. Current Anthropology. 36: 473–494. Spencer, H. 1891. Essays: Scientific, Political, and Speculative (3 volumes). London: Williams & Norgate. 26 Strassman, B. 1991. The Function of Menstrual Taboos among the Dogon Defense against Cuckoldry? Human Nature, 2: 89–131. Thornhill, N. W. 1990. Evolutionary significance of incest rules, Ethology and Sociobiology, 11: 113-129. Tooby, J. & Cosmides, L.. 1989. Evolutionary psychology and the generation of culture, part I: theoretical considerations, Ethology and Sociobiology, 10: 29- 49. Trivers, R.L. 1971. The evolution of reciprocal altruism. Quarterly Review of Biology, 46: 35–57. Turner, J. H. & A. Maryanski. 1979. Functionalism. Menlo Park CA., Benjamin/Cummings. van den Berghe, P.L. 1981. The Ethnic Phenomenon. New York: Elsevier. Westermarck, Edward. 1894. The History of Human Marriage. London: Macmillan. Wilson, E. O. 1975. Sociobiology, Cambridge MA, Harvard University Press. Wolf, Arthur P. 1970. “Childhood Association and Sexual Attraction: A Further Test of the Westermarck Hypothesis. American Anthropologist 72(3): 503-515. Young, H. P. 1998. Individual strategy and social structure : an evolutionary theory of institutions. Princeton, N.J. : Princeton University Press,.