The Evolution of Human Ultra-sociality
Peter J. Richerson
Division of Environmental Studies
University of California, Davis
Davis, California 95616
Department of Anthropology
University of California, Los Angeles
Los Angeles, California 90024
Version 2.01 June, 1997. In press: I. Eibl-Eibisfeldt and F. Salter, eds. Ideology, Warfare, and
Indoctrinability. Please do not cite without authors’ permission.
1.1 Human sociality in comparative perspective
E.O. Wilson (1975) described humans as one of the four pinnacles of social evolution. The
other pinnacles are the colonial invertebrates, the social insects, and the non-human mammals.
Wilson separated human sociality from that of the rest of the mammals because, with the
exception of the social insect like Naked Mole Rats, only humans have generated societies of a
grade of complexity that approaches that of the social insects and colonial invertebrates. In the
last few millennia, human societies have even begun to exceed, in numbers of individuals and
degree of complexity, the societies of ants, termites, and corals.
Human social complexity is based on quite different principles than the ultra-sociality of
any other species. In all other known cases, the constituent individuals of societies are either
genetically identical, as in the colonial invertebrates, or closely related, as in the social insects and
non-human mammals. In the prototypical social insect colony, elaborate societies are built around
a single reproductive queen, who suppresses the reproduction of workers in the colony. The
offspring of the queen are siblings that share many genes in common, and they make up the work
force of the colony. W.D. Hamilton (1964) argued in his famous pair of papers that worker
sterility and relatedness to each other and the queen are fundamental to insect sociality. It is
interesting to note (Wilson and Hölldobler 1988) that the phylogenetic rarity of eusociality in
insects is in sharp contrast to its ecological ubiquity. Ants and termites are among the most
abundant insects, and social bees and wasps are by no means rare. As with other social animals
the range and adaptive diversity of the human species has increased, and our population densities
have exploded, as our societies have grown more sophisticated. Thus, the cooperation,
coordination, and division of labor that characterize ultra-social animals appear to be highly
adaptive in many contexts. At the same time, evolutionary processes can rarely overcome the
conflicts between individual reproductive success and the cooperation and self-sacrifice required
to produce the colony-level benefits that ultra-sociality produces.
Humans are, arguably, a new page in the natural history of animal cooperation (Campbell
1983). Our reproductive biology is similar to the other social mammals. Among our close
relatives, the apes and monkeys, genetic relatedness and reciprocal altruism support a diverse
array of small-scale societies, but no other spectacular ones. Humans have built extremely
complex societies by some mechanism or mechanisms different than any other know highly social
species. At the same time, there are remarkable parallels between human and ape social behavior
(deWaal 1982) and material culture (McGrew 1992), not to mention many convergences between
humans and other social and tool-using species. Consistent with classical comparative anatomy
and modern molecular studies, human behavior is clearly recently derived from ape behavior.
There is room for only relatively few modifications of the behavior of the last common ancestor of
chimpanzees and humans.
In this chapter, we draw together the results of some of our own theoretical work on the
role of the processes of cultural evolution in the evolution of human sociality, the contributions of
human sociobiologists, the general comparative data on the form of human societies, and some
critical experimental data from psychology to develop a hypothesis about how human ultra-
sociality might have arisen by the Darwinian coevolution of culture and genes. We believe that
evidence from several independent theoretical and empirical domains support the hypothesis.
1.2 The argument in outline
Our basic claim in this chapter is that cultural evolutionary processes drove the evolution
of human ultra-sociality. The cultural transmission of ordinary adaptive information has
advantages in the highly variable environments of the Pleistocene, and the hominine line diverged
from the last common ape ancestor in developing a massive dependence on adaptive social and
technological traditions. Some of these processes have the effect of making group selection on
cultural variation possible and the use of cultural cues to structure populations common. As
cultural group selection began to produce primitive patterns of ingroup cooperation and outgroup
hostility, human cognitive capacities and emotional responses, presumably coded in large measure
by genes, responded to adapt people to living in culturally defined cooperative groups.
Cultural transmission is a little like the haplodiploidy of the social insects. It tweaks the
evolutionary process in a way that happens to make the evolution of ultra-sociality possible. It is
quite unlike haplodiploidy in that it leaves the genetic system of inheritance itself unaltered. Thus
human ultra-sociality arose by adding a cultural system of inheritance to a genetic one that
normally supports small-scale societies based on kinship and reciprocity.
Our account is a modernization of Darwin's (1874) argument in Chapters III-V of the
Descent, where he outlines his theory of the evolution of the psychological basis human ultra-
sociality. His account turns on a mixture of what we would call today reciprocity based on
foresight of consequences, aided by group selection (see also Alexander 1977, 93 ff.). For the
usual reasons, genetic group selection is likely to be weak in humans (but see Wilson and Sober
1994, and commentators). Group selection on cultural variation is a different story.
2.0 Kin Selection, Reciprocity, and Punishment In Human Societies
Most evolutionary biologists believe that all known cases of non-human animal
cooperation can be explained by kinship and reciprocity. Both are certainly important in human
Kin selection is fundamental to the evolution of cooperation in many animal societies. Not
only is the eusociality of the social insects built upon kinship, but so are the much more modest
social systems of the non-human primates. In the case of our closest relatives, the chimpanzee, the
core of the social system is sets of related males (Goodall 1986). A population composed of
relatively small, relatively exogamous kin groups with limited reproductive suppression of
subordinate members of societies is a rather typical pattern for social mammals.
There is little doubt that cooperation with kin is 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 famous examples, potential husbands
and wives are raised in close companionship as children, much as siblings normally are, even
though unrelated in fact. Co-resident agemates apparently have an innate algorithm that invokes
an mating avoidance mechanism. The kin recognition system fails in the rather unusual
circumstances discovered by Shepher and Wolf, but in normal families will function properly as an
incest avoidance mechanism.
Daly and Wilson (1988) explore a useful set of data on patterns of homicide. Family
members, living in close proximity, have the maximum opportunity to kill one another, and family
homicides are a large fraction of the total. Nevertheless, there is a striking tendency for
consanguineal relatives to refrain from killing each other, compared to affinal family members. For
example, child abuses and child murders are very disproportionately committed by step-parents.
The homicide data, like that from incest avoidance, strongly suggest that humans have deep-
seated psychological mechanisms for kin recognition and for kin appropriate behavior, as
predicted by Hamilton's theory of inclusive fitness. One famous example of human altruism argued
to contradict the theory, Polynesian adoption practices, turns out to strongly support it (Silk,
1980). Kin altruism no doubt has deep roots in the vertebrate order, and human examples of
nepotistic behavior can be multiplied at will.
Trivers (1971) described a mechanism whereby long-continued interactions between
partners might lead to the development of sufficient trust in reciprocation of acts of aid to make
such acts routine. Axelrod and Hamilton (1981) analyzed the case of pairwise interactions of
players of prisoner's dilemma in their important contribution to the theory of reciprocity. In their
model, the interaction has a given probability of continuing for another turn. In this situation, it is
easy for strategies like "tit-for tat," ("cooperate on the first round of the game, and then do
whatever your partners did on the previous round of the game") to evolve. If two players of this
strategy meet, they cooperate from the first. If tit-for-tat players play against an unconditional
non-cooperator, they receive the worst possible sucker's payoff on the first round of the game, but
are only victimized once. If payoffs to cooperation are relatively high and games go on for many
rounds, then selection can prevent unconditional defectors from invading a population of tit-for-
Most important for our purposes here, Alexander (1987) has proposed that Axelrod’s and
Hamilton’s result can be extended to very large societies by what he calls indirect reciprocity. If
one of us helps you, you may help some third person who in turn helps the other of us. Rather
than being restricted to pair-wise interactions, perhaps reciprocity can encourage cooperation
among large, diffuse networks of reciprocators. Primatologists (Byrne and Witten 1988; Dunbar
1992; Kummer et al., in press) provide some support for this idea in the form of what is called the
“Social Intelligence Hypothesis.” Perhaps primate encephalization is generally an adaptation to
managing more complex societies, and humans are merely the most extreme case. Dunbar’s data
suggest that brain size is correlated with group size in non-human primates. Intuitively, the trouble
with supposing that intelligence alone can increase cooperation is that smart individuals can
compute selfish advantage and learn sophisticated deceptive tactics just as readily as they can
compute their enlightened self interest via cooperation. Intelligence may be necessary to manage
human cooperation, but is it sufficient to produce it?
We have studied mathematical models of the effects of group size on the evolution of
strategies like tit-for tat. In one series of models, we considered unstructured groups in which all
individuals in the group simultaneously played repeated turns of prisoner’s dilemma (Boyd and
Richerson 1988).We generalized the idea of tit-for-tat to large groups by studying rules of the
form "cooperate if k out of the other n individuals cooperated on the last iteration of the game."
The tit-for-tat results do not generalize to large groups. As group size increases, it rapidly gets
very hard to get reciprocity to increase when rare in a population dominated by unconditional
defection. This makes sense. In a situation where interacting groups are of size ten and your rule
is cooperate if 5 others do, it is highly improbable the five others with the same rule will be in any
given group when rare. Alexander’s idea involves indirect reciprocity flows along fairly stable
networks (what goes around eventually comes around). Adding a ring structure to groups to
organize reciprocity in larger groups helps, but not much (Boyd and Richerson 1989).
It is often suggested that punishment of rule-breakers plays an important role in the
maintenance of cooperation. The defection of tit-for-tat strategists when their partners defect is a
kind of punishment, but in large groups it is very inefficient since it leads to the breakdown of
cooperation for cooperators as well as defectors. It should be much more efficacious to punish
defectors by discriminating retribution and go on cooperating with other cooperative individuals.
We used the repeated prisoner dilemma system to model how punishment might evolve (Boyd and
Richerson 1992). Each round of the game is divided into two parts. In the first part individuals in
a group of size n either cooperate or defect. In the second stage of the round some strategies will,
at some cost, punish those who defected in the first part of the round. The results of this
theoretical investigation suggest that a division of labor can arise in which a few punishers can
induce a larger number of reluctant cooperators to cooperate. At equilibrium, under a range of
conditions, a group tends to have one punisher, just the number needed to induce cooperation in
most groups. This result is reminiscent of the famous bigman political systems of New Guinea,
and common in tribal societies elsewhere. We need not imagine that “punishers” are brutal thugs
dictating cooperation by physical coercion. “Bigmen” may be, as Eibl-Eibesfeldt remarked at the
conference, prosocial leaders who pay the costs of persuading and cajoling to induce cooperation
from others, with perhaps the distant threat of physical harm.
Punishment suggests the possibility of the evolution of “moralistic” strategies in which
punishers punish reluctant cooperators not only for defecting themselves, but also if they do not
punish others who fail to cooperate and even those who fail to punish those that don’t punish.
Such strategies can certainly induce cooperation, but they can also make any arbitrary behavior an
ESS that happens to become the target of a moralistic strategy.
This effect is in turn only an extreme example of the way games of coordination arise in
models of cooperation. In games of coordination, it is best to do whatever everyone else is doing.
You may hate daylight savings time, but it is quite inconvenient to arrange life around standard
time when everyone else is using daylight savings. As models of cooperation become more
complex, there are typically many ESSs, introducing a coordination effect. Moralistic strategies
can potentially make even absurd things, like wearing neckties to work, evolutionarily stabile.
Cultural evolution is perhaps rapid enough to evolve to different coordination equilibria in
different local populations, leading to group selection among societies with different equilibria. It
is likely that genetically determined variation in coordination equilibria exists only at the species
level, making interdemic group selection on the basis of different coordination equilibria unlikely.
There is no doubt that humans are capable of using rules about appropriate behavior in
partners as a basis for social decisions and actions. Cosmides (1989) conducted a series of
experiments to test the hypothesis that at least one of the kinds of problems that people evolved
to solve are those that involve social contracts. She supposed that human cognition includes the
rule "if you take the benefit, then you must pay the cost." Of many simple rules that humans might
use in social or more general decision-making this one is uniquely suited to detecting defection in
games like repeated prisoners dilemma. She then tested subjects ability to solve a logical problem
when couched as a social contract compared to their ability to solve it couched as some simple
non-social problem. For example the rule "if a man eats cassava, he has a tattoo on his face" can
be framed as a social rule (only men who have undergone the painful ritual of tattooing can eat
cassava) or as a mere contingent empirical fact (tribes that eat cassava also happen to tattoo).
Framed as a cheater detection problem success rates were in excess of 70%, while less than 25%
solved the problem in its empirical guise. Cosmides interprets these results to mean that people
have an innate decision-making module devoted to detecting cheaters on social contracts.
Whatever one might think of the innate interpretation of the experiment, it would certainly appear
that humans are very efficient decision-makers when it comes to detecting defectors on a social
In addition to recognizing cheaters in social games, people seem psychologically prepared
to exact retribution. Daly and Wilson (1988) report high rates of spousal homicide in some
modern communities. Suspicion of infidelity and sexual rivalry are very common motives for
murder. Knauft (1987) notes that in small-scale communities lacking any formal authority, murder
is quite common. In the absence of a satisfactory rule of law (which poor people can’t afford, and
some simple societies don’t have) individuals are certainly capable of costly acts of punishment.
Of course, the ultimate threat of physical coercion backs up formal law.
The capacity of people to behave as reluctant cooperators is strikingly demonstrated in
experimental settings such as those constructed by Milgram (1974) and Nuttin and Beckers
(1975). In one of Nuttin and Beckers’ experiments, Belgian college students (male) were asked by
Beckers (an attractive young woman experimenter) to give a speech in favor of a hated exam
system, to which they were subjected. A strenuous protest movement against the system was
supported by virtually all students. In the most extreme experimental condition used, all 11
students assigned to experimental treatment requesting students to give a television speech
favoring the exam system complied with Beckers’ request for cooperation. 22 fellow students
were asked how many students would comply with such a request, the majority thought less than
5%, and the most cynical guessed 30%!
3.0 Kin Selection, Reciprocity and Punishment Are Not Enough
There is no doubt that kinship, reciprocity and punishment--family, friends and righteous
anger--are important components of human social life. There is little doubt that they account for
the eusociality of insects and the small-scale cooperation of non-human primates. However, it is
hard to imagine how to make a case for these processes, unaided, creating human ultra-sociality.
The comparative evidence strongly suggests that the conventional mechanisms aren’t
sufficient. If kin selection, reciprocity, and punishment were sufficient to create ultra-sociality in
local populations with open mating and without reproductive suppression of workers, then ultra-
sociality ought to be common in our relatives, given its ecological success in our case. Unusual
adaptations suggest unusual environments or unusually evolutionary processes.
The common theoretical problem that afflicts all three processes is that they all inherently
favor small groups, absent special mechanisms like sterile worker castes. Kin altruism should
favor the most closely related individuals differentially. Sibships and even first or second cousin-
ships are composed of relatively few people. All else equal, it is much easier to organize
reciprocal cooperation in smaller groups than in larger ones. Even if cooperation in large units
gets started somehow, the relative ease of organizing nepotistic cooperation and small cabals of
reciprocators will tend to subvert larger scale organizations. It would seem that socially intelligent
individuals might organize many pairwise reciprocal arrangements, but not engage in much
Punishment by a leader has natural limits in the ability of an individual to monitor and
coerce reluctant cooperators. Also, the cooperation produced by leaders creates a public good for
all participants. If leaders can control group size in their own interest, they are not likely to select
the group size that maximizes overall benefits. If the marginal cost of having more than a few
individuals in the group exceeds the benefits, leaders will keep groups small. In fact, in the classic
iterated prisoner’s dilemma model to which we added punishment, the per capita benefits do not
increase as a function of group size, and potential leaders would prefer to be loners and have to
coerce no one. The asocial heroes that Clint Eastwood often plays in films illustrate this result.
His characters can coerce cooperation from whole towns, but they only do so when circumstances
make exit impossible. At the other extreme, it is easy to imagine situations in which punishment is
cheap, and in which punishers can reward themselves disproportionately from the fruits of
cooperation. Leaders would then be motivated to assemble a large number of exploitable
retainers. In this situation, coerced individuals will exit if they possibly can. What is lacking in the
punishment systems, and in related moralistic strategies, is any obvious internal mechanism to
regulate them in the group interest, even approximately. The comparative evidence (Eibl-
Eibesfeldt 1989: 297 ff.) is not supportive of the idea that coercion can generate ultra-sociality.
Coercion and deference to coercion are very widespread in animal societies in the form of
dominance hierarchies and similar principles of social organization. Dominant animals sometimes
act as prosocial leaders. Human dominance behavior shows many continuities with animal
dominance, and if dominance behavior alone could build large societies, they ought to be common
in other social animals.
Our hypothesis, then, is that these conventional mechanisms for the evolution of
cooperation in humans act quite strongly, but only to favor the same small-scale cooperation that
they do in other mammals. They are a major source of conflict in human societies (Campbell,
1983). Process of cultural group selection build large scale ultra-social institutions which very
frequently compete with kinship and friendship for the loyalties of people. This conflict has no
easy resolution because the culture-driven institutions and psychological mechanisms that support
ultra-sociality have been added in on top of a mating system, family system, and face to face
reciprocity system that derives directly from life in small primate troops, as zoologists have noted
Micro-sociological data seem to support the conflict hypothesis. Predatory crime, fraud,
embezzlement, anti-trust violations and the like, when they are cooperative at all, are typically
organized on a small scale. Pairs of robbers may engage in their enterprises with some confidence
that they have solved the prisoners dilemma problem.. Larger scale criminal enterprises, like the
Mafia and the Colombian drug cartels, resemble legal ultra-social institutions. They use principles
of recruitment like ethnicity and ritualistic resocialization to extend their scale of organization far
beyond what a single kin group or band of close friends can accomplish. Olson (1982) gives
evidence that the postwar economic success of the countries defeated in W.W.II can be explained
as the result of the temporary absence of small-scale organizations exploiting large-scale
institutions. In Germany and Japan, the pre-war totalitarian regimes destroyed or co-opted many
small-scale social organizations. Then, denazification and demilitarization destroyed the
organizations created by the totalitarian regimes. The constitutions imposed on the two countries
by the Allies reconstructed national scale organizations, and for some period of time these
operated free of the influence of a multitude of small-scale interest groups. The economies of the
Allied victors grew more slowly because "special interest" organizations remained intact,
demanding expensive and inefficient subsidies and other aid at the expense of the nation as a
4.0 Genetic Group Selection Implausible
Several prominent modern Darwinians [W.D. Hamilton (1975), E.O. Wilson (1975: 561-
2), R.D. Alexander (1987: 169), and Eibl-Eibesfeldt (1982)] have given serious consideration to
group selection as a force in the special case of human ultra-sociality. They are impressed, as we
are, by the organization of human populations into units which engage in sustained, lethal combat
with other groups, not to mention other forms of cooperation. The trouble with a straightforward
group selection hypothesis is our mating system. We do not build up concentrations of intra-
demic relatedness like social insects, and few demic boundaries are without considerable
intermarriage. Moreover, the details of human combat are more lethal to the hypothesis of genetic
group selection than to the human participants. For some of the most violent groups among
simple societies, wife capture is one of the main motives for raids on neighbors. In precontact
New Guinea, warfare typically results in the social breakup of a defeated group, and the dispersal
of its members to other groups where the defeated have bonds of friendship and kinship (Soltis et
al. 1995). Pastoralists in Eurasia generated complex intermarrying confederations across ethnic
lines to the extent that it takes considerable detective work to be reasonably certain that groups
like Attila’s Huns were mainly a Turkic group (Maenchen-Helfen 1973). The human groups that
compete are demographically very open, and violent conflict increases migration rates. It is hard
to see how genetic differences can be maintained in the face of such migration rates.
5.0 Cultural Evolutionary Processes and Ultra-sociality
Are there cultural evolutionary processes that can support the evolution of norms and
sentiments that permit people to live in large cooperative groups, even in the teeth of small-scale
loyalties based on kinship and reciprocity? Humans are unique in the scale of their use of learning
from others as a mode of adaptation. Cultural transmission is like genetic transmission in many
respects, although the differences are just as important (Cavalli-Sforza and Feldman 1981; Boyd
and Richerson 1985). We are also unique among vertebrates in the scale of our social
cooperation, the degree to which social groups are marked by symbolic means, and apparently
unique among all animals in the mode by which we construct them. Thus, it is worth pursuing the
hypothesis that the association of culture and cooperation is not accidental. We think that there is
good evidence that cultural evolutionary processes did play a role loosely analogous to
haplodiploidy and reproductive suppression of workers in engendering human ultra-social
5.2 Cooperation and cultural group selection
5.2.1 Theoretical models
Intuitively, group selection on culturally transmitted variation is more plausible than in the
genetic case because of differences in the way culture is transmitted. One such difference is that
humans may survey several individuals in the course of acquiring a cultural trait. We can have
more than two cultural “parents.” In variable environments with migration between
subpopulations, there are liable to be many erroneous ideas introduced into the population by
outsiders whose behavior is appropriate for a different environment. We can expect that the
adaptive forces--even natural selection--to act on culture to keep advantageous traits in high
frequency so long as the amount of migration is not overwhelming. Theoretical models show that
using a conformist transmission rule of the form “when in Rome, do as the Romans do” is
advantageous in this circumstance. Individuals benefit by using the information about the
commonness or rarity of a cultural trait to bias their adoption choice. This transmission rule in
turn has the effect of reducing variation within populations and protecting between-population
difference. Even with fairly substantial migration rates and relatively weak conformity, the
differences between groups is maintained in the models (Boyd and Richerson 1985: Ch. 7).
Cultural group selection meets the test of theoretical plausibility, a test that most evolutionary
biologists believe genetic group selection flunks.
In addition to between-group differences stabilized by conformity, games of coordination,
the coordination element in many games of cooperation, and moralistic punishment all lead to
multiple evolutionarily stable equilibria. If cultural evolution is generally more rapid than genetic
evolution, and there is reason to assume that this is one of the basic adaptive advantages of
learning by imitation (Boyd and Richerson 1985: Ch. 4), then cultural metapopulations may find
more alternative equilibria than metapopulations without culture, and there will be more variation
at the group level on which selection might play.
5.2.2 Empirical data
The cultural group selection hypothesis predicts that social units that are co-socialized
should be the groups that cooperate. Some evidence supports this prediction. Otterbein’s (1968)
cross-cultural study of warfare revealed that societies engaged in internal warfare had small-scale
fraternal interest groups composed of related males as the dominant element of social
organization. Societies that suppressed internal warfare and conducted mostly external warfare
emphasize socialization of warriors into age sets, police societies, or other military institutions
that cross-cut units of male relatedness. When the cultural group that is co-socialized is restricted
to genetic kinsmen, we recover, approximately, as expected, the pattern predicted by ordinary kin
selection theory. When a larger group is the focus of socialization for military virtues, larger scale
fighting units emerge.
Elsewhere we have reviewed the psychological evidence that humans actually use
conformist rules in cultural transmission. Despite a large literature on conformity, more work is
needed on the extent to which we actually use it in cultural acquisition. (Boyd and Richerson
It is also possible to test the empirical plausibility of the cultural group selection process
with a direct microevolutionary analysis. Soltis et al., (1995) used information on group
extinctions by warfare in Highland New Guinea to provide an estimate of the plausibility and
power of group selection. The New Guinea data provide relatively extensive and sophisticated
ethnographic accounts for a series of simple tribal societies before contact with colonial powers.
Although horticultural rather than hunting and gathering societies, the sorts of processes operative
in Highland New Guinea are roughly similar to those that might have operated on simple societies
generally over the long period of time during which human ultra-sociality arose.
To work effectively, cultural group selection by the conformity mechanism requires a
number of conditions to be effective. For example, new groups must be formed mainly by the
splitting of old groups, not by drawing random samples of migrants to a new group. The former
mode of group multiplication preserves variation between groups, whereas the latter destroys it.
Group extinctions due to chronic warfare between village communities are quite common
in New Guinea. For the five populations for which quantitative estimates could be derived from
the literature, rates ranged from 1.6% to 31.3% per generation. Taking into consideration the
quality of the data used in the estimates, the long-term, area wide rate was probably between 10%
and 20%. The New Guinea data strongly support the model’s assumed group reproduction
processes. New groups always arise by the fissioning of existing groups, usually large, successful
ones. There is ample evidence for cultural differences between groups, although there is little
evidence in Highland New Guinea ethnography documenting differences that persist for long
periods of time or that are directly connected to extinctions. Other sets of data do strongly
suggest that group differences in competitive ability are common (e.g. Barth 1981; Knauft 1985,
Taking the New Guinea rates at face value, our calculations suggest that cultural group
selection can cause an initially rare, favored institution to become common in about a millennium.
This rate is fast on the time scales of genetic macroevolution, and is roughly consistent with rates
of evolution of human political institutions under pre-state conditions. The archaeological and
linguistic evidence from Western North America, where hunting and gathering subsistence is the
rule, suggests that major events causing population displacements did indeed have approximately
millennial time scales (Bettinger and Baumhoff 1982). If this process began working sometime
back in the Pleistocene, then there will have been some tens of millennia, perhaps as many as a
few hundred, for this process to have affected human social cognition by genetic coadaptations to
life under group selected cultural adaptations.
We know from simpler cases that cultural evolution can drive a genetic response in cases
like the evolution of adult lactose absorption in dairying populations on the millennial time scale
(Durham 1991: Ch. 5). If a process like cultural group selection worked over substantial periods
of time, and especially if it recruited coercive strategies to effectively punish defectors from social
contracts, it is reasonable to expect that human decision-making strategies should be adapted to
life in a prosocial world.
Batson’s (1991) experimental studies of the psychology of human altruism provide good
evidence that our cognitions and emotions have been molded by some form of group selection.
His studies address the structure of motivations that produce helping behavior using laboratory
experiments. The experiments were designed to distinguish between an empathy-altruism
hypothesis and various forms of covertly egoistic motives for helping behavior. According to the
empathy-altruism hypothesis, once a potential helper’s empathy for a sufferer is engaged, helping
behavior is motivated, at least in measurable part, by a genuinely unselfish desire to actually
relieve the victim’s suffering. Rational actor theorists propose that instead of a desire to help for
its own sake, apparent altruists give aid in the expectation of some form of personal reward.
Egoistic hypotheses are numerous. Individuals may help another in need because they seek to gain
rewards or avoid punishment. These expected rewards and punishments may be external or
internal. That is, people may expect others to give rewards or administer punishments, or it may
be a matter of conscience, self-administered rewards and punishments. Expected external rewards
or punishments may be material (soldier’s pensions) or social (enhanced prestige). Of course,
actual human motivations may be complex, and egoistic motivations of several kinds may jointly
act with altruistic ones. Batson’s question is whether any significant non-egoistic motivation can
The experiments elicited an empathetic response on the part of some experimental
subjects, for example by asking them to take the point of view of the experiment’s victim, while
non-empathetic controls were asked to view the situation objectively. Then, the experimental
conditions were manipulated to control for one or another egoistic motivation and test whether
subjects in the empathy condition still willing to provide aid to the experiments victim. For
example, one egoistic motivation for aiding a victim might be to relieve one’s own suffering at
having to watch the victim suffer. To test this possibility, “Elaine,” the sham victim in one
experiment, was to suffer a series of moderately painful shocks. The experimental subjects were
told at the beginning of the sham experiment that Elaine is unusually sensitive to shocks due to a
traumatic childhood experience, and finds them exceedingly uncomfortable. The experimenter
expresses concern about this, and offers the real subjects the chance to continue the “experiment”
in place of Elaine. The shocks will be uncomfortable for them, but not nearly so painful as for
Elaine. If helping is motivated by the desire to avoid viewing someone else suffering, Batson
reasoned that allowing the experimental subjects to escape (watch only two trials of the
experiment on Elaine instead of all ten) should reduce the tendency to help, whereas if subjects
had a genuine desire to help Elaine, even subjects allowed escape should offer to help. In this
experiment, difficult escape had a dramatic effect on helping in the low empathy case, raising the
proportion helping from about 0.2 to 0.6. In the high empathy condition there was no significant
difference in helping; in the easy escape condition the proportion offering to help was 0.9, and in
the difficult escape, 0.8. While the egoistic motivation clearly had some effect, the helping in the
empathy condition is consistent with the empathy-altruism hypothesis.
Batson even produced evidence that people’s desires to aid others go beyond conscience
and pride, internal psychological punishments and rewards for helping. In experiments in which
the desire to help was aroused and then frustrated by someone else getting to help, an internal
reward system would elevate the mood of those who got to help, but not those that did not.
Contrary to this hypothesis, subjects who saw help provided, but didn’t have to provide it
themselves, had the greatest mood increase, and those that were prevented from helping when no
one else did had the lowest mood. Once empathy is engaged, people apparently have a genuinely
unselfish desire to help.
A skeptic about the effects of group selection does have rejoinders to these experiments.
Genuine proximal altruism could be part of a strategy of advertisement of one’s willingness to
engage in acts helping kin and reciprocal partners, and perhaps be made consistent with some
form of the social intelligence hypothesis. However, the cultural group selection hypothesis
provides the most natural evolutionary explanation for Batson’s results. If cultural groups have
long been subject to group selection, then fellow group members ought to engage in empathy-
altruism even when kinship is remote and expectations of reciprocity limited. The question is,
what arouses our empathy?
5.3 Ideology and symbol system evolution: Ingroup marking
It is a striking feature of human ultra-sociality that groups are explicitly defined and
marked by symbolic boundaries. Some of these marks are relatively simple badges such as body
ornamentation and dialects. Others are complex ritual systems accompanied by elaborately
rationalized ideologies. It is a commonplace that social relations are regulated by norms embedded
in a group’s sanctified belief system (Rappaport 1979). Even in simple hunting and gathering
societies, the groups so marked are rather large, as we already noted. In this section we argue that
symbolic cultural marks are potent factors dividing ingroups from outgroups, sharpening the
boundaries between social units subject to cultural group selection, and regulating empathy and
We have studied models of a processes we term “indirect bias” (Boyd and Richerson
1985: Ch. 8; 1987). In indirect bias, individual imitators use one trait in potential cultural models
to weight that individual in the socialization process. Young people picking adult role models on
the basis of charm, prestige, power, or whatever is a commonplace example. Once a role model is
chosen, youngsters may pick up a number of other ideas, norms, skills, and attitudes without
further bias. If any of these other traits happen to correlated with indices of prestige or charm that
youngsters are using to choose role models, then the correlated variants will increase in
The evolution of symbolic, stylistic markers of group boundaries is an interesting special
case of cultural evolution under indirect bias. We (Boyd and Richerson 1987) investigated a
model of the evolution of “ethnic” markers motivated by these observations. Supposed that there
are two episodes of transmission. In the first episode, children learn some neutral marker trait, like
a speech dialect, from their parents and other local adults. In the second, they learn a subsistence
trait and modify their dialect by selecting among a broader set of adult models. In the first episode
there is just blind copying. In the second, models are weighted by both their symbolic trait and
their subsistence success. Juvenile imitators prefer models with a dialect like that they learned as
children, but also prefer successful models. The broader set of models in the second episode is
meant to mimic the effects of cultural diffusion from one environment to another. The question is,
in an environment in which the best subsistence strategy is variable from place to place, can a
symbolic marker reducing the effect of cultural diffusion of locally maladaptive traits from
neighboring environments arise? The answer is yes. If there is a sharp environmental gradient, a
dialect difference will emerge and continue to get more extreme until the degree of cultural
isolation is sufficient to allow the population to optimize the mean subsistence behavior. Thus, to
the picture of a culturally group selected hominid from the previous section, we can also, in
theory, imagine that the boundaries between groups are formally marked by symbol or style
5.3.2 Empirical evidence
Archaeology suggests that the acquisition of a fully symbolic cultural system led to a
considerable increase in human adaptive sophistication (Bettinger 1991: 203-8). One notable
feature of the human species, increasingly apparent throughout the Pleistocene, is our very wide
geographical range. In Europe, stylistic features of culture marking apparent ethnic differences
first appear at the Upper Paleolithic Transition, about 35,000 ybp. Upper Paleolithic peoples
apparently had more regionally differentiated subsistence strategies and greater population
densities than the Middle Paleolithic populations they replaced. Upper Paleolithic peoples also
extended the human range into such isolated land areas as Australia and the New World.
There is a fair amount of data on the microevolutionary properties of indirect bias. First, it
is highly plausible that people use indirect bias effectively for everyday culture acquisition. This is
quite well documented in the applied literature on the diffusion of innovations (Rogers 1983). It
also seems that in many societies, culturally defined prestige, which people use as indicators for
indirect bias judgments, is correlated with wealth and reproductive success (Irons 1979;
Borgerhoff Mulder 1988). As the theory suggests, everyday fitness advantages can be obtained by
using indirect bias. Second, ethnographic investigations suggest that the symbolic, stylistic
markers in the form of body adornment and the like are used to communicate and negotiate
personal and social identities relative to others. Wiessner (1984) makes a strong case for such
usage in the Kalahari San. These data, together with the appearance of stylistic artifacts at the
Upper Paleolithic Transition, argue that expressive symbolic displays have been part of human
strategies for managing cultural transmission, among other problems of social life, for a
respectable period. Third, there is also evidence, again Wiessner’s (1983) San data is an example,
that ethnolinguistic boundaries are symbolically marked, and that stylistic marks of group
membership are highly salient. Wiessner collected arrow points from a number of Bushmen
groups, including groups unknown to the Kalahari San and in fact ethnically rather different from
them, and asked San men for comments on the distinctive styles. Confronted by points of an
unknown style, Kalahari San guess that their makers are very different people from themselves.
Within groups, on the other hand, exchange of stylistically familiar beadwork and other valuables
is used to build up a notion of the Kalahari San social universe and to build a web of relationships
that link people within ethnolinguistic units.
At the proximal psychological level, the “minimal group” experimental system developed
by Tajfel ( Tajfel et al. 1971) is interesting regarding the cognitive mechanisms involved in the use
of symbols to demarcate groups, and the actions people take based on group membership. It is a
common finding in social psychological experiments, as in real life, that members of groups favor
one another and discriminate against outgroups. The social psychologists in Tajfel’s tradition were
interested in separating the effects of group membership per se from the personal attachments that
form in groups. Turner (1984) contrasts two sorts of hypotheses to explain group oriented
behavior. It could be that functional social groups are composed entirely of networks of
individuals that are linked by personal ties, objective shared fate, or other individual-centered ties,
much as in the social intelligence hypothesis. Groups, on this hypothesis are “some collection of
individuals characterized by mutual interpersonal attraction reflecting some degree of
interdependence and mutual need-satisfaction.” The contrasting hypothesis is that identity symbols
alone are sufficient to induce humans to accept membership in a group, acting positively toward
ingroup members and negatively toward outgroups.
In his prototypical experiments, Tajfel, et al. (1971) told subjects that they were
participating a test of aesthetic judgment. They were shown pictures of paintings by Klee and
Kandinsky, and asked to indicate which they preferred. Then the subjects were divided into two
groups, supposedly on the basis of their art preference, but in fact at random. The subject’s task
was then to divide a sum of money among members of one’s own group or the other group.
Subjects discriminated in favor of the sham ingroup members. The interpretation of these results
was that indications of membership, even in a new, externally imposed, and very abstract group,
was sufficient to motivate ingroup favoritism on the part of subjects.
Turner et al. (1983) were interested an extreme case of categorization generating group-
oriented cooperative behaviors. Could such behavior be motivated by an external categorization
of a group in which the group was categorized as those everyone, including group members,
disliked? Children were shown their picture paired with others and asked to rate the others as
liked or disliked. Then children were ostensibly placed in groups by explicitly random criteria, or
on the basis of liking or on the basis of disliking in different treatments. In still other treatments,
children were not explicitly placed into groups. Then children had to make private decisions about
rewards to allocate to others. Being explicitly assigned to a group had a marked effect, as in the
classic Tajfel design. The most striking contrast was that in the ungrouped condition, disliked
children favored liked others. However, when they were explicitly grouped as to liked or disliked,
disliked subjects, led to believe that they were now members of a group of pariah disliked others,
discriminated almost as much as liked-group members in favor of their fellow pariahs (recall
classic films like The Seven Samurai and The Dirty Dozen, where the plot turns on the bonding
among a group of initially hostile outcasts). Turner (1984) argues that such experiments show that
even shared membership in a group of unattractive others can induce a commitment to the group.
Categorization per se does have a strong effect, independent of inter-personal bonds.
Other social psychologists have objected that Tajfel and Turner have overemphasized the
raw categorization effect in interpreting their experiments. Rabbie (1991) argues that no
experimental system implying that people belong to a group can control away people’s
expectation that the group membership imposed by experimenters implies instrumental
interdependence. From an evolutionary point of view, Rabbie’s point is well taken; people
presumably only react to symbolic badges of group membership because in the evolutionary past
they generally signal politically important social units. In the politically complex world outside the
lab, where many groupings are abstractly possible, people no doubt attempt to make sensible
decisions about which to take seriously in any given circumstance. Nevertheless, it is striking in
humans how a rather abstract, large, impersonal, marked groups (Protestant Irish, Serb, Jew,
German, etc.) can attract great emotional salience and motivate desperate deeds of great risk to
participants. Long periods of relative dormancy seem to leave largely intact the potential
emotional power of groups that can plausibly claim deep historic roots. Individual decision-
makers must also contend with the awful possibility that an aggressive outgroup may suddenly,
for reasons of its own, target one as belonging to a previously weakly relevant group, as has
happened recently to Bosnian Muslims.
A considerable amount of evidence from several domains ranging from comparative
natural history to psychology suggests that the processes of cultural evolution permitted group
selection on cultural variation and the marking of group boundaries by cultural symbols.
Archaeological evidence suggests that symbolic behavior is at least a few tens of thousands of
years old. The hunting and gathering societies under which these processes operated were
composed were very small, overlapping with typical primate bands in size. However, ethnographic
studies show that the bands themselves are typically quite fluid, and that “tribal” ethnolinguistic
units of 500-1500 people (Birdsell 1968) are an important part of the human social adaptation.
Hunting and gathering societies of the ethnographic record appear to expend considerable effort
to maintain the ethnolinguistic tribe as an institution. Yengoyan (1968) noted that Australian
marriage systems were most complex in the arid, sparsely populated interior. He argues that the
complex section systems of the interior ensured that people sought mates from geographically
great distances, which kept the dispersed tribe a functional demographic and ceremonial unit. In
the harsh, variable environment of Central Australia, the ties at the tribe level were used to gain
aid in times of emergency. Although tribal politics among hunters and gatherers is usually quite
simple, some cooperation at the level of the ethnolinguistic tribe is very common. While kinship,
reciprocity, and dominance organize higher primate troops into units as large as hunter-gatherer
bands, there is in the non-human species no social unit corresponding to the ethnolinguistic tribe.
We suppose that the symbolically marked, cooperating ingroup adaptation was based on
psychological decision-making system analogous to Chomsky’s view of language as an innate set
of algorithmic principles, whose parameters are set culturally by the specific language learned
(Pinker 1994: 111-2). The propensity to cooperate with sympathy inspiring ingroup members, and
to use symbolic markers to define ingroups, is a like the innate principles of language. The specific
markers, size of group(s), and internal structure of the group(s) to which individuals belong are
culturally variable parameters that generate a specific functioning instance of the innate
propensities. Like innate language structures and adult lactose absorption, the social decision-
making principles most likely coevolved as adaptations to an emerging cultural environment in
which cooperation within culturally marked groups was becoming important. Success in
intergroup competition came to depend upon within-group cooperation; an evolutionary arms
race arose. The scale of cooperation-to-compete might escalate until ecological rather than
evolutionary constraints bring a halt. Once the barriers of scale imposed by kinship and reciprocity
are breached by cultural group selection, it is not clear what the next natural evolutionary limit to
scale of cooperation is. This picture squares with experimental evidence such as Batson’s and
Tajfel’s that perhaps give us a look at the innate principles of human social action, and it explains
how human societies can organize cooperation on a far larger scale than our primate relatives.
It can be objected that the subjects of these experiments already have considerable cultural
sophistication when the participate in such experiments. The same objection applies to the
experiments such as Cosmides on cheater detection. Fair enough, some human universals likely
involve conservative cultural features as well as genetic elements. But it is also true that
chimpanzees would not behave in the same way in any of these experiments no matter how hard
experimenters tried, nor would our Australopithecine ancestors of 5 million years ago. Even if we
cannot yet be sure exactly what the innate principles are that divide us from other apes we can be
sure that they exist, as remarkably human-like as chimp behavior is in so many ways.
Human populations grew explosively in the Holocene when the development of food plant
production removed previously tight constraints on human densities. With ecological constraints
removed, the cooperation-to-compete arms race renewed its progress. Very large, hierarchically
organized societies grew up in the agricultural regions of the world. This phase of social evolution
is still an ongoing process. The last 10,000 years have witnessed the repeated emergence of ever
more powerful societies, but so sign of any perfected equilibrium adaptation.
Complex societies make lavish use of symbolic boundary marking, and are capable of
impressive feats of cooperation, such as road nets, irrigation systems, and, unfortunately, armies.
The tendency of ethnic identities to retain great importance in even the most modern societies was
a surprise to many social scientists (Glazer and Moynihan 1975). But complex societies also
require (or at any rate always have had) a substantial degree of organized coercive control.
Ethnographically known hunting and gathering societies have only the weakest forms of
leadership and little tendency for leaders to be able to lead lives of privilege at the expense of
followers. Human social instincts are certainly not adapted to live in societies numbering in the
millions organized by elites that establish a great social distance between themselves and ordinary
citizens. The means by which ancient social instincts and modern cultural institutions conspire to
create the impressive coordination, cooperation and division of labor of complex societies remains
to be told. We suggest that a close examination of the mechanisms will reveal a series of only
partially successful work-arounds of social instincts deriving from adaptation to life in much
simpler, but still highly cooperative, symbolically marked groups. To the conflicts between the
very small-group altruism normally favored by selection on genes and the ingroup altruism that
coevolved with culture in the Pleistocene, the post-Pleistocene social insitutions of complex
societies add another layer of determinants of behavior that frequently conflict with all our innate
Alexander, R.D. (1987). The biology of moral systems. Aldine de Gruyter, New York.
Axelrod, R., and Hamilton, W.D. (1981). The evolution of cooperation. Science 211, 1390-6
Barth, F. (1981). Features of person and society in Swat: collected essays on Pathans. Routledge
and Kegan Paul, London.
Batson, C.D. (1991). The altruism question: toward a social psychological answer. Lawrence
Erlbaum, Hillsdale NJ.
Bettinger, R.L. and Baumhoff, M.A.. (1982). The Numic spread: Great Basin cultures in
competition. American Antiquity, 47, 485-503.
Bettinger, R.L. (1991). Hunter-gatherers: archaeological and evolutionary theory. Plenum, New
Birdsell, J.B. (1968). Some predictions for the Pleistocene based on equilibrium systems of
Recent hunter-gatherers. In Man the Hunter (ed. R.B. Lee and I. DeVore), pp. 229-240.
Borgerhoff Mulder, M. (1988). Behavioral ecology of traditional societies. Trends in Ecology and
Evolution, 3, 260-4.
Boyd, R., and Richerson, P.J. (1985). Culture and the evolutionary process. University of
Chicago Press, Chicago.
Boyd, R. and Richerson, P.J. (1987). The evolution of ethnic markers. Cultural Anthropology, 2,
Boyd, R, and Richerson, P.J. (1988). The evolution of reciprocity in sizable groups. Journal of
Theoretical Biology, 132, 337-56.
Boyd, R. and Richerson, P.J. (1989). The evolution of indirect reciprocity. Social Networks 11,
Boyd, R. and Richerson, P.J. (1992). Punishment allows the evolution of cooperation (or anything
else) in sizable groups. Ethology and Sociobiology 13, 171-95.
Bryne, R. and A. Whitten (eds.). (1988). Machiavellian intelligence: social expertise and the
evolution of the intellect in monkeys, apes, and humans. Oxford University Press, Oxford.
Campbell, D.T. (1983) The two distinct routes beyond kin selection to ultrasociality: Implications
for the Humanities and Social Sciences. In: The Nature of Prosocial Development: Theories
and Strategies D. Bridgeman (ed.), pp. 11-39, Academic Press, New York.
Cavalli-Sforza, L.L. and Feldman, M.W. (1981). Cultural transmission and evolution: a
quantitative approach. Princeton University Press, Princeton.
Cosmides, L. (1989). The logic of social exchange: has natural selection shaped human reason?
Studies with the Wason selection task. Cognition, 31, 187-276.
Daly, M. and Wilson, M. 1988. Homicide. Aldine de Gruyter, New York.
Darwin, C. (1874 ). The descent of man, 2nd. ed. American Home Library, New York.
Dunbar, R.I.M. (1992). Neocortical size as a constraint on group size in primates. Journal of
Human Evolution. 22, 469-93.
Durham, W.H. (1991). Coevolution: genes, culture, and human diversity. Stanford University
Eibl-Eibesfeldt, I. (1982). Warfare, man’s indoctrinability, and group selection. Zeitschrift für
Tierpsychologie, 67, 177-198.
Eibl-Eibesfeldt, I. (1989). Human ethology. Aldine de Gruyter, New York.
Glazer, N., and Moynihan, D.P. (1975). Ethnicity: theory and experience. Harvard University
Goodall, J. (1986). The chimpanzees of Gombe: patterns of behavior. Cambridge Mass.: Harvard
University Press, Cambridge.
Hamilton, W.D. 1964. Genetical evolution of social behavior I, II. Journal of theoretical Biology,
Hamilton, W.D. (1975). Innate social aptitudes of man: An approach from evolutionary genetics.
In Biosocial anthropology (ed. R.Fox), pp. 133-155. Malaby, London.
Harcourt, A.H. (1992). Cooperation in conflicts: commonalities between humans and other
animals. Politics and the Life Sciences, 11, 251-9.
Irons, W. (1979). Cultural and biological success. In Evolutionary biology and human social
behavior (ed. N.A. Chagnon and W. Irons), pp. 257-272. Duxbury Press, North Scituate,
Knauft, B.M. (1985). Good company and violence: sorcery and social action in a Lowland New
Guinea society. University of California Press, Berkeley.
Knauft, B.M. (1987). Reconsidering violence in simple human societies: Homicide among the
Gebusi of New Guinea. Current Anthropology, 28, 457-500.
Knauft, B.M. (1993). South Coast New Guinea cultures: history, comparison, dialectic.
Cambridge University Press, Cambridge.
Kummer, H., Daston, L., Gigerenzer, G., and Silk, J. (in press). The social intelligence hypothesis.
In Human by nature (P. Weingard, S. Massen, S. Mitchell, and P. Richerson, eds.).
Lawrence Erlbaum, Hillsdale, NJ.
Maenchen-Helfen, O.J. (1973). The world of the Huns. University of California Press, Berkeley.
McGrew, W.C. (1992). Chimpanzee material culture: implications for human evolution.
Cambridge University Press, Cambridge
Milgram, S. (1974). Obedience to authority: an experimental view. Harper and Row, New York.
Nuttin, J.M., Jr., and Beckers, A. (1975). The illusion of attitude change: towards a response
contagion theory of persuasion. Academic Press, London.
Olson, M. (1982). The rise and decline of nations: economic growth, stagflation, and social
rigidities. Yale University Press, New Haven.
Otterbein, K.F. (1968). Internal war: A cross-cultural study. American Anthropologist, 70, 277-
Pinker, S. (1994). The Language instinct: how the mind creates language. William Morrow, New
Rabbie, J.M. (1991). Determinants of instrumental intra-group cooperation. In Cooperation and
prosocial behavior (ed. R.A. Hinde and J. Groebel), pp.238-262. Cambridge University
Rappaport, R.A. (1979). Ecology, meaning, and religion. North Atlantic Books, Richmond, CA.
Rogers, E.M. (1983). Diffusion of innovations, 3rd edition. Macmillan, New York.
Sepher, J. 1983. Incest, the biosocial view. Academic Press, New York.
Soltis, J., R. Boyd, R., and Richerson, P.J. (1995). Can group-functional behaviors evolve by
cultural group selection? An empirical test. Current Anthropology, 36, 437-94.
Tajfel, H., Flament,C., Billig, M.G., and Bundy, R.P. (1971). Social categorization and intergroup
behaviour. European Journal of Social Psychology, 1, 149-75.
Trivers, R.L. (1971). The evolution of reciprocal altruism. Quarterly Review of Biology, 46, 35-
Turner, J.C. (1984). Social identification and psychological group formation. In The social
dimension: European developments in social psychology, Vol. 2 (ed. H. Tajfel), pp. 518-
538. Cambridge Univesity Press, Cambridge.
Silk, J.B. 1980. Adoption and kinship in Oceania. American Anthropologist, 82, 79-820.
deWaal, F.B.M. (1992) .Chimpanzee politics. Jonathan Cape, London.
Westermarck, E. (1894). The history of human marriage. Macmillan, London.
Wiessner, P. 1983. Style and social information in Kalahari San projectile points. American
Antiquity, 48, 253-76.
Wiessner, P. (1984). Reconsidering the behavioral bais for style: A case study among the Kalahari
San. Journal of Anthropological Archaeology, 3, 190-234.
Wilson, D.S. and Sober, E. (1994). Reintroducing group selection to the human behavioral
sciences. Behavioral and Brain Sciences, 17, 585-654.
Wilson, E.O. 1975. Sociobiology: the new synthesis. Harvard University Press, Cambridge.
Wilson, E.O. and B. Hölldobler. 1988. Dense heterarchies and mass communication as the basis
of organization in ant colonies. Trends in Ecology and Evolution, 3, 65-7.
Wolf, A.P. (1970). Childhood association and sexual attraction: A further test of the Westermarck
hypothesis. American Anthropologist, 72, 503-515.
Yengoyan, A.A. 1968. Demographic and ecological influences on Aboriginal Australian marriage
systems. In Man the hunter ( ed. R.B. Lee and I. DeVore), pp. 185-199. Aldine, Chicago.
Acknowledgments: We thank the organizers of the Ringberg conference, Ιrenäus Eibl-
Eibesfeldt and Frank Salter for the opportunity to participate in this project. Sandy Harcourt,
Mary Jackman, Michael McGuire, Lore Ruttan, Dan Sellen, Joseph Soltis, and David Wilson
provided most helpful comments on the manuscript. Supported in part by a grant from the Alfred
P. Sloan Foundation