. 2
( 13)


We can summarize our results as follows. First, the canonical model
of self-regarding behavior is not supported in any of the societies
studied. In the ultimatum game, for example, in all societies either re-
sponders, proposers, or both behaved in a reciprocal manner. Second,
there is considerably more behavioral variability across groups than
28 Gintis, Bowles, Boyd, and Fehr

had been found in previous cross-cultural research. While mean ulti-
matum game offers in experiments with student subjects are typically
between forty-three and forty-eight percent, the mean offers from pro-
posers in our sample ranged from twenty-six to ¬fty-eight percent.
While modal ultimatum game offers are consistently ¬fty percent
among university students, sample modes with the data range in this
study ranged from ¬fteen to ¬fty percent. Rejections were extremely
rare, in some groups (even in the presence of very low offers), while in
others, rejection rates were substantial, including frequent rejections of
hyper-fair offers (i.e., offers above ¬fty percent). By contrast, the Machi-
guenga have mean offer of twenty-six percent but no rejections. The
´ ´
Ache and Tsimane distributions resemble inverted American distribu-
tions. The Orma and Huinca (non“Mapuche Chileans living among
the Mapuche) have modes near the center of the distribution, but
show secondary peaks at full cooperation.
Third, differences between societies in ˜˜market integration™™ and ˜˜coopera-
tion in production™™ explain a substantial portion (about ¬fty percent) of the
behavioral variation between groups. The higher the degree of market
integration and the higher the payoffs to cooperation, the greater the
level of cooperation and sharing in experimental games. The societies
were rank-ordered in ¬ve categories”market integration (how often
do people buy and sell, or work for a wage?), cooperation in produc-
tion (is production collective or individual?), plus anonymity (how
prevalent are anonymous roles and transactions?), privacy (how easily
can people keep their activities secret?), and complexity (how much
centralized decision-making occurs above the level of the household?).
Using statistical regression analysis, only the ¬rst two characteristics
were signi¬cant, and they together accounted for about ¬fty percent of
the variation among societies in mean ultimatum game offers. Fourth,
individual-level economic and demographic variables do not explain
behavior either within or across groups. Finally, the nature and degree
of cooperation and punishment in the experiments is generally consis-
tent with economic patterns of everyday life in these societies.
The ¬nal point of this experiment is in some respects the most im-
portant for future research. In a number of cases, the parallels between
experimental game play and the structure of daily life were quite strik-
ing. Nor was this relationship lost on the subjects themselves. The
Orma immediately recognized that the public goods game was similar
to the harambee, a locally-initiated contribution that households make
when a community decides to construct a road or school. They dubbed
Moral Sentiments and Material Interests 29

the experiment ˜˜the harambee game™™ and gave generously (mean ¬fty-
eight percent with twenty-¬ve percent full contributors).
Among the Au and Gnau, many proposers offered more than half
the total amount and many of these hyper-fair offers were rejected!
This re¬‚ects the Melanesian culture of status-seeking through gift giv-
ing. Making a large gift is a bid for social dominance in everyday life in
these societies, and rejecting the gift is a rejection of being subordinate.
Among the whale-hunting Lamalera, sixty-three percent of the pro-
posers in the ultimatum game divided the total amount equally, and
most of those who did not offered more than ¬fty percent (the mean
offer was ¬fty-seven percent). In real life, a large catch”always the
product of cooperation among many individual whalers”is meticu-
lously divided into predesignated proportions and carefully distri-
buted among the members of the community.
Among the Ache, seventy-nine percent of proposers offered either
forty or ¬fty percent, and sixteen percent offered more than ¬fty per-
cent, with no rejected offers. In daily life, the Ache regularly share
meat, which is distributed equally among all households irrespective
of which hunter made the catch.
In contrast to the Ache, the Hadza made low offers and had high re-
jection rates in the ultimatum game. This re¬‚ects the tendency of these
small-scale foragers to share meat but with a high level of con¬‚ict and
frequent attempts of hunters to hide their catch from the group.
Both the Machiguenga and Tsimane made low ultimatum game
offers, and there were virtually no rejections. These groups exhibit little
cooperation, exchange, or sharing beyond the family unit. Ethnograph-
ically, both groups show little fear of social sanctions and care little
about ˜˜public opinion.™™
The Mapuche™s social relations are characterized by mutual suspi-
cion, envy, and fear of being envied. This pattern is consistent with
researchers™ interviews with the Mapuche following the ultimatum
game. Mapuche proposers rarely claimed that their offers were in¬‚u-
enced by fairness but rather by a fear of rejection. Even proposers who
made hyper-fair offers claimed that they feared the remote possibility
of spiteful responders, who would be willing to reject even ¬fty-¬fty

1.9.2 Cultural Evolution
Suppose a group, in the name of promoting group harmony, has
adopted the norm of peaceful adjudication of disputes. If the members
30 Gintis, Bowles, Boyd, and Fehr

are self-interested, no third party will intervene in a dispute between
two members to thwart a violent interaction and punish its perpetra-
tors. By contrast, a group with a suf¬cient fraction of reciprocators will
intervene, allowing the norm to persist over time, even in the face of
the indifference of the self-interested and the opposition of an appre-
ciable fraction of troublemakers. Thus, strong reciprocity can stabilize
prosocial norms that otherwise could not be sustained in the group.
Conversely, suppose in the name of preventing invidious distinc-
tions, a group has adopted a work norm that discourages members
from supplying effort above a certain approved level. Such a norm is,
of course, ¬tness-reducing for the group™s members. Indeed, if mem-
bers are self-interested, some will violate the norm, and no others will
intervene to protect it. The ¬tness-reducing norm will thus disappear.
However, a small fraction of strong reciprocators who accept the norm
and who punish its violators can stabilize the norm even when many
would prefer to violate it.
Our point here is simple. For most of human history (until a few
thousand years ago), there were no schools, churches, books, laws, or
states. There was, therefore, no centralized institutional mechanism for
enforcing norms that affect the members of a group as a whole. Strong
reciprocity evolved because groups with strong reciprocators were
capable of stabilizing prosocial norms that could not be supported
using principles of long-term self-interest alone, because it is gener-
ally ¬tness-enhancing for an individual to punish only transgressions
against the individual himself and then only if the time horizon is suf-
¬ciently lengthy to render a reputation for protecting one™s interests.
On the other hand, the same mechanisms that have the ability to en-
force prosocial norms can almost as easily enforce ¬tness-neutral and
antisocial norms (Edgerton 1992; Boyd and Richerson 1992; Richerson
and Boyd 2003).
In this framework, prosocial norms evolve not because they have su-
perior ¬tness within groups, but because groups with prosocial norms
outcompete groups that are de¬cient in this respect. It is not surpris-
ing, for instance, that the ˜˜great religions™™ ( Judaism, Christianity, Bud-
dhism, Islam, Hinduism, and so forth) stress prosocial norms”such as
helping one™s neighbors, giving each his due, turning the other cheek,
and the like.
There is considerable evidence for the operation of natural selection
in cultural evolution (Richerson and Boyd 2003). For instance, religious
practice differences entail fertility and survival differentials (Roof and
Moral Sentiments and Material Interests 31

McKinney 1987), and the organization of human populations into
units which engage in sustained, lethal combat with other groups leads
to the survival of groups with prosocial organizational and participa-
tory forms. Soltis, Boyd, and Richerson (1995) reviewed the ethnogra-
phy of warfare in simple societies in highland New Guinea. The
pattern of group extinction and new group formation in these cases
conforms well to a cultural evolution model. The strength of cultural
group selection in highland New Guinea was strong enough to cause
the spread of a favorable new social institution among a metapopula-
tion in about 1,000 years. Cases of group selection by demic expansion
are quite well described, for example the spread of the southern Suda-
nese Nuer at the expense of the Dinka (Kelly 1985), the expansion of
the Marind-anim at the expense of their neighbors by means of large,
well-organized head-hunting raids at the expense of their neighbors,
including the capture and incorporation of women and children
(Knauft 1993), and the Hispanic conquest of Latin America (Foster

1.10 Applications to Social Policy

Economic policy has generally been based on a model of the self-
regarding individual. It would be surprising if our model of strong
reciprocity did not suggest signi¬cant revisions in standard economic
policy reasoning, and indeed it does. This section includes several
applications of the strong reciprocity model to social policy. In fact,
only a relatively weak version of strong reciprocity enters into policy
analysis. All that is required is that agents be conditional cooperators
and altruistic punishers in public and repeated situations where repu-
tations can be established”an assumption amply justi¬ed by the be-
havioral evidence. Speci¬cally, it is unimportant for these analyses
whether strong reciprocity is the product of purely cultural or gene-
culture coevolutionary dynamics”whether this behavior is truly al-
truistic or includes some dif¬cult-to-observe personal payoff (such as
costly signaling, as suggested by Smith and Bliege Bird in chapter 4),
or whether it is fundamentally adaptive or maladaptive.
Elinor Ostrom argues in chapter 9 that common pool resource
management has often failed when based on the standard model of
incentives, whereas a more balanced program of local community
management and government regulation”often the former alone”
can contribute to effective conservation and egalitarian distribution of
32 Gintis, Bowles, Boyd, and Fehr

common pool resources. This alternative policy framework ¬‚ows natu-
rally from the strong reciprocity model and depends on the presence of
a fraction of strong reciprocators in the population for its effectiveness.
As Christina Fong, Samuel Bowles, and Herbert Gintis show in
chapter 10, approaches to egalitarian income redistribution are also
strengthened by the use of the strong reciprocity model. During the
last few decades of the twentieth century in the United States, there
emerged an unprecedented malaise concerning the system of egalitar-
ian redistribution in public opinion. Many interpret this shift, which
has led to important changes in the social welfare system, as a resur-
gence of self-interest on the part of the country™s nonpoor and of racist
attitudes on the part of the majority white citizenry. Fong, Bowles, and
Gintis present a body of evidence that disputes this view and argue in
favor of model of voter behavior based on strong reciprocity.
In chapter 11, Truman Bewley uses strong reciprocity to model un-
employment in the macroeconomy of the United States. Bewley tackles
one of the oldest, and most controversial, puzzles in economics: why
nominal wages rarely fall (and real wages do not fall enough) when
unemployment is high. He does so in a novel way, through interviews
with over 300 businessmen, union leaders, job recruiters, and unem-
ployment counselors in the northeastern United States during the
early 1990s recession. Bewley concludes that employers resist pay cuts
largely because the savings from lower wages are usually outweighed
by the cost of reducing worker morale: pay cuts are seen by workers
as an unfriendly and unfair act, and employees retaliate by working
less hard and less in line with managements™ goals. Bewley thus shows
that even the most standard of economic problems, that of wage deter-
mination, cannot be understood outside the framework of an empirical
and behavioral approach to individual behavior.
Nowhere has the standard model of the self-regarding actor had
more in¬‚uence than in legal theory and the politics of legislation.
Beginning with the work of economist Ronald Coase (1960) and devel-
oped by the legal scholar Richard Posner (1973), ˜˜Law and Economics™™
has become a potent analytical framework for studying the effect of
legislation on social welfare. While we do not doubt the value of this
work, its abstraction from reciprocity and other non“self-regarding
motives limits its general relevance. In chapter 12, Dan M. Kahan
addresses the relevance of reciprocity to law and public policy. He sug-
gests that individuals will often contribute voluntarily to collective
goods so long as they believe that most others are willing to do the
Moral Sentiments and Material Interests 33

same. Promoting trust, in the form of reason to believe that fellow citi-
zens are contributing their fair share, is thus a potential alternative to
costly incentive schemes for solving societal collective action problems.
Indeed, conspicuous penalties and subsidies, reciprocity theory implies,
might sometimes aggravate rather than ameliorate collective action
problems by giving citizens reason to doubt that other citizens are con-
tributing voluntarily to societal collective goods. He illustrates these
conclusions by analyzing several regulatory problems”including tax
evasion, the location of toxic waste facilities, and the production of
information and technology.
In the ¬nal chapter of this volume, Samuel Bowles and Herbert Gin-
tis offer a larger and more synthetic vision of what a deeper apprecia-
tion of moral sentiments might imply for social structure and policy.
They argue that the moral sentiments documented and analyzed in this
book lead us to a new view of social communities and an understand-
ing of why the two preeminently anonymous modern institutions”
the market and the state”only incompletely addresses modern social
If Bowles and Gintis are right in asserting that communities work
well relative to markets and states where the tasks are qualitative and
hard to capture in explicit contracts, and the con¬‚icts of interest among
the members are limited, it seems likely that extremely unequal soci-
eties will be competitively disadvantaged in the future because their
structures of privilege and material reward limit the capacity of com-
munity governance to facilitate the qualitative interactions that under-
pin the modern economy. Political democracy, policies that limit the
extent of social and economic inequality, and widespread civil liberties
may thus not only be desirable in terms of political ethics, but may in
fact be necessary to harness moral sentiments to future economic and
social development around the world.


1. We say an action is altruistic when it confers bene¬ts to other members of a group at a
cost to the actor. Note that this de¬nition says nothing about the intentions of the actor.
Note also that an action can be altruistic yet increase the subjective utility of the actor. In-
deed, any voluntary, intended act of altruism will have this property.
2. Since we care about behavior rather than its subjective correlates, throughout this
chapter we use the term ˜˜self-regarding™™ rather than ˜˜self-interested.™™ For instance, if one
truly cares about others, it may be self-interested to sacri¬ce on their behalf, even though
it is manifestly non“self-regarding to do so.
34 Gintis, Bowles, Boyd, and Fehr

3. While the term ˜˜strong reciprocity™™ is new, the idea certainly is not, having been
studied by Homans (1958), Gouldner (1960), Moore Jr. (1978), Frank (1988), and Hirshlei-
fer and Rasmusen (1989), among others.
4. The adaptive signi¬cance of the human ability to detect cheaters was stressed by
Cosmides and Tooby (1992) who, in contrast with our usage, consider this capacity as in-
dividually ¬tness-enhancing rather than altruistic. The precommitment to punish trans-
gressors has been insightfully analyzed by Hirshleifer (1987) and Frank (1988).

5. Classical group selection involves the altruistic behavior having ¬tness costs as com-
pared with behavior of non-altruistic group members, but these costs being more than
offset by the higher ¬tness of groups with many altruists, as compared with groups in
which altruism is rare or absent.
6. By multilevel selection (Keller 1999), we mean that selection operates at some level other
than that of the gene or individual. For instance, the social organization of a beehive con-
tributes to the ¬tness of individual bees, which leads to the growth of beehives.

7. The observed behavior was predicted by Akerlof (1982).
8. For additional experimental results and analysis, see Bowles and Gintis (2002) and
Fehr and Gachter (2002).


Akerlof, George A. ˜˜Labor Contracts as Partial Gift Exchange,™™ Quarterly Journal of Eco-
nomics 97, 4 (November 1982): 543“569.
Alcock, John. Animal Behavior: An Evolutionary Approach. Sunderland, MA: Sinauer, 1993.
Alexander, Richard D. The Biology of Moral Systems. New York: Aldine, 1987.
Andreoni, James. ˜˜Cooperation in Public Goods Experiments: Kindness or Confusion,™™
American Economic Review 85, 4 (1995): 891“904.
Andreoni, James, and John H. Miller. ˜˜Giving According to GARP: An Experimental Test
of the Consistency of Preferences for Altruism,™™ Econometrica 70, 2 (2002): 737“753.
Andreoni, James, Brian Erard, and Jonathan Feinstein. ˜˜Tax Compliance,™™ Journal of Eco-
nomic Literature 36, 2 ( June 1998): 818“860.
Axelrod, Robert, and William D. Hamilton. ˜˜The Evolution of Cooperation,™™ Science 211
(1981): 1390“1396.
Blount, Sally. ˜˜When Social Outcomes Aren™t Fair: The Effect of Causal Attributions on
Preferences,™™ Organizational Behavior & Human Decision Processes 63, 2 (August 1995):

Boehm, Christopher. ˜˜The Evolutionary Development of Morality as an Effect of Domi-
nance Behavior and Con¬‚ict Interference,™™ Journal of Social and Biological Structures 5
(1982): 413“421.
Boone, James L., and Karen L. Kessler. ˜˜More Status or More Children? Social Status, Fer-
tility Reduction, and Long-Term Fitness,™™ Evolution & Human Behavior 20, 4 ( July 1999):
Moral Sentiments and Material Interests 35

Boorman, Scott A., and Paul Levitt. The Genetics of Altruism. New York: Academic Press,
Bowles, Samuel. ˜˜Individual Interactions, Group Con¬‚icts, and the Evolution of Prefer-
ences,™™ in Steven N. Durlauf and H. Peyton Young (eds.) Social Dynamics. Cambridge,
MA: MIT Press, 2001, 155“190.
Bowles, Samuel, and Herbert Gintis. ˜˜Homo Reciprocans,™™ Nature 415 (10 January 2002):
”””. ˜˜The Evolution of Strong Reciprocity: Cooperation in Heterogeneous Popula-
tions,™™ Theoretical Population Biology 65 (2004): 17“28.
Bowles, Samuel, Jung-kyoo Choi, and Astrid Hopfensitz. ˜˜The Co-evolution of Individ-
ual Behaviors and Social Institutions,™™ Journal of Theoretical Biology 223 (2003): 135“147.
Boyd, Robert, and Peter J. Richerson. Culture and the Evolutionary Process. Chicago: Uni-
versity of Chicago Press, 1985.

”””. ˜˜Punishment Allows the Evolution of Cooperation (or Anything Else) in Sizeable
Groups,™™ Ethology and Sociobiology 113 (1992): 171“195.

”””. ˜˜Group Bene¬cial Norms Can Spread Rapidly in a Cultural Population,™™ Journal
of Theoretical Biology 215 (2002): 287“296.
Bowles, Samuel, Herbert Gintis, Samuel Bowles, and Peter J. Richerson. ˜˜Evolution of
Altruistic Punishment,™™ Proceedings of the National Academy of Sciences 100, 6 (March
2003): 3531“3535.
Brown, Donald E. Human Universals. New York: McGraw-Hill, 1991.
Camerer, Colin, and Richard Thaler. ˜˜Ultimatums, Dictators, and Manners,™™ Journal of
Economic Perspectives 9, 2 (1995): 209“219.
Coase, Ronald H. ˜˜The Problem of Social Cost,™™ Journal of Law and Economics 3 (October
1960): 1“44.
Coleman, James S. Foundations of Social Theory. Cambridge, MA: Belknap, 1990.
Cosmides, Leda, and John Tooby. ˜˜Cognitive Adaptations for Social Exchange,™™ in Jer-
ome H. Barkow, Leda Cosmides, and John Tooby (eds.) The Adapted Mind: Evolutionary
Psychology and the Generation of Culture. New York: Oxford University Press, 1992, 163“

Crow, James F., and Motoo Kimura. An Introduction to Population Genetic Theory. New
York: Harper & Row, 1970.
Dawes, Robyn M., and Richard Thaler. ˜˜Cooperation,™™ Journal of Economic Perspectives 2
(1988): 187“197.
Dawkins, Richard. The Sel¬sh Gene. Oxford: Oxford University Press, 1976.

”””. The Sel¬sh Gene, 2nd Edition. Oxford: Oxford University Press, 1989.
Durham, William H. Coevolution: Genes, Culture, and Human Diversity. Stanford: Stanford
University Press, 1991.

Edgerton, Robert B. Sick Societies: Challenging the Myth of Primitive Harmony. New York:
The Free Press, 1992.
36 Gintis, Bowles, Boyd, and Fehr

Eibl-Eibesfeldt, I. ˜˜Warfare, Man™s Indoctrinability and Group Selection,™™ Journal of Com-
parative Ethnology 60, 3 (1982): 177“198.
Falk, Armin, Ernst Fehr, and Urs Fischbacher. ˜˜Testing Theories of Fairness and
Reciprocity-Intentions Matter,™™ 2002. University of Zurich.
Fehr, Ernst, and Bettina Rockenbach. ˜˜Detrimental Effects of Incentives on Human Altru-
ism?™™ Nature 422 (March 2003): 137“140.
Fehr, Ernst, and J. List. ˜˜The Hidden Costs and Returns of Incentives: Trust and Trust-
worthiness among CEOs,™™ 2002. Working Paper, Institute for Empirical Research, Uni-
versity of Zurich.
Fehr, Ernst, and Simon Gachter. ˜˜Cooperation and Punishment,™™ American Economic Re-
view 90, 4 (September 2000a): 980“994.
”””. ˜˜Do Incentive Contracts Crowd Out Voluntary Cooperation?™™ 2000b. Working
Paper No. 34, Institute for Empirical Research, University of Zurich.

”””. ˜˜Altruistic Punishment in Humans,™™ Nature 415 (10 January 2002): 137“140.
Fehr, Ernst, and Klaus M. Schmidt. ˜˜A Theory of Fairness, Competition, and Coopera-
tion,™™ Quarterly Journal of Economics 114 (August 1999): 817“868.
Fehr, Ernst, Simon Gachter, and Georg Kirchsteiger. ˜˜Reciprocity as a Contract Enforce-
ment Device: Experimental Evidence,™™ Econometrica 65, 4 ( July 1997): 833“860.

Fehr, Ernst, Georg Kirchsteiger, and Arno Riedl. ˜˜Does Fairness Prevent Market Clear-
ing?™™ Quarterly Journal of Economics 108, 2 (1993): 437“459.

”””. ˜˜Gift Exchange and Reciprocity in Competitive Experimental Markets,™™ European
Economic Review 42, 1 (1998): 1“34.
Feldman, Marcus W., and Lev A. Zhivotovsky. ˜˜Gene-Culture Coevolution: Toward a
General Theory of Vertical Transmission,™™ Proceedings of the National Academy of Sciences
89 (December 1992): 11935“11938.
Feldman, Marcus W., Luca L. Cavalli-Sforza, and Joel R. Peck. ˜˜Gene-Culture Coevolu-
tion: Models for the Evolution of Altruism with Cultural Transmission,™™ Proceedings of
the National Academy of Sciences 82 (1985): 5814“5818.
Fix, Alan. Migration and Colonization in Human Microevolution. Cambridge: Cambridge
University Press, 1999.

Foster, George M. Culture and Conquest: America™s Spanish Heritage. New York: Wenner-
Gren, 1960.
Frank, Robert H. Passions Within Reason: The Strategic Role of the Emotions. New York:
Norton, 1988.
Frey, Bruno. ˜˜A Constitution for Knaves Crowds Out Civic Virtue,™™ Economic Journal 107,
443 ( July 1997): 1043“1053.
”””. ˜˜The Cost of Price Incentives: An Empirical Analysis of Motivation Crowding
Out,™™ American Economic Review 87, 4 (September 1997): 746“755.
”””. Not Just for the Money: An Economic Theory of Personal Motivation. Cheltenham,
UK: Edward Elgar, 1997.
Moral Sentiments and Material Interests 37

Friedman, Milton. Essays in Positive Economics. Chicago: University of Chicago Press,
Frohlich, Norman, and Joe Oppenheimer. ˜˜Choosing Justice in Experimental Democra-
cies with Production,™™ American Political Science Review 84, 2 ( June 1990): 461“477.
Fudenberg, Drew, and Eric Maskin. ˜˜The Folk Theorem in Repeated Games with Dis-
counting or with Incomplete Information,™™ Econometrica 54, 3 (May 1986): 533“554.
Gachter, Simon, and Ernst Fehr. ˜˜Collective Action as a Social Exchange,™™ Journal of Eco-
nomic Behavior and Organization 39, 4 ( July 1999): 341“369.
Gintis, Herbert. ˜˜Strong Reciprocity and Human Sociality,™™ Journal of Theoretical Biology
206 (2000): 169“179.
”””. ˜˜Solving the Puzzle of Human Prosociality,™™ Rationality and Society 15, 2 (May
2003): 155“187.
”””. ˜˜The Hitchhiker™s Guide to Altruism: Genes, Culture, and the Internalization of
Norms,™™ Journal of Theoretical Biology 220, 4 (2003): 407“418.
Gintis, Herbert, Eric Alden Smith, and Samuel Bowles. ˜˜Costly Signaling and Coopera-
tion,™™ Journal of Theoretical Biology 213 (2001): 103“119.
Glaeser, Edward, David Laibson, Jose A. Scheinkman, and Christine L. Soutter. ˜˜Measur-
ing Trust,™™ Quarterly Journal of Economics 65 (2000): 622“846.

Gouldner, Alvin W. ˜˜The Norm of Reciprocity: A Preliminary Statement,™™ American So-
ciological Review 25 (1960): 161“178.

Greenberg, M. S., and D. M. Frisch. ˜˜Effect of Intentionality on Willingness to Reciprocate
a Favor,™™ Journal of Experimental Social Psychology 8 (1972): 99“111.
Guth, Werner, and Reinhard Tietz. ˜˜Ultimatum Bargaining Behavior: A Survey and
Comparison of Experimental Results,™™ Journal of Economic Psychology 11 (1990): 417“449.
Hamilton, W. D. ˜˜The Genetical Evolution of Social Behavior,™™ Journal of Theoretical Biol-
ogy 37 (1964): 1“16, 17“52.
Hechter, Michael, and Satoshi Kanazawa. ˜˜Sociological Rational Choice,™™ Annual Review
of Sociology 23 (1997): 199“214.
Henrich, Joseph, and Robert Boyd. ˜˜Why People Punish Defectors: Weak Conformist
Transmission Can Stabilize Costly Enforcement of Norms in Cooperative Dilemmas,™™
Journal of Theoretical Biology 208 (2001): 79“89.
Henrich, Joe, Robert Boyd, Samuel Bowles, Colin Camerer, Ernst Fehr, and Herbert
Gintis. Foundations of Human Sociality: Ethnography and Experiments in Fifteen Small-scale
Societies. Oxford: Oxford University Press, 2004.
Henrich, Joe, Robert Boyd, Samuel Bowles, Colin Camerer, Ernst Fehr, Herbert Gintis,
and Richard McElreath. ˜˜Cooperation, Reciprocity and Punishment in Fifteen Small-scale
Societies,™™ American Economic Review 91 (May 2001): 73“78.
Hertwig, Ralph, and Andreas Ortmann. ˜˜Experimental Practices in Economics: A Meth-
odological Challenge for Psychologists?™™ Behavioral and Brain Sciences 24 (2001): 383“451.
Hirschman, Albert. ˜˜Against Parsimony,™™ Economic Philosophy 1 (1985): 7“21.
38 Gintis, Bowles, Boyd, and Fehr

Hirshleifer, David, and Eric Rasmusen. ˜˜Cooperation in a Repeated Prisoners™ Dilemma
with Ostracism,™™ Journal of Economic Behavior and Organization 12 (1989): 87“106.
Hirshleifer, Jack. ˜˜Economics from a Biological Viewpoint,™™ in Jay B. Barney and William
G. Ouchi (eds.) Organizational Economics. San Francisco: Jossey-Bass, 1987, 319“371.
Homans, George C. ˜˜Social Behavior as Exchange,™™ American Journal of Sociology 65, 6
(May 1958): 597“606.
Hume, David. Essays: Moral, Political and Literary. London: Longmans, Green, 1898(1754).
Keckler, C. N. W. ˜˜Catastrophic Mortality in Simulations of Forager Age-of-Death:
Where Did all the Humans Go?™™ in R. Paine (ed.) Integrating Archaeological Demography:
Multidisciplinary Approaches to Prehistoric Populations. Center for Archaeological Investiga-
tions, Occasional Papers No. 24, 205“228.
Keller, Laurent. Levels of Selection in Evolution. Princeton, NJ: Princeton University Press,

Kelly, Raymond C. The Nuer Conquest: The Structure and Development of an Expansionist
System. Ann Arbor: University of Michigan Press, 1985.

Knauft, Bruce. ˜˜South Coast New Guinea Cultures: History, Comparison, Dialectic,™™
Cambridge Studies in Social and Cultural Anthropology 89 (1993).
Ledyard, J. O. ˜˜Public Goods: A Survey of Experimental Research,™™ in J. H. Kagel and
A. E. Roth (eds.) The Handbook of Experimental Economics. Princeton, NJ: Princeton Univer-
sity Press, 1995, 111“194.
Loewenstein, George. ˜˜Experimental Economics from the Vantage Point of View of
Behavioural Economics,™™ Economic Journal 109, 453 (February 1999): F25“F34.
Lumsden, C. J., and E. O. Wilson. Genes, Mind, and Culture: The Coevolutionary Process.
Cambridge, MA: Harvard University Press, 1981.
Maynard Smith, John. ˜˜Group Selection,™™ Quarterly Review of Biology 51 (1976): 277“283.
Monroe, Kristen Renwick. The Economic Approach to Politics. Reading, MA: Addison Wes-
ley, 1991.
Moore, Jr., Barrington. Injustice: The Social Bases of Obedience and Revolt. White Plains:
M. E. Sharpe, 1978.
Orbell, John M., Robyn M. Dawes, and J. C. Van de Kragt. ˜˜Organizing Groups for Col-
lective Action,™™ American Political Science Review 80 (December 1986): 1171“1185.
Ostrom, Elinor, James Walker, and Roy Gardner. ˜˜Covenants with and without a Sword:
Self-Governance Is Possible,™™ American Political Science Review 86, 2 ( June 1992): 404“417.

Posner, Richard. Economic Analysis of Law. New York: Little, Brown, 1973.
Price, G. R. ˜˜Selection and Covariance,™™ Nature 227 (1970): 520“521.
”””. ˜˜Extension of Covariance Selection Mathematics,™™ Annals of Human Genetics 35
(1972): 485“490.
Richerson, Peter J., and Robert Boyd. The Nature of Cultures. Chicago: University of Chi-
cago Press, 2003.
Moral Sentiments and Material Interests 39

Roof, Wade Clark, and William McKinney. American Mainline Religion: Its Changing Shape
and Future. New Brunswick, NJ: Rutgers University Press, 1987.
Roth, Alvin E., Vesna Prasnikar, Masahiro Okuno-Fujiwara, and Shmuel Zamir. ˜˜Bar-
gaining and Market Behavior in Jerusalem, Ljubljana, Pittsburgh, and Tokyo: An Experi-
mental Study,™™ American Economic Review 81, 5 (December 1991): 1068“1095.
Sato, Kaori. ˜˜Distribution and the Cost of Maintaining Common Property Resources,™™
Journal of Experimental Social Psychology 23 ( January 1987): 19“31.
Smith, Adam. The Theory of Moral Sentiments. Indianapolis: Liberty Fund, 1982(1759).
Smith, Adam. The Wealth of Nations. New York: Prometheus Books, 1991(1776).
Sober, Elliot, and David Sloan Wilson. Unto Others: The Evolution and Psychology of Unself-
ish Behavior. Cambridge, MA: Harvard University Press, 1998.
Soltis, Joseph, Robert Boyd, and Peter J. Richerson. ˜˜Can Group-functional Behaviors
Evolve by Cultural Group Selection: An Empirical Test,™™ Current Anthropology 36, 3
( June 1995): 473“483.
Stephens, W., C. M. McLinn, and J. R. Stevens. ˜˜Discounting and Reciprocity in an Iter-
ated Prisoner™s Dilemma,™™ Science 298 (13 December 2002): 2216“2218.
Taylor, Michael. Anarchy and Cooperation. London: John Wiley and Sons, 1976.
Titmuss, R. M. The Give Relationship. London: Allen and Unwin, 1970.
Trivers, R. L. ˜˜The Evolution of Reciprocal Altruism,™™ Quarterly Review of Biology 46
(1971): 35“57.
Williams, G. C. Adaptation and Natural Selection: A Critique of Some Current Evolutionary
Thought. Princeton, NJ: Princeton University Press, 1966.
Wilson, Edward O. Sociobiology: The New Synthesis. Cambridge, MA: Harvard University
Press, 1975.
Yamagishi, Toshio. ˜˜The Provision of a Sanctioning System as a Public Good,™™ Journal of
Personality and Social Psychology 51 (1986): 110“116.
”””. ˜˜The Provision of a Sanctioning System in the United States and Japan,™™ Social
Psychology Quarterly 51, 3 (1988a): 265“271.

”””. ˜˜Seriousness of Social Dilemmas and the Provision of a Sanctioning System,™™
Social Psychology Quarterly 51, 1 (1988b): 32“42.
”””. ˜˜Group Size and the Provision of a Sanctioning System in a Social Dilemma,™™ in
W. B. G. Liebrand, David M. Messick, and H. A. M. Wilke (eds.) Social Dilemmas: Theoret-
ical Issues and Research Findings. Oxford: Pergamon Press, 1992, 267“287.
II The Behavioral Ecology of
2 The Evolution of
Cooperation in Primate

Joan B. Silk

Primates do not donate to National Public Radio or give blood, but
they do perform a variety of altruistic behaviors. That is, they act in
ways that reduce their own ¬tness, but increase the ¬tness of their
partners. For example, male chimpanzees form alliances and patrol
the borders of their territories, sometimes launching lethal attacks on
members of other communities (Goodall et al. 1979; Nishida, Hiraiwa-
Hasegawa, and Takahata 1985; Boesch and Boesch-Achermann 2000;
Watts and Mitani 2001); vervet monkeys give alarm calls when they
detect predators (Struhsaker 1967; Seyfarth et al. 1980); captive cebus
monkeys and chimpanzees allow others to share their food (de Waal
1997a, 1997b, 2000); macaque females defend juveniles from harass-
ment by other group members (Chapais 1992); langurs and howlers
spend considerable amounts of time carrying other females™ infants
(Paul 1999); and monkeys in a number of species spend ten to twenty
percent of their waking hours removing dirt, debris, and ectoparasites
from the hair and skin of other group members (Dunbar 1991).
Over the last twenty-¬ve years, primatologists have collected large
quantities of information about the distribution of these charitable
activities. Evolutionary theory predicts that altruism will occur when
bene¬ts increase the actor™s own inclusive ¬tness (Hamilton 1964) or
when bene¬ts are exchanged by reciprocating partners (Trivers 1971;
Axelrod and Hamilton 1981). Thus, examinations of kinship and reci-
procity have dominated efforts to account for the distribution of al-
truistic behavior among primates (Gouzoules and Gouzoules 1987;
Dugatkin 1997; Silk 1987, 2002). Data that do not conform to predic-
tions derived from these models have been discounted, denied, or sim-
ply ignored because they do not ¬t into our theoretical paradigms.
However, empirical and theoretical work in experimental economics
suggests that humans cooperate when standard evolutionary theory
44 Silk

tells us that they should not. Efforts to develop systematic explanations
of human behavior that explain these anomalies have generated new
models of the motives that give rise to human cooperation, including
strong reciprocity (Gintis 2000, this volume).
The goal of this chapter is to review what we know about the evolu-
tionary forces that underlie cooperation in primate groups and to eval-
uate the possibility that the motives that give rise to strong reciprocity
in humans also produce cooperation in primate groups. The literature
provides very strong evidence that kin selection plays a fundamental
role in the lives of nonhuman primates”shaping social organization,
dispersal strategies, dominance hierarchies, and patterning of af¬lia-
tive interactions. There is reasonably good evidence of reciprocity and
interchange within nonhuman primate dyads, but very little system-
atic evidence of punishment. Experimental studies indicate that coop-
eration is contingent on the nature of previous interactions among
partners, but the proximate mechanisms that generate these contingen-
cies are largely unknown.
Analyses of the evolutionary mechanisms underlying cooperation in
primates rely on assumptions about the relative magnitude and nature
of the effects of these kinds of behaviors on individual ¬tness. In fact, it
is virtually impossible to quantify the effects of a single behavioral act
or social interaction on lifetime ¬tness. This problem is common to al-
most all studies of the adaptive function of social behavior in animals.
We rely on what Grafen (1991) calls the ˜˜phenotypic gambit,™™ the as-
sumption that the short-term bene¬ts that individuals derive from
social interactions are ultimately translated into long-term differences
in ¬tness. Animals who are regularly supported in agonistic con-
frontations, protected from harassment, or allowed to share access to
desirable resources are expected to gain short-term bene¬ts that are
ultimately translated into ¬tness gains.

2.1 The Evolution of Altruism by Kin Selection

In general, natural selection favors the evolution of behaviors that
increase an individual™s relative ¬tness. Altruistic behaviors that re-
duce individual ¬tness contradict this logic. The theory of kin selection,
developed by the late W. D. Hamilton, relies on the insight that rela-
tives share some of their genetic material because they have a common
ancestor (Hamilton 1964). If individuals behave altruistically toward
their relatives, then they have some chance of conferring bene¬ts upon
The Evolution of Cooperation in Primate Groups 45

individuals who carry copies of their own genes. The likelihood of this
happening is based upon the genetic relationship between the actor
and the recipient. Hamilton demonstrated that altruistic behaviors will
be favored by selection when the costs of performing the behavior, c,
are less than the bene¬ts, b, discounted by the coef¬cient of relatedness
between actor and recipient, r. The coef¬cient of relatedness is the av-
erage probability that two individuals acquire the same allele through
descent from a common ancestor. This principle, which is generally
called Hamilton™s Rule, is stated as: rb > c.
Two basic insights can be derived from Hamilton™s Rule. First, it is
clear that when r ¼ 0, this inequality cannot be satis¬ed. This means
that unconditional altruism (via kin selection) will be restricted to kin
(r > 0). Second, costly altruism will be limited to close kin, as the con-
ditions for Hamilton™s Rule become progressively more dif¬cult to
satisfy as costs rise. During the 1950s, the famous British evolutionary
biologist, J. B. S. Haldane, is said to have worked through these simple
calculations on the back of an envelope one evening in a pub and
announced that he would give up his life to save two brothers or eight
Multi-level selection models (Wilson 1997) provide an alternative
mathematical representation of the processes that underlie Hamilton™s
model of kin selection. In the inclusive ¬tness approach, ¬tness
effects are accounted to the bodies in which the genes causing the
effects are expressed, while in the multi-level selection approach,
¬tness effects are partitioned into within-group and between-group
components (Reeve and Keller 1999). The two approaches are math-
ematically equivalent, but their heuristic value may vary in different

2.2 Kin Recognition

The coef¬cient of relatedness, r, is the critical element for determining
an adaptive course of action in social interactions that are in¬‚uenced
by kin selection (Hamilton 1987). In order to meet the conditions of
Hamilton™s Rule, animals must make sure that they limit altruistic
behavior toward kin (r > 0). For species in which kin are clustered in
discrete locations, such as burrows or nests, spatial location may pro-
vide suf¬cient information for kin discrimination (Blaustein, Bekoff,
and Daniels 1987). For other animals, however, the problem is more
complicated. Hamilton (1987) predicted that the ability to identify
46 Silk

kin would be most fully developed in species that live in social
groups”when there are opportunities for costly behaviors (such as
egg dumping), and when passive, context-dependent mechanisms for
distinguishing kin from nonkin are not likely to be effective.
Primates ¬t all three of these conditions. Most primates live in large
and relatively stable social groups (Smuts et al. 1987). Even the most
solitary primates, like orangutans and galagos, have regular interac-
tions with familiar conspeci¬cs (Bearder 1987; Galdikas 1988), and kin-
ship may structure their communities (Wimmer, Tautz, and Kappeler
2002; Radespiel et al. 2003). Primates engage in a variety of ¬tness-
reducing behaviors, including infanticide (van Schaik and Janson
2000); severe intragroup aggression (McGrew and McLuckie 1986);
and intense feeding competition (Dittus 1979, 1988). Most primates
live for extended periods of time in groups that include both relatives
and nonrelatives, so context-driven mechanisms for distinguishing kin
are likely to be of limited use. Thus, primates are expected to exhibit
¬nely developed kin recognition abilities.
A number of different perceptual mechanisms underlie kin recogni-
tion in animal species. For example, sea squirts are able to recognize
other sea squirts that carry the same allele on the hypervariable his-
tocompatability locus (Pfennig and Sherman 1995). Some animals, in-
cluding most mammals, are thought to learn who their relatives are
during the course of development, drawing cues about kinship from
patterns of association and interactions.
Close association early in life is generally thought to be the basis for
kin recognition in primate groups (Bernstein 1991; Walters 1987). Nep-
otistic biases in association and interaction provide accurate and useful
cues that monkeys use to identify their maternal relatives. An infant
may learn who its relatives are by observing its mother™s pattern of in-
teraction and association with other group members. Similarly, a juve-
nile learns who its younger siblings are by watching its mother interact
with her newborn infants.
Early association allows for recognition of maternal kin, but not
paternal kin. Close associations between males and females are uncom-
mon in most nonmonogamous primate species, limiting infants™ abili-
ties to learn who their fathers are. Other proxies for paternity are
prone to error. For example, in multi-male species, male rank is often
correlated with reproductive success, but the association is far from
perfect. In some species that form one-male groups, such as patas and
blue monkeys, incursions by nonresident males may occur during the
The Evolution of Cooperation in Primate Groups 47

mating season (Cords 1987). Even in pair-bonded species, like gibbons
and callicebus monkeys, females sometimes mate with males from out-
side their groups (Mason 1966; Palombit 1994; Reichard 1995).
For species in which a single male monopolizes mating opportu-
nities, age may be a good cue of paternal kinship (Altmann 1979).
Among baboons in Amboseli, Kenya, high-ranking males monopolize
access to females (Altmann et al. 1996) and agemates are therefore
likely to be paternal half-siblings. Adult females interact at higher
rates with agemates than others, generating signi¬cant differences in
the rate of interactions between paternal half-siblings and unrelated
females (Smith, Alberts, and Altmann 2003). Similar patterns character-
ize female rhesus macaques on Cayo Santiago (Widdig and Nurnberg
2001). In Amboseli, male baboons also recognize their own offspring.
Adult males selectively support their own genetic offspring in agonis-
tic encounters (Buchan et al. 2003). It is not clear what cues males use
to identify their offspring. They may rely on their previous mating his-
tory, females™ responses to males after they give birth, phenotypic cues,
or some combination of these factors.
There is also tantalizing evidence that monkeys and apes may actu-
ally be able to recognize paternal kin based on phenotypic cues alone.
In baboons and rhesus macaques, females distinguished among age-
mates, showing slight preferences for paternal half-siblings over non-
kin (Smith, Alberts, and Altmann 2003; Widdig and Nurnberg 2001).

2.3 Social Organization Facilitates Kin Selection

The structure of social groups in many primate species facilitates the
evolution of cooperation via kin selection (Silk 2002). Virtually all mon-
keys and apes live in stable social groups. Primate infants are com-
pletely dependent on their mothers (and sometimes their fathers) for
support at birth, but become gradually more independent as they ma-
ture. Bonds between mothers and their offspring commonly continue
beyond weaning, which marks the end of nutritional dependence. In
some species, such as pair-bonded siamangs and owl monkeys, fathers
are active participants in offspring care. In some species, such as mar-
mosets and tamarins, older offspring act as ˜˜helpers at the nest™™ and
their support enhances parental reproductive success (Garber 1997).
Extended family ties are presumably the product of kin selection.
Dispersal patterns play an important role in the evolution of cooper-
ation via kin selection. In all primate species, members of one or both
48 Silk

sexes disperse from their natal groups (Pusey and Packer 1987). While
natal dispersal presumably evolved to prevent inbreeding (Pusey and
Wolf 1996), the patterns of dispersal may re¬‚ect selective pressures that
favor kin-selected altruism (Wrangham 1980). In many primate spe-
cies, members of only one sex (usually males) disperse, while members
of the other sex remain in their natal group throughout their lives
(Pusey and Packer 1987). When only one sex disperses, members of
the nondispersing (philopatric) sex live among kin of varying degrees
of relatedness. Thus, in baboon, macaque, and vervet groups, females
grow up within a complex network of maternal and paternal kin:
mother, grandmother, sisters, brothers, aunts, uncles, and cousins.
In these species, maternal kin spend much of their time in close prox-
imity, and virtually all behaviors that are generally classi¬ed as altruis-
tic, including grooming, food sharing, benign alloparenting, and alarm
calling show matrilineal kin biases (reviewed by Bernstein 1991; Silk
1987; Gouzoules and Gouzoules 1987; Walters 1987; Silk 2002). We do
not know whether the distribution of these behaviors ¬ts predictions
derived form Hamilton™s Rule because the costs and bene¬ts associ-
ated with these behaviors have not been measured. Nonetheless, the
matrilineal bias in social behavior seems likely to be the product of kin
More compelling evidence comes from studies of coalition forma-
tion”interactions in which one individual intervenes on behalf of an-
other in an ongoing agonistic interaction. Monkeys who intervene
in ongoing disputes put themselves at some risk, as monkeys are
equipped with sharp teeth that they sometimes use to bite their
opponents. Primates can be wounded in these disputes, sometimes
seriously. Thus, coalitions provide ˜˜the clearest evidence of primates
engaging in behavior that bene¬ts another at some risk and/or cost to
self™™ (Bernstein 1991).
Monkeys, particularly females, often intervene in ongoing disputes
in support of their relatives. Females are signi¬cantly more likely to
support kin than nonkin in aggressive disputes (Berman 1983a, 1983b,
1983c; Chapais 1983; Cheney 1983; Datta 1983a, 1983b; Kaplan 1977,
1978; Kurland 1977; Massey 1977; Silk 1982; Silk, Alberts, and Altmann
2004), particularly against higher ranking opponents (Chapais 1983,
Chapais, Girard, and Primi 1991; Cheney 1983; Hunte and Horrocks
1987; Kurland 1977; Netto and van Hooff 1986; Pereira 1989; Silk 1982;
Walters 1980; Watanabe 1979). Since allies run some risk of being
threatened, chased, attacked, or injured when they intervene against
The Evolution of Cooperation in Primate Groups 49

higher-ranking monkeys, females are evidently willing to take greater
risks on behalf of kin than on behalf of nonkin.
Support has both short-term and long-term consequences. In the
short term, animals that obtain support are more likely to win dis-
putes and less likely to become involved in escalated attacks. In the
long term, support facilitates rank acquisition (see Chapais 1992 for
a detailed analysis of this process). Infants are protected by their
mothers and close female kin when they are threatened by other group
members, particularly females that are of lower rank than their own
mothers (Berman 1980; Datta 1983a; Cheney 1977; de Waal 1977; de
Waal and Luttrell 1985; Horrocks and Hunte 1983; Johnson 1987; Lee
1983a, 1983b; Lee and Oliver 1979; Paul and Kuester 1987; Pereira
1989; Walters 1980). As they grow older, young juveniles obtain sup-
port when they challenge peers whose mothers are lower-ranking than
their own mothers and when they challenge adults who are subordi-
nate to their own mothers. Initially, juveniles can defeat older and
larger juveniles only when their own mothers are nearby (Datta 1983a,
1983b; Horrocks and Hunte 1983; Walters 1980). Eventually, imma-
tures are able to defeat all group members who are subordinate to their
own mothers, even when their mothers are not in the vicinity. Since
juveniles are able to defeat everyone that their own mothers can defeat
(but not their mothers themselves), offspring acquire ranks just below
their mothers.
The same process, repeated over generations and across families,
generates matrilineal dominance hierarchies in which all members of
the same matriline occupy contiguous ranks. Moreover, all members
of a given matriline rank above or below all the members of other
matrilines. Matrilineal dominance hierarchies have now been docu-
mented in at least seven species of macaques, baboons, and vervet
monkeys (Chapais 1992). These dominance hierarchies are remarkably
linear and stable over time, although the mechanisms that maintain
this stability are not well understood (Silk, Alberts, and Altmann
2004). These arrangements have important ¬tness consequences for
females: high-ranking females typically mature at earlier ages, give
birth to healthier infants, and have shorter interbirth intervals than
low-ranking females (reviewed by Silk 1987, 1993; Harcourt 1987).
Primates are discriminating nepotists. Thus, Japanese macaques and
rhesus macaques treat distant kin much like nonkin (Kapsalis and Ber-
man 1996a; Chapais et al. 1997). It is not clear whether monkeys do not
recognize distant relatives as kin (Kapsalis and Berman 1996a) or if
50 Silk

support for distant kin fails to meet the criteria for altruism speci¬ed
by Hamilton™s Rule.
Nepotism is also contingent on the circumstances. Among Japanese
macaques, younger sisters commonly rise in rank over their older sis-
ters. This process is sometimes contentious, and younger sisters ˜˜tar-
get™™ their older sisters for rank reversals. When females intervene in
disputes involving their older sisters and subordinate nonkin, they are
as likely to intervene against their sisters as they are to support them.
In contrast, when females intervene in con¬‚icts involving kin that
are not targeted for rank reversals, females are much more likely to in-
tervene on behalf of their relatives than their opponents (Chapais,
Prud™homme, and Teijeiro 1994). Thus, females ˜˜apparently solve the
con¬‚ict of interest between egotism and nepotism by maximizing their
own rank among their kin on the one hand, and by maximizing the
rank of their kin in relation to non-kin on the other™™ (Chapais 1995:
When females disperse and males remain in their natal group, there
are parallel opportunities for kin-selected altruism among males. Male
philopatry is associated with strong male bonds among chimpanzees
(Goodall 1986), muriquis (Strier 1992, 2000), spider monkeys (Syming-
ton 1990), Costa Rican squirrel monkeys (Boinski 1994), and in some
populations of red colobus monkeys (Struhsaker 2000; but also see
Starin 1994). For those interested in the evolutionary roots of human
behavior, male bonding in chimpanzees is of particular interest. Male
chimpanzees spend much of their time in the company of other males.
They groom one another, hunt together, share meat, and collectively
patrol the borders of their territories (Goodall 1986; Mitani, Merri-
wether, and Zhang 2000; Simpson 1973; Watts 2000; Wrangham and
Smuts 1980). In some populations, pairs or trios jointly control access
to receptive females and share matings (Watts 1998).
Primatologists have generally assumed that kin selection underlies
male cooperation, but af¬liative and cooperative behavior is not linked
to matrilineal kinship in two Ugandan chimpanzee communities (Gold-
berg and Wrangham 1997; Mitani, Merriwether, and Zhang 2000).
However, there is a strong tendency for males to form close ties to age-
mates (Mitani et al. 2002). Thus, it seems possible that paternal kinship
could underlie cooperative activity among chimpanzees.
When both sexes disperse, opportunities for kin selection to operate
are more limited, but may still be important. Red howlers provide a
particularly compelling example of this phenomenon. The number
of females in red howler groups is con¬ned within narrow limits”
The Evolution of Cooperation in Primate Groups 51

groups with too few females are unable to defend their territories,
while groups with too many females face competition for food and be-
come more attractive targets for male takeovers, which leads to infanti-
cide (Pope 2000b).
Therefore, when groups reach the optimal size, maturing females
must disperse. Dispersal is very costly for females, particularly when
local habitats are fully saturated. Some females never succeed in estab-
lishing new groups and those that do succeed begin to reproduce later
than females who remain in their natal groups. The high costs of dis-
persal generate intense competition among females over recruitment
opportunities for their daughters. Adult females actively harass matur-
ing females in an effort to force them to emigrate. Females actively in-
tervene on behalf of their daughters in these contests (Crockett 1984;
Crockett and Pope 1993). In most cases, ˜˜only the daughters of a single
presumably dominant adult female are successful at remaining to
breed™™ (Pope 2000a).
Kin selection also shapes the life histories of male red howlers. Males
gain access to breeding females in a variety of ways. When habitats are
not crowded, they may join up with migrant females and help them es-
tablish new territories. But as habitats become more saturated, males
can only gain access to breeding females by taking over established
groups and evicting male residents. This is a risky strategy because
males are often injured in takeover attempts (Crockett and Pope 1988).
Moreover, males tend to remain in their natal groups longer, helping
their fathers resist takeover attempts. Thus, when habitats are satu-
rated, single males are at a distinct disadvantage in obtaining access to
breeding females.
Competition among males generates powerful incentives for cooper-
ation. Thus, single males form coalitions and cooperate in efforts to
evict male residents from bisexual groups. After they have established
residence, males collectively defend the group against incursions by
extragroup males. However, cooperation involves clear ¬tness costs
because only one male fathers infants within the group. Not surpris-
ingly, coalitions that are made up of related males last on average
8.2 years, while coalitions among unrelated males last only 2.3 years
(Pope 1990). Coalitions composed of kin are also less likely to experi-
ence the dominance changes that often lead to infanticide than are coa-
litions composed of unrelated males (Pope 1990).
In summary, there seems to be little doubt that kin selection plays
an important role in the evolution of cooperation in primate groups.
Our efforts to evaluate the extent of kin selection are limited by the
52 Silk

dif¬culty of quantifying the effects of social behavior on ¬tness and
our limited knowledge of paternal kinship.

2.4 Reciprocity in Primate Groups

Reciprocal altruism provides another vehicle for cooperation in pri-
mate groups (Axelrod and Hamilton 1981; Trivers 1971). Primates
easily meet the necessary conditions for reciprocal altruism: they recog-
nize their partners as individuals and have frequent opportunities to
interact with group members. Moreover, they seem to be able to moni-
tor and remember their partners™ responses and adjust their subse-
quent behavior accordingly.
Although primates are prime candidates for reciprocal altruism,
there is much less evidence of reciprocity than of kin selection
(Seyfarth and Cheney 1988; Noe and Hammerstein 1995). This may be
due to the fact that it is dif¬cult to detect reciprocal altruism in nature
(Seyfarth and Cheney 1988). We can tabulate the frequency and dura-
tion of services performed within dyads, but we cannot translate these
values directly into ¬tness units and calculate the balance between
bene¬ts given and received. This is particularly complicated when
exchanges involve different currencies or when reciprocity is delayed
over time. Even if we ¬nd tight associations between altruism given
and received among partners, it is possible that the association is caus-
ally linked to a third variable that we have not taken into account, such
as kinship (Hemelrijk and Ek 1991). In naturalistic settings, it is often
dif¬cult to determine whether the delivery of bene¬ts is contingent on
Much of what primatologists have written about reciprocity involves
grooming. Grooming is an obvious candidate for reciprocal exchanges
because it is common and involves complementary roles”I™ll scratch
your back if you scratch mine. Grooming is the most common form of
social behavior among nonhuman primates, occupying up 20 percent
of every day (Dunbar 1991). The functions of grooming are not fully
understood. Grooming is thought to be bene¬cial to the recipient be-
cause ectoparasites”such as ticks, lice, and bot¬‚ies”are removed and
wounds are cleaned (Saunders 1988; Henzi and Barrett 1999). This sug-
gests that grooming would be concentrated on regions of the body that
animals cannot reach themselves, and this is often the case (Perez and
Vea 2000). However, grooming solicitations do not correspond per-
fectly to accessibility, and this suggests that other factors may also be
The Evolution of Cooperation in Primate Groups 53

in play. Anyone observing monkeys grooming one another would sus-
pect that grooming is intensely pleasurable”animals who are being
groomed seem to be utterly relaxed. In fact, grooming lowers heart
rates and raises levels of beta-endorphins (Aureli and Smucny 2000).
Grooming may also have social functions (Dunbar 1988, 1991), pro-
viding a means to reinforce social bonds and cultivate valuable social
While grooming seems to be bene¬cial to recipients, those who
provide these services incur some costs. At the very least, the groomer
expends time and energy in servicing its partner. The groomer may
also become more vulnerable to attacks by predators or other group
members because vigilance is reduced during grooming (Cords 1995;
Maestripieri 1993).
If grooming is the product of reciprocal altruism, then grooming
(among nonkin) should be limited to reciprocating partners. Several
lines of evidence suggest that this may be the case. First grooming in
large groups is restricted to a relatively limited subset of potential part-
ners. For example, female baboons in the Okavango Delta of Botswana
groomed on average only eight of the other eighteen adult females in
their group; and most females concentrated their grooming on an even
smaller number of females (Silk, Cheney, and Seyfarth 1999). In gen-
eral, the extent of selectivity is related to the number of available part-
ners. In small groups, females distribute their grooming evenly across
the group, but as groups grow larger, grooming is less evenly allocated
across potential partners (Silk, Cheney, and Seyfarth 1999). This may
re¬‚ect cognitive constraints on females™ ability to keep track of large
numbers of relationships (Henzi and Barrett 1999) or ecological con-
straints that limit the amount of time that females can afford to spend
grooming (Dunbar 1991; Henzi, Lycett, and Weingrill 1997).
Is grooming reciprocated? A de¬nitive answer to this question is
surprisingly elusive. Among male chimpanzees, there are positive cor-
relations between the amount of grooming given and received, but
grooming is not evenly balanced within most dyads (Watts 2000).
There are also cases in which grooming is evenly balanced within
the majority of dyads. Thus, adult female baboons in the Okavango
Delta tended to groom each of their partners as often as their partners
groomed them (Silk, Cheney, and Seyfarth 1999). Similarly, in white-
faced capuchins, grooming is evenly balanced within the majority of
dyads (Manson et al. 1999). In some cases, grooming roles are alter-
nated within bouts (Barrett et al. 1999; Muroyama 1991), but others in
54 Silk

which grooming tends to be reciprocated over longer time periods
(Manson et al. 2004).
There are many groups in which grooming is unbalanced within
dyads, and disparities in grooming given and received are sometimes
linked to dominance rank. In some groups, high-ranking partners re-
ceive more grooming than they give each of their partners (Chapais
1983; Fairbanks 1980; Seyfarth 1980; Silk 1982; Sambrook, Whiten, and
Strum 1995; Stammbach 1978; Watts 2000; Manson et al. 2004), in other
groups high-ranking partners give more grooming than they receive
in return (Altmann, Myles, and Combes 1998; O™Brien 1993; Di Bitetti
1997; Linn et al. 1995; Parr et al. 1997). Most primatologists assume
that these imbalances exist because grooming is exchanged for other
commodities such as coalitionary support (Seyfarth 1977), food (de
Waal 1997a), tolerance (Silk 1982; Fairbanks 1980), access to attractive
infants (Muroyama 1994; Henzi 2001), or maintaining group cohesion
(Altmann, Myles, and Combes 1998).
Seyfarth (1977) was the ¬rst to suggest that monkeys might ex-
change grooming for support in agonistic interactions. His argument
was based on the notion that high-ranking animals make powerful co-
alition partners. He reasoned that females might groom higher-ranking
monkeys who would in return provide support for them when they
were harassed by other group members. Grooming and support are
positively correlated among vervets in Amboseli (Seyfarth 1980) and
white-faced capuchins in Costa Rica (Perry 1996). However, kinship
was not known for these two groups, and the observed correlation be-
tween grooming and support might actually arise because females
selectively support and groom close kin (Hemelrijk and Ek 1991). This
is apparently the case among rhesus macaques on Cayo Santiago
where grooming and support are correlated among related females
but not among unrelated females (Kapsalis and Berman 1996b). How-
ever, grooming and support are correlated among male bonnet ma-
caques (Silk 1992) and male chimpanzees (Mitani, Merriwether, and
Zhang 2000) and these results are not confounded by maternal kinship.
Researchers have failed to ¬nd consistent associations between groom-
ing and support in several cases (Fairbanks 1980; Silk 1982; de Waal
and Luttrell 1986; Silk, Alberts, and Altmann 2004).
Although the naturalistic data provide only tepid support for Sey-
farth™s model, two experimental studies demonstrate a direct link
between grooming and support in Old World monkeys. Using tape-
recorded vocalizations of females™ screams, which signal distress and
are often used to recruit support, Seyfarth and Cheney (1984) showed
The Evolution of Cooperation in Primate Groups 55

that free-ranging vervet females were more attentive to screams of
unrelated females if they had been groomed by the screaming female
shortly before they heard the scream than if they had not been
groomed by her. Similarly, Hemelrijk (1994) arti¬cially induced ¬ghts
among unrelated female macaques housed temporarily in groups of
three. When ¬ghts between two females occurred, aggressors some-
times received support from the third female. Support was more likely
to be given to the aggressor if she had previously groomed the poten-
tial supporter.
Grooming may also be used to obtain other valuable bene¬ts. Fe-
male bonnet macaques are less likely to be harassed while they are
grooming higher-ranking females than when they are grooming lower-
ranking females (Silk 1982), and grooming may confer protection. Fe-
male monkeys may also use grooming to obtain access to infants. For
reasons that are not altogether clear, female monkeys are strongly
attracted to newborn infants (Paul 1999; Maestripieri 1994; Silk 1999).
Females gather around new mothers, attempting to smell, nuzzle,
touch, and inspect the genitals of newborn infants. Macaque and ba-
boon mothers do not seem to welcome this interest in their infants,
even though most of the interactions seem relatively benign. In these
species, new mothers are often approached and groomed at higher
rates than they are at other times (Altmann 1980) and some researchers
suggest that females trade grooming for access to newborn infants
(Muroyama 1994; Henzi 2001).

2.5 Food Sharing

Food sharing plays a fundamental role in the organization of tradi-
tional human societies (Foley 1987). While gathered foods are gener-
ally redistributed only to family members, meat is typically shared
with all members of the group. In primates, which rely mainly on plant
foods, food sharing is generally uncommon and limited to offspring
(Foley 1987; McGrew 1992). Chimpanzees represent a major exception
to this rule”males hunt regularly and successfully, and share ac-
cess to their kills (Boesch and Boesch-Achermann 2000; Goodall 1986;
Mitani and Watts 2001; Stanford et al. 1994). This has generated con-
siderable interest in the dynamics of hunting and food sharing among
In chimpanzees, hunting is usually a collective activity. At some sites,
hunters take different roles in stalking, ambushing, and snatching prey
(Boesch 1994; Boesch and Boesch 1989; Boesch and Boesch-Achermann
56 Silk

2000). At other sites, hunting involves no obvious coordination (Stan-
ford 1996; Busse 1978; Goodall 1986; Uehara et al. 1992). Surprisingly,
there is little consensus about why male chimpanzees hunt. In some
primates, predatory activity increases when plant foods become scarce
(Dunbar 1983; Foley 1987). Chimpanzees rely heavily on ripe fruit, and
they may hunt to compensate for seasonal shortages of their preferred
foods (Teleki 1973; Takahata, Hasegawa, and Nishida 1984; Stanford
1996, 1998). In contrast, at Ngogo males hunt most when food is most
abundant (Watts and Mitani 2002).
Hunting seems to have a social component as well. Males are most
likely to hunt when they are in large groups, and hunting success gen-
erally increases with party size (Stanford 1996; Watts and Mitani 2002).
This suggests that males may hunt to obtain meat that they can trade
for sexual access to females (Stanford 1996, 1998; Stanford et al. 1994)
or they may use meat to cultivate social bonds with other males
(Nishida et al. 1992; Boesch and Boesch-Achermann 2000; Mitani and
Watts 2001).
Careful analyses of the distribution of fruit, hunting effort, and food
sharing at Ngogo, a site in the Kibale Forest of Uganda (Mitani and
Watts 2001) suggest that hunting may enhance the quality of social
bonds among males. In Ngogo, chimpanzees hunt most often when
fruit is most abundant (Watts and Mitani 2002), ruling out the possibil-
ity that males hunt to compensate for food shortages. Males did not
share selectively with sexually receptive females and receptive females
did not mate selectively with males who shared food with them, sug-
gesting that males do not trade meat for sex at Ngogo. However, males
did share meat selectively with males who shared meat with them and
with males who regularly supported them in agonistic interactions.
Moreover, males who hunt together also tend to groom one another
selectively, support one another, and participate in border patrols to-
gether (Mitani, Merriwether, and Zhang 2000). Frequent participation
in border patrols is, in turn, linked to male mating success (Watts and
Mitani 2001). It is not clear whether the patterns detected at Ngogo
characterize chimpanzees at other sites.
In captivity, food sharing extends to provisioned plant foods. De
Waal (1997a) observed chimpanzees for several hours before and after
they were fed fresh cuttings of leaves and branches, delicacies that
the chimpanzees clearly relished. Those that possessed leaves and
branches were more likely to share their booty with animals that had
previously groomed them than with animals who had not groomed
The Evolution of Cooperation in Primate Groups 57

them in the past few hours. Moreover, if there had been no grooming
before provisioning, the possessor was more likely to respond aggres-
sively to efforts to take food from their pile. The possessor™s largesse
was not simply a result of being groomed”the chimps limited their
generosity to the animals that had just groomed them. Furthermore,
the possessor™s attitude was not simply a re¬‚ection of the quality of
the relationship between the two animals”the chimps were more
likely to share with those that groomed them than those that they had
groomed themselves. However, the magnitude of the effect of prior
grooming was in¬‚uenced by the nature of the relationship between
the two individuals”for pairs that rarely groomed, sharing was
strongly contingent on recent grooming, while for pairs that groomed
at higher rates, recent grooming had a smaller impact on sharing.
De Waal and his colleagues have also studied the mechanisms
underlying food sharing in captive capuchin monkeys. In one set of
experiments, a pair of familiar monkeys was held in adjacent cages
separated by wire mesh (de Waal 1997a). The holes in the mesh were
large enough to allow the monkeys to reach into the adjacent cage and
take food items. The experimental design was simple. First, one mon-
key was given food. Later, the other monkey was given food. All trans-
fers of food in both phases of the experiment were monitored by the
In this experimental situation, a considerable amount of food
changed hands. Owners virtually never handed food to their partners
or pushed it through the holes in the wire mesh, but they often sat
very near the mesh partition with their food. When they did so, the
monkey in the adjacent cage was able to reach through the wire mesh
and take pieces of food, often from within the owner™s reach and in
plain sight. For some animals, the rate of transfer from the owner to its
partner in the ¬rst phase of the experiment was positively correlated
with the rate of transfer when their roles were reversed in the second
phase of the experiment. However, there was a wide range of values
in the correlation coef¬cients across individuals. Transfer rates were
affected by the quality of social relationships among females, as dyads
that tended to associate frequently and ¬ght infrequently had higher
transfer rates than dyads that associated less often and fought more
De Waal (1997a) initially used the term ˜˜sharing™™ to refer to these
food transfers, but subsequently suggested that ˜˜facilitated taking™™
might be a better label for them (de Waal 2000). He points out that the
58 Silk

capuchins rarely gave food to their partners directly (de Waal 1997b),
even though they did little to protect their food from theft. Thus, capu-
chins may be strongly motivated to be near particular partners, and
food transfers may be an inadvertent side effect of their sociability.
The fact that the quality of social bonds in¬‚uences food transfer rates
suggest that the capuchins may not share in a strictly contingent man-
ner (de Waal 1997b). De Waal (2000) conducted a second set of experi-
ments in which two females housed in adjacent cages were given food
at the same time, but the food items differed in their desirability.
Females spent more time near the mesh partition when a monkey was
in the adjacent cage than when it was empty, but they dropped less
food near the partition when it was occupied by another monkey.
Moreover, females tended to spend less time near the partition (and
within their partner™s reach) when they had more desirable foods than
their partners. Thus, females seem to be drawn to favored companions,
but are also wary of losing desirable food items to them. Observed
rates of food transfer are the product of a compromise between these
competing motivations (de Waal 2000).
Using a different experimental paradigm, de Waal and Berger (2000)
explored capuchins willingness to participate in cooperative tasks. As
in the previous experiment, monkeys were held in adjacent cages sepa-
rated by a wire mesh partition. Here, the monkeys had to pull a coun-
terweighted bar to bring a tray holding a baited food bowl within
reach. De Waal and Berger examined the monkeys™ participation in
this task under three different conditions. In the solo condition, only
one food bowl was baited and a single monkey was able to pull the
bowl to within reach. In the cooperative condition, only one food bowl
was baited, but it required joint action by both monkeys to pull the
bowl within reach. In the mutualistic condition, both bowls were
baited and it required joint action by both monkeys to pull the bowl
within reach. Monkeys were equally successful in the solo and mutual-
istic conditions, pulling the food bowl forward approximately 85 per-
cent of the time. Monkeys succeeded on the cooperative task only 40
percent of the time. However, when monkeys did succeed on the coop-
erative task, more food was transferred than in the successful solo
trials. Moreover, a larger fraction of food transfers were tolerated (in
sight and reach of the owner) than in solo trials.
De Waal and Berger (2000) argue that these experiments show that
˜˜capuchins cooperate even if it is obvious that only one of them, and
which one, will be rewarded,™™ and that that capuchins ˜˜exchange labor
The Evolution of Cooperation in Primate Groups 59

for payment.™™ Yet given the small size of the cages, the capuchins
marked af¬nity for their partners, and the messiness of their eating
habits, both parties may be relatively certain that they will obtain food
if they cooperate in pulling the bowl forward. Moreover, it is not clear
that food transfers re¬‚ect an exchange of payment for labor. Even in
solo trials, some food is transferred and the incremental effects of coop-
eration on food transfers and tolerance is relatively small. In solo trials
seven to nine pieces of food are transferred on average and 58 percent
of those transfers are tolerated by the owner. In cooperative trials,
these numbers increase only slightly”nine to eleven pieces are taken
and 65 percent of these transfers are tolerated.
Primatologists have recently begun to explore the psychological pre-
dispositions that underly exchanges in primate groups. One of the key
assumptions of reciprocity is that animals must be able to evaluate
the value of the commodities or services that are being exchanged.
Brosnan and de Waal (2003) conducted an intriguing experiment to
explore how monkeys assess ˜˜value.™™ In these experiments, capuchins
were trained to exchange tokens for food rewards. When a monkey
handed a token to the experimenter, it was given a piece of food. The
experimenters then conducted a series of trials in which the subjects
observed transfers involving other individuals. In some cases, mon-
keys saw others receive food without any exchange of tokens, and in
some cases they saw other monkeys receive a higher quality food re-
ward than they received themselves when they exchanged tokens for
foods. Monkeys who observed others obtain rewards without ex-
change or obtain higher quality rewards than they received were sig-
ni¬cantly more likely to refuse the food rewards that they obtained
themselves”sometimes ¬‚inging food back at the experimenters. Mon-
keys virtually never refused rewards unless they observed others who
had gotten a better deal. The authors suggest that monkeys displayed
an aversion to inequality, although this interpretation has been ques-
tioned (Henrich 2004). At the very least, the data suggest that monkeys
have some ability to evaluate the value of commodities and react nega-
tively when they perceive that an exchange is disadvantageous to
The psychology underlying exchange has also been explored in cap-
tive tamarins. Hauser et al. (2003) created an experimental paradigm in
which one individual could pull a tool that would provide food for its
partner but no food for itself. The researchers trained several tamarins
to be ˜˜unconditional altruists™™ who always pulled and others to be
60 Silk

˜˜unconditional defectors™™ who never pulled. They paired these trained
animals with untrained animals to determine whether the untrained
tamarins would adjust their behavior in a contingent way. Tamarins
pulled more when paired with unconditional altruists than when
paired with unconditional defectors, indicating that cooperation was
contingent on the behavior of the partner. However, Stevens and
Hauser (2004) emphasize that the tamarins cooperated only half the
time and that cooperation with unconditional altruists declined over
the course of the experiments. They conclude that tamarins do not
˜˜demonstrate robust reciprocity™™ and conclude that ˜˜cognitive limita-
tions such as temporal discounting, numerical discrimination, and
memory make reciprocity dif¬cult for animals™™ including nonhuman

2.6 Evolutionary Mechanisms Underlying Reciprocity in Primates

Balanced exchanges between partners and interchange across curren-
cies are often interpreted as evidence that monkeys practice reciprocal
altruism. De Waal has questioned this interpretation, suggesting that
balanced exchanges might simply arise from mutual tolerance or high
rates of association between partners rather than from contingent
exchanges that require careful record keeping (de Waal and Luttrell
1988; de Waal 1997b; de Waal 2000; de Waal and Berger 2000): ˜˜If
members of a species were to direct aid preferentially to close associ-
ates, a reciprocal distribution would automatically result due to the
symmetrical nature of association™™ (de Waal 2000). De Waal calls this
˜˜symmetry-based reciprocity™™ and suggests that proximity should be
controlled in analyses of reciprocity (de Waal and Luttrell 1988).
There are both logical and empirical reasons to doubt that
symmetry-based reciprocity accounts for the distribution of altruistic
behavior in primate groups. Symmetry-based reciprocity implies that
proximity can be treated as an independent variable that is not affected
by the nature of interactions between individuals. It seems more likely
that association patterns re¬‚ect the nature of af¬liative relationships
between individuals. Thus, animals preferentially associate with those
that tolerate, groom, and help them; they do not preferentially tolerate,
help, and groom those that they just happen to associate with. Second,
it seems unlikely that symmetry-based reciprocity would be stable
against invasion by cheaters. Those who accepted help from close
associates but did not return it would be at a distinct advantage. In
The Evolution of Cooperation in Primate Groups 61

fact, there is no evidence for symmetry-based reciprocity in primate
groups. Signi¬cant correlations between bene¬ts given and received
are maintained, even when proximity is controlled statistically (de
Waal and Luttrell 1988). Moreover, several experimental studies dem-
onstrate contingencies between bene¬ts given and subsequently
received (Seyfarth and Cheney 1984; Hemelrijk 1994; de Waal 1997a,
1997b, 2000).
De Waal™s (2000) observations of ¬‚uctuations in the rate of food
transfer within dyads over the course of successive experiments led
him to suggest that reciprocity may be based on a tendency to mirror
the social predispositions of partners, responding positively to positive
social overtures and negatively to negative social overtures: ˜˜If facili-
tated taking is mediated by such general social predispositions, this
would mean that, rather than keeping track of exact amounts of
given and received food, the monkeys follow a simple tolerance-
breeds-tolerance scheme™™ (de Waal 2000, 260). Attitudinal reciprocity
is assumed to be less cognitively demanding than ˜˜calculated reci-
procity,™™ which relies on precise quanti¬cation of bene¬ts given and
received in different currencies.
Attitudinal reciprocity is analogous to strong reciprocity because
both processes focus on the proximate motives that generate cooper-
ation and assume that reciprocity could occur without concern for
long-term consequences. However, it is not clear how evolution could
sustain attitudinal reciprocity (or strong reciprocity) in primate groups.
It seems likely that individuals who systematically returned somewhat
less than they received would bene¬t at the expense of their partners.
To avoid this, costs and bene¬ts must be translated into affect, a pro-
cess that may hide the calculus of reciprocal altruism, but does not
eliminate it.

2.7 Punishment

Strong reciprocity relies on the tendency to punish noncooperators.
Among nonhuman primates there is considerable evidence of negative
reciprocity. Thus, animals use aggression or other forms of costly sanc-
tions to shape the behavior of group members (Clutton-Brock and
Parker 1995a, 1995b) or to exact revenge (de Waal and Luttrell 1988;
Silk 1992). But there is very little evidence that monkeys and apes use
aggression or negative sanctions to shape the behavior of third parties
or to punish deviation from social norms.
62 Silk

Several researchers have reported episodes of aggressive behavior
that could be interpreted as punishment. In the Mahale Mountains of


. 2
( 13)