The Evolution of Prestige: Freely Conferred Deference as a Mechanism for Enhancing the Benefits of Cultural Transmission

Abstract

This paper advances an "information goods" theory that explains prestige processes as an emergent product of psychological adaptations that evolved to improve the quality of information acquired via cultural transmission. Natural selection favored social learners who could evaluate potential models and copy the most successful among them. In order to improve the fidelity and comprehensiveness of such ranked-biased copying, social learners further evolved dispositions to sycophantically ingratiate themselves with their chosen models, so as to gain close proximity to, and prolonged interaction with, these models. Once common, these dispositions created, at the group level, distributions of deference that new entrants may adaptively exploit to decide who to begin copying. This generated a preference for models who seem generally "popular." Building on social exchange theories, we argue that a wider range of phenomena associated with prestige processes can more plausibly be explained by this simple theory than by others, and we test its predictions with data from throughout the social sciences. In addition, we distinguish carefully between dominance (force or force threat) and prestige (freely conferred deference).

Full text

The evolution of prestige
Freely conferred deference as a mechanism for
enhancing the benefits of cultural transmission
Joseph Henrich
a,
*, Francisco J. Gil-White
b
a
University of Michigan Business School, 701 Tappan Drive, D3276, Ann Arbor, MI 48109-1234, USA
b
Department of Psychology, University of Pennsylvania, 3815 Walnut Street, Philadelphia, PA 19104-6196, USA
Received 5 December 1999; accepted 15 November 2000
Abstract
This paper advances an ``information goods'' theory that explains prestige processes as an emergent
product of psychological adaptations that evolved to improve the quality of information acquired via
cultural transmission. Natural selection favored social learners who could evaluate potential models
and copy the most successful among them. In order to improve the fidelity and comprehensiveness of
such ranked-biased copying, social learners further evolved dispositions to sycophantically ingratiate
themselves with their chosen models, so as to gain close proximity to, and prolonged interaction with,
these models. Once common, these dispositions created, at the group level, distributions of deference
that new entrants may adaptively exploit to decide who to begin copying. This generated a preference
for models who seem generally ``popular.'' Building on social exchange theories, we argue that a
wider range of phenomena associated with prestige processes can more plausibly be explained by this
simple theory than by others, and we test its predictions with data from throughout the social sciences.
In addition, we distinguish carefully between dominance (force or force threat) and prestige (freely
conferred deference). D2001 Elsevier Science Inc. All rights reserved.
Keywords: Status; Prestige; Prestige-biased transmission; Cultural transmission; Social learning; Dual
inheritance theory
Acknowledged expertise attracts, though perhaps only temporarily, what we may term a
following of dependent persons. These persons will be welcomed as a principal source of
prestige Ð as a capital benefit of the hunter's expertise. Nor is this expertise necessarily
* Corresponding author.
E-mail addresses: henrich@umich.edu (J. Henrich), fjgil@psych.upenn.edu (F.J. Gil-White).
Evolution and Human Behavior 22 (2001) 165 ± 196
1090-5138/01/$ ± see front matter D2001 Elsevier Science Inc. All rights reserved.
PII: S 1 090-5138(00)00071-4

reduced or dissipated through having to share it with other persons attached to him (Paine,
1973, p. 308 of the Naskapi).
1. Introduction
This paper explores the evolution and psychology of noncoerced, interindividual, within-
group, human status asymmetries Ð or prestige. We distinguish prestige from dominance and
argue that each has its own distinct psychology, selected for by distinct evolutionary
pressures. In the first part, we explain how the selective forces unleashed by the evolution
of certain forms of cultural transmission generated a prestige psychology that further
enhanced our cultural learning capabilities. In the second part, we test a series of predictions
derived from our evolutionary analysis using data from across the social sciences.
1.1. What is status?
Social scientists have not converged on a common taxonomy of social asymmetries,
classifying them inconsistently and variously as ``status,'' ``prestige,'' ``power,'' or ``dom-
inance'' differences. One scheme divides status into ``legal,'' ``traditional,'' and ``charis-
matic'' (Weber, 1946, 1958). Another into ``prestige,'' ``dominance,'' and ``wealth'' (Goode,
1978). Archaeologists divide status into ``ascribed'' (e.g., chiefdoms and states) and
``achieved'' (e.g., ``big man'' societies; Renfrew & Bahn, 1996, pp. 187± 188). Other scholars
see no more than one status dimension, whether they call it ``status,'' ``power,'' or ``prestige''
(e.g., Leach, 1977, p. 10; Ryckman, Rodda, & Sherman, 1972; Shils, 1970, pp. 424± 427), or
they make finer distinctions but nevertheless claim conceptual unity for the assembled whole
(e.g., Cartwright, 1959; French & Raven, 1959). Many evolutionary scholars tend to see all
human status as homologous to nonhuman dominance (Barkow, 1975, 1989; Ellis, 1995; Hill,
1984a, 1984b), even though some human status processes are absent in nonhumans. This
leads to calling status ``dominance'' even when no force or force threat is involved (e.g.,
Bernhardt, 1997, p. 45; Gibb, 1954, pp. 220±221).
Status can be viewed as either a hierarchy of rewards or as a hierarchy of displays Ð or
both simultaneously. Status as rewards implies a hierarchy of privilege. High status entails
greater access to desirable things, that access typically is not actively resisted by inferiors.
There will be occasional fights, but not typically. If those who usually get their way have to
fight for this every single time, we may speak of a tabulation of frequent winners and losers,
but not of a status hierarchy. For status, properly speaking, we require a relatively stable
acquiescence (begrudging or not) from the ``have-nots'' (Weisfeld & Beresford, 1982). This
understanding is signaled in ethological behavior, such that higher status individuals Ð
identified by the flow of benefits Ð are typically the receivers, rather than the givers, of
deference displays.
Although nonhuman status is still poorly understood, a single process appears at least
strongly predominant: agonism (aggression, intimidation, violence, etc. Ð that is, force or
force threat) The resulting social asymmetries are referred to as ``dominance hierarchies'' in
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196166

the ethological and behavioral ecology literatures. The privileges that accrue to dominant
individuals are (1) in males, preferential reproductive access to females, food, and spaces, as
well as a disproportionate amount of grooming from others; (2) in females, preferential access
to food and spaces, and disproportionate grooming. Despite some controversy, the evidence
suggests that dominance correlates with fitness (Cowlishaw & Dunbar, 1991; Ellis, 1995).
The stability of dominance is often reinforced through ``reminders'': submissive behaviors
(e.g., grooming, submissive displays, yielding space, etc.) from subordinate to superior,
whether or not induced through intimidation by the latter.
In humans, in contrast, status and its perquisites often come from nonagonistic sources Ð
in particular, from excellence in valued domains of activity, even without any credible claim
to superior force. For example, paraplegic physicist Stephen Hawking Ð widely regarded as
Einstein's heir, and current occupant of Newton's chair at Cambridge University Ð certainly
enjoys very high status throughout the world. Those who, like Hawking, achieve status by
excelling in valued domains are often said to have ``prestige.''
In the Amazon, several researchers have observed two avenues to status and leadership in
small-scale societies: ``force'' and ``persuasion'' (Krackle, 1978). ``Forceful'' leaders are
domineering headmen who maintain their position through fear, threat, and compulsion (see
also Maybury-Lewis, 1965, pp. 215± 240, 1967, pp. 175 ±178). ``Persuasive'' leaders depend
on their influence and the consent of their followers and lack the force to obligate (see also
Arvelo de Jime
Ânez, 1971, pp. 239± 243; Clastres, 1998; Goldman, 1979; Huxley, 1966, pp.
66±73; Le
Âvi-Strauss, 1944). These two styles of leadership, involving either persuasion or
force, correspond to our two types of status: prestige and dominance.
1.2. Summary of our argument
The evolution of the human cultural capacity Ð that is, for intergenerationally stable, high
fidelity, social transmission Ð created a new selective environment in which mutations
improving the reproductive benefits of such transmission were favored. Our ancestral
psychology evolved (within physical and phylogenetic constraints) into an increasingly
well-organized and specialized battery of biases jointly designed to extract reproductive
benefit from the flow of socially transmitted information. Prestige processes emerge from this
evolved social learning psychology.
Cultural transmission is adaptive because it saves learners the costs of individual learning.
Once some cultural transmission capacities exist, natural selection favors improved learning
efficiencies, such as abilities to identify and preferentially copy models who are likely to
possess better-than-average information. Moreover, selection will favor behaviors in the
learner that lead to better learning environments, e.g., gaining greater frequency and intimacy
of interaction with the model, plus his/her cooperation. Copiers thus evolve to provide all
sorts of benefits (i.e., ``deference'') to targeted models in order to induce preferred models to
grant greater access and cooperation. Such preferred models may be said to have prestige
with respect to their ``clients'' (the copiers).
The above implies that the most skilled/knowledgeable models will, on-average, end up
with the biggest and most lavish clienteles, so the size and lavishness of a given model's
clientele (the prestige) provides a convenient and reliable proxy for that person's information
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 167

quality. Thus, selection favors clients who initially pick their models on the basis of the
current deference distribution, refining their assessments of relative model worth as
information becomes available through both social and individual learning. This strategy
confers a potentially dramatic adaptive savings in the start-up costs of rank-biased social
learning. Finally, because high-quality information (``expertise,'' ``performative skills,''
``wisdom,'' ``knowledge'') brings fitness-enhancing deferential clients, models have an extra
incentive to outexcel each other.
2. What is prestige?
Since the common meaning of ``prestige'' corresponds closely with our posited domain of
psychological and social causation, with its associated ethological displays and emergent
population-level phenomena, we will stay close to the common meaning.
prestige
1: standing or estimation in the eyes of people; weight or credit in general opinion.
2: commanding position in people's minds. syn see INFLUENCE Ð Merriam
Webster's Collegiate Dictionary (1994, p. 923)
Notice that the synonym is ``influence'' Ð not ``authority,'' ``power,'' or ``dominance.''
Someone with prestige is listened to, their opinions are heavily weighed (not obeyed) because
the person enjoys credit, estimation, or standing in general opinion.
prestigious
2: having prestige Ð HONORED.
honor
1: a good name or public esteem: REPUTATION: a showing of usually merited respect.
2: PRIVILEGE. 3: a person of superior standing. 4: one whose worth brings respect or
fame: CREDIT. syn HOMAGE, REVERENCE, DEFERENCE: mean respect/esteem
shown to another. HONOR means the recognition of one's right to great respect or any
expression of such recognition. HOMAGE adds the implication of accompanying praise.
REVERENCE implies profound respect mingled with love, devotion, or awe.
DEFERENCE implies a yielding or submitting to another's judgment or preference out
of respect or reverence (ibid.)
Nothing above suggests prestigious people are feared. Rather, others believe they have
earned the right, if not to be obeyed, at least to have their opinions and desires considered
more closely than those of ordinary people. They are also excused from certain
obligations and obtain certain privileges. The words ``respect,'' ``awe,'' ``devotion,''
``reverence,'' and ``love'' all connote that inferiors do not begrudge superiors and willingly
confer such benefits.
Contrasting dominance to prestige sharpens our understanding of both. Agonistic
encounters, real or implied, stabilize rank hierarchies in many primate species. ``Fear''
estimates the costs of challenging superiors. Subordinates signal acquiescence by averting
their gaze and avoiding superiors Ð for to stare is to challenge (Goodall, 1986; Schaller,
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196168

1963). Deference is often transitive
1
(if A defers to B and B defers to C, then A defers to C;
see e.g., Strayer & Cummins, 1980). Finally, repeated losses by the high ranking lead to
changes in the rank order.
The ethology of dominance in chimpanzees Ð our closest relatives Ð consists of five
categories of behavior, each involving several action patterns (culled from Goodall, 1986):
Subordinates:
1. Proximity management: baseline avoidance of higher-ups.
2. Submissive behavior after agonistic interaction (e.g., gaze avoidance, hunched
shoulders, turning body away, lowered head, etc.).
3. Occasional submissive behaviors without context or provocation.
Superiors:
4. Grandstanding (higher frequency of aggressive displays than subordinates in order to
signal their position and have it confirmed).
Everybody but the alpha:
5. Occasional challenges to the rank ordering (i.e., agonistic encounters initiated or
resisted by the subordinate).
We now compare the above description with an ethnographic description of a society
with clear prestige, but apparently no dominance, hierarchies. The Semai are an
indigenous, Malaysian people who are famous for their nonviolence, acephalous structure,
and their prickly, independent individuals who will not be bossed around (Dentan, 1979).
The absence of dominance hierarchies is maintained and guaranteed by the readiness with
which diffuse punishment by third parties descends on those who would arrogate
themselves authority.
At first it seems as if Semai communities are run by a council of elders .... The elders in fact
have no authority to enforce their decisions, however, and the variety of ways in which the
Semai calculate age often makes it hard to tell just who the elders are .... The fact that the
Semai respect the elders does not mean that they have to obey them .... A Semai takes heed
of what his elders say. In the Semai phrase, he ``hears'' them. He does not interrupt while they
are speaking, nor does he address them familiarly .... On the other hand, after listening
respectfully to them, he may reject their advice. If they press the point, he may say, ``I don't
hear you.'' Although a senior may have great influence over some of his juniors, he cannot
order them to do anything they do not want to do (Dentan, 1979, pp. 65± 66).
The Semai have three different ways of reckoning age, and the resulting ambiguity allows
them considerable freedom in choosing their ``elders,'' giving a person ``considerable leeway
in deciding just whom he wants to respect'' (Dentan, 1979, p. 67). ``Elders'' apply their
1
Although in chimpanzees, at least, the transitivity seems to be between hierarchical levels and not between
individuals (Goodall, 1991, p. 125). That is, coalitions of two or more individuals may obtain a certain rank, so
some of the elements in the transitive set may be groups, rather than individuals.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 169

influence gently (Dentan, 1979, p. 69) and must never arrogate authority, otherwise other
individuals will cease ``hearing'' the elders. Self-deprecation, for example, is a common
rhetorical technique used by elders to assure listeners that they are not trying to compel
compliance. Despite this, or perhaps because of it, respected elders manage to exert
considerable influence in Semai society.
Dentan's (1979) ethological observations are limited, so we combine them with our own
informal observations of prestige interactions Ð these, of course, await more ethological
fieldwork for confirmation. Our tentative ethology of prestige is as follows:
Low-status ``clients'':
1. Proximity management: they are responsible for the overwhelming majority of the
total time that they spend in interactions with superiors.
2. Dyadic interaction: eyes and posture directed towards the superior. Relative to the
superior, they make few utterances. If superior pauses, even for an unusually long
period of time (as conversational pauses go), subordinate is unlikely to take the floor.
3. Public interaction: sing praises of superior and escalate these as others escalate. Offer
denials of the self-deprecating offerings of superiors.
Superiors:
1. Proximity management: responsible for a minority of total time spent with
subordinates.
2. Dyadic interaction: free posture but no grandstanding, undue raising of the voice, or
making violent gestures with the arms and body. Speaks most of the time. If
subordinate is speaking and there is a pause, superior is more likely to take the floor.
3. Public interaction: self-deprecating denials of whatever praises are extended, with
expressions of gratitude. Posture is confident but does not become a swagger, and at
key moments (e.g., beginning or end of the interaction, whenever the public roars
approval, etc.) posture may become positively servile (bows, etc.).
Some elements above may be culturally specific, but they do not cease to be prestige.We
expect functional similarities in prestige ethologies despite such local variations.
Prestige rests on merit in the eyes of others (rather than force deployed against them), and
promotes the admiration of inferiors (not their fear), a desire for proximity (not distance), and
periods of sustained observation (not furtive glances). Transitivity is weaker than in
dominance because prestige hierarchies can be domain-specific. For example, if I defer to
you because of your superior computer skills and you defer to Bob because he is an excellent
grass hockey player, I may not give Bob any special deference if grass hockey is not my
thing. Also, clients can shop around for the model offering the best copying ``deal'' (a
function not only of information quality, but of the price of access, and may thus be different
for individuals choosing at different times). Finally, changes in prestige rank result from
``students'' surpassing former ``teachers,'' which has nothing to do with losing violent
contests (except when prestige comes from one's prowess in combat Ð but even here, one's
rank falls by losing to one's enemies or sanctioned competitors, not to members of one's
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196170

clientele). High-prestige individuals can lose status by behaving as if entitled to the attentions
of clients (thereby raising their price of access) rather than grateful for freely conferred client
deference. This points to the power of market choice that clients exercise. As the summary on
Table 1 shows, there are many important differences between prestige and dominance
hierarchies, even at a broad, first approximation. If dominance and prestige ethologies are
different, we suspect that their underlying psychologies are also different, as are the selection
pressures that produced them.
Before presenting the theory, we have two clarifications. First, our focus is not on
prestigious and dominant individuals, but on prestige and dominance processes. Although
one may find prestige and dominance status in the same individual, the fact remains that
qualitatively different stimuli elicit prototypical prestige and dominance responses. To lose
sight of this distinction is to obscure our argument. Modern societies have complex,
institutional hierarchies, so individuals often enjoy both dominance and prestige status.
Presidents and school teachers, for example, have real control over rewards and punishments
on the one hand, but may also be perceived as doing a good job, on the other. What justifies
our distinction is not that individuals must always have only one or the other form of status,
but that it is possible for humans to have only one or the other because the prototypical
stimuli and underlying psychologies are fundamentally different (e.g., Stephen Hawking, for
pure prestige, and a high-school bully, for pure dominance). Although acquiring prestige may
confer a capacity for force threat, in turn evoking dominance psychology in subordinates, it is
the pathway that is relevant here. Some can scare others only if they first excel at something,
then gain a large following or an institutional rank. Stephen Hawking may strike fear in
graduate students on whom he can inflict real costs with some measure of institutional power,
but this is because he first achieved prominence through excellence (and Star Trek fans, over
whom he has no institutional power, still respect and adore him). His avenue to status was a
pure prestige process, even if in the end it gives him a measure of dominance in a small circle.
Finally, prestige-deference evoked by one's great skill or knowledge in inflicting costs (e.g.,
Table 1
Summary of ethological differences between dominance and prestige
Dominance Prestige
Low-status individual
Approaches often (proximity) no yes
Prolonged stares no yes
Occasional attack (to challenge) yes no
Fears high-status individual yes no
High-status individual
Charges and attacks yes no
Is frequently imitated no yes
Swaggers a lot a little; sometimes
displays subdominant
ethology
Received gifts/services yes yes
Transitivity more less
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 171

violence) should not be confused with the dominance-deference created by the fear that one
may become a victim of that violence.
Second, our explanation focuses on particular forms of direct social learning, which we
collectively term infocopying. This category encompasses all forms of acquiring information
directly from another, and includes, but is not limited to, ``true imitation'' (acquiring the
details of motor patterns via direct observation; see Tomasello, 1994) and ``goal emulation''
(inferring behavioral goals via direct observation). Infocopying excludes indirect social
learning processes, such as ``social facilitation'' or ``local enhancement,'' where learners
have a higher probability of reinventing something due to close proximity to a competent
performer and the materials involved. Infocopiers may also unconsciously acquire manner-
isms, consciously acquire verbal knowledge and arguments, and consciously or uncon-
sciously imitate action patterns. Finally, much infocopying does not approximate information
``replication'' (as implied by ``copying''). For example, infocopying includes ``influence''
that occurs when a person expresses a position P at strength S that moves the strength of my
own agreement with P closer to S. Such processes are fully included in this theory (justified
below, also see Boyd & Richerson, 1985, pp. 70± 71).
3. Prior evolutionary theories
3.1. Theories conflating prestige and dominance
Barkow (1975, 1989) argues that prestige is homologous to nonhuman dominance. The
social rankings imposed through force and force threat in nonhumans became the rankings
of merited deference that we see in human societies. Missing from Barkow's account,
however, are the relevant selection pressures that would have achieved this exaptation. His
attempt to supply them involves female mate choice. We sketch his model (Barkow, 1989,
pp. 186±188):
Importance of male parental investment !female preference for males with greater
subsistence skills, and male preference for the same in females !both males and females
compete intrasexually to increase skills related to resource acquisition !selection for
skill !exaptation of dominance into prestige
Every link makes sense, except for the last one: how does selection for skill transform
dominance into prestige? Barkow says that males with greater skills are capable of higher
levels of investment, and they are also those with better cultural capacities (i.e., greater innate
ability in social transmission). Females, then, in choosing skilled males, select males with
greater cultural capacities. But the last link remains unexplained. Why is prestige an
exaptation from dominance? And what exactly is prestige? Female preference? Or greater
cultural capacity?
Barkow (1989, p. 150) solves the problem by assumption: ``Let us assume (assumption 1)
that, among our protocultural ancestors, those males best able to learn and to transmit
protocultural information tended to be the most able to provide parental investment and also
to be the highest in status.'' Since those best able to transmit were earlier assumed to be the
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196172

most highly skilled, the above stipulation amounts to assuming that the skilled have status.
But this is what has to be explained (for a similar model, see Hill, 1984a, 1984b).
3.2. Status arises from social exchange
Some evolutionary psychologists explain nonagonistic status as exchange between
individuals with differing assets, skills and/or resources (Patton, 2000; Pinker, 1997, p.
499, Tooby & Cosmides, 1996). Whether or not a behavior is directed to us, if the behavior
benefits us, we should (if we can at a reasonable cost) do things that increase the likelihood
that the actor will repeat it. This logic explains how trading relationships were bootstrapped
from behaviors that originally were not directed towards trade but constituted instead
``positive externalities'' Ð that is, behaviors incidentally beneficial to others (Tooby &
Cosmides, 1996). To the extent that some trading relationships involve one partner obtaining
tangible goods (e.g., food, tools, coalitional support) at least partly in return for giving
deference (e.g., by allowing the other to choose the best sleeping sites, sharing choice foods,
excusing him from certain obligations, etc.), this view helps explain the origins of
nonagonistic status. Hereafter, we call this the ``tangible goods'' hypothesis.
In our view, one must pay close attention to the goods being exchanged, for important
selection pressures may be particular to certain goods. Explaining many ethological displays,
attention and memory biases, emotional states, and patterns of influence and imitation
evident in human status processes requires special attention to the trade in deference for high-
quality information, for these processes would not arise by trade in deference for tangible
goods alone.
If individuals have valuable assets, it makes sense to exchange deference for them. But
why defer to them when they cease to produce tangibles (e.g., old men who are no longer able
to produce)? Why remember what they say about a wide range of topics well outside of the
exchange? Why copy their behaviors and opinions? And why is human status so different
from that of other social species? Why does not social exchange among baboons and
chimpanzees, who vary greatly in their individual skills and productive capacities, produce
human-like status hierarchies? The hypothesis below addresses these questions.
4. Evolving prestige: the information goods theory
4.1. The importance of social learning
Prestige is a consequence of the evolution of direct social learning capacities in the human
lineage Ð abilities that are quite distinct, from the social learning abilities of other species
(Boyd & Richerson, 1985; Durham, 1991; Tomasello, 1994). For our argument, the most
important difference between human and nonhuman social learning is that humans possess
various forms of direct social learning, or ``infocopying'' that most other animals completely
or almost completely lack (Tomasello, 1994, p. 304; Tomasello, Kruger, & Ratner, 1993). In
true imitation, for example, a human imitator can copy the behavior or behavioral strategy of
a model, including the motor patterns and objectives. So, for example, when a human child
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 173

learns to throw a ball, she tries to copy the model's arm motions and footwork as well as the
objective (usually, getting the ball accurately to the receiver). In contrast, other forms of social
learning common in nonhumans do not involve the transmission of both goals and motor
patterns, e.g., local enhancement,social facilitation,emulation learning, etc. (Whiten &
Ham, 1992). In local enhancement, for example, the learner's proximity to a skilled
individual (and any necessary materials) increases the chance of reinventing Ð rather than
directly acquiring Ð the behavior.
4.2. Ranking and deference
If chimpanzees lack infocopying abilities, these social learning capacities probably arose in
our lineage some time after the hominid± chimpanzee split.
2
However, our story begins with
two adaptations that probably preceded the rise of infocopying: (1) skill ranking of
conspecifics and (2) discriminatory deference. Pigeons and macaques (Giraldeau & Lefebvre,
1986, 1987; Stammbach, 1988), and possibly many group-living species, can rank con-
specifics in terms of foraging success. ``Scroungers'' in these species can identify successful
food producers and maintain proximity to them in order to feed from their food finds.
Probably to increase their success, scroungers also give deference to the producers. For
example, some macaques (Stammbach, 1988) not only maintain close proximity to successful
foragers, but also preferentially groom producers even when such producers have low
dominance rank. Note that trading deference for scrounging opportunities has nothing to do
with social learning, and may thus predate infocopying. In fact, macaques show no
infocopying abilities.
With the evolution of infocopying, selection favors articulation of these capacities with
preexisting ranking abilities and deference biases. Ranking abilities allow infocopiers to
target their infocopying efforts preferentially toward models with high-quality skills.
Deference biases allow imitators to buy proximity to their targets, improving their copying
reliability and fidelity. If infocopying evolved first, then selection would have favored
ranking abilities and deference biases, since discriminatory, sycophantic infocopiers would
have outcompeted infocopiers without these capacities. However, these likely cognitive
preadaptations add plausibility to our account.
If rank-biased social learning is such a useful adaptation, why is it not more common? The
reason is that ranking abilities do not improve social learning unless the variation in skill or
knowledge of models can be tapped. For example, suppose chimpanzees, by watching
conspecifics, can infer a connection between reeds and getting termites (emulation learning),
but, lacking true imitation, they cannot acquire the details of anyone's technique. Since the
association between reeds and termites is constant across all models, and what vary are the
details of technique (which cannot be acquired without direct social learning), social learners
2
It is not clear whether chimpanzees are entirely lacking in true imitative capacities (see Boesch & Tomasello,
1998; Whiten, 1998). However, the key difference may be quantitative, rather than qualitative. As a species'
reliance on true imitation increases, the relative benefit from improved copying abilities increases. But, if a species
only rarely uses its imitative abilities, or only in some narrow behavior domain (like termite fishing), the benefits
of improved imitation may not exceed the costs.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196174

derive no benefit from ranking the termite-fishing success of others. Thus, we predict that
rank-biased social learning should only be found in group-living animals with direct social
learning capacities (infocopying).
These observations apply generally to prestige-biased transmission processes. In ``influ-
ence,'' where the behavior of a source moves me closer to her position, some information
needs to be directly transmissible Ð otherwise all we have is the association of a person with
a position. Unless information related to the variation in strength of commitment to a position
or belief across different individuals can be directly transmitted, ranking their success in some
domain cannot result in influence.
Next question: if infocopying capacities are so adaptive, why are they not more common?
Boyd and Richerson (1996) have modeled this problem, and convincingly argued that if
group members are all individual learners, mutant infocopiers cannot invade even though
infocopying is evolutionarily stable once common. Thus, a fitness valley impedes the
evolution of direct social-learning capacities, so they should be rare in nature.
4.3. Picky infocopiers and rank-biased transmission
Coevolutionary models of individual learning and infocopying show that individuals ought
to retain some reliance on individual learning because (1) anything acquired socially can be
refined through individual learning and (2) temporally and spatially varying environments
will act to devalue culturally transmitted information (Boyd & Richerson, 1985; Henrich &
Boyd, 1998). At the population-level, individual learning, infocopying, ranking abilities, and
deference combine to produce what we call prestige-biased guided variation. To clarify this
process, we construct it in steps, first considering the articulation of individual learning with
rank-biased true imitation, and then adding discriminatory deference.
Although infocopying does not make individual learning obsolete, selection favors a
strong reliance on infocopying in a wide range of environments (Boyd & Richerson, 1985,
Chap. 4, 1988; Henrich & Boyd, 1998). Humans are thus default infocopiers, usually trying
first to learn directly from others instead of ``reinventing the wheel'' (a great cost savings),
and only then seeking improvements through individual learning.
3
Default infocopying relies
on social transmission when the costs of individual experimentation are greatest (thus
outcompeting strategies with a heavier initial reliance on experimentation) but preserves
the benefits of acquiring information about model quality.
Preadaptations to rank others (e.g., according to their food-finding abilities), combined
with infocopying capacities, result in ``picky'' copiers. The details remain unknown, but both
empirical and theoretical considerations suggest that people rank potential models using a
number of cues. The first cue is the amount of freely conferred benefits and displays (i.e.,
deference) an individual receives Ð see below. Second, infocopiers assess the competence
(skill/knowledge) of potential models in culturally valued domains (e.g., hunting, basketball,
3
Coevolutionary models of social and individual learning (Henrich & Boyd, 1998) show that mutants who
rely occasionally on individual learning as their first guess Ð instead of social learning Ð outcompete those who
always rely on socially transmitted information. This is because a small reliance on individual learning prevents
individuals from getting stuck on the wrong behavior in spatially and temporally varying environments.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 175

big yam growing, etc.) Ð often by observing simple indices such as hunting returns,
rebounds, or yam size. Third, infocopiers assess cues of a potential model's state of health,
such as clear skin (no growths, infections, discolorations, etc.), bright eyes, shiny hair, and a
lack of disfiguring injuries. Finally, by incorporating the age and sex of potential models into
their assessment, infocopiers increase their likelihood of acquiring information relevant to
their role and stage in life.
By combining an assessment of model competence in culturally valued domains with their
overall state of health (direct proxies of fitness), infocopiers can both adapt to novel
environments where required competencies differ (ice fishing, stone tool making, and bitter
manioc processing), and ignore the maladaptive behaviors of individuals competent in
cultural domains with net fitness costs. In our ancestral environment, this state of health
assessment acted to cull out cultural domains that tended to reduce the fitness of those who
adopted them as valued, and promote the spread of cultural domains whose practitioners see
fitness benefits. We do not expect this to be true in many contemporary societies where
modern institutions and divisions of labor have decoupled the effect of investing in a valued
domain on one's state-of-health proxies. For example, an expert in theoretical physics may
not look any less healthy than a farmer. Modern societies can therefore spawn endless valued
domains where higher investments correlate negatively with reproductive success. However,
in small-scale societies lacking division of labor and supporting institutions, arcane endeavors
that compromise food production are likely to make practitioners appear unhealthy compared
to their neighbors.
Individual, trial-and-error learning may also act to maintain the statistical connection
between ``culturally valued domains'' and adaptive behavioral domains in ancestral environ-
ments. Individuals receiving poor marginal rates of return for investments in skill are likely to
invest instead in other domains yielding ``sellable products'' (e.g., fish, medicinal herbs,
canoes), deference benefits, or simply direct benefits (lots of fatty meat and sex). Given the
ineffectiveness of contraception in the ancestral environments, excellent performers in valued
domains would be transforming their trade-benefits into reproductive success.
Age and sex should also be important ranking criteria. Premodern societies have a strong,
sex-based division of labor, so children should prefer somewhat older, same-sex models. This
allows children to acquire gender-relevant skills and scaffold themselves to increasingly
complex skills Ð for copying models that are too advanced will often result in failure.
Each generation, as people copy highly ranked models, the mean behavior of the
population will move quickly Ð relative to genetic evolution and ordinary ``guided variation''
(Boyd & Richerson, 1985) Ð towards the most adaptive behavioral repertoire currently
represented. As copying errors and individual experiments add new variation, the group's
mean behavior will move toward the locally most adaptive configuration. However, because
infocopying is generalized rather than targeted at specific traits (defended below), maladap-
tive traits possessed by the highly ranked models may hitchhike along with adaptive ones.
4.4. Picky infocopiers ``kiss up'': the evolution of deference to skilled models
Infocopying benefits from the cooperation of the chosen model because perceptual access
is a minimum requirement. Picky infocopiers who induce models to grant better perceptual
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196176

access (and perhaps provide direct instruction) have an advantage, especially when acquiring
more complex behaviors and strategies. For example, only if the skilled hunter likes you can
you accompany him on the hunt and observe his practices up close (perhaps receiving
verbally transmitted tips), and thereby acquiring his superior knowledge/skills (e.g., tracking
practices, approach methods, spoor interpretation, bow handling techniques, etc.). In order to
gain this kind of preferential access, infocopiers become valuable interactants by ``kissing
up.'' Infocopiers have evolved to do all sorts of things that models were already adapted to
like or seek in potential interactants, such as being especially trustworthy, offering all sorts of
help without expecting anything in return, deferring to the model's judgment, being nice and
helpful to the model's children, exempting the model from certain obligations vis-a
Á-vis the
copier, etc.
One may object that many behaviors can be acquired without close proximity, so buying
access with deference incurs unnecessary costs. This objection assumes that infocopiers know
at what times they should be watching, i.e., that they have figured out exactly which of the
model's many behaviors contribute to his/her success. However, the success of the model,
like most behavioral outcomes, is likely the result of very complex interactions among a large
number of variables. Being a good hunter, say, probably depends not only on specific skills
such as making a good bow, knowing how to aim, etc., but also on tracking knowledge/skills,
animal behavioral knowledge, approach and pursuit techniques, prey preference, as well as on
more indirect factors such as sleeping properly, keeping a certain diet (e.g., eating lots of
vitamin A-rich foods to maintain good eyesight), and observing certain habits, etc. Thus,
given the prohibitive acquisition, storage, and analysis costs involved in teasing out precisely
which behavioral combinations actually lead to desired results, evolution would instead make
copiers rely on a general copying bias (Boyd & Richerson, 1985, Chap. 8).
In any case, many behaviors and skills Ð even those obviously related to a model's
success Ð cannot be copied without close proximity to, and interaction with, the model.
Hunting is one of them. Tracking abilities alone require the development of a counterintuitive
``scientific'' skepticism, involving hypothesis-testing, that cannot easily be learned without
social assistance (Liebenberg, 1990). Furthermore, the evolution of language liberated a great
deal of information for social transmission that is difficult to infer through mere audio-visual
perception. The disbursement of this language-bound information can be tightly controlled by
the model and thus creates even stronger selection pressures for copier deference.
4.5. Broadcasting expertise: deference as an honest signal
``Status'' equals the amount of deference received. In our species, it appears that those with
nonagonistic status are also those with skill/knowledge in valued domains of behavior. Thus,
if a model's skill stimulates observers' deference (to buy proximity), then deference and skill
will correlate Ð something Barkow (1989) assumed but which the present hypothesis can
explain. This correlation provides yet another opportunity to save on information-gathering
costs, because new entrants can exploit the distribution of deference as a low-cost cue to
identify worthy models. This selects for infocopiers who leapfrog directly to the more
fawned-over models and avoid the start-up costs of gathering and processing information
about relative model quality. In addition, such infocopiers avoid the costs of not having
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 177

adaptive information (specific traits or information about model quality) during the initial
stages of learning.
The distribution of deference is a reliable and honest signal of relative model worth
because such signals are costly to fake. Sycophants cannot deceive their competitors by
deferring to someone they would rather not copy without increasing their total deference costs
and losing some access to their preferred models. Sycophants also cannot easily conceal
deference directed to the desired model, for this entails a bias for private deference and
therefore a reduction in total deference, and hence in less access. Moreover, models should
prefer public displays of deference in order to broadcast their prestige and attract more clients.
Thus, default infocopiers, when unsure about model quality, evolved to prefer ``models
with the largest and most lavish clienteles.'' As information about relative skill differences
becomes available, infocopiers can switch their focus if a superior model is found.
Concurrently, they improve their skills somewhat via individual learning.
If every naõÈve starts out with a model perceived to be (as per deference received) the most
skilled individual, then everybody's initial goal is the best, or close to the best, currently
available transmissible idea, behavior, or strategy. If copying is reasonably reliable, effective
cultural traits can spread rapidly. Thus, prestige-biased guided variation allows populations to
approach adaptive optima much faster than they would under Boyd and Richerson's (1985)
guided variation, which lacks the information-channeling force of prestige distributions.
A hunting example will illustrate the above argument. Novice hunters may assess who is
most successful among experienced hunters by comparing daily returns. However, picking a
model on the basis of this 1-day sample is risky, for hunting success will exhibit much short-
term variance. Only hunting returns averaged over a great many days will reliably predict
hunting skill. Thus, novices are initially better off selecting models who are already favored
by others. Later, after they have accumulated their own long-term samples, they can refine
these borrowed judgments. Hunting returns are hard to fake Ð and if they bring prestige, they
will be advertised Ð so information-gathering costs are substantially reduced for novices.
4
4.6. Coevolution of copiers and their models
Some further implications of our theory arise from considering the coevolution between
copiers and models. In prestige, clients choose whom they defer to, so a kind of ``market''
results. Like ``firms,'' models compete for ``customers'' (the copiers) who shop around for
the best deal. Models should be sensitive to how ``profit curves'' change with added clients,
for these ``firms'' can have too many customers. A hunter's fitness initially increases as more
clients raise total deference. So the hunter may prefer having 3 sycophants to 1, but would
he want 20? Large hunting parties may scare off potential prey. Thus, good hunters should
raise the cost of access by acting more arrogantly as clientele size approaches the optimum.
On the other hand, if no practical limit on optimal clientele size exists (e.g., great
storytellers), or if means other than arrogance will limit clientele-size (e.g., bodyguards),
4
Although we have been using hunting as our reference example, and hunting is typically a male activity,
these arguments apply equally well to females and to the whole range of typically female activities such as food
gathering and child care.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196178

then increases in arrogance should not accompany growing prestige. Alternatively, if one's
benefits do not come directly and primarily from client deference, the prestigious may learn
that arrogance is not too costly (e.g., some sports stars). Finally, models should prefer above-
average learners because they advertise the model's quality and provide a potential source of
valuable information.
For clients, the benefits of access diminish rapidly with increasing clientele size.
Competing with more clients may mean less individual attention from the model, so copiers
may prefer less popular, lower-quality models with cheaper prices of access.
Careful modeling will be required to understand the full implications of this process.
However, we submit that the more competition there is for clients, the ``nicer'' models should
be, because infocopiers evaluate their clientele size, and the price of joining Ð not just skill.
The prestige market thus seems more like an ecology of resource patches (the models) where
consumers (the clients) distribute themselves according to the patches' richness and the
competition for access. Some readers may reason that if client choice dynamics are fast
relative to those governing (1) the entrance of new naõÈve clients and (2) improvements in
model skill and price of access, the system should quickly reach an ``ideal free distribution,''
where every ``deal'' is the same for the next entrant into the system. At this point, the
distribution of deference no longer provides information, so naõÈve clients should be
indifferent as to their choice of model and no stable selection pressure results for an initial
bias to prefer models with intense clienteles.
We think this intuition is wrong. If clients are going to pay for their models, they will do
so only for above-average models because average skills can be obtained from one's parents
without deference payments. So clients will be distributed only over above-average
performers, and thus a bias for clientele intensity still narrows the naõÈve entrant's choices
to models skilled enough to have clients. Second, if one person occasionally monopolizes all
clients, a prestige-bias takes naõÈve entrants straight to the best copying ``deal.'' Finally, client
choice is probably not so fast that the death, injury, and skill improvements of models do not
jolt the system out of equilibrium for significant periods. As in modern markets, equilibria
are never achieved and consumers should compare quality and prices, or risk missing a
rising star.
We believe this prestige market allows us to explain what appears as a counterintuitive
fact: some high-status humans display subdominant ethology (deep bows, bringing one's
hands to the center of the body, lowering the head, and generally appearing bashful). In
contrast, these behaviors are not observed in dominant nonhumans. Self-deprecation is also
common in prestige: those receiving applause and awards will publicly ``doubt'' their
worthiness and attribute the gesture more to client generosity than personal prowess (in
the US, Oscar-winning movie stars thank everyone, and stand-up comedians credit their
success to their ``great'' audience).
Some elements of dominant pride displays (staring, squared shoulders, chest out, erect
posture, stiff-legged gait, etc.) are sometimes evident in prestigious individuals, and may act
as an advertisement of status (and, by implication, skill), thus alerting potential followers.
However, they are diminished, less common, and apparently unappealing in their strong
versions (cf. Goode, 1978, pp. 21±22). Perhaps, this is because: (1) to the extent that pride
signals a dominant individual, it scares off clients and (2) because clients have choices,
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 179

models may learn to avoid behaviors Ð beyond any basic attenuations already achieved by
selection Ð that make them less competitive in the prestige market.
4.7. Costly signaling, public goods, and prestige
Commonly, those who supply public goods at personal cost earn prestige. Our model does
not, by itself, account for this. Deference will not buy altruism, should it depend on it,
because those who ``cheat'' (do not defer) have higher fitness; giving deference in exchange
for public goods has a second-order free-rider problem (Boyd & Richerson, 1992). Moreover,
one should not want to copy obviously costly behaviors. However, perhaps excelling in
domains generating public goods helps broadcast the altruist's copy-worthy skills. Everytime
a hunter shares a big kill with the group, his prowess is more likely to be noticed and
remembered by others. All other things equal, models who excel in domains with better
broadcast opportunities will gain more clients, more prestige, and more fitness-enhancing
deference. Consequently, there should be a preference for behaviors that facilitate such
broadcast opportunities Ð and providing public goods often furnishes such an opportunity
(Smith & Bliege Bird, 2000).
5. Predictions and evidence
Here we present predictions derived from the information goods theory, and the available
evidence. Not all predictions below are unique to this theory but, as a group, they cannot be
better accounted for by any competing model.
5.1. General predictions about prestige, skill, and age
5.1.1. Skilled individuals have higher status
Individuals will seek out and pay deference to highly skilled individuals in exchange for
copying access. The relation between skill and status is also predicted by the tangible goods
hypothesis discussed earlier.
In geographically diverse societies, numerous ethnographers have observed the relation-
ship between hunting skill and status. For example, among the Kuna, an island-living
population that hunts and farms on Panama's Caribbean coast, a lifetime record of tapir kills
is kept (i.e., remembered) for each male. Males with the most tapir kills (a measure of hunting
skill) receive higher status (Ventocilla et al., 1995, pp. 39 ± 40). Among the Naskapi, hunting
knowledge about animal migratory patterns, feeding cycles, tracking, etc., confers prestige
(Moore, 1957). Among the Cubeo, Goldman (1979, p. 57) writes, ``hunting, in summary, is a
distinctive pursuit and marks one for prominence'' (jaguar teeth are used to make girdles,
which mark one's high status). Among the Siriono (Bolivian foragers), Holmberg (1985, p.
58) notes, ``If a man is a good hunter, his status is apt to be high.'' Stearman (1984) further
confirmed this for the recently settled Yuquõ
Â, who are probably closely related cultural
relatives of the Siriono. Among the hunters in the Kalahari, ``Although no one is in
command, an informal leadership may develop and parties tend to form around good
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196180

hunters'' Liebenberg (1990, p. 55). After reviewing the literature on foraging societies, Kelly
(1995) argues that hunting large game is always a highly valued activity, and hunting skill is a
primary means to acquire prestige.
5
Excelling in certain other domains is also commonly associated with status throughout the
ethnographic literature, particularly in simpler societies. These include combat (e.g., Yano-
mano: Chagnon, 1992; Achuar: Patton, 2000), oration (e.g., Semai, Dentan, 1979, p. 69;
Benkulu: Fessler, 1999; Kuna: Howe, 1986; !Kung: Lee, 1979, pp. 343±344), and healing/
supernatural knowledge (Lee, 1979, pp. 343±344; Simmons, 1945). Counting systems to
keep track of an individual's successes in combat (i.e., their ``kills'') are common (e.g.,
Bateson, 1958, p. 48; Murphy, 1960). Farming and herding skill are an important criterion for
status in small farming communities in rural New Zealand and, moreover, people feel that this
is ``natural'' and in need of no justification (Hatch, 1992, pp. 89 ±90, 102, 108 ±109).
Coleman (1961) asked adolescent males what qualities were necessary to be respected by
male peers, and by female peers. Students ranked athletic skill at the top, and scholastic skill
at either the bottom or second-to-last, for both questions (the list of 6 options also included
``being in the leading crowd,'' ``having a nice car,'' etc). This shows (1) the importance of
skill and (2) the effects of local culture in making some domains relevant.
5.1.2. Older individuals will tend to get more prestige than younger ones
Age is a proxy for skill/knowledge/success; the longer someone has lived, the more and
better skills/knowledge he/she has likely accumulated. Simply living longer is a complex
``skill'' with acquirable components. Deference toward elders allows proximity and thereby
promotes the acquisition of useful information. This reasoning predicts a general correlation
between age and prestige, and also that elderly individuals will maintain their status well past
their prime. In contrast, the status of elderly nonhuman primates usually falls in tandem with
their coercive powers.
This prediction does not hopelessly confound prestige with dominance, or with conven-
tional reciprocity, because, though older people receive deference, (1) some older individuals
cannot deploy force or force threat, (2) many elderly are unable to reciprocate good turns in
tangible currencies (or at least will not do so for much longer), and (3) one can determine
whether deference towards elders has prestige vs. dominance ethology/psychology.
The best current evidence for the relationship between age and prestige comes from
behavior towards the elderly. Despite being unable to contribute to the household or
community economy, or reciprocate in tangible benefits, the very old will nevertheless
often become high in status. This should not be the case if all nonagonistic status is
reducible to exchange for tangible goods. Simmons (1945) provides evidence for this in an
extensive cross-cultural survey of the role of the aged in 72 simple societies, 69 of which are
known from ethnographic rather than historical sources. Of these 69, 46 include explicit
ethnographic mention of (often quite extreme) respect, deference, reverence, homage, or
5
For additional examples see: Ache
Â: Hawkes, 1991; !Kung: Lee, 1979; Naskapi: Moore, 1957; Shuaranahua:
Siskind, 1973; Cashinahua: Kensinger, 1995; Efe: Bailey, 1991; Shavante: Maybury-Lewis, 1967; Meriam: Bliege
Bird & Bird, 1997.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 181

obeisance to the aged. From 5 more, the high status of the elderly can be inferred (e.g., a
requirement that chiefly roles be filled by elders). Simmons (p. 79) observes that ``the most
striking fact about respect for old age is its widespread occurrence ... practically universal in
all known societies.''
Simmons (ibid.) also notes that ``There have usually been extenuating circumstances,
qualifying conditions, and ... a `prime of life' in old age Ð when prestige has been attained;
and other circumstances under which it has been denied or has practically disappeared.''
Tellingly for our hypothesis, the most important moderating variable seems to be the
elderly's obvious skills/knowledge or lack thereof. For virtually all of the sample, there is
ethnographic mention of recognized bodies of knowledge that only the aged possess, or
possess in obvious superabundance relative to younger people (magic, lore, hunting skills,
calendrical, and traditional knowledge, medicine, etc.). Respect towards individual elderly
persons varies considerably, and those with acknowledged expertise are most highly
respected. Neglect of the aged appears invariably to follow senility and decrepitude, which
make information transfer either moot or impossible. When variability in respect is extreme,
with some elderly but not others suffering serious neglect, ethnographers typically report that
the aged can only escape neglect if they possess valuable knowledge and skills. Simmons
(1945, pp. 50±51) concludes:
Most primitive societies have insured some respect for the aged Ð often remarkable
deference ... at least until they have become so ``overaged'' that they are obviously
powerless and incompetent ... respect for old age has, as a rule, been accorded to persons on
the basis of ... their extensive knowledge, seasoned experience, expert skill, power to work
magic, exercise of priestly functions, control of property rights, or manipulation of family
prerogatives ... their skill in games, dances, songs, and storytelling.
Despite their physical weakness the elderly certainly hold much political sway (sometimes
total). A full 52 (out of 72) societies in the sample boasted aged chiefs, and many require
advanced age for the role. There is also widespread participation of the aged in councils and
they tend to be generally influential even when there is no institutional office to fill.
Silverman and Maxwell (1978), like us, see a link between expertise and deference to the
aged. In a randomly selected sample of 34 small-scale societies they found only two with no
ethnographic mention of deference towards the elderly.
5.1.3. Individuals with perceived skill/knowledge receive privileges, and are excused from
certain social obligations
Those with real or perceived skill will see an asymmetrical flow of ``perks'' in their favor.
The tangible goods hypothesis also makes this prediction.
A variety of ethnographic data confirms this prediction. Bateson (1958, p. 91), for
example, found a case among the Iatmul in which a man ``had sufficient standing in the
community to marry his own wife's mother, and this while his wife was still alive and
married to him. He was a great sorcerer and at the same time a great debater and fighter. It
was nobody's business to say him nay ....'' Similarly, Hawkes (1990, 1991) reports that
Ache
Âmales more frequently ``overlook'' sexual liaisons between their wives and highly
skilled hunters.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196182

In small-scale societies, the elderly tend to be prestigious, perhaps due to their accumulated
experience, and this often translates into specific institutionalized perquisites and norm
exceptions. The following age-perk illustrations are culled from Simmons (1945). Aged
Omaha were no longer obliged to scarify themselves when someone died. Among Tasma-
nians, old people get the best food. Beer drinking was formerly an exclusive right of Akamba
``grandfathers,'' and the wood from a certain ``spirit tree'' could only be used by an old man
or woman. Old Todas were accorded special privileges in the ``catching of buffalo'' at the
funeral services. Aged Ainu had the exclusive privilege of conversing with foreigners. Aged
Crow were excused from unpleasant tasks, and at the Sun dance ceremonies they were free to
move at will.
Some experiments support the idea that there is a psychological bias to exempt
prestigious individuals from some social sanctions and confer perquisites on them. For
example, Bickman (1971) showed that subjects are more honest towards high-status
individuals, i.e., those wearing business suits. Ungar (1981), with a similar manipulation,
found that when high-status individuals offer excuses (claiming that somebody else is to
blame) they are derogated less than low-status individuals, even though nobody is fooled
about who is to blame.
5.1.4. Skillful individuals evoke prestige-deference and are well thought of in their social
group
Those with real or perceived skill are popular. Their followers will seek to preferentially
maintain proximity and interaction with them. The tangible goods model predicts this, but our
theory also predicts that even if skilled individuals do not confer tangible benefits they will
remain attractive, as information goods may remain available.
Psychological studies have repeatedly shown a correlation between skill and companion
desirability. Gross and Johnson (1984) measured performance in 12 athletic skills (including
running, swimming, basketball, and soccer abilities), and preferences for work and playmates.
For 69 boys (ages 9±13), their performance scores in 9 of 12 skill areas was positively
correlated with their preference scores; while among 39 girls, the same was true of 7 out of 12
skills (also see Moore & Fall, 1970; Thomas & Chissom, 1973).
Economic experiments using real monetary rewards show that more skilled individuals
receive benefits from less skilled individuals. For example, (Ball, Eckel, Grossman, & Zane,
unpublished observation) manipulated apparent skill by making the results of an apparent
``trivia quiz'' known to participants. In subsequent bargaining interactions, prices were higher
when high-status sellers faced low-status buyers and lower when low-status sellers faced
high-status buyers relative to controls. Performance on the trivia quiz is the only information
participants have about each other, so players seem motivated to defer (at a real financial cost)
to more skilled individuals.
5.2. Predictions about imitation, biased transmission, and influence
5.2.1. People preferentially copy skilled/successful individuals
Miller and Dollard (1941) claimed that the ``prestigious'' Ð defined as skilled or
successful individuals in the immediate circumstances Ð are preferentially copied. Rosen-
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 183

baum and Tucker (1962) tested this by asking pairs of subjects to pick winners in horse races.
Subjects showed a strong propensity to imitate the choices (i.e., horse ``A'' or ``B'') of models
with a ``high competence'' (those with a high frequency of correct answers), even though
each subject faced a different race (see also Baron, 1970). Qualitatively similar results have
been repeatedly obtained (Chalmers, Horne, & Rosenbaum, 1963; Greenfeld & Kuznicki,
1975; Kelman, 1958; Mausner, 1954; Mausner & Bloch, 1957).
Age and competence interact to influence copying bias. In a study using second-graders,
Brody and Stoneman (1985) showed that age and competence interact to influence the
copying bias. In order of decreasing preference, second-graders copy: same-age/high-
competence, younger/high-competence, same-age/low-competence, younger/low-compe-
tence. When competence information on an unrelated task was available, it was a much
stronger determinant of imitation than age. Brody and Stoneman (1981) also show that
children will preferentially imitate older and same age models over younger models when the
models are observed side-by-side in the ``favorite food game.''
Several studies have found that investment decisions are based on copying good
performers. In a multi-round investment experiment involving substantial stakes, Kroll and
Levy (1992) unexpectedly found that when the performance of players (MBA students) was
posted between rounds for all to see, participants ``mimicked'' the behavior of top-
performers, even though players were rewarded competitively. In contrast to a previous
experiment in which nothing was posted, copying high performers here allowed the whole
group to move much closer to the optimal allocation behavior. Similarly, Offerman and
Sonnemans (1998) show that players making investment decisions tended to copy the beliefs
of successful investors (based on past performance) about the current state of the environ-
ment, even though imitators knew that good investors possessed the same information as they
did (also see Pingle, 1995; Pingle & Day, 1996).
5.2.2. Prestigious individuals are influential, even beyond their domain of expertise
Like behavioral traits, the ideas, values, and opinions of prestigious individuals are also
likely to be copied (Gibb, 1954, p. 252). Because figuring out which combinations of
ideas, beliefs, and behaviors make someone successful is costly and difficult, selection
favored a general copying bias, which also tends to make prestigious individuals generally
influential (as people copy and internalize their opinions). Furthermore, much of the
information that leads to success in one domain will often be transferable to others. This is
probably why acquiring skill in one domain (e.g., a martial art) is often touted as
promoting success in many other areas. For example, problem-solving methods, goal-
achieving strategies, eye± hand coordination, control over one's emotions, etc., are useful
across several domains. Thus, prestigious individuals tend to be influential. No competing
theory makes this prediction.
Ethnographic accounts of the influence of prestigious individuals are common. Goldman
(1979, p. 155) writes that a Cubeo headman ``is the chairman of discussions and arbiter of
disputes. He has no authority to order punishment, although his opinion carries weight.''
Among the Meriam of the Torres Strait, great turtle hunters are permitted to speak and are
listened to more than others, despite the fact that their skill in hunting turtles gives no
direct indication of their skill in public affairs or politics (Smith & Bliege Bird, 2000).
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196184

Skilled !Kung orators, arguers, ritual specialists, and hunters ``may speak out more than
others [in group discussion], may be deferred to by other discussants, and one gets the
feeling that their opinions hold a bit more weight than the opinions of other discussants''
(Lee, 1979, p. 343).
Ryckman et al. (1972) demonstrated prestige influence by examining how ``locus of
control'' (``internal'' vs. ``external'') interacts with a prestigious source in attitude change.
``Externals'' are fatalistic, whereas ``internals'' feel ``in control'' due to their skill and/or
determination. Using the topic of ``student activism,'' researchers elicited the opinions of
subjects in three conditions: (1) after hearing the opinion of an expert in student activism
(``relevant condition''), (2) after hearing the opinions of an expert on the Ming Dynasty
(``irrelevant condition''), and (3) after hearing no opinions (the control). They found that
``externals'' (80% of the general population), in both the relevant and irrelevant expert
conditions, and in contrast to the control, shifted their attitudes toward those of the prestigious
individual. The effect size was identical in the relevant and irrelevant conditions. In contrast,
neither expert had any effect on the opinions of internals. The researchers report that the lack
of an effect for internals may result from their apparent awareness of the point of the
experiment and their strong distaste for such manipulations. Ritchie and Phares (1969)
obtained a similar result using the topic of national budget priorities. Rosenbaum and Tucker
(1962) also found that competent models are copied even when their competence is unrelated
to the copier's circumstances.
A study by Tannenbaum (1956) investigated the effect of a prestigious source on attitude
(opinion) change and found important effects. The sources used were: (1) a prominent
individual, (2) a prominent newspaper, and (3) a prominent social group. His 3 3 design
tested for how the same source would bias subjects who (1) had prior respect for the source,
(2) were previously neutral towards it, or (3) had prior disrespect for it, on an attitude item
they (1) previously agreed with, (2) were previously neutral towards, or (3) previously
disagreed with. He tested the effects of the source's positive and negative attitudes and
found that subjects' attitudes were pulled closer to those of the source, even when subjects'
prior opinions were contrary. Methodological problems of earlier research were avoided here
(Asch, 1948).
Social learning should dominate when individual learning is difficult or costly, and this
extends to prestige effects on influence. Studies show source expertise has greater impact
when distraction is high (Kiesler & Mathog, 1968) and when topic-relevant knowledge is
low (Wood & Kallgren, 1988). When the amount of scrutiny devoted to an argument is
manipulated by presenting message recipients with written vs. taped arguments, source
expertise has a greater impact on attitudes (Andreoli & Worchel, 1978). When arguments
are ambiguous (difficult to evaluate), expert sources are more persuasive (Chaiken &
Maheswaran, 1994). And, when messages are run faster to impair ease of processing,
listeners rely more on source expertise than argument quality (Moore, Hausknecht, &
Thamodaran, 1986).
The attractiveness construct shows very little consistency across researchers, and it does
not always involve varying physical attractiveness while controlling for other sources of
likeability (Petty & Wegener, 1998, p. 345). In one study that avoids these problems,
Haiman (1949) showed that a model's prestige and perceived competence significantly
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 185

affected opinions, while perceived physical attractiveness, ``likeability,'' and ``fair-mind-
edness'' did not.
5.2.3. Prestigious individuals are memorable
People should more accurately remember what prestigious individuals do and say
compared to nonprestigious individuals.
Holtgraves, Srull, and Socall (1989) found that subjects recalled better the conversational
contributions of an individual when they were told he was ``the boss.'' In the high-status
condition, subjects were told that one of the individuals was ``the boss.'' In the equal-status
condition, both interactants were identified as ``co-executives.'' Subjects recalled what the
``boss'' said significantly better than what the others said in both treatments. Since ``the boss''
here is not the experimental subject's boss, the effect probably results from inferring that his
superior skills made him someone's boss Ð thus, it's prestige rather than dominance vis-a
Á-vis
the subject.
6
5.2.4. Prestigious individuals, but not dominant ones, are preferentially copied in many
behavioral domains
Those who are freely deferred to Ð even when their skills have not been directly evaluated
by the copier Ð should be copied because of the superior information they typically possess.
However, the same should not be true for dominant individuals, unless individuals are
copying the combat/dominance skills in an effort to become dominant. If nonconformity with
the dominant individual's behavior is taken as a challenge, individuals should copy, but only
in the presence of the dominant individual, and only to appease him (a matter of compliance
rather than internalization; see Kelman, 1958). Prestige-biased information transfers should
have greater postinteraction stability than dominance-induced or pure-tangible-goods-induced
transfers. No competing theory makes this prediction. Some work suggests that dominant
sources have influence, but it does not examine the degree to which it is compliance rather
than internalization (Petty & Wegener, 1998, p. 346).
In his review of the literature on the Diffusion of Innovations, Rogers (1995) argues that
``local opinion leaders'' strongly influence diffusion processes. These leaders are: (1) locally
high in social status (e.g., high status within the village or village cluster), (2) well respected
(indicating prestige rather than dominance), (3) widely connected, and (4) effective social
models. After local opinion leaders adopt a novel practice or technology, it tends to spread
much more rapidly, whereas prior to this it may not have spread at all. Rogers argues that
``social imitation'' is the heart of the diffusion process.
Labov (1972, 1980) shows that dialect change is led by individuals with high status in their
local community. In Philadelphia, upper-class working women pioneer novel sound changes
6
Holtgraves et al. (1989) also manipulated when subjects received the status information: before or after the
reading or observing task (they tested both recall and recognition using both observation of live action and reading
tasks). Contrary to the researchers' initial hypotheses, but in accordance with the predictions of the information
goods theory, high status showed a significant retention effect in the ``before,'' but not in the ``after,'' condition.
Only in the before condition can status information have an effect on attention and encoding processes with
consequent results in memory stability.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196186

that spread through the social strata. Similarly, in Martha's Vineyard, most do not notice
dialect differences between themselves and mainlanders. Yet, they have granted considerable
social status to local fishermen Ð who exemplify the local spirit of resistance and tradition Ð
which has led to inadvertent copying of distinctive linguistic markers.
Using a task that involved guiding a marble through a maze, Bauer, Schlottmann, Bates,
and Masters (1983) showed that female undergraduates preferentially copy the ``style'' of a
prestigious model over a lower-status model, where the time taken to navigate the maze is the
dependent measure. The ``high'' prestige condition subjects observed a ``professionally
attired PhD,'' while in the ``low'' prestige condition subjects observed an ``immature-looking
college student'' (both female). Both models took about 70 s to perform the task, but the
prestigious models displayed a ``slow and deliberate'' style. Subjects copied this style, and as
a consequence, performed much more slowly (in time to completion) than subjects with the
low-prestige model (or in the no-model control). The work confirms previous research on the
effect of prestigious models (Bandura & Kupers, 1964; Harvey & Rutherford, 1960;
Lefkowitz, Blake & Mouton, 1955).
5.3. Predictions about ethological patterns
5.3.1. Prestigious individuals will be gazed at more
For effective copying, clients must gaze at prestigious models longer and more frequently.
No competing theory predicts this.
In the Bauer et al.'s (1983) maze navigation task described above, the researchers also
recorded the amount of time that subjects spent watching each of the models (a measure they
termed ``visual fixation''). Bauer et al. do not provide the data, but they specify that the high-
prestige model was watched significantly more than the low-prestige model. They also note a
significant correlation between ``time to completion'' of the maze (the modeled behavior) and
time spent staring at the model. Thus, those who watch the model longer are also more likely
to acquire her behavior (all participants were female).
5.3.2. Absent other information, individuals should infer prestige status from ethology
In the absence of personal knowledge and experience, people should pick out prestigious
members of a social group simply by observing the ethological displays of group members.
5.3.3. The ethologies and other behaviors elicited by dominant and prestigious individuals
will contrast in predictable ways
These are enumerated as follows: (1) prestigious individuals get direct and plentiful
attention, dominant individuals get furtive glances, (2) in prestige, but not in dominance-
hierarchies, lower-ranked individuals will seek close contact with the higher-ranked, who will
be preferentially copied, and (3) prestigious individuals receive more freely conferred gifts
(and other tokens) than dominant individuals.
Ethological studies of children have explored questions that bear directly on our
predictions. Hold (1976) explored the relationship between status and attention, imitation,
and other pertinent variables. Similarly, Abramovitch (1976) tested the relationship between
dominance and proximity, on the one hand, and dominance and attention, on the other.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 187

In Hold's (1976) ethological study of preschoolers, ``attention'' is a proxy for social rank.
Hold (p. 179) argues that attention structure is ``the best framework for analyzing social rank as
it takes into account all leadership styles.'' However, we wish to distinguish between
``leadership styles,'' and attention can arise in different ways. In dominance, agonistic
encounters will attract much attention because they convey information about changes in the
dominance hierarchy (Joan Silk, personal communication). But most of the time, a dominant
individual gets mostly furtive glances from a safe distance Ð they certainly do not absorb
much attention in long stares from up close. On the contrary, subordinates tend to turn their
whole bodies submissively away from such higher-ups to avoid any appearance of confronta-
tion. In prestige, we predict exactly the opposite: high-status individuals should get sustained
and generalized attention, and clients should keep close proximity whenever possible.
In contrast to the attention measure(s) used by other researchers, Hold's (1976) measure
captures prestige not dominance. Every 5 min, Hold sampled every child and recorded
whether he was ``the center of attention'' (COA) Ð that is, ``being looked at by three or more
children simultaneously,'' who are, ``standing within one meter of the child under observa-
tion, and their bodies and heads had to be oriented in the direction of the subject child'' (Hold,
1976, p. 180). The probability that three individuals will simultaneously find themselves
watching a dominant individual, with their bodies oriented toward him, within a meter of the
dominant individual seems rather low.
7
Therefore, COA, as an indirect measure of prestige,
should strongly and positively covary with several of Hold's other measures, namely, ``being
imitated,'' ``being obeyed,'' ``receiving presents,'' and ``Is told, shown, asked'' (interactional
preference). Table 2a shows the substantial correlations we found for each of these.
Importantly, COA does not correlate with ``aggressor,'' but does correlate with ``protec-
tor.'' Prestigious individuals need not be ``aggressors'' (that is a dominance strategy), but they
possess the social influence required to ``protect'' the lower-ranked. Second, COA is
uncorrelated with ``imitator.'' The more prestige (i.e., skill-derived status) you have, the
fewer skilled models will be available for imitation. Third, COA is either uncorrelated or
negatively correlated, with ``retreater'' and ``onlooker.'' More prestige means you do not need
to retreat; your status leads others to defer to you and this includes restraining their
aggression. And, with higher prestige comes greater number of interactions because you
are sought more often by others.
COA should correlate strongly with an underlying prestige variable, but so should other
prestige-relevant variables. To explore this, we performed a principal components analysis on
Hold's (1976) data (see Table 2b for results) and found that Factor 1 (F1) accounts for 51% of
the variation, and is by far the most important factor. Factor 2 accounts for only 24% of the
variation. F1 seems to capture the hypothesized prestige variable. As expected, the ``center-of-
attention'' variable is most strongly positively correlated with F1, along with ``is imitated,''
and ``present receiver'' (all > .90). ``Is obeyed,'' ``protector,'' and ``is asked, told, shown''
7
Some nonhuman primate subordinates groom dominant individuals, perhaps to avoid future aggressions.
This may be confused with prestige-caused proximity maintenance. However, subordinate groomers do not
maintain proximity to the dominant individual after the interaction, they do not fix their eyes on the dominant
except when grooming, they do not stare, and they do not orient themselves toward the dominant, except in
whatever positions are necessary for grooming.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196188

were correlated at >.80. We also predict the negative relationship between F1 and both
``retreater'' and ``onlooker.'' Prestigious individuals are rarely forced to retreat, owing to their
influence and the desire of others to please them. Their salience prevents them from being
merely an ``onlooker.'' We remain surprised that ``is Avoided'' contradicts our predictions.
Abramovitch's (1976, p. 158) study of dominance interactions in children ages 3 to 6, in
contrast to Hold (1976), operationalized rank as follows:
the number of individuals with whom fights were won and lost .... Rank determinants were
made on the basis of ``property fights,'' struggles to gain or to retain an object (....) A
property fight was defined as an agonistic or quasi-agonistic encounter in which two
individuals actively ``tussled'' or fought over the same object (cf. McGrew, 1972). The child
who obtained or retained the object was scored as the winner ....
Table 2
(a) Correlations of the center of attention variable with 13 measures
Observed behavior category
Correlation with center
of attention measure
Significance
Pvalue
2. Aggressor .016 .9744
3. Protector .795 .03
4. Is imitated .841 .0142
5. Is obeyed .930 .0009
6. Present receiver .794 .0305
7. Is told, shown, asked .912 .0021
8. Is avoided .615 .152
9. Imitator .128 .7975
10. Friendly child .058 .9071
11. Seeking reassurance .135 .7853
12. Retreater .745 .055
13. Onlooker .625 .1425
Correlations clearly predicted by the information goods theory are in bold.
(b) Theoretical predictions and F1 from principal components analysis
Predicted relationship
with prestige
Actual correlations
with F1
1. Center of attention + + .94
2. Aggressor .0 .18
3. Protector + .84
4. Is imitated + + .93
5. Is obeyed + + .88
6. Present receiver + + .92
7. Is told, shown, asked + + .83
8. Is avoided .76
9. Imitator .31
10. Friendly child .0 .01
11. Seeking reassurance .0 .02
12. Retreater .85
13. Onlooker .70
The ``+ + '' symbols predict a strong correlation. The `` + '' symbol indicates a weaker positive correlation. The
``.0'' predicts no relationship. The `` '' indicates a negative correlation.
Observed behavior category
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 189

This is more likely to pick out dominance rather than prestige. As for attention,
Abramovitch (p. 154) throughout claims to have recorded ``glances,'' and in her introduction
notes that subordinates attend to alphas but avoid face-to-face contact and practice gaze
aversion in order to avoid staring. These twin facts suggest she recorded glances, not stares, but
she does not say explicitly how ``glances'' were operationalized. With regard to proximity, it
``was investigated by analyzing the spacing between individuals of various ranks rather than by
looking at actual amounts of space held by particular individuals.'' This measure allows her to
see whether subordinates are avoiding those of higher rank. In contrast to Hold, Abramovitch
found that the high-ranked got more glances, and were avoided more, by subordinates.
8
To this evidence we add that recent ethnographic work on the ethology of status in
Benkulu, Sumatra reveals that people with institutional offices (i.e., with real power over
punishments and rewards, and thus analogous to dominant individuals in nonhuman primate
hierarchies) receive displays from subordinates like those offered to dominant individuals by
nonhuman primates (Fessler, 1999). However, these same displays are not offered to those
whose prominence derives solely from their individual above-average achievements, such as
good poets (Fessler, personal communication).
5.3.4. Prestige rankings are socially transmitted
When information about the relative merits of individuals is hard to collect, people rely on
the ``judgments of others,'' often coded in markers such as medals, a university degree, etc.
No competing theory predicts this.
Evidence for the importance of socially transmitted prestige ranking comes from three
sources. First, the ``expectation states'' literature has found that status differences are
``instantaneously'' created in task groups, rather than evolving out of face-to-face interaction
among participants. People may rely on markers that stand for social judgments, not personal
evaluations (Berger, Rosenholtz, & Zelditch, 1980). Experimental manipulations provide
similar evidence, for subjects often have only the researchers' word on the relative statuses of
available models (e.g., ``leading economist'' vs. ``college sophomore''; Ritchie & Phares,
1969), and yet the predicted effects obtain. Finally, Hatch (1992) found that New Zealand
farmers in small communities rely on the assessments of other farmers, as well as on
individual judgment, to rank farmers according to their skill (which confers local prestige).
Nonfarmers, who lack much direct evidence of farming skill, rely almost entirely on the
socially acquired judgments of farmers to develop their prestige hierarchies, which corre-
spond closely to those formed by the farmers.
5.4. Distinguishing the information goods from other social exchange theories
Social exchange between individuals with asymmetrical endowments of abilities, skills,
resources, or knowledge can produce status differences. Such asymmetrical endowments may
result from genetic inheritance, cultural inheritance, or noncultural environmental differences.
8
Russon and Waite (1991) studied the relationship between dominance rank and imitation in babies (11± 16
months). Although their findings support our theory, we feel the subjects are too young to draw any conclusions.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196190

In general, this logic should apply to many social species, not only humans. However,
because of our unique social learning capacities, humans can trade an asset that most other
species cannot: directly acquirable information. The exchange of information goods results in
status processes characterized by particular psychological, ethological, and sociological
features that would not arise through the exchange for tangible goods alone. Table 3
compares the tangible and information goods models. Both models make many of the same
predictions, but there are interesting differences as well Ð and the preceding section provides
evidence that addresses these differences. Focusing on the exchange for deference benefits for
information goods helps us understand many important human status processes that an
exclusive attention to ``tangible'' tradables would leave unexplained.
This is not the last word on human status. The evolutionary logic of social exchange,
deployed in the analysis of specific kinds of assets, resources, and endowments, will continue
Table 3
Comparison of predictions derived from the information goods theory of nonagonistic status with those arising
from the tangible goods approaches to social exchange
Predictions
Tangible
goods
Information
goods
Positive relationship between status and
skill/knowledge
yes yes
Asymmetrical flow of deference benefits yes yes
Influential opinion of skilled/knowledge
individual
yes yes
Characteristic ethological displays
a
yes
Proximity maintenance when there are
no tangible goods to be had
a
yes
Individuals of perceived skill statements
are more memorable
a
yes
Prestigious individual's are gazed at more,
even outside dyadic interactions
a
yes
Differences between the two models
Opinions adopted from prestigious
individuals are both internalized and
fairly stable.
no yes
Status accorded by age, and to very old
(even when economically marginal)
no yes
Biased imitation (including imitation
outside the model's expert domain)
no yes
Children should bestow prestige-deference
on and preferentially copy models that
are somewhat older and of the same sex
no yes
Noncultural and cultural species (those
with direct social learning) will have
qualitatively different status hierarchies
no yes
a
We can imagine possible ways that the tangible goods approach could generate these predictions. But, to our
knowledge, no one has explored them in detail or made them explicit, and the predictions are not as
straightforward as they are for the information goods model.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 191

to generate additional insights into the psychology, ethology, and sociology of both human
and nonhuman status processes. For example, the nature of assets like coalitional support,
storable food, fungible currencies, and rare linguistic competencies may have generated their
own cascades of psychological adaptations, characteristic ethologies, and sociological
phenomena. And, while they may not have yet spurred genetic adaptations, more recent
cultural developments like long-term debt, large-scale private property, political institutions,
and inheritance laws probably evoke novel recombinations of our old status processes,
thereby giving rise to new sociological phenomena.
6. Conclusion
We have presented a theory for the evolution of prestige and prestige-biased cultural
transmission. We have argued that prestige, in contrast to dominance, is a second avenue to
human status and status-competition, which results from group living plus direct social
learning capacities. From this theory, we have generated a number of testable predictions and
have begun to review evidence from throughout the social sciences to substantiate them. In
future work, we will test our theory by exploring the importance of hunting and related
behaviors to prestige in foraging societies, the emotions involved in prestige and dominance
hierarchies, and the evolution of social hierarchy (from the prestige economies of ``big men''
to the prestige goods of chiefdoms).
Acknowledgments
We thank Nick Blurton-Jones for encouraging this project as it emerged from his
seminar on foragers. We especially thank Robert Boyd, who carefully read several drafts
and coached our project at various stages. We also got invaluable help from Natalie
Smith, Martin Daly, Dan Fessler, Francisco Gil Diaz, Melanie Green, Allen Johnson, Joe
Manson, Clark McCauley, Richard McElreath, Susan Perry, Pete Richerson, Joan Silk, and
Margo Wilson.
References
Abramovitch, R. (1976). The relation of attention and proximity to rank in preschool children. In: M. R. A.
Larsen, & R. R. Larsen (Eds.), The social structure of attention ( pp. 153 ± 176). London: Wiley.
Andreoli, V., & Worchel, S. (1978). Effects of media, communicator, and message position on attitude change.
Public Opinion Quarterly,42, 59± 70.
Arvelo de Jime
Ânez, N. (1971). Political relations in a tribal society: a study of the Ye'cuana Indians of Venezuela.
Doctoral dissertation, Cornell University, Ithaca, NY.
Asch, S. E. (1948). The doctrine of suggestion, prestige, and imitation in social psychology. Psychological Review,
41, 258± 290.
Bailey, R. C. (1991). The behavioral ecology of Efe Pygmy Men in the Ituri Forest Zaire. Ann Arbor, MI:
University of Michigan Press.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196192

Bandura, A., & Kupers, C. J. (1964). Transmission patterns of self-reinforcement through modeling. Journal of
Abnormal and Social Psychology,69, 1±9.
Barkow, J. H. (1975). Prestige and culture: a biosocial interpretation. Current Anthropology,16, 553± 572.
Barkow, J. H. (1989). Darwin, sex, and status: biological approaches to mind and culture. Toronto: University of
Toronto Press.
Baron, R. A. (1970). Attraction toward the model and model's competence as determinants of adult imitative
behavior. Journal of Personality and Social Psychology,14, 345± 351.
Bateson, G. (1958). Naven, a survey of the problems suggested by a composite picture of the culture of a New
Guinea tribe drawn from three points of view. Stanford, CA: Stanford Univ. Press.
Bauer, G. P., Schlottmann, R. S., Bates, J. V., & Masters, M. A. (1983). Effect of state and trait anxiety and
prestige of model on imitation. Psychological Reports,52, 375 ± 382.
Berger, J., Rosenholtz, S. J., & Zelditch, M. (1980). Status organizing processes. Annual Review of Sociology,6,
679± 708.
Bernhardt, P. C. (1997). Influences of serotonin and testosterone in aggression and dominance: convergence with
social psychology. Current Directions in Psychological Science,6, 44± 48.
Bickman, L. (1971). The effect of social status on the honesty of others. Journal of Social Psychology,85, 87 ± 92.
Bikhchandani, S., Hirschleifer, D., & Welch, I. (1998). Learning from the behavior of others: conformity, fads, and
informational cascades. Journal of Economic Perspectives,12, 151 ± 170.
Bliege Bird, R. L., & Bird, D. W. (1997). Delayed reciprocity and tolerated theft. Current Anthropology,38,
49± 78.
Boesch, C., & Tomasello, M. (1998). Chimpanzee and human cultures. Current Anthropology,39, 591± 614.
Boyd, R., & Richerson, P. J. (1985). Culture and the evolutionary process. Chicago, IL: University of
Chicago Press.
Boyd, R., & Richerson, P. J. (1988). An evolutionary model of social learning: the effects of spatial and temporal
variation. In T. R. Zentall & B. G. Galef (Eds.) Social learning: psychological and biological perspectives (pp.
29± 48-xi, 357). Hillsdale, NJ: Lawrence Erlbaum Associates.
Boyd, R., & Richerson, P. J. (1992). Punishment allows the evolution of cooperation (or anything else) in sizable
groups. Ethology and Sociobiology,13, 171± 195.
Boyd, R., & Richerson, P. J. (1996). Why culture is common, but cultural evolution is rare. Proceedings of the
British Academy,88, 77± 93.
Brody, G. H., & Stoneman, Z. (1981). Selective imitation of same-age, older, and younger peer models. Child
Development,52, 717± 720.
Baharody, G. H., & Stoneman, Z. (1985). Peer imitation: an examination of status and competence hypotheses.
Journal of Genetic Psychology,146, 161± 170.
Cartwright, D. (1959). A field theoretical conception of power. In: D. Cartwright (Ed.), Studies in social power
(pp. 150 ± 167 ). Ann Arbor, MI: University of Michigan.
Chagnon, N. A. (1992). Yanomamo. Fort Worth, TX: Harcourt Brace Jovanovich College Publishers.
Chaiken, S., & Maheswaran, D. (1994). Heuristic processing can bias systematic processing: effects of source
credibility, argument ambiguity, and task importance on attitude judgment. Journal of Personality and Social
Psychology,66, 460± 473.
Chalmers, D. K., Horne, W. C., & Rosenbaum, M. E. (1963). Social agreement and the learning of matching
behavior. Journal of Abnormal and Social Psychology,66, 556 ± 561.
Clastres, P. (1998). Chronicle of the Guayaki Indians. New York, NY: MIT Press.
Coleman, J. S. (1961). The adolescent society. New York, NY: Free Press.
Cowlishaw, G., & Dunbar, R. I. M. (1991). Dominance rank and mating success in male primates. Animal
Behavior,41, 1045± 1056.
Dentan, R. K. (1979). The Semai. New York, NY: Holt, Reinhart, and Winston.
Durham, W. H. (1991). Coevolution: genes, culture, and human diversity. Stanford, CA: Stanford Univ. Press.
Ellis, L. (1995). Dominance and reproductive success among nonhuman animals: a cross-species comparison.
Ethology and Sociobiology,16, 257± 333.
Fessler, D. M. T. (1999). Toward an understanding of the universality of second order emotions. In: A. L. Hinton
(Ed.), Biocultural approaches to the emotions (pp. 75± 116). New York, NY: Cambridge Univ. Press.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 193

French, J. R. P., & Raven, B. (1959). The bases of social power. In: D. Cartwright (Ed.), Studies in social power
(pp. 150 ± 167 ). Ann Arbor, MI: University of Michigan.
Gibb, C. A. (1954). Leadership. In: G. Lindzey (Ed.), Handbook of social psychology ( pp. 877 ± 920). Reading,
MA: Addison-Wesley.
Giraldeau, L.-A., & Lefebvre, L. (1986). Exchangeable producer and scrounger roles in a captive flock of feral
pigeons: a case for the skill pool effect. Animal Behavior,34, 797 ± 803.
Giraldeau, L.-A., & Lefebvre, L. (1987). Scrounging prevents cultural transmission of food finding behavior in
pigeons. Animal Behavior,35, 387± 394.
Goldman, I. (1979). The Cubeo. Urbana, IL: University of Illinois Press.
Goodall, J. (1986). The chimpanzees of Gombe. Cambridge, MA: Belknap Press of Harvard Univ. Press.
Goodall, J. (1991). Gombe chimpanzee politics. In: G. Schubert, & R. D. Masters (Eds.), Primate politics
(pp. 105± 137). Lanham, MD: University Press of America.
Goode, W. J. (1978). The celebration of heroes: prestige as a social control system. Berkeley, CA: University of
California Press.
Greenfeld, N., & Kuznicki, J. T. (1975). Implied competence, task complexity, and imitative behavior. Journal of
Social Psychology,95, 251± 261.
Gross, A. M., & Johnson, T. C. (1984). Athletic skill and social status in children. Journal of Social and Clinical
Psychology,2, 89± 96.
Haiman, F. S. (1949). An experimental study of the effects of ethos on public speaking. Speech Monographs,16,
190± 202.
Harvey, O. J., & Rutherford, J. (1960). Status in the informal group: influence and influencability at different age
levels. Child Development,31, 377± 385.
Hatch, E. (1992). Respectable lives: social standing in rural New Zealand. Berkeley, CA: University of
California Press.
Hawkes, K. (1991). Showing off: tests of an hypothesis about men's foraging goals. Ethology and Sociobiology,
12, 29± 54.
Henrich, J., & Boyd, R. (1998). The evolution of conformist transmission and the emergence of between-group
differences. Evolution and Human Behavior,19, 215± 241.
Hill, J. (1984a). Human altruism and sociocultural fitness. Journal of Social and Biological Structures,7,
17± 35.
Hill, J. (1984b). Prestige and reproductive success in man. Ethology and Sociobiology,5, 77± 95.
Hold, B. C. L. (1976). Attention structure and rank specific behaviour in preschool children. In: M. R. A. Chance,
& R. R. Larsen (Eds.), The social structure of attention ( pp. 177 ± 202). London: Wiley.
Holmberg, A. R. (1985/1959). Nomads of the long bow: the Siriono of Eastern Bolivia. Prospect Heights, IL:
Waveland Press.
Holtgraves, T., Srull, T., & Socall, D. (1989). Conversational memory: the effects of speaker status on memory for
the assertiveness of conversation remarks. Journal of Personality and Social Psychology,56, 149±160.
Howe, J. (1986). Kuna gathering. Austin, TX: University of Texas Press.
Huxley, F. (1966). Affable savages. New York, NY: Capricorn.
Kelly, R. L. (1995). The foraging spectrum: diversity in hunter ± gatherer lifeways. Washington, DC:
Smithsonian Press.
Kelman, H. C. (1958). Compliance, identification, and internalization: three processes of attitude change. Journal
of Conflict Resolution,2, 51 ± 60.
Kensinger, K. (1995). How real people ought to live: the Cashinahua of Eastern Peru. Prospect Heights, IL:
Waveland Press.
Kiesler, S. B., & Mathog, R. (1968). The distraction hypothesis in attitude change. Psychological Reports,23,
1123± 1133.
Krackle, W. H. (1978). Force and persuasion: leadership in an Amazonian society. Chicago, IL: University of
Chicago Press.
Kroll, Y., & Levy, H. (1992). Further tests of the separation theorem and the capital asset pricing model. American
Economic Review,82, 664± 670.
Labov, W. (1972). Sociolinguistic patterns. Philadelphia, PA: University of Pennsylvania Press.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196194

Labov, W. (1980). The social origins of sound change. In: W. Labov (Ed.), Locating language in time and space
(pp. 251 ± 265 ). New York, NY: Academic Press.
Leach, E. R. (1977/1954). Political systems of highland Burma. London: London School of Economics and
Political Science.
Lee, R. B. (1979). The !Kung San: men, women, and work in a foraging society. Cambridge: Cambridge Univ. Press.
Lefkowitz, M., Blake, R. R., & Mouton, J. S. (1955). Status factors in pedestrian violations of traffic signals.
Journal of Abnormal and Social Psychology,51, 704±706.
Levi-Strauss, C. (1994). The social and psychological aspects of chieftainship in a primitive tribe: The Nambi-
kuara of northwestern Mato Grosso. In: R. Cohen, & J. Middleton (Eds.), Comparative political systems.
Garden City, NY: Natural History Press.
Liebenberg, L. (1990). The art of tracking. Cape Town: David Philip Publishers.
Mausner, B. (1954). The effect of prior reinforcement on the interaction of observe pairs. Journal of Abnormal
Social Psychology,49, 5± 68.
Mausner, B., & Bloch, B. L. (1957). A study of the additivity of variables affecting social interaction. Journal of
Abnormal and Social Psychology,54, 250± 256.
Maybury-Lewis, D. (1965). The savage and the innocent. Cleveland, OH: World.
Maybury-Lewis, D. (1967). Akwe-Shavante society. Oxford: Clarendon Press.
McGrew, W. C. (1972). An ethological study of children's behavior. New York, NY: Academic Press.
Miller, N. E., & Dollard, J. (1941). Social learning and imitation. New Haven, CT: Yale Univ. Press.
Mish, F. C. (Ed). (1994). Merriam Webster's Collegiate Dictionary. Springfield, MA: Merriam Webster.
Moore, D. L., Hausknecht, D., & Thamodaran, K. (1986). Time compression, response opportunity, and persua-
sion. Journal of Consumer Research,13, 85± 99.
Moore, G. C., & Fall, H. B. (1970). Functional classification for physical education in the upper elementary grades
by peer assessment. Research Quarterly,41, 519± 522.
Moore, O. K. (1957). Divination: a new perspective. American Anthropologist,59, 69± 74.
Murphy, R. F. (1960). Headhunter's heritage: social and economic change among the Mundurucu Indians.
Berkeley, CA: University of California.
Offerman, T., & Sonnemans, J. (1998). Learning by experience and learning by imitating others. Journal of
Economic Behavior and Organization,34, 559±575.
Paine, R. (1973). Animals as capital: comparisons among northern nomadic herders and hunters. In: B. Cox
(Ed.), Cultural ecology: readings on the Canadian Indians and Eskimos ( pp. 301 ± 317). Toronto: Mclelland
& Stewart.
Patton, J. Q. (2000). Reciprocal altruism and warfare: a case from the Ecuadorian Amazon. In: L. Cronk,
N. Chagnon, & W. Irons (Eds.), Adaption and Human Behavior: An anthropological perspective
(pp. 417 ± 436 ). New York: Aldine de Gruyter.
Petty, R. E., & Wegener, D. T. (1998). Attitude change: multiple roles for persuasion variables. In: D. T. Gilbert, S.
T. Fiske, & L. Gardner (Eds.), The handbook of social psychology (pp. 323 ± 390). Boston, MA: McGraw-Hill.
Pingle, M. (1995). Imitation vs. rationality: An experimental perspective on decision-making. Journal of Socio-
Economics,24, 281 ± 315.
Pingle, M., & Day, R. H. (1996). Modes of economizing behavior: experimental evidence. Journal of Economic
Behavior & Organization,29, 191± 209.
Pinker, S. (1997). How the mind works. New York, NY: W.W. Norton & Co.
Renfrew, C., & Bahn, P. (1996). Archaeology: theories, methods and practice. London: Thames & Hudson.
Ritchie, E., & Phares, J. E. (1969). Attitude change as a function of internal external control and communicator
status. Journal of Personality,37, 429± 443.
Rogers, E. M. (1995). Diffusion of innovations. New York, NY: Free Press.
Rosenbaum, M. E., & Tucker, I. F. (1962). The competence of the model and the learning of imitation and
nonimitation. Journal of Experimental Psychology,63, 183± 190.
Russon, A. E., & Waite, B. E. (1991). Patterns of dominance and imitation in an infant peer group. Ethology and
Sociobiology,13, 55± 73.
Ryckman, R. M., Rodda, W. C., & Sherman, M. F. (1972). Locus of control and expertise relevance as determi-
nants of changes in opinion about student activism. Journal of Social Psychology,88, 107± 114.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196 195

Schaller, G. B. (1963). The mountain gorilla: ecology and behavior. Chicago, IL: University of Chicago Press.
Shils, E. A. (1970). Deference. In: E. O. Laumann, P. M. Siegel, & W. R. Hodge (Eds.), The logic of social
hierarchies (pp. 420± 448). Chicago, IL: Markham Publishing.
Silverman, P., & Maxwell, R. J. (1978). How do I respect thee? Let me count the ways: deference towards elderly
men and women. Behavior Science Research,2, 91± 108.
Simmons, L. W. (1945). The role of the aged in primitive society. London: Yale Univ. Press.
Siskind, J. (1973). To hunt in the morning. Oxford: Oxford Univ. Press.
Smith, E. A., & Bliege Bird, R. L. (2000). Turtle hunting and tombstone opening: public generosity as costly
signaling. Evolution and Human Behavior,21, 245±261.
Stammbach, E. (1988). Group responses to specially skilled individuals in a Macacca fascicularis group. Beha-
viour,107, 241±266.
Stearman, A. M. (1984). The Yuqui connection: another look at Siriono deculturation. American Anthropologist,
86, 630± 650.
Strayer, F. F., & Cummins, M. S. (1980). Aggressive and competitive social structures in captive monkey groups.
In: D. R. Omark, F. F. Strayer, & D. G. Freedman (Eds.), Dominance relations: an ethological view of human
conflict and social interaction. Edinburgh: Livingstone.
Tannenbaum, P. H. (1956). Initial attitude toward source and concept as factors in attitude change through
communication. Public Opinion Quarterly,20, 413± 425.
Thomas, J. R., & Chissom, B. S. (1973). Differentiation between high and low sociometric status for sixth-grade
boys using selected measures of motor skill. Child Study Journal,3, 125 ± 130.
Tomasello, M. (1994). The question of chimpanzee culture. In: R. Wrangham, F. De Waal, W. McGrew, & P. G.
Heltne (Eds.), Chimpanzee cultures (pp. 301 ± 317). Cambridge, MA: Harvard Univ. Press.
Tomasello, M., Kruger, A. C., & Ratner, H. H. (1993). Cultural learning. Brain and Behavioral Science,16,
95± 552.
Tooby, J., & Cosmides, L. (1996). Friendship and the banker's paradox: other pathways to the evolution of
adaptations for altruism. Evolution of Social Behaviour Patterns in Primates and Man. Proceedings of the
British Academy,88, 119± 143-vi, 297.
Ungar, S. (1981). The effects of status and excuse on interpersonal reactions to deviant behavior. Social Psychol-
ogy Quarterly,44, 260.
Ventocilla, J., Herrera, H., Nunez, V., & Hams, R. (1995). Plants and animals in the life of the Kuna. Austin:
University of Texas Press.
Weber, M. (1946). For Max Weber: essays in sociology. New York, NY: Oxford Univ. Press.
Weber, M. (1958). The Protestant ethic and the spirit of capitalism. New York, NY: Scribner.
Weisfeld, G. E., & Beresford, J. M. (1982). Erectness of posture as an indicator of dominance or success in
humans. Motivation and Emotion,6, 113± 131.
Whiten, A. (1998). Imitation of the sequential structure of action by chimpanzees (Pan troglodytes). Journal of
Comparative Psychology,112, 270 ± 282.
Whiten, A., & Ham, R. (1992). On the nature and evolution of imitation in the animal kingdom: reappraisal of a
century of research. In: P. J. B. Slater, J. S. Rosenblatt, C. Beer, & M. Milinskypp (Eds.), Advances in the study
of behavior (pp. 239 ± 283 ). New York, NY: Academic Press.
Wood, W., & Kallgren, C. A. (1988). Communicator attributes and persuasion: recipients access to attitude-
relevant information in memory. Personality and Social Psychology Bulletin,14, 172±182.
J. Henrich, F.J. Gil-White / Evolution and Human Behavior 22 (2001) 165±196196