Five Misunderstandings About Cultural Evolution
Joseph Henrich & Robert Boyd & Peter J. Richerson
# Springer Science + Business Media, LLC 2008
Abstract Recent debates about memetics have revealed some widespread misunderstandings
about Darwinian approaches to cultural evolution. Drawing from
these debates, this paper disputes five common claims: (1) mental representations are
rarely discrete, and therefore models that assume discrete, gene-like particles (i.e.,
replicators) are useless; (2) replicators are necessary for cumulative, adaptive
evolution; (3) content-dependent psychological biases are the only important
processes that affect the spread of cultural representations; (4) the "cultural fitness"
of a mental representation can be inferred from its successful transmission; and (5)
selective forces only matter if the sources of variation are random. We close by
sketching the outlines of a unified evolutionary science of culture.
Keywords Dual inheritance theory. Memes . Cultural evolution . Epidemiology of
representations . Cultural transmission . Replicators
Recent debates about the utility of "memes" have revealed some fundamental
misunderstandings about the nature of cultural evolution. Memeticists and their
many critics seem to share the view that evolutionary principles can only be applied
to cultural evolution if culture can be thought of as arising from the transmission of
Hum Nat
DOI 10.1007/s12110-008-9037-1
J. Henrich (*)
Departments of Psychology and Economics, University of British Columbia, 2136 West Mall,
Vancouver, BC V6T 1Z4, Canada
e-mail: joseph.henrich@gmail.com
R. Boyd
Department of Anthropology, University of California-Los Angeles, Los Angeles, CA 90095, USA
e-mail: rboyd@anthro.ucla.edu
P. J. Richerson
Department of Environmental Science and Policy, University of California-Davis, One Shields Ave,
Davis, CA 95616, USA
gene-like replicators (Atran 2001; Blackmore 1999; Dennett 1995). The memeticists
believe that such particles (or at least close approximations) exist, and thus
Darwinian reasoning--which has proven so useful in biology--can be applied to
culture. Their critics argue that replicating particles do not exist, and therefore, that it
is inappropriate to apply Darwinian ideas to culture. While we think that culture is
clearly a Darwinian process (Mesoudi et al. 2004, 2006b), we argue that both camps
have been misguided by an overly enthusiastic analogy between genes and culture.
Because much of culture can be understood in the most general sense as
information stored in human brains--information that got into those brains by
various mechanisms of social learning--we think that population-dynamic concepts
and evolutionary models are extremely useful for understanding how such processes
work. But, and this is a big "but," we maintain that constructing appropriate models
of cultural evolution demands that close attention be paid to the psychological and
social processes involved. From this broader approach, both the memeticists and
their critics labor under a number of recurrent misunderstandings about cultural
evolution. Here we focus on these five:
1. Mental representations are rarely discrete, and therefore models that assume
discrete, gene-like particles (i.e., replicators) are useless (Atran 2001).
2. Replicators are necessary for cumulative, adaptive evolution (Dawkins 1976,
1982).
3. Content-dependent psychological biases are the only important processes that
affect the spread of cultural representations (Sperber 1996).
4. The "cultural fitness" of a mental representation can be inferred from its
successful transmission through the population.
5. Selection can only occur if the sources of variation are random (Pinker 1997).
These assertions are often used to dismiss whole categories of thinking about
cultural evolution. For example, some anti-memeticists have suggested that if there
are no cultural replicators, or if selection requires random variation, researchers
interested in the distribution of representations can ignore cultural evolutionary
models that assume discrete traits (Atran 2001; Boyer 1994). Or, as some
memeticists have suggested, if cultural replicators exist and are operating in
cumulative evolution, one can ignore a lot of complicated mathematical theorizing--
it's just natural selection, after all (Blackmore 1999; Dennett 1995). However, none
of these claims is correct. In the rest of this paper, we will try to convince you of
these facts.
Discrete Replicator Models of Cultural Inheritance can be Useful Even
if Mental Representations are Never Discrete
A great deal of work on cultural evolution assumes that cultural traits can be
modeled as discrete, gene-like entities that are faithfully transmitted from one
individual to another. Memeticists like Blackmore (1999) and Aunger (2002) believe
cultural representations or, as they prefer, memes must be particulate for cumulative
cultural change to occur. Cultural evolutionary theorists (e.g., Boyd and Richerson
1995; Cavalli-Sforza and Feldman 1981; Henrich and Boyd 1998; Rogers 1989)
Hum Nat
have devoted much effort analyzing models of cultural evolution in which cultural
traits are assumed to be discrete--although it is sometimes overlooked that these
theorists have also spent a substantial amount of effort analyzing the evolution of
continuous (non-discrete) cultural traits.
Cognitive anthropologists have criticized such "replicator approaches," arguing
that such thinking is at variance with two observations. First, Atran (2001, 2002) has
suggested that there is no evidence that the mental representations that underpin
cultural traits are discrete, gene-like entities. Instead, he argues that mental
representations are continuously graded entities. Second, Sperber (1996), Atran
(2001), and Boyer (1999) emphasize that, unlike genes, ideas are not transmitted
intact from one brain to another. Instead, the mental representations in one brain
generate observable behavior, a "public representation" in Sperber's terminology.
Someone else then observes this public representation, and then (somehow) infers
the underlying mental representation necessary to generate a similar public
representation. The problem is that there is no guarantee that the mental
representation in the second brain is the same as it is in the first. Any particular
public representation can potentially generate an infinite number of mental
representations in other minds. Mental representations will be replicated from one
brain to another only if most people induce a unique mental representation from a
given public representation. Moreover, inferential processes often systematically
transform mental representations, so that unlike genetic transmission, cultural
transmission is highly biased toward particular representations. Following Sperber
(1996), we call the representations favored by processes of psychological inference
(including storage and retrieval) "cognitive attractors."1
While the nature of the cognitive processes that give rise to social learning are
very much a matter of debate (e.g., Rosenthal and Zimmerman 1978; Tomasello
1996; Whiten 2000), we think it is quite likely that the general picture painted by
Sperber, Boyer, and Atran is correct--cultural transmission does not involve the
accurate replication of discrete, gene-like entities. Nonetheless, we also believe that
models which assume discrete replicators that evolve under the influence of naturalselection-like
forces can be useful. In fact, we think such models are useful because
of the action of strong cognitive attractors during the social learning.
The reason is simple: cognitive attractors will rapidly concentrate the cultural
variation in a population. Instead of a continuum of cultural variants, most people
will hold a representation near an attractor. If there is only one attractor, it will
dominate. However, if, as seems likely in most cases, attactors are many, other
selective forces will then act to increase the frequency of people holding a
representation near one attractor over others. Under such conditions, even weak
selective forces ("weak" relative to the strength of the attractors) can determine the
final distribution of representations in the population.
Henrich and Boyd (2002) analyze a simple mathematical model to show that this
verbal reasoning is cogent. In this paper we represent each individual's mental
1
In recent years our views and those of Sperber, Atran, and Boyer have largely converged. However,
perhaps owing to their important contributions on numerous fronts, a legacy of published claims about the
problems with formal models and the nature of cultural traits continues to sow confusion among many,
especially those not well equipped to digest mathematical models.
Hum Nat
representation as a numerical value (x) between zero and one. For example, x might
represent an individual's beliefs about the moon. Individuals with x=0 perceive the
moon as a self-aware, conscious entity with goals, emotions, and motivations--thus
the moon's behavior can be understood using folk psychology (Leslie 1994). In
contrast, individuals with x=1 see the moon as simply a big rock, lacking goals,
consciousness, and emotions. These individuals attribute the moon's color, shape,
and movement to the effects of non-agentic interactions with light and the gravity of
other mindless bodies, governed by physical laws that operate throughout the
universe. Now, it is possible to imagine moon-concepts that mix these poles (0  x  1).
One could believe, for example, that the moon's movement and shape are out of its
control (governed by physical laws), while its color or hue expresses its mood, which
in turn influences the weather. Or, perhaps the moon's color is 23% controlled by its
emotions and 77% controlled by the laws of light refraction. One might also believe
that on Tuesdays and Thursdays the moon is a goal-oriented agent; on Mondays,
Wednesdays, and Fridays the moon is a big rock; and on the weekends these two
alternate minute by minute. Such beliefs might seem odd to us because they violate
intuitive expectations, which is why cognitive attractors might transform them. In
contrast to intermediate concepts (x values), x=1 or 0 is "easier to think." The formal
model described here uses a one-dimensional representation of x, but this easily
extends to the n dimensions needed to capture the above example.
Individuals in the model acquire their mental representations by observing the
behavior of others. Two cognitive mechanisms affect this learning process. First,
inferential transformation captures the manner in which cognitive processes of
acquisition, storage, and retrieval alter mental representations in ways to favor some
representations over others--cognitive attactors. Because the two extreme representations--"moon
as person" and "moon as rock"--are easier to think, they act as
cognitive attractors in our example. Individuals who observe behaviors that result
from intermediate representations tend to infer mental representations closer to one
of the two attractors. The second process, selective attention, captures the tendency
for individuals to pay particular attention to some individuals more than others. For
example, it could be that in a modern environment, where the representations
favored by science are prestigious, people who hold the "moon as rock"
representation are more successful than those who hold the alternative, and thus
they attract more attention (and are more likely to be learned from). Finally we
assume the effects of inferential transformation are much stronger than the effects of
selective attention.
Figure 1 shows what happens to the distribution of mental representations. In the
underlying simulation, we assumed every mental representation is equally common
initially (this has no impact on the results). The effects of inferential transformation
dominate the early part of the trajectory, rapidly causing almost everyone to have a
representation close to one of the two attractors. Once everyone is clustered around
one of the two attractors, the rest of the trajectory is dominated by the effects of
selective attention. In Henrich and Boyd (2002) we showed analytically that, as long
as there are multiple attractors, the resulting population dynamics and the final
distribution of mental representations are closely approximated by a discrete-trait
replicator dynamics model in which the discrete traits are the strong attractor
locations. This result is confirmed by the simulations results shown in Fig. 1.
Hum Nat
Two conclusions are important here: First, the selective processes (i.e., paying
attention to certain individuals) that generate cumulative adaptive evolution do not
depend on replication, fidelity, or longevity. This model shows that a replicatorapproximating
process can arise and lead to cumulative adaptation even when
representations are non-discrete and are transformed during every acquisition. You
do not need to assume gene-like replicators exist to deploy replicator dynamics.
Second, we showed that the stronger the inferential transformations, the better the
replicator-dynamics approximation. Therefore, contrary to the common assumption
that a rich cognitive architecture relegates the selective process to a limited
importance, we showed that such assumptions imply that selective processes will be
critical to understanding the epidemiology of representations.
The above claims should not be interpreted as asserting that understanding
cognitive attractors and our rich cognitive architecture is unimportant for
understanding cultural evolution. In the above model, it is the attractors that create
quasi-discrete representations for selective forces to act on. What the analysis does
show is that if one believes human cultural transmission is substantially influenced
by potent attractors and rich cognition, one must also recognize the potential
importance of selective forces (even weak ones) in influencing cultural evolutionary
outcomes and the potential utility of replicator models, even though actual mental
representations may not be discrete.
Claidiere and Sperber (2007) have clarified and extended the scope of the above
findings using additional simulations. Their exploration confirms that replicator
dynamics remains a good approximation even when (1) attractors are moderately
strong and (2) selective forces are not at their maximum at any of the attractors. In
this second situation, the strong attractors become replicator-like and the attractor-
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time
Average value of
representations in the domain of attraction for the "0-attractor"
Average value of
representations
in the domain of
the "1-attractor"
Predictions of
replicator dynamics
Mean value of representation
for one run
Mean value of representation
for over 10 runs
Representationvalue
Fig. 1 Results from the simulation model described in the text. The two strong cognitive attractors are
located at 1 and 0 (along Representation value). The overall evolution of the population is well
approximated by a discrete model in which only weak selective forces are present
Hum Nat
replicator with the greatest impact on selective forces (e.g., that which most
increases success) ultimately spreads through the population in a process captured by
replicator dynamics. However, their analysis also emphasizes that adding substantial
noise (stochasticity) to the transmission process makes replicator dynamics a poor
approximation, although selective forces remain important in understanding the
evolutionary dynamics and final distribution of representations. Such situations
require the classical continuous-trait models of cultural transmission (Boyd and
Richerson 1985; Cavalli-Sforza and Feldman 1981). In the next section, we show
how natural selection has equipped learners to cut through such transmission noise
using conformist and blending algorithms, and we discuss how such learning biases
favor the emergence of cumulative adaptive cultural evolution.
Replicators are not Necessary for Cumulative, Adaptive Cultural Evolution
Much confusion about cultural evolution traces to Dawkins's (1976, 1982) argument
that discrete, accurately copied, long-lived "replicators" are necessary for cumulative,
adaptive evolution. Dawkins argues that self-replicating entities are a
requirement for cumulative evolution and must have the following characteristics:
Fidelity. The copying must be sufficiently accurate that even after a long chain
of copies the replicator remains almost unchanged.
Fecundity. At least some varieties of the replicator must be capable of
generating more than one copy of themselves.
Longevity. Replicators must survive long enough to affect their own rate of
replication.
This argument has been repeated and elaborated by Dennett (1995), Blackmore
(1999), Aunger (2002), among others, and has convinced many people that discrete,
gene-like particles are a requirement for adaptive cultural evolution.
While we agree that the existence of replicators is sufficient for cumulative
adaptive evolution, they are not necessary. Any process of cultural transmission that
leads to accurate replication of the average characteristics of the population will
work. Accurate replication at the level of the gene (or meme) will have this effect,
but accurate replication at the population level can arise for other reasons as well.
Here are two examples.
Henrich and Boyd (2002) analyze a discrete trait model with very inaccurate
transmission. They assume that there are two mental representations, A and B. As
before, mental representations are transmitted when one individual observes the
behavior of a second individual and attempts to infer the underlying mental
representation that gave rise to that behavior. Now, however, we assume that this
process is very inaccurate--individuals make the wrong inference with probability
m. Formally, m plays a role identical to mutation in a genetic model. Genes are
replicators because m is tiny, say 10­6
. Here we are going to assume that m is a big
number like 0.2. When m=0.5 there is no transmission at all, so m=0.2 represents
very low fidelity transmission. Thus, if nothing else were going on, cumulative
adaptive evolution would be extremely unlikely. However, we also assume that
individuals have a psychological propensity for conformist transmission, an
Hum Nat
assumption that is both theoretically and empirically well grounded (see below).
Suppose that each learner selects n different individuals to learn from. For each
individual, the learners attempt to infer what the underlying mental representation is
(either A or B) but make an error with probability m for each inference. Based on
these inferences, they then adopt what they think is the most common representation
in their sample. For example, suppose a learner selects five individuals. Three of
these five hold mental representation A, and the remaining two hold B. If our learner
estimates all five accurately, he will adopt A. If he gets one of two holding B
incorrect (and the rest correct), he will still adopt A. But, if he gets one of the three
holding A wrong, he will adopt B. Our results show that conformist transmission
effectively corrects even large errors in transmission, even when the inferential/
transmission channel is 60% noise. The reason for this is simple: errors have a
bigger effect on populations in which one mental representation is common than in
populations in which both mental representations have similar frequencies. However,
when one representation is common, the conformist effect is also stronger and thus
systematically corrects for the effect of errors. Although this model is limited to two
traits, there is no reason to suspect that the insights derived are similarly limited.
Conclusion: fidelity of replication is not required for cumulative adaptation.
Here is a second example. For more than 25 years cultural evolutionary theorists have
analyzed blending models of cultural evolution (e.g., Cavalli-Sforza and Feldman 1981;
Boyd and Richerson 1985:71­79). In such models, no mental representations are
replicated, but nonetheless cumulative evolution is possible. To see this, suppose that
in deciding what length to make his arrow, a hunter samples n models from a larger
population and adopts as his mental representation (his arrow length) the average of
the lengths of the n models. Suppose n=3, and the arrow lengths of the three models
are 16 cm, 20 cm, and 21 cm. This means the hunter adopts an arrow length of 19 cm.
Note, this 19-cm meme is not represented among the n individuals sampled--there is
no replication, fecundity, or longevity. If we further assume that in selecting their n
models, individuals preferentially focus on the best hunters, and that proximity to the
optimal arrow length (say 20 cm) contributes to a hunter's success (on average), then
blending will generate adaptive evolution on arrow length. Laboratory experiments
involving incentivized decisions indicate that such success-biased blending processes
are likely a reasonable approximation for some form of cultural transmission
(McElreath et al. 2005; Mesoudi 2008; Offerman and Sonnemans 1998).
Neither of these mechanisms results in the same kind of relatively "frictionless"
adaptation as genetic replication. Highly accurate, unbiased, genetic replication
allows minute selective forces to generate and preserve adaptations over millions of
years. Error-prone cultural replication, even when "corrected" by a conformist bias,
imposes modest, but still significant forces on the cultural composition of the
population. Similarly, blending inheritance rapidly depletes the variation in a
population necessary for selective processes such as prestige-biased transmission to
have an effect. But, because the inferential processes that underlie cultural
transmission are noisy, it is likely that they can maintain lots of variation. However,
this also means that they are likely to create evolutionary forces that act to change
the mean, and thus compete with selective forces.
The contrasts between cultural and genetic evolution provide more reasons, not
fewer, for analyzing formal cultural evolutionary models. The forces that are
Hum Nat
important for understanding cultural evolution (such as non-random errors and
blending) are likely not the same forces that are important for understanding genetic
evolution. Population-based models of cultural evolution can be useful, but they
have to pay careful attention to these differences.
Content-Dependent Psychological Biases are not the Only Processes that Affect
the Spread of Cultural Traits
Content bias, or how the content of memes "fit into" the cognitive structure of
human minds, is not the only important process for understanding cultural evolution
(Boyer 1994, 1999).2
Genetically maladaptive memes about religion, food taboos,
ghosts, and so on, may readily spread because of their ability to exploit aspects of
human psychology in ways that make them more likely to be acquired, stored, and
transmitted. However, summarizing evidence from across the social sciences,
Henrich and Gil-White (2001) show that humans are quite selective in picking the
individuals they will learn from, or be influenced by ("imitate," if you will). Human
psychology seems geared up to selectively extract useful (locally adaptive)
information from the individual(s) most likely to possess such information. Skill,
success, and prestige all make individuals substantially more likely to be learned
from, or imitated. This psychological propensity for "model selectivity" seems to
operate across most, if not all, domains of culture (Henrich and McElreath 2006;
Henrich and Henrich 2007), from dialects and word choice to political opinions,
suicide, food preferences, and technical innovations (such as using fertilizer). The
classic literature on the diffusion of innovations (Rogers 1995) is a rich source of
examples. This means that a meme's mimetic fitness (vs. genetic fitness) will depend
jointly on how attractive its content is to human brains and how it affects an
individual's likelihood of being selected as a cultural model by other individuals.
Consider the following example. In a fishing village on an Indonesian island, an
old man is out fishing at night in small boat. The next morning he is found dead in
his boat, which is filled with a massive catch. A rumor begins to spread that a
demon-fish, common in local mythology, sucked out the man's soul because he was
fishing at night. Individuals who believe this rumor stop fishing at night (which is
often the most productive time to fish). For simplicity, we assume that individuals
either believe the meme, or not. The variable p gives the frequency of individuals in
this large village who believe the fish-demon has returned and do not fish at night.
From the meme's perspective, the relative mimetic fitness of the demon-fish belief is
wf=a+, while the relative mimetic fitness of not believing the rumor is n=+.
The parameter  is the baseline mimetic fitness,  is the strength of the meme's
content bias, and  is the cost to an individual who bears it in terms of their
likelihood of being selected as a cultural model by a learner. An individual's
likelihood of being selected as a model is affected because not-fishing at night means
2
Broadly, "content bias" refers to any situation in which a meme's representational content influences its
likelihood of transmission. Such biases arise from the interaction of the representational content of the meme
and human psychologies. While this includes reliably developing aspects of human psychology (e.g., incest
aversion favoring favor ubiquitous oedipal narratives; Johnson and Price-Williams 1996), it also includes
the "fit" between different memes, or different experiences and certain memes (cultural psychologies).
Hum Nat
fewer fish to sell, and thus less extra money for clothing, sugar, house maintenance,
throwing feasts, and the children's health needs--all of which may make one more
likely to be selected as a cultural model. Putting these expressions into standard
replicator dynamics gives us
p  p 1  p  g   
where p is the change in the frequency of rumor believers. This equation, as it
stands, tells us that there are two potential stable equilibria: either everyone will
come to believe the fish-demon story and cease all night fishing, or the success costs
of not fishing will dominate and the rumor will not be favored in the long run.
Clearly, just because the demon-fish story is fun to tell, easy to remember, built on
widely believed local mythologies, and interacts with innate inferential machinery in
interesting ways (Boyer 2001) does not guarantee it will spread if possessing the
belief makes one less likely to be selected as a cultural model.
The point of this example is to show that the human mind's tendency to focus
attention preferentially on certain individuals (independent of mimetic content)
means the usual approach to memetic reproduction is insufficient. It further means
that whether a particular genetic-fitness-reducing meme can spread, and how far it
will spread, depends on the details--the dynamics of which are best understood by
formally modeling the social and psychological processes involved. No categorical
claims based on hand-waving arguments about the relationship between genetic and
mimetic fitness are likely to hold, as Rogers (1989) demonstrates. For example, just
because something is transmitted "horizontally" within a generation tells us nothing
the genetic adaptiveness of those memes. We should also note at this point that the
appropriateness of tracking fitness from the perspective of the meme (assigning
fitnesses to alternative memes) or to individuals (or groups) is merely a modeling
convenience. For example, just as with genetic evolution, it is not "more correct" to
view fitness in association with memes, individuals, or groups. As with genetic
models, the above model can be fully derived from the perspective of individuals,
rather than memes, by specifying the individual's tendency to transmit particular
ideas, rather than from the meme's ability to transmit itself. Different fitness tracking
systems may allow certain aspects of the problem to be studied more or less
effectively, but they are all formally identical at some level (McElreath and Boyd
2007).
Successful Diffusion is not a Measure of Fitness
Authors who adopt the selfish meme concept often give us no causal idea of what
actually bestows different "fitnesses" of alternative memes. How do we know
whether a bit of a tune or a catch phrase is a fit meme? Often, it seems, only by
asking whether the meme has successfully spread.
This is dangerous territory. Used in this way, natural selection is a useless, or even
misleading, tautology. For example, a recessive gene causing a severe vision
disorder called achromatopsia has spread to roughly 30% of the population on the
Micronesian island of Pingelap. Sufferers of achromatopsia cannot see well under
any circumstances, and they are especially disadvantaged in the bright sunlight of a
Hum Nat
tropical island. Nonetheless, there is no doubt that this gene spread on Pingelap
because people who carried it had more descendants than those who didn't carry the
gene. If we were to infer the relative fitness of the achromatopsia and normal alleles
from this spread, we would conclude that the achromatopsia allele had higher fitness.
However, this doesn't mean that achromatopsia was favored by selection, because
the achromatopsia didn't cause their increased reproductive success. Rather, it seems
that the gene was carried by members of a chiefly lineage whose social position
allowed them to survive the aftermath of a severe typhoon that struck the island during
the 1700s--it likely spread by a combination of drift and a chance covariation with
social status. The same kinds of phenomena are likely at work in cultural evolution.
Otherwise-deleterious or unattractive ideas and practices often spread because they
happen to be statistically correlated with attractive individuals or successful groups.
Why did the English language rapidly spread across North America during the
eighteenth and nineteenth centuries? Certainly not because it is an intrinsically more
attractive mode of expression than Cherokee or Apache. Rather, it spread because it
happened to be associated with the military advantages, technological innovations, and
infectious diseases that allowed English speakers to conquer the native cultures of North
America. Similarly, the Western business suit has also spread across the world in the
twentieth century not, we conjecture, because the four-in-hand tie is intrinsically more
attractive than its many alternatives, but because it happens to be associated with the
economic and military prowess of the West.
Evolutionary biologists escape this circularity in defining fitness because they
have independent means of predicting which genetic variants are more fit. Peter and
Rosemary Grant's (1986) famous studies of the evolution of beak depth in
Galapagos finches illustrate how this works. During a severe drought, the birds
evolved stouter beaks. We know this change is due to selection because the
investigation showed that (1) large, tough seeds predominated during the drought,
(2) finches with stouter beaks were better able to crack larger seeds, and (3) beak
stoutness is heritable. Similarly, we know that the human pelvis was shaped by
selection because we understand the biomechanics of bipedal locomotion.
Evolutionary biologists are also in the habit of subdividing their concepts--
selection especially--to create a rather diverse family of sub-concepts. These include
classics like Darwin's two kinds of sexual selection as well as modern concepts like
frequency- and density-dependent selection. The reason is that experimentalists are
typically concerned, like the Grants, with concrete details. The concrete cases of
selection involve everything that happens to heritably varying organisms as their daily
lives unfold. An incredible variety of things can and does happen, and evolutionary
biologists collect similar ones together using a rough-and-ready taxonomy to cope
with the otherwise overwhelming diversity. Notice that we have been doing the same
thing with the psychological forces that affect the distribution of representations.
Attractors are different from conformity, and both are different from prestige-based
imitation (see Richerson and Boyd 2005:69 for a taxonomy of forces).
These principles should also apply to the study of memes. The rapid spread of the
New World's sweet potato throughout highland New Guinea during the 1700s is
easy to understand. Sweet potatoes have higher yields and grow at higher altitudes
than yams, the previous staple. People noticed these properties and avidly adopted
the new crop. Here we have a causal theory that links evolved psychology (people
Hum Nat
like to be well fed) with the preference for one cultural variant over another. In many
cases, however, it is difficult to predict which representations will spread because we
do not understand much about the underlying psychological or ecological processes
(but see, for example, Martindale 1975; Rogers 1995; Taylor 1996). Why do we like
particular musical forms or literary devices? Why do some religious beliefs spread
while others fail? Why do some religious beliefs spread in some groups (e.g.,
Christianity in Polynesia) even while they decline in their homelands (Christianity in
Europe)?
Even for technological traits, there are many puzzles such as the fact that
throughout New Guinea the idea of fletching arrows has never caught on, while just
across the Torres Strait in Australia the idea of bows and arrows en toto never
spread, or why the Tasmanians abandoned or never adopted bone tools, cold-weather
clothing, barbed spears, and fishing during their ten thousand years of isolation
(Henrich 2004). These questions are not unanswerable in principle, but meme
theory, as it stands, seems ill-equipped to tackle them. We believe that constructing a
full-fledged theory of cultural evolution requires considering a longish list of
psychological, social, and ecological processes that interact to generate the
differential "fitness" of cultural variants.
Selection Does not Require Random Variation
Many people have argued that selection cannot affect cultural evolution because
cultural variation, unlike genetic mutations, is not based on random copying errors.
Instead, the argument goes, cultural changes are systematic, driven by attempts to
innovate or by the cognitive machinery by which individuals make inferences about
the beliefs of others, and this means selective processes are not important. For
example, Pinker (1997:209) makes this argument in the following passage:
A meme impels its bearer to broadcast it, and it mutates in some recipients: a
sound of a word, or a phrase is randomly altered. Perhaps, as in Monty Python's
The Life of Brian, the audience of the Sermon on the Mount mishears "Blessed
are the peacemakers" as "Blessed are the cheesemakers." The new version is
more memorable and comes to predominate in the majority of minds. It too is
mangled by typos and speakos and hearos, and the most spreadable ones
accumulate, gradually transforming the sequence of sounds. Eventually, they
spell out "That's one small step for man, one giant leap for mankind." I think
you'll agree that this is not how cultural change works. A complex meme does
not arise by the retention of copying errors.
We will agree that Pinker provides a pithy example showing why selection isn't
everything. The problem is that he then concludes that it is nothing. If selection does
not explain complex design in cultural evolution by itself, then it is of no importance.
This is mistaken. There is no doubt that, as people acquire and modify beliefs, ideas,
and values, the variation that is generated can be highly non-random, and these nonselective
processes shape cultural variation. But so what? Selection occurs anytime
there is heritable variation that affects survival or reproduction (transmission). It
does not matter whether the variation is random. In cultural evolution, unlike genetic
Hum Nat
evolution, natural selection may compete with other important directional processes
created by human psychology. In any given case, whether one or another force will
predominate is an empirical issue.
We also think that Pinker overestimates the importance of conscious problemsolving
in innovation (also see Mesoudi et al. 2006a, b). Pinker (1997:209) writes
that innovation occurs when "some person knuckles down, racks his brain, musters
his ingenuity, and composes or writes or paints or invents something." This
pervasive "Myth of the Heroic Inventor," as scholars of the history of technology
called it (Basalla 1988; Diamond 1997), fails to sufficiently recognize (1) the central
importance in the history of science and technology of luck, happenstance, and
recombination, and (2) that most great inventors actually make only incremental
additions to the existing or emerging capacities or understandings of their times. We
leave a complete defense of these views to the existing historical works that have
confronted this in detail (e.g., Basalla 1988; Diamond 1997; Hager 2007; Meyers
2007; Sneader 2005; Williams 1987) but provide five illustrative examples here.
1. James Watt "invented" the steam engine in 1769 after repairing a Newcomen
steam engine constructed 57 years earlier. This engine was modified from
Thomas Savery's design of 1698, the components of which trace to seventeenthcentury
Europe and thirteenth-century China. After dissecting the steam engine,
famed historian Joseph Needham concluded that "No single man was the father
of the steam engine; no single civilization either" (quotation from Basalla 1988).
2. The discovery of penicillin, and the dawn of the age of antibiotics, began when
Alexander Fleming returned from holiday to find that his Petri dishes had been
contaminated with mold. Seeking to clean up his chronically messy laboratory,
he dumped the whole batch of dishes into a laboratory sink where they sat until
he retrieved an unsubmerged disk to show a visitor. He happened to notice that
while the mold was growing fine, the staph was dead. Penicillin was discovered
as a result of luck and messiness.
3. Establishing the germ theory of disease required obtaining pure cultures of
bacteria. In the nineteenth century, dozens of researchers were trying to figure
out how to do this, without success. Robert Koch solved the problem when,
while cleaning up his laboratory, he ran across a half of a boiled potato that had
been carelessly left for a few days. Koch noticed the growth of discrete reddish
dots at different places on the white potato and realized that one needed a solid,
not a liquid, medium. He went on to firmly link specific pathogens with specific
diseases, and to develop his four postulates for making this link based on
cultivating a pure culture (Hager 2007). None of this could have occurred
without the carelessly left potato.
4. Edison's "invention" of the incandescent light bulb only improved on many
other such bulbs patented between 1841 and 1878 by a wide variety of
inventors. Of course, if you are from Britain, Sir Joseph W. Swan is the inventor
of the incandescent light bulb, whereas if you are from Russia, it's A. N.
Lodygin (Conot 1979).
5. The Wright brothers' invention of the airplane built on existing manned gliders
and unmanned powered airplanes. Their contribution was a recombinant of
existing lines of technology (Diamond 1997).
Hum Nat
Pinker errs in conflating exemplars of a work within a tradition with the tradition
itself. Mozart composed innovative symphonies but he did not invent the symphony.
Watt built innovative steam engines but he did not invent the steam engine.
We emphasize, however, that even if Pinker's view of innovation is correct, this
does not mean (1) that selective forces cannot operate (since they require only
variation, not random variation), and (2) that cultural evolution cannot be understood
and modeled as a population process. What is important is that one has to construct
specific models of cultural evolution, based on what is known of the underlying
individual-level decision processes.
Charting a Course: Foundations for a Unified Science of Cultural Phenomena
In this final section we briefly sketch some of the essential components for a
successful research program in cultural evolution and human behavior. We do not
strive here to take a full accounting of all the important and necessary domains of
inquiry (see Mesoudi et al. 2006b; Richerson and Boyd 2005), but only to highlight
certain areas.
Rich Psychology
Two key components of psychology are of most direct relevance to understanding
cultural evolution. The first involves understanding how cognition directs social
learning toward particular individuals or ideas, beliefs, and so on, and how cognition
extracts, or makes use of, the socially available information in a population. For
example, evolutionary theory applied to social learning predicts that individuals
should use model-based cues of skill, success, health, prestige, and self-similarity (e.g.,
sex and ethnicity) to figure out who to pay particular attention to for cultural learning
(Boyd and Richerson 1985: chapter 8; Henrich and McElreath 2006; Henrich and
Gil-White 2001). Similarly, theory also indicates that individuals should, in the
absence of decisive social information from skilled (or successful, etc.) individuals
or high-quality environmental information, rely on copying the majoritarian behavior--
conformist transmission (Boyd and Richerson 1985: chapter 7; Henrich and Boyd
1998). Recent work has also suggested that, to avoid exploitation during cultural
learning by models that seek to convey one representation while actually holding
another, natural selection has equipped learners to rely on inferentially potent
displays when acquiring memes that can be cheaply transmitted using verbal (or
other symbolic) communication. Inferentially potent displays are actions that would
likely only be performed by those models who actually hold (believe in) the memes
they have expressed verbally (Henrich 2007). Both experimental and field evidence
support these different theoretical predictions to varying degrees (Henrich and
Henrich 2007).
The second component of psychology involves inferential, storage, and recall
processes (Sperber 1996). How do cognitive processes organize and interpret
information coming in from the social world? The idea here is to open the black box
of imitation. In acquiring something like tool-making skill, how do individuals
decompose a continuous stream of behavior into steps? How do individuals infer the
Hum Nat
goals of the individual they attempt to imitate? How do the building blocks of
inference (e.g., theory of mind, nave physics, folk biology) shape the inferences
individuals draw from observing these selected cultural models (Atran 1998, 2002;
Boyer 2001)? Given that public representations of underlying mental representations
are nearly always incomplete, how do inferential processes reconstruct mental
representations? How do inference processes deal with the range of different public representations
produced by a single individual? How do culturally acquired representations
influence subsequent learning processes?
Population Processes
Knowledge of psychological mechanisms and cognitive structures is insufficient to
predict the epidemiology of cultural representations in most cases. Understanding
the population-level consequences of individuals, each possessing learning psychologies
and interacting, requires the construction of formal cultural evolutionary
models. Even with simple psychological assumptions such models have proven
useful in understanding a wide range of phenomena (Boyd and Richerson 1985),
including (1) the origins of ethnic groups (Boyd and Richerson 1987; McElreath
et al. 2003), (2) evolution of economic specialization and the emergence of largescale
cooperation (Boyd and Richerson 1992; Henrich and Boyd 2001, in press;
Panchanathan and Boyd 2004), (3) conditions for technological accumulation
(Henrich 2004; Shennan 2001), (4) emergence of a culture of honor (McElreath
2003), and (5) dynamics of the diffusion of innovations (Henrich 2001).
Cobbling up from psychological mechanisms to population processes is also
increasingly informing research on larger-scale cultural evolutionary processes.
Important work arising from evolutionary archaeology and behavioral ecology is
exploring how to use archaeological, ethnographic, and historical data to reconstruct
cultural lineages, assess linkages between different cultural traits, recognize adaptive
processes, and predict migration patterns (Bentley et al. 2007; Collard et al. 2006;
Eerkens and Lipo 2005; Lipo et al. 2006; Shennan in press).
Ecological-Economic Processes
The fitness of cultural variants may be determined entirely by psychological forces,
but more commonly different variants have consequences in the environments in
which people live. These consequences will often interact with psychological forces
(Baum 2005). People will find some cultural variants useful in one environment and
another useful in a different environment; reinforcement at the individual level will
create content-based biases favoring different variants in different environments.
Similarly, economic success often translates into prestige and model-based cultural
transmission biases, and different activities lead to economic successes in different
economies. But residual effects not accounted for by psychology are also liable to be
common. The many forms of natural selection are candidates to influence cultural
evolution and to produce cultural fitnesses that are close analogs to genetic fitnesses.
But these effects are importantly different from those generated by psychological
processes (Richerson and Boyd 2005).
Hum Nat
Evolutionary and Culture-Gene Coevolutionary Origins
What are the evolutionary origins of the psychological capacities that give rise to
cultural evolution? Understanding the origins of the psychological mechanisms
discussed above goes hand-in-hand with hypothesizing what the details of those
mechanisms might be. To date, we and our colleagues have explored the evolution
of, and trade-offs regarding, parent-offspring transmission (McElreath and Strimling
in press), conformist transmission, prestige-biased transmission, and ethnic biases
(McElreath et al. 2003). We have also sought to understand why human-like cultural
and cognitive abilities are so rare in nature (Boyd and Richerson 1996).
In our view, one of the most important, and least explored, avenues of
evolutionary inquiry in human behavior and psychology are the "Baldwinian"
processes that arise from the interaction of cultural and genetic transmission.
Cultural traditions manifestly change the environments faced by human genes
(Durham 1991; Henrich and Henrich 2007; Laland et al. 2000; McElreath et al.
2003; Richerson and Boyd 1998, 2000, 2005). This opens novel evolutionary
pathways that are not available to species that are not heavily reliant on social
learning for acquiring phenotype (Mesoudi and Laland 2007). Human teeth, lack of
body hair, digestive processes, malaria resistance, and manual dexterity certainly
cannot be understood without realizing that genes responded to the cultural
transmission of the use of clothing, fire, agriculture, and tools (Wrangham et al.
1999). Similarly, culture has likely shaped cognition, both directly and indirectly by
changing the selective environment faced by genes. Despite numerous physiological
examples and gene-culture coevolution and a rock-solid theoretical foundation,
mainstream evolutionary psychology (e.g., Pinker 1997; Tooby and Cosmides 1992)
has largely ignored gene-culture coevolution.
Methodological Pluralism
The theoretical and empirical demands of this program exceed those available in any
one discipline. Theoretically, tools have been drawn from population genetics,
communication theory, epidemiology, learning theory, statistics, and evolutionary
game theory. In the future, insight may come from fields as diverse as information
theory and statistical mechanics. Empirically, our program demands the integration
of both observational and experimental data from human biology, psychology,
economics and anthropology (e.g., Henrich et al. 2004; Mesoudi 2008), as well as
studies of processes of long-term change from paleoecology, history, and
archaeology (e.g., Henrich 2004; Richerson and Boyd 2005; Shennan 2003, in
press).
Conclusion
We believe that the Darwinian approach differs from traditional social sciences
approaches in ways that are not yet fully appreciated. All five misunderstandings we
describe here have a common theme. They result from a tendency to think
Hum Nat
categorically rather than quantitatively. Take the meme controversy. The disputants
take the main issue to be whether or not culture is highly analogous to genes. If so,
then their evolution is to be explained by Darwinian fitness; if not, Darwinism is
useless. If we are correct, this debate is a red herring. The proper approach is to
recognize that the analogy between genes and culture is quite loose, and to build up
a theory of cultural evolution that takes into account the actual properties of the
cultural system (Mesoudi et al. 2006b). Culture has a much richer array of
psychological processes with population level consequences than is the case for
genes. But neither particular psychological forces nor the integrated effect of all such
forces in any way rules out a role for natural selection, or vice versa. The matter
turns entirely on how the numbers work out in the particular case at hand. Because
its most complex examples are confined to our species, culture can hardly prove to
be as diverse in its outcomes as organic evolution. However, we expect that it will
turn out to be a baroque system. The balance of evolutionary forces on culture no
doubt changed with the advent of mass literacy and mass media; no doubt
economically important traits differ from symbolic ones; and so forth. To paraphrase
something J. B. S. Haldane is supposed to have said: Culture is not only queerer than
we imagine but, as of this moment, queerer than we can imagine.
Acknowledgments We thank Natalie Henrich for her comments as well as the three insightful
anonymous reviewers. Boyd and Henrich thank the Institute for Advanced Study in Berlin, where much of
this paper was written. Boyd's contribution was partially funded by a grant from the MacArthur
Foundation and Henrich's contribution was partially funded by the National Science Foundation.
References
Atran, S. (1998). Folk biology and the anthropology of science: cognitive universals and cultural
particulars. Behavioral and Brain Sciences, 21, 547­609.
Atran, S. (2001). The trouble with memes: inference versus imitation in cultural creation. Human Nature,
12, 351­381.
Atran, S. (2002). The religious landscape. Cambridge: Cambridge University Press.
Aunger, R. (2002). The electric meme: A new theory of how we think. New York: Free Press.
Basalla, G. (1988). The evolution of technology. New York: Cambridge University Press.
Baum, W. M. (2005). Understanding behaviorism: Behavior, culture, evolution. Oxford: Blackwell.
Bentley, R. A., et al. (2007). Regular rates of popular culture change reflect random copying. Evolution
and Human Behavior, 28, 151­158.
Blackmore, S. (1999). The meme machine. Oxford: Oxford University Press.
Boyd, R., & Richerson, P. J. (1985). Culture and the evolutionary process. Chicago: University of
Chicago Press.
Boyd, R., & Richerson, P. J. (1987). The evolution of ethnic markers. Cultural Anthropology, 2, 27­38.
Boyd, R., & Richerson, P. J. (1992). Punishment allows the evolution of cooperation (or anything else) in
sizable groups. Ethology & Sociobiology, 13, 171­195.
Boyd, R., & Richerson, P. J. (1995). Why does culture increase adaptability? Ethology & Sociobiology, 16,
125­143.
Boyd, R., & Richerson, P. J. (1996). Why culture is common, but cultural evolution is rare. Proceedings of
the British Academy, 88, 77­93.
Boyer, P. (1994). The naturalness of religious ideas. Berkeley: University of California Press.
Boyer, P. (1999). Cognitive tracks of cultural inheritance: how evolved intuitive ontology governs cultural
transmission. American Anthropologist, 100, 876­889.
Boyer, P. (2001). Religion explained: The evolutionary origins of religious thought. New York: Basic
Books.
Hum Nat
Cavalli-Sforza, L. L., & Feldman, M. (1981). Cultural transmission and evolution. Princeton: Princeton
University Press.
Claidiere, N., & Sperber, D. (2007). The role of attraction in cultural evolution. Journal of Cognition and
Culture, 7, 89­111.
Collard, M., Shennan, S. J., & Tehrani, J. J. (2006). Branching, blending, and the evolution of cultural
similarities and differences among human populations. Evolution and Human Behavior, 27, 168­184.
Conot, R. E. (1979). A streak of luck. New York: Seaview Books (distributed by Simon and Schuster).
Dawkins, R. (1976). The selfish gene. Oxford: Oxford Unversity Press.
Dawkins, R. (1982). The extended phenotype. Oxford: Oxford University Press.
Dennett, D. (1995). Darwin's dangerous idea. London: Penguin Press.
Diamond, J. M. (1997). Guns, germs, and steel: The fates of human societies. New York: W.W. Norton & Co.
Durham, W. H. (1991). Coevolution: Genes, culture, and human diversity. Stanford: Stanford University
Press.
Eerkens, J. W., & Lipo, C. P. (2005). Cultural transmission, copying errors, and the generation of variation in
material culture and the archaeological record. Journal of Anthropological Archaeology, 24, 316­334.
Grant, P., & Grant, R. (1986). Ecology and evolution of Darwin's finches. Princeton, NJ: Princeton
University Press.
Hager, T. (2007). The demon under the microscope. New York: Three Rivers Press.
Henrich, J. (2001). Cultural transmission and the diffusion of innovations: adoption dynamics indicate that
biased cultural transmission is the predominate force in behavioral change and much of sociocultural
evolution. American Anthropologist, 103, 992­1013.
Henrich, J. (2004). Demography and cultural evolution: why adaptive cultural processes produced
maladaptive losses in Tasmania. American Antiquity, 69, 197­214.
Henrich, J. (2007). The evolution of costly displays, cooperation, and religion: Inferentially potent
displays and their implications for cultural evolution. Economics and Evolution working paper series.
Evolutionary Economics Group, MPI Jena, Germany. <Available at https://papers.econ.mpg.de/evo/
discussionpapers/2007­21.pdf>
Henrich, J., & Boyd, R. (1998). The evolution of conformist transmission and the emergence of betweengroup
differences. Evolution and Human Behavior, 19, 215­242.
Henrich, J., & Boyd, R. (2001). Why people punish defectors: weak conformist transmission can stabilize
costly enforcement of norms in cooperative dilemmas. Journal of Theoretical Biology, 208, 79­89.
Henrich, J., & Boyd, R. (2002). On modeling cultural evolution: why replicators are not necessary for
cultural evolution. Journal of Cognition and Culture, 2(2), 87­112.
Henrich, J, & Boyd, R. (In press). Division of labor, economic specialization, and the evolution of social
stratification. Current Anthropology, in press.
Henrich, J., & Gil-White, F. (2001). The evolution of prestige: freely conferred deference as a mechanism
for enhancing the benefits of cultural transmission. Evolution and Human Behavior, 22, 165­196.
Henrich, J., & McElreath, R. (2006). Dual inheritance theory: The evolution of human cultural capacities
and cultural evolution. In R. Dunbar, & L. Barrett (Eds.) Oxford Handbook of evolutionary
psychology (pp. 555­570). Oxford: Oxford University Press.
Henrich, J. et al. (Eds.) (2004). Foundations of human sociality: Economic experiments and ethnographic
evidence from fifteen small-scale societies. Oxford: Oxford University Press.
Henrich, N. S., & Henrich, J. (2007). Why humans cooperate: A cultural and evolutionary explanation.
Oxford: Oxford University Press.
Johnson, A. W., & Price-Williams, D. R. (1996). Oedipus ubiquitous: The family complex in world folk
literature. Stanford: Stanford University Press.
Laland, K. N., Odling-Smee, J., & Feldman, M. W. (2000). Niche construction, biological evolution, and
cultural change. Behavioral and Brain Sciences, 23, 131­175.
Leslie, A. M. (1994). ToMM, ToBY, and agency: Core architecture and domain specificity. In L. A.
Hirschfeld, & S. A. Gelman (Eds.) Mapping the mind: Domain specificity in cognition and culture
(pp. 119­148). Cambridge: Cambridge University Press.
Lipo, C. P. (Ed.) (2006). Mapping our ancestors. Piscataway, NJ: Aldine Transaction.
Martindale, C. (1975). Romantic progression: The psychology of literary history. Halsted: Washington
Hemisphere.
McElreath, R. (2003). Reputation and the evolution of conflict. Journal of Theoretical Biology, 220, 345­357.
McElreath, R., & Boyd, R. (2007). Modeling the evolution of social behavior. Princeton: Princeton
University Press.
McElreath, R., & Strimling, P. (2008). When natural selection favors imitation of parents. Current
Anthropology, in press.
Hum Nat
McElreath, R., Boyd, R., & Richerson, P. J. (2003). Shared norms and the evolution of ethnic markers.
Current Anthropology, 44, 122­129.
McElreath, R., et al. (2005). Applying evolutionary models to the laboratory study of social learning.
Evolution and Human Behavior, 26, 483­508.
Mesoudi, A. (2008). The cultural transmission of Great Basin projectile technology: an experimental
simulation. American Antiquity, 73, 3­28.
Mesoudi, A., & Laland, K. N. (2007). Culturally transmitted paternity beliefs and the evolution of human
mating behaviour. Proceedings of the Royal Society of London. B: Biological Sciences, 274, 1273­
1278.
Mesoudi, A., Whiten, A., & Laland, K. N. (2004). Perspective: is human cultural evolution Darwinian?
Evidence reviewed from the perspective of The Origin of Species. Evolution, 58, 1­11.
Mesoudi, A., Whiten, A., & Laland, K. N. (2006a). A science of culture: clarifications and extensions.
Behavioral and Brain Sciences, 29, 366­383.
Mesoudi, A., Whiten, A., & Laland, K. N. (2006b). Towards a unified science of cultural evolution.
Behavioral and Brain Sciences, 29, 329­347.
Meyers, M. A. (2007). Happy accidents: Serendipity in modern medical breakthroughs. New York:
Arcade.
Offerman, T., & Sonnemans, J. (1998). Learning by experience and learning by imitating others. Journal
of Economic Behavior and Organization, 34, 559­575.
Panchanathan, K., & Boyd, R. (2004). Indirect reciprocity can stabilize cooperation without the secondorder
free rider problem. Nature, 432, 499­502.
Pinker, S. (1997). How the mind works. New York: W. W. Norton.
Richerson, P., & Boyd, R. (1998). The evolution of ultrasociality. In I. Eibl-Eibesfeldt, & F. K. Salter
(Eds.) Indoctrinability, ideology and warfare (pp. 71­96). New York: Berghahn Books.
Richerson, P., & Boyd, R. (2000). Complex societies: the evolutionary dynamics of a crude
superorganism. Human Nature, 10, 253­289.
Richerson, P., & Boyd, R. (2005). Not by genes alone: How culture transformed human evolution.
Chicago: University of Chicago Press.
Rogers, A. (1989). Does biology constrain culture? American Anthropologist, 90, 819­831.
Rogers, E. M. (1995). Diffusion of innovations. New York: Free Press.
Rosenthal, T. L., & Zimmerman, B. J. (1978). Social learning and cognition. New York: Academic Press.
Shennan, S. (2001). Demography and cultural innovation: a model and its implications for the emergence
of modern human culture. Cambridge Archaeology Journal, 11, 5­16.
Shennan, S. (2003). Genes, memes, and human history: Darwinian archaeology and cultural evolution.
London: Thames & Hudson.
Shennan, S. (in press). Pattern and process in cultural evolution. Berkeley: University of California.
Sneader, W. (2005). Drug discovery: A history. Chichester, England, and Hoboken, NJ: Wiley.
Sperber, D. (1996). Explaining culture: A naturalistic approach. Oxford, UK, and Cambridge, MA:
Blackwell.
Taylor, G. (1996). Cultural selection. New York: Basic Books.
Tomasello, M. (1996). Do apes ape? In C. M. Heyes, & B. G. Galef Jr. (Eds.) Social learning in animals:
The roots of culture (pp. 319­346). San Diego: Academic Press.
Tooby, J., & Cosmides, L. (1992). The psychological foundations of culture. In J. Barkow, L. Cosmides,
& J. Tooby (Eds.) The adapted mind (pp. 19­136). New York: Oxford University Press.
Whiten, A. (2000). Primate culture and social learning. Cognitive Science, 24(3), 477­508.
Williams, T. (1987). The history of invention. New York: Facts on File.
Wrangham, R. W., et al. (1999). The raw and the stolen: cooking and the ecology of human origins.
Current Anthropology, 40, 567­594.
Joseph Henrich (Ph.D. UCLA, 1999) holds the Canada Research Chair in Culture, Cognition, and
Coevolution in the Departments of Psychology and Economics at the University of British Columbia. His
research combines behavioral and cognitive experiments, in-depth field ethnography, and evolutionary
modeling to explore the coevolutionary emergence of cooperative institutions, prosocial motivations, religions,
and complex cultural adaptations. See his website at http://www.psych.ubc.ca/~henrich/home.html.
Hum Nat
Robert Boyd received his bachelor's degree in physics from the University of California at San Diego and
a Ph.D. in ecology from UC Davis. He has taught at Duke and Emory universities and has been at UCLA
since 1986. With Herb Gintis, Rob currently co-directs the MacArthur Research Network on the Nature
and Origin of Preferences. His research focuses on population models of culture. Rob has also co-authored
an introductory textbook in biological anthropology, How Humans Evolved, with his wife, Joan Silk. He
and Joan have two children and live in Los Angeles. His hobbies are rock climbing and bicycling.
Peter J. Richerson received undergraduate and graduate degrees in entomology and zoology at the
University of California, Davis. He is currently Distinguished Professor in the Department of
Environmental Science and Policy at UC Davis. His research focuses on the processes of cultural
evolution, most of it co-authored with Robert Boyd. Their 1985 book applied the mathematical tools used
by organic evolutionists to study a number of basic problems in human cultural evolution. His recent
publications have used theoretical models to try to understand some of the main events in human
evolution, such as the evolution of the advanced capacity for imitation (and hence cumulative cultural
evolution) in humans, the origins of tribal and larger-scale cooperation, and the origins of agriculture. He
collaborates with Richard McElreath and Mark Lubell in an NSF-funded research group devoted to the
study of cultural transmission and cultural evolution in laboratory systems.
Hum Nat