USING THE METHODS OF EXPERIMENTAL SOCIAL PSYCHOLOGY TO STUDY CULTURAL EVOLUTION

Journal of Social, Evolutionary, and Cultural Psychology
www.jsecjournal.com - 2007, 1 (2): 35-58.

Original Article


USING THE METHODS OF EXPERIMENTAL SOCIAL
PSYCHOLOGY TO STUDY CULTURAL EVOLUTION


Alex Mesoudi
University of British Columbia


Abstract
Cultural psychology, and other social sciences (e.g. cultural anthropology, sociology),
seek to document cultural variation, yet have difficulty providing strong empirical tests of
explanations for that variation. It is argued here that an effective means of testing
hypotheses regarding the origin of, and persistence and change in, cultural variation is by
simulating cultural transmission in the lab using certain methods from experimental
social psychology. Three experimental methods are reviewed: the transmission chain
method, the replacement method, and the constant-group method. Although very few
studies have explicitly simulated specific cross-cultural patterns, much potential exists for
future investigations. This integration of small-scale experimental simulations and large-
scale observational or historical data is facilitated by an evolutionary framework for the
study of culture, and has a precedent in the biological sciences, where experiments are
used to simulate and explain the processes of biological evolution.

Keywords: Culture, cultural transmission, cultural psychology, cross-cultural
psychology, social psychology, cultural evolution, evolutionary psychology, Bartlett,
evolutionary synthesis
.
Introduction


This new forum, the Journal of Social, Evolutionary and Cultural Psychology,
seems an ideal place to discuss a particular interaction between these three fields that has
hitherto received little attention. The central message of this paper is that certain methods
of experimental social psychology can usefully inform our understanding of cultural
variation and cultural change, by simulating that cultural change under controlled
conditions in the psychology lab. These methods can enrich cultural psychology as well
as other social sciences such as anthropology and sociology by allowing strong empirical
tests of specific hypotheses regarding the origin of cultural variation and the mechanisms
underlying the persistence of and change in that variation.

Culture is here defined as the body of information, such as knowledge, beliefs,
attitudes, norms, skills, etc., that is passed from individual to individual via social
learning/cultural transmission, is stored in brains or artifacts (e.g. books), and may be

AUTHOR’S NOTE: Please address all correspondence via email to Alex Mesoudi
(mesoudi@interchange.ubc.ca).
©2007 Journal of Social, Evolutionary, and Cultural Psychology

35

---

Cultural Evolution Experiments
expressed as behavior, language or artifacts (e.g. tools). This is contrasted with
information that is passed on genetically in DNA, and information that is recreated or
rediscovered independently in each individual (through individual or asocial learning).
This culturally transmitted information can vary spatially, with different culturally
transmitted traditions characterizing different regions or social groups, or temporally,
wherein cultural information changes over time and may in some cases accumulate
successive modifications over multiple generations (Boyd & Richerson’s [1996]
“cumulative culture” or Tomasello, Kruger, & Ratner’s [1993] “ratchet effect”). This
paper focuses on spatial (geographic or ethnographic) cultural variation, although
temporal cultural change is discussed in the next section as a means of explaining spatial
variation.

The study of cultural variation has traditionally been the purview of anthropology
and sociology (e.g. Benedict, 1934; Geertz, 1973; Kroeber, 1948; Kroeber & Kluckohn,
1952; Kuper, 1999; Murdock, 1932; Steward, 1955; Tylor, 1871). Specifically,
ethnographers have compiled ethnographic databases (e.g. the Ethnographic Atlas:
Murdock, 1967; or the Human Relations Area Files: Murdock et al., 1987) that chart
enormous cultural variation across many societies world-wide, from means of
transportation to premarital sex taboos to social stratification. Similar databases map
variation in the almost 7000 languages worldwide (Grimes, 2002). Perhaps more
pertinent for this journal, the growing field of cross-cultural psychology1 (Choi, Nisbett,
& Norenzayan, 1999; Heine & Norenzayan, 2006; Lehman, Chiu, & Schaller, 2004;
Nisbett, Peng, Choi, & Norenzayan, 2001) has documented cultural variation in
numerous psychological characteristics and in various societies world-wide, including
aspects of perception (Masuda & Nisbett, 2001; Segall, Campbell, & Herskovits, 1963),
economic decision-making (Henrich et al., 2005), and attributions and motivations
towards self and others (Choi & Nisbett, 1998; Heine, Lehman, Markus, & Kitayama,
1999).

Anthropology and sociology typically have little interaction with psychology,
and when they do it is with cognitive, rather than social, psychology (e.g. D'Andrade,
1995). Cross-cultural psychology also typically has a foundation in cognitive psychology
or perception, where individuals’ performance on cognitive or perceptual tasks is
compared across cultures. However, these disciplines are neglecting another experimental
psychology literature - that of a particular tradition within social psychology regarding
cultural transmission - that can enrich their fields by explaining the causes of cultural
variation, by simulating how and why that variation changes or persists over time. First,
however, we will examine existing theories and methods that have been used to explain
cultural variation.

Explaining Cultural Variation


One problem with the charting of cultural variation within cross-cultural
psychology and cultural anthropology is that it tends to be descriptive rather than
explanatory. Although the mapping of cultural variation is crucial, an equally important

1 I refer here to mainstream cross-cultural psychology, rather than the alternative Vygotskyan strand of
cultural psychology (e.g. Shweder, 1990; Vygotsky, 1978). This is not to say that the experiments outlined
here have no relevance to the latter, merely that constraints on space prevent a more detailed exploration of
how they specifically relate to the Vygotskyan sub-field.
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


36

---

Cultural Evolution Experiments
next step is to explain how that variation came about and what mechanisms are
responsible for its persistence.

Past attempts to explain cultural variation within anthropology and sociology
have often been incomplete and unsatisfactory. For example, diffusionists (Graebner,
1905, 1911; Rogers, 1995) attempt to explain all cultural variation in terms of the
transmission of traits between different groups, without providing rigorous, empirically
supported explanations for why some innovations spread, and others do not spread, from
one group to another. Materialists (e.g. Harris, 1989), meanwhile, assume that all cultural
variation is functional or adaptive, which fails to explain maladaptive or arbitrary cultural
variation (Durham, 1992; Mesoudi, Whiten, & Laland, 2004; Richerson & Boyd, 2005).
Similar problems face the argument by some evolutionary psychologists (e.g. Gangestad,
Haselton, & Buss, 2006; Tooby & Cosmides, 1992) that cultural variation is generated by
biologically evolved responses to immediate environmental variation. This “evoked
culture” argument is also weakened by the extensive evidence for the transmission of
cultural information independently of genes (Mesoudi, Whiten, & Laland, 2006;
Richerson & Boyd, 2005).

It is problematic to assume a priori that all cultural variation can be explained
with a single, vaguely specified cause, such as cultural transmission/diffusion, individual
learning/adaptation or evoked biologically evolved responses to the environment. This is
not to say that any of these explanations are incorrect, merely that they are not mutually
exclusive. Any single pattern of cultural variation may be the result of any or all of these
to differing degrees. What is needed are empirical methods to partition specific patterns
of cultural variation according to their underlying causes, as well as explore in more
detail the mechanisms of each (e.g. whether cultural transmission is horizontal, oblique or
vertical [Cavalli-Sforza & Feldman, 1981], or whether it exhibits conformist, anti-
conformist or prestige-based biases [Boyd & Richerson, 1985]).

One way of testing potential explanations for cultural variation is by examining
the history of that variation, in order to look at its origin and persistence. For example,
Nisbett, Peng, Choi and Norenzayan (2001) argued that many of the differences in
psychological attributes observed by cross-cultural psychologists between East Asians
and Western Europeans can be traced back to differences in systems of thought and social
organization between ancient Chinese and ancient Greek societies, the former being
collectivist and the latter individualist. These different forms of social organization were
in turn linked to differences in subsistence: the ancient Chinese depended upon large-
scale agriculture that fostered cooperation, while the ancient Greeks relied on more
individualistic hunting and fishing. However, as Nisbett et al. (2001) themselves note,
these links are speculative, and many questions remain regarding the mechanisms
underlying the persistence of these differences over millennia.

A more rigorous and formal means of testing hypotheses regarding the origin of
cultural variation is by using the comparative method with large samples of ethnographic,
historical and/or archaeological data, to show when and where cultural variation first
arose and how it changed over time. For example, the comparative method can be used to
identify specific environmental factors that correlate with a particular cultural variant, as
used by Mace and Pagel (1994) to show that camel herding in East Africa was adopted
(either culturally acquired or independently invented) only in dry climates. Cladistic
methods such as those employed by Mace and Pagel can be used to further identify
whether two populations share a particular cultural trait because they both inherited it
from a common ancestral population, or whether it was independently invented in each
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


37

---

Cultural Evolution Experiments
population (for overviews of such work, see Lipo, O’Brien, Collard, & Shennan, 2006;
Mace & Holden, 2005).

Historical methods, however, feature several limitations that restrict their
usefulness for explaining the origin and persistence of cultural variation (Mesoudi, in
press). Historians and archaeologists cannot ‘re-run’ history multiple times to explore the
effects of contingency on cultural change; they cannot manipulate variables in different
runs to explore the effects of those variables; they cannot assign groups or societies
randomly to different experimental and control conditions; and they rarely have access to
complete, uninterrupted and unbiased data. Experimental methods allow us to do all of
these – they allow multiple replications, the manipulation of variables, the random
assignment of groups to different conditions, and the generation of complete datasets. In
the next section it is argued that one potentially powerful way of testing hypotheses
regarding the origin and persistence of cultural variation is by simulating cultural
transmission in the psychology lab, using experimental methods developed within social
psychology.

Mathematical models and computer simulations offer many of the same
advantages as experimental methods, and have similarly been used to explore and explain
historical (Turchin, 2003), archaeological (Henrich, 2004; Kohler & Gumerman, 2000;
Neiman, 1995; Shennan & Wilkinson, 2001) and sociological (Bentley, Hahn, &
Shennan, 2004; Henrich, 2001) patterns of cultural change and variation, as well as
cultural change more generally (Boyd & Richerson, 1985; Cavalli-Sforza & Feldman,
1981). More detailed discussion of mathematical and computer models is beyond the
scope of this paper, beyond noting that such methods are perfectly complementary to the
experimental methods discussed here. While mathematical and computer models can
often be more rigorous and comprehensive than experiments (allowing the manipulation
of many more parameters and conditions than is possible with experiments, for example),
mathematical models and computer simulations are only ever as good as their
assumptions, and experiments provide a useful means of testing those assumptions with
real people. Of course, experimental methods feature several limitations of their own,
most obviously a lack of external validity compared with observational and historical
methods, and these limitations are discussed in the final section. First, however, let us
examine these experimental methods in more detail.

Using Social Psychology Methods to Simulate Cultural Transmission


Social psychology has a set of experimental methods that allow researchers
interested in cultural variation, such as anthropologists, sociologists and cross-cultural
psychologists, to empirically test hypotheses regarding the mechanisms responsible for
the origin and persistence of that cultural variation. Essentially, these methods allow
researchers to simulate cultural change under controlled conditions in the laboratory. The
following sections discuss three of these methods – the transmission chain method, the
replacement method, and the constant-group method.

The transmission chain method


This method, originally developed by Bartlett (1932), is similar to the children’s
games “Chinese Whispers” or “Broken Telephone,” and involves passing material
(usually written text) relevant to the hypothesis being tested along chains of participants.
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


38

---

Cultural Evolution Experiments
The first participant in the chain reads or hears the material and, following a short delay
or distracter task, attempts to recall it. The resultant recall is then given to the second
participant, who does the same. Their recall is in turn passed to the third participant, and
so on along the chain (Figure 1). The changes that occur to the material as it is
transmitted along the chain, and/or the different degradation rates of different types of
material, can then reveal systematic biases in cultural transmission. Although highly
simplified compared with real human culture, the transmission chain method affords a
high degree of experimental control and, as will be seen below, has the potential to
provide important contributions to the study of cultural transmission and cultural
variation. Indeed, it has been described by Plotkin (1995, p. 219) as “close to an
experiment tailor-made for those interested in culture.”

Figure 1 – Experimental design of a typical transmission chain study. The original material relevant
to the hypothesis under investigation is given to the first participant (circles indicate participants) in
each of four parallel chains (A-D). This material is then passed along four further
participants/generations (1-4) in each chain. Changes in accuracy, quantity or content of the
material can be assessed in each generation, to test for specific biases in cultural transmission.


Bartlett (1932) reported results for various types of material: two folk tales, “The
War of the Ghosts” (from Native American culture) and “The Son Who Tried to Outwit
His Father” (from the Congo); passages describing a cricket match, an air raid, and how
to play tennis; a joke; two arguments; and a series of pictures. The participants were
mostly Cambridge undergraduates, with some undergraduates from India. A general
finding was that the material rapidly became considerably shorter in length and lost much
of its detail, with only the overall gist being preserved. A second general finding was that
participants tended to distort the material to make it more coherent and consistent with
their own pre-existing knowledge. “The War of the Ghosts,” for example, contained
many supernatural elements that were nonsensical to the British participants and were
subsequently removed or replaced with more familiar events. These two processes, loss
of detail and rationalization, led Bartlett to propose that remembering is primarily a
reconstructive process and hardly ever a process of exact replication. Only the gist or
overall impression of the material is preserved and is rebuilt around pre-existing
knowledge structures, or schemas. Bartlett also found that the folk stories were
transmitted with greater accuracy than any of the other material, which he explained by
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


39

---

Cultural Evolution Experiments
arguing that people already possess implicit story schemas that contain the structure of a
typical folk tale, thus aiding recall.

In the 30 years following Bartlett’s (1932) original studies, several transmission
chain studies were carried out which shared Bartlett’s general methodology but varied in
the material used and participants tested. The transmission of stories was studied using
chains of children (Northway, 1936) and adults of different professions (Maxwell, 1936)
and nationalities (Talland, 1956). Allport and Postman (1947), Ward (1949), and Hall
(1951) studied the transmission of pictorial rather than written stimuli. The results of
these studies largely confirmed Bartlett’s original findings of a general reduction in the
length or complexity of the material, that many of the details are lost, and only the overall
gist or impression is preserved.

Although Bartlett’s (1932) work was groundbreaking, it can be criticized on a
number of grounds, as can many of the subsequent studies listed above. There was little
attempt at any quantitative or statistical analyses, the methodology was poorly described
and unstandardized, and inferences were often drawn on single chains that are vulnerable
to individual idiosyncrasies, making results difficult to generalize and replicate. More
recent studies, however, have updated the transmission chain method according to
modern standards of experimental psychology, such as the reporting of standardized
experimental procedures, the use of statistics, and sample sizes large enough to yield
significant results.

For example, Bangerter (2000) used the transmission chain method to test
whether participants’ gender stereotypes would distort a scientific text describing
conception, passing it along twenty replicate chains each containing four participants. It
was found firstly that the sperm and ovum described in the text were anthropomorphised,
moving from the object to the subject position of sentences, and secondly, the sperm
tended to be given an active role and the ovum a passive role, which Bangerter (2000)
attributed to gender stereotyping. In a similar study, Kashima (2000) had twelve chains
each comprising five generations transmit a single story containing both stereotype-
consistent (SC) and stereotype-inconsistent (SI) behaviour. Although the first two
generations were more likely to recall SI than SC information, in the final two
generations this trend was reversed, with better recall of SC information. This effect was
due to the different degradation rates of the two types of material: although SI
information was initially recalled more accurately, it then underwent faster degradation
than the SC information, so that by the last two generations it had fallen below the SC
recall level. Kashima’s (2000) study is important because it shows that the transmission
chain method can reveal effects in later generations that cannot be predicted by patterns
of individual recall (i.e. the first generation), demonstrating the benefit of this
methodology over standard single-generation memory studies.

Mesoudi and Whiten (2004) used the transmission chain method to test Schank
and Abelson’s (1977) hypothesis that people’s knowledge of routine events, such as
going to a restaurant, is represented in hierarchically organized “action scripts.” The
overall goal of the script (e.g. to go to a restaurant) constitutes the highest level of this
hierarchy. This can be subdivided into a series of constituent subroutines (e.g. enter the
restaurant, sit down, order food, eat the food, pay, and leave), each of which can in turn
be subdivided into a series of low-level actions (e.g. pick up the menu, read it, decide on
the food, and tell the waiter). Mesoudi and Whiten (2004) tested this hypothesis in the
context of cultural transmission by passing descriptions of three everyday events (going
to a restaurant, getting up, and going shopping) expressed entirely in terms of low-level
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


40

---

Cultural Evolution Experiments
actions along ten parallel chains each comprising four participants. Consistent with
Schank and Abelson’s (1977) script hypothesis, the low-level actions were gradually
subsumed into their higher-level parent goals as they were transmitted along the chains,
demonstrating that the participants were spontaneously imposing a hierarchical structure
onto the event knowledge and transforming the events into higher levels of that structure.
So, rather than the participants just ‘simplifying’ the material (Bartlett, 1932), Mesoudi
and Whiten (2004) demonstrated that this simplification occurs in a systematic manner
according to a hierarchical bias in cultural transmission.

Finally, Mesoudi, Whiten and Dunbar (2006) found that information regarding
social relationships and social interactions (e.g. a student having an affair with her
professor, or asking for directions) was transmitted with significantly greater accuracy
and in significantly greater quantity than information regarding non-social interactions
(either an individual interacting with the physical environment or interactions within a
physical system). This is consistent with evolutionary hypotheses which argue that
primate intelligence evolved to solve social problems (Byrne & Whiten, 1988; Dunbar,
2003; Whiten & Byrne, 1997), with one consequence of this ‘Machiavellian intelligence’
or ‘social brain’ hypothesis being that individuals should preferentially attend to, process
and transmit social information over non-social information. Mesoudi et al. (2006)
therefore argued that this bias for social information represents a biologically evolved
bias in cultural transmission.

In summary, Bartlett’s (1932) transmission chain method constitutes a simple yet
effective means of testing hypotheses regarding cultural transmission under controlled
experimental conditions. Although early studies featured a number of methodological
shortcomings, more recent studies have demonstrated that the transmission chain method
can be updated to meet modern standards of experimental psychology, specifically by the
use of multiple parallel chains, quantitative and statistical analyses, and properly
standardized and controlled methodology.

Bartlett (1932) was also aware of the wider implications of the transmission
chain method, stating that “elements of culture, or cultural complexes, pass from person
to person within a group, or from group to group, and, eventually reaching a thoroughly
conventionalized form, may take an established place in the general mass of culture
possessed by a specific group” (Bartlett 1932, p. 118). Hence Bartlett argued, as is argued
here, that the results of these small-scale laboratory studies regarding memory biases
along linear chains of a few individuals can potentially be extrapolated up to explain
certain large-scale patterns in cultural variation. Although this has yet to be tested
formally, we can speculate that the kinds of information that are preserved well in the
experimental simulations, such as folk tales (Bartlett, 1932) or information regarding
social relationships (Mesoudi, Whiten, & Dunbar, 2006), should be more stable over time
and more prevalent than certain other types of information. Indeed, we might informally
note that folk tales have indeed been preserved over many generations precisely because
of their schema-like properties (Rubin, 1995), while the popular mass media tends to
favor social gossip over non-social factual information (Mesoudi, 2005), although more
systematic tests are needed to investigate such links more fully.

The replacement method


One limitation of the transmission chain method lies in the linearity of the one-
to-one chains. Actual cultural transmission may frequently involve more than one model
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


41

---

Cultural Evolution Experiments
and more than one receiver, and a slightly different method is needed to study these more
group-based aspects of transmission. The replacement method, originally proposed by
Gerard, Kluckhohn and Rapoport (1956), involves establishing a norm or bias in a group
of participants, and one by one replacing the participants with new, untrained participants
(Figure 2). Each replacement represents one cultural “generation.” The degree to which
the norm remains in the population during successive replacements/generations
represents a measure of its successful transmission to the new members.

Figure 2 – Experimental design of a typical replacement method study. Each generation (1-4)
comprises a four-member group, who typically must perform some task or solve a problem. After
each generation, one member of the group is replaced with a new participant (e.g. in generation 2,
A is removed and E is added). The degree to which norms or strategies persist despite replacement
represents a measure of cultural transmission.


Jacobs and Campbell (1961) used the replacement method to study the
persistence due to conformity of an artificially exaggerated perceptual judgment of the
“auto-kinetic effect,” a perceptual illusion in which a stationary point of light is perceived
as constantly moving slightly when viewed in an otherwise pitch-black room. In earlier
work by Sherif (1936), groups of participants were shown this illusion and the group
members asked one by one to estimate publicly how much they thought the light was
moving. Each group was, in fact, composed of only one genuine participant, the rest
being confederates of the experimenter who had been instructed to give unrealistically
exaggerated estimates of the light’s movement. Sherif’s (1936) now-classic finding was
that most of his participants gave estimates similar to those of the confederates despite
that estimate being obviously false, illustrating the powerful effect of conformity in group
settings. Jacobs and Campbell (1961) repeated Sherif’s (1936) experiment with the
addition that, after each round of estimating, one group member was replaced with a new,
naïve participant. Significant evidence of the inculcated norm was found for about four or
five generations after the replacement of all of the confederates, after which the
perceptual judgment returned to that exhibited by naïve control groups. This
demonstrates some degree of transmission but no long-term persistence.

Weick and Gilfillan (1971) used the replacement method to study the persistence
of problem-solving strategies. Participants in a group had to individually call out
numbers, without being able to hear the other members’ numbers, so that the sum total of
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


42

---

Cultural Evolution Experiments
all group members’ numbers equaled a specified target value. Groups were taught either
hard or easy strategies for coordinating their responses, and group members were
periodically replaced with new participants. It was found that the easy strategies persisted
for about eight generations after the last trained member had been replaced, while the
difficult strategies were hardly transmitted at all, demonstrating that high fidelity
transmission of problem-solving strategies is possible when the strategy is both effective
and easy to implement.

More recently, Baum, Richerson, Efferson and Paciotti (2004) used the
replacement method to study cultural transmission in an anagram-solving task. Groups of
participants could choose to solve an anagram printed on either red or blue card. The red
anagrams gave a small but immediate payoff, whereas the blue anagrams gave a larger
payoff but were followed by a “time-out,” during which no anagrams could be solved. By
manipulating the length of this time-out, the experimenters were able to determine which
of the two anagrams gave the highest overall payoff; when the blue time-out was short,
blue was optimal, and when the blue time-out was long, red was optimal. Every 12
minutes one member of the group was replaced with a new participant. It was found that
the optimal choice of anagram type tended to emerge in the groups, and was transmitted
through successive generations, with existing group members instructing new members
in the optimal choice by transmitting either accurate or inaccurate information about
payoffs, or (less frequently) through coercion.

Most relevant to the present discussion, however, is a set of studies by Insko et
al. (1980; 1982; 1983), which used the replacement method to simulate the trading of
goods between groups, with the aim of testing competing anthropological theories
regarding the emergence of particular forms of social organization in traditional societies.
In the most relevant study, Insko et al. (1983) tested two alternative theories regarding the
origin of “chiefdoms,” which are collections of small bands or villages that engage in
mutual trade. Service (1975) argued, based on archaeological and ethnographic data, that
chiefdoms emerge when three conditions are present: (i) inequality of resources, such that
different villages produce different products for trade; (ii) geographic centrality, such that
one village occupies a position of economic superiority (i.e. it does not depend on the
other groups for the production of viable goods) and communicative superiority (i.e. all
trade must be conducted through that dominant group); and (iii) geographic
circumscription, such that it is more advantageous for villages to trade with each other
than to relocate. An alternative theory proposed by Carneiro (1970) argued that
chiefdoms emerge instead when conditions (i) and (iii) exist but additionally where one
village has a military or coercive advantage over other villages, such that it can confiscate
goods without voluntary trade.

Insko et al. (1983) simulated these different scenarios for state formation in the
social psychology lab using the replacement method. Three groups each comprising four
participants were taught to produce different paper models, for which they could earn
money. Earnings were maximized if paper models from different groups were combined,
encouraging trade. In the “Service” condition, one group was placed in a position of
economic and communicative superiority by making its products more valuable than
those of the other two groups, and forcing all trade to be conducted through it. The other
groups, however, retained control over their products and were free to withhold them or
negotiate their trade value. In the “Carneiro” condition, the dominant group was allowed
to freely confiscate and redistribute the other groups’ products as it saw fit. In both
conditions one member of each group was replaced with a new participant after each
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


43

---

Cultural Evolution Experiments
round of trading, with a total of nine replacements/generations, in order to simulate the
overlapping generations and intergenerational cultural transmission of traditions observed
in traditional societies. It was found that there was significantly less production and lower
earnings in the Carneiro condition than in the Service condition for all groups, both
subordinate and dominant. In addition, significantly more social unrest was observed in
the Carneiro groups: the subordinate groups were less accepting of the dominant group’s
authority, less motivated, and more frequently attempted strikes, slowdowns and
sabotage. Although not explicitly acknowledged by Insko et al. (1983), we can predict on
the basis of these findings that voluntaristic Service-style societies would out-compete
and outlast the less productive and more unstable coercive, Carneiro-style societies,
through a process of cultural group selection2 (Boyd & Richerson, 1985; Richerson &
Boyd, 2005). Hence the simulations support Service’s (1975) explanation for the origin
of hierarchically organized chiefdoms, although as Insko et al. (1983) note, further
simulations varying many other parameters and conditions are needed to fully resolve this
debate.

In summary, the replacement method offers a useful complement to the
transmission chain method. Whereas the transmission chain method has been mainly used
to study the transmission of verbal material along one-to-one chains, the replacement
method has been used to study the emergence and persistence of group-wide behavioral
traditions that may persist despite changes in the composition of groups. This is a likely
situation in business organizations and traditional societies, making it ideally suited to
simulating certain economic, sociological or ethnographic data. Indeed, this is
demonstrated by Insko et al. (1983), who tested a specific anthropological hypothesis
regarding the emergence of hierarchically organized chiefdoms.

The constant-group method


A third and final method looks at cultural transmission within groups of
participants with no replacement, where group membership is constant. Here, individuals
within a group perform some task or play a game, and are given the option of copying
one or more other group members (Figure 3). This method can be used to test claims
about when people choose to copy others as opposed to learn on their own, and who they
choose to copy. A large body of theoretical work suggests that such factors can
significantly affect large-scale cultural dynamics and play a major role in generating
different patterns of cultural variation. For example, copying other individuals at random

2 The theory of “cultural group selection” derives from group selection theory in evolutionary biology (e.g.
Wynne Edwards, 1962), popular in the first half of the 20th century, which holds that biological evolution
occurs through competition between different groups of individuals, such as flocks of birds or even entire
species. This competition leads to the differential survival and reproduction of successful groups, and the
spread of group-beneficial traits such as in-group altruism. Group selectionist arguments have been largely
discredited within biology as being extremely unlikely because groups rarely persist for long enough to be
subject to selection, do not reproduce with high fidelity, and are vulnerable to exploitation by selfish
individuals, or free-riders, moving between groups. Natural selection is now predominantly seen as occurring
at the level of the gene (Hamilton, 1964; Dawkins, 1976), favouring the spread of traits that are beneficial to
genes, such as kin-based altruism. Richerson and Boyd (2005), however, have argued that group selection is a
viable theory in cultural, rather than biological, evolution, where different social groups compete and
differentially reproduce, because of social mechanisms like conformity, which reduces within-group variation
and keeps groups from breaking up, and cultural transmission, which allows groups to “reproduce” their
values and beliefs through time.
Journal of Social, Evolutionary, and Cultural Psychology – ISSN 1933-5377 – Volume 1(2). 2007.


44

---