[Paleopsych] Boston Consulting Group: Brain Size, Group Size, and Language

Premise Checker checker at panix.com
Sun Jan 1 02:43:54 UTC 2006

Brain Size, Group Size, and Language


Evidence in primates suggests that the size of social groups is
constrained by cognitive capacity as measured by brain size. After a
point, the number and nature of group relationships becomes too
complex and groups tend to grown unstable and fission. Based on these
projections, human beings should reach a "natural" cognitive limit
when group size reaches about 150. There is extensive empirical
evidence of social groupings of about this size in the
anthropological literature. It is suggested that language arose as a
means of enabling social interactions in large groups as a more
efficient substitute for one-on-one social grooming in primates.

Brain Size, Group Size, and Language

Why do people have such big brains? After all, it is very expensive
to maintain this organ - while it only accounts for about 2% of adult
body weight, the human brain consumes about 20% of total energy
output. Professor Robin Dunbar has advanced a theory relating brain
size in primates to the size of social groups and to the evolution of
language in humans. This work supports some of the suppositions of
the "Machiavellian Intelligence Hypothesis" which states that
intelligence first evolved for social purposes (see Byrne, Richard
&Whiten, Andrew (1988) Machiavellian Intelligence . Oxford: Clarendon
Press.) This rebuts competing arguments linking brain size to more
sophisticated food gathering and extended home range size.

Brain size as a determinant of group size in primates

Dunbar plots data on brain size (the measure he uses is the ratio of
the neocortex - the "thinking" part of the brain - to total brain
size) versus observed social group sizes for 36 genera of primates.
He obtains a very good fit for the data (r-squared = 0.764, P

Log10(N) = 0.093 + 3.389Log10(CR)

Where N is the mean group size, and CR is the neocortex ratio. The
results are plotted on a log-log scale.

The fact that brain size in primates is closely related to group size
implies that animals have to be able to keep track of an increased
information load represented by these larger social groups. Note that
the relationship of brain size to group size in the model is not
linear, which it would be if the cohesion of the group depended only
on each individual's relationship to all the other members of the
group. The fact the relationship is logarithmic implies that the task
of information processing is more complex: each animal has to keep
track not just of its own relationships to every member of the group
but also the third party relationships among other group members. At
some point, the complexity of these relationships exceeds the
animals' mental ability to deal with them. Several primate societies,
such as chimpanzees, are known to become unstable and to fission when
the group size exceeds a certain level. It may be that there is upper
limit on group size set by cognitive constraints.

Implications for human group size
While the "natural" group size for humans is not known, the size of
our brains is (neocortex ratio = 4.1). Plotting this as the
independent variable on Dunbar's regression line yields a group size
of 147.8 for homo sapiens.

Is there any empirical evidence for natural human group sizes of
about 150? Based on his scan of the anthropological literature,
Dunbar concludes that there is. One source of evidence is from modern
day hunter-gatherer societies (whose way of life best approximates
that of our late Pleistocene ancestors of 250,000 years ago - the
period when our current brain size is thought to have evolved). Based
on this data for groups in Australia, the South Pacific, Africa,
North and South America, there appear to be three distinct size
classes: overnight camps of 30-50 people, tribes of 500-2,500
individuals, and an intermediate group - either a more permanent
village or a defined clan group - of 100-200 people. The mean size of
this intermediate group in Dunbar's (admitted small) sample is 148.4,
which matches remarkably well with the prediction of the neocortex
size model. This grouping also had the lowest coefficient of
variation, which we would expect if this group size truly subject to
an internal constraint (i.e., cognitive capacity), whereas smaller
and larger groupings are more unstable. While these intermediate size
groups may be dispersed over a wide area much of the time, they
gather regularly for group rituals and develop bonds based on direct
personal contact. These groups come together for mutual support in
times of threat.

Other examples of communal groups of this size abound. The
Hutterites, a fundamentalist group that lives and farms communally in
South Dakota and Manitoba, regards 150 as the upper limit for the
size of a farming community. When the group reaches this size, it is
split into two daughter communities. Professional armies, dating from
Roman times to the modern day, maintain basic units - the "company" -
that typically consists of 100-200 soldiers. Modern psychological
studies also demonstrate the size of typical "friendship networks" in
this same range. These examples provide further evidence of natural
group size constraints. Once the number of individuals rises much
beyond the limit of 150, social cohesion can no longer be maintained
exclusively through a peer network. In order for stability to be
maintained in larger communities, they invariably require some sort
of hierarchical structure.

Groups and the evolution of language

Dunbar points out that primate groups are held together by social
grooming, which is necessarily a one-on-one activity and can absorb a
good deal of the animals' time. In order to maintain these bonds in
groups of 200 individuals would require us to devote about 57% of the
day to social grooming. Dunbar proposes that the maintenance of these
social bonds in humans was made possible through the evolution of
language, which emerged as a more efficient means for "grooming" -
since one can talk to several others at once. Dunbar's model predicts
a conversation group size for humans (as a substitute for grooming)
of 3.8. He then sites evidence that this is indeed about the size
actually observed in human conversation groups. Conversations tend to
partition into new conversational cliques of about four individuals.
Furthermore, studies have shown that a high percentage of ordinary
conversations (over 60%) is devoted to discussing personal
relationships and social experience - i.e., gossip.

Based on Robin Dunbar 1992, 1993
Contributed by David Gray, 2000

Dunbar, R.I.M. (1992), 'Neocortex Size as a Constraint on Group Size
in Primates', Journal of Human Evolution 20, 469-493

Dunbar, R.I.M. (1993), 'Coevolution of neocortical size, group size
and language in humans', Behavioral and Brain Sciences 16, 681-735.

     * Group size effects the dynamics of social networks - a
community ethos is more likely to arise in human groups smaller than 150
     * Network formation depends on social interaction - effective
networks arise from regular personal contact that creates a shared
sense of community
     * Networks can be costly to maintain - time and resources are
required to maintain the social ties that support a network
     * Hierarchy becomes important as group size grows - more complex
societies require authoritarian structures to clarify and enforce
social relationships

Social networks, primates, intelligence, group size, gossip,
grooming, hunter-gatherer societies, Hutterites, army company,
fission, bonds, friendship, hierarchy, peers, authority, evolution, language

More information about the paleopsych mailing list