Humans live in much larger groups than other primates. Language may have evolved as a form of grooming to allow us to live with so many people.

HUMANS are fascinated by what other people get up to. The success of the tabloid dailies is ample proof of that. But why should we find tittle-tattle about the private lives of minor celebrities, royalty and politicians of such overwhelming interest that it can drive the starving children of Somalia and the war-ravaged cities of former Yugoslavia off the front pages of even the most sedate of newspapers? The answer seems to lie in the evolutionary roots of our large brains. 

As far as absolute brain size goes, humans don’t have the largest brains in the animal kingdom; that privilege goes to whales. But, relative to body size, primates have much larger brains than any other animals, and we humans, not surprisingly, have the biggest brains of all—about six times larger than you would expect for a mammal of our size. 

Why do primates have such big brains? 
There are two general kinds of theories. The more conventional one is that they need big brains to help them to find their way about the world and solve problems in their daily search for food. 
The alternative type of theory is that the complex social world in which primates live has provided the impetus for the evolution of large brains. The main version of the social intelligence theory, sometimes known as the Machiavellian intelligence hypothesis, has the merit of identifying the thing that sets primates apart from all other animals—the complexity of their social relationships. 

Primate societies seem to differ from those of other animals in two key respects. The first is the dependence on intense social bonds between individuals, which gives primate groups a highly structured appearance. Primates cannot join and leave these groups as easily as animals in the relatively amorphous herds of migrating antelope or the swarms of many insects. Other species may have groups that are highly structured in this way—elephants and prairie dogs are two obvious examples—but these animals differ from primates in a second respect. This is that the primates use their knowledge about the social world in which they live to form more complex alliances with each other than do other animals. 

Keeping track of relationships

This social intelligence hypothesis is supported by a strong correlation between the size of the group, and hence complexity of the social world, and the relative size of the neocortex— the outer surface layer of the brain that is mainly responsible for conscious thinking— in various species of nonhuman primates. This result seems to reflect a limitation on the number (and/or quality) of relationships that an animal of a given species can keep track of simultaneously. Just as a computer’s ability to handle complex tasks is limited by the size of its memory and processor, so the brain’s ability to manipulate information about the constantly changing social domain may be limited by the size of its neocortex.                      

In evolutionary terms, the correlation in Figure 1 suggests - that it was the need to live in larger groups that has driven the evolution of large brains in primates. After all, group size is flexible, and often variable, within a given population but increases in brain size would only come about if they offered an evolutionary advantage to a species. There are several reasons why particular species might want to live in larger groups, not least protection against predators. And it is conspicuous that the primates which both live in the largest groups and have the biggest neocortices are species such as baboons, macaques and chimpanzees that spend most of their time on the ground and live either in relatively open habitats, such as savannah woodlands, or on the forest edge where they are exposed to much higher risk from predators than most forest-dwelling species. 

This relationship between neocortex and group size in the nonhuman primates begs an obvious question. What size of group would we predict for humans, given our unusually large neocortex? Extrapolating from Figure 1 gives a size of 148. But is there any evidence to suggest that groups of this size actually occur in humans? On the face of it, things do not look promising. After all, in the modern world, we live in huge nation states containing tens of millions of individuals. However, we have to be a little more subtle: the relationship for nonhuman primates shown in Figure 1 is concerned with the number of individuals with whom an animal can maintain a coherent relationship over time. 
It is quite obvious that those of us living in, for example, Britain do not have personal relationships with every one of the other 55 million inhabitants. Indeed, the vast majority of these people are born, live and die without ever knowing each other’s names, let alone meeting. The existence of such large groupings is certainly something we have to explain, but they are something quite different to the natural groupings we see in primates. 

One place we might look for evidence of “natural” human group sizes is among pre-industrial societies, and in particular among hunter-gatherers. 
Most hunter-gatherers live in complex societies that operate at a number of levels. The smallest groupings occur at temporary night camps and have between 30 and 50 individuals. These are relatively unstable, however, with individuals or families constantly joining and leaving as they move between different foraging areas or water holes. The largest grouping is normally the tribe itself, usually a linguistic grouping that defines itself rather strictly in terms of its cultural identity. Tribal groupings typically number between 500 and 2500 men, women and children. 
These two layers of traditional societies are widely recognised in anthropology. In between these two layers, however, is a third group often discussed, but seldom enumerated. Sometimes, it takes the form of “clans” that have ritual significance, such as the periodic celebration of coming-of-age ceremonies. Sometimes, the clan is based on common ownership of a hunting area or a set of water holes. 

Finding the right groups

For the few cases where census data are available, these clan groups turn out to have a mean size of about 153. Figure 2 shows all the data I have been able to collate from the literature on traditional societies. In addition to traditional hunter- gatherers I also included settled hunter-gatherers and simple of the village and clan-like groupings for these societies fall between 100 and 230, which is within the range of variation that, statistically, we would expect from the prediction of 148. In contrast, the mean sizes of overnight camps and tribal groupings all fall outside these statistical limits. 
But what about our own society? Is there anything to suggest that the figure of 150 might be a relevant social unit? The answer is yes. Once you start to look for them, groups of about this size turn up everywhere. In most modern armies, for example, the smallest independent unit is the company, which normally consists of three fighting platoons of 30 to 40 men each plus the commander’s staff, making a total of 130-150 men. Even the basic fighting unit of the Roman Army during the Republic was of similar size, containing exactly 130 men. The same figure turns up in businesses. A rule of thumb commonly used by managers is that companies smaller than 150 work fine on a person-to-persor basis, but once they grow large than this they need a formal hierarchy if they are to work efficiently. Sociologists have known since the 1950s that there is a critical threshold in the region of 150 to 200, with larger companies suffering a disproportionate amount of absenteeism and sickness. In 1989, Tony Becher of the education department of the University of Sussex published a survey of 12 disciplines in both the sciences and the humanities. He found that the number of researchers whose work an individual was likely to pay attention to was between 100 and 200. Once a discipline becomes larger than this, it seems that it fragments into two  or more subdisciplines.
There are other interesting examples. Neolithic villages from the Middle East around 6000 BC typically seem to have ^contained 120 to 150 people, judging by the number of dwellings. The Hutterites, a group of contemporary North American religious fundamentalists who live and farm communally, regard 150 as the maximum size for their communities. What is interesting is the reason they give for splitting communities at this size. They find that when there are more than ut 150 individuals, they cannot control the behaviour of members by peer pressure alone. 

Cementing relationships 
These are just a few of the examples from a great deal of sociological and anthropological evidence which suggest that “natural” groupings of about 150 individuals do occur in human societies, and that it is possible for individuals within such groups to maintain a coherent set of relationships through time by direct contact between individuals. The fact that the “natural” group might be as large as 150 raises some interesting questions about how such large groups are kept together. Among primates in general, and the Old World monkeys and apes in particular, social grooming is the main mechanism used to cement relationships between individuals. Moreover, the amount of time devoted to grooming other individuals is proportional to group size, at least within the Old World monkeys and apes. Those living in large groups of about 55 animals (essentially baboons and chimpanzees) spend up to 20 per cent of their day in this activity. Extrapolating from these data, we find that the amount of time required to maintain cohesion within a group of 150 would be something in the region of 35 to 45 per cent. 
It is doubtful whether any species that has to search for food daily could afford to spend almost half its day in social interaction. How then could the ancestral humans have ensured the proper degree of bonding within such large groups? 

Cost of grooming

The main problem with social grooming is that it is costly in terms of time. A monkey can only groom one individual at a time, and grooming cannot easily be combined with eating, travelling between feeding sites or other essential activities. The baboons and chimpanzees, for whom grooming consumes 20 per cent of their time, are probably already operating at the limits of their capacity. Any increase in group size can only come through a shift in gear in the mechanism used for social bonding. For humans, that shift in gear may have come in the form of language. 
Language has two interesting properties compared to grooming: you can talk to several people at once and you can talk while travelling, eating or working in the fields. Conventional wisdom has always supposed that language evolved to enable humans to exchange information about food sources and to aid cooperation during hunting. But it is difficult to see why humans should be any more in need of this than other primates or, if hunting is the issue, the social carnivores such as lions and wolves. A more plausible suggestion is that language evolved to enable humans to integrate a larger number of individuals into their social groups. 
Some evidence to support this comes from the number of people it is possible to have a conversation with. If language really is just a form of vocal grooming designed to service a larger number of relationships, then its efficiency relative to grooming should mirror the ratio of the size of human groups to the largest found among primates—55 in chim- panzees and baboons. If human groups are typically about 148, then speech should be 148/55 times more efficient than grooming as a bonding mechanism. Given that grooming is limited to a one-to-one relationship, this should mean that language should make a 1:2.7 ratio possible. In other words, while monkeys are limited to grooming in pairs, human conversation groups should consist of an average of 3.7  individuals (one speaker 2-7 listeners). And indeed do.
The sizes of conversation groups in a student refectory, for example, consist of an average of 3-4 individuals, with a striking tendency for groups larger than four to fragment into two or more smaller conversation subgroups. Interestingly, there turns out to be an accustical limit on conversation group sizes. Measurements of sound level of speech over distance indicate an upper group for conversation of about five people. If the group gets larger than this, individuals are forced to stand in too large circle to be able to hear everyone clearly. Thus, the characteristics of speech seem to be closely ried to the size of the interaction group required to maintain cohesion. 

Importance of a good gossip

There is, of course, another way in which language allows us to integrate a large number of social relationships, and that is by allowing us to exchange information about other individuals who are not present. In other words, by talking to one person, we can find out a great deal about how other individuals are likely to behave, how we should react to them when we actually meet them and what kinds of relationships they have with third parties. All these things allow us to coordinate our social relationshhips within a group more effectively. And this is likely to be especially important in the dispersed groups that are characteristic of humans. 
This could explain our fascination for social gossip in the newspapers, and why gossip about relationships accounts for an overwhelming proportion of human conversations. Even conversation in such august places as university coffee rooms tend to wing back and forth between academic issues and gossip about individuals. To get some idea of ow important gossip is, we monitored conversations in a university refectory, scoring the topic at 30-second intervals. Social relationships and personal experiences accounted for about 70 per cent of conversation time. About half of this was devoted to he relationships or experiences of third parties (people not present). Males, however, tended to talk more about their own relationships and experiences whereas females tended to talk most about other people’s. 
This is interesting because it can be interpreted as suggesting hat language evolved in the conext of social bonding between females. Most anthropologists have assumed that it evolved in the context of male-male relationships, during hunting for example. The suggestion that female-female bonding, based on knowledge of the relationships of
other individuals, was more important fits much better with views about the structure of nonhuman primare societies where relationships between females are all- important. 
 

That conversations allow us to exchange information about people who are not present is vitally important. It allows us to teach others how to relate to individuals they have never seen before. Combined with the fact that language also makes it easy to categorise people into types, we can learn how to relate to classes of individuals rather than being restricted to single individuals as primates are in grooming. For example, we can give types of individuals special markers, such as dog collars, white lab coats or large blue helmets, which allow us to behave appropriately towards them even though we have never met before. Without that knowledge, it would take us days to work out the basis of a relationship. 
Classifications and social conventions allow us to broaden the network of social relationships by making networks of networks, and this in turn allows us to create very large groups indeed. Of course, the level of the relationship is necessarily rather crude but at least it allows us to avoid major social faux pas at the more superficial levels of interaction when we first meet someone. Significantly, when it comes to really intense relationships that are especially important to us, we invariably abandon language and revert to that old-fashioned primate form of direct interaction—mutual mauling. 

What we seem to have here, then, is a new theory for the evolution of language that also seems to account for a number of other facets of human behaviour. The theory explains why gossip about other people is so fascinating; it explains why human societies are so often hierarchies; it predicts the small size of conversation groups; it meshes well with our general understanding of why primates have larger brains than other mammals, and it agrees with the general view that language only evolved with the appearance of Homo sapiens. 
What it does not explain, of course, is why our ancestors should have needed to live in groups of about 150. It is unlikely that this has anything to do with defence against predators (the main reason why most nonhuman primates live in groups) because human groups far exceed the sizes of all other primate groups. But it might have something to do with the management or defence of resources, particularly dispersed resources such as water holes that nomadic hunter-gatherers might have had to depend on at certain times of the year.              
 Robin Dunbar is professor of biological anthropology at University College London.