The human brain is a social brain, according to Robin Dunbar, who challenges the conventional assumption that language evolved with men to enable them to hunt. Instead, he argues, it emerged with women to allow them to gossip.

Language, it has sometimes been said, is the defining character of our species. It sets us apart from the brute beasts. Yet our interest in it has been surprisingly myopic. Linguists have chiselled away at grammar, psycholinguists at how we produce and recognise speech sounds, sociolinguists have explored sex and class differences in pronunciation and word use, and neuroanatomists have worried about which particular bits of the brain control which bits of the process. There has been almost no interaction between these groups and no one seems to have thought of asking any of the really big questions, like why we have language in the first place. It seems to be assumed that, well I it just is.

Language, so one version of the argument runs, is simply a byproduct of having a large brain. The only problem with this claim is that brain tissue turns out to be exceptionally expensive to maintain. Replenishing neurotransmitters and maintaining the ionic pumps are so costly of energy that the brain consumes ten times as much as you would expect for the equivalent volume of muscle tissue. Maintaining such a costly organ can hardly be considered a mere byproduct: the counter-selection pressures against it would be enormous and it would require very significant selective advantages for an increase in brain size to evolve against this gradient.

The conventional assumption about language is that its primary function is information processing and exchange. By that is invariably meant information about the external world - instructions on how to do things, explanations for how the world works, plans for future action. Archetypally, language evolved to enable our forefathers (and I use the word advisedly) to coordinate and plan their hunts. Indeed, the unwritten assumption running through psychology and the neurosciences is that the brain as a whole evolved to process and manage environmental and technical information.

³ÉÈËVRÊÓƵ

But, once again, we are stuck with a curious anomaly. Why should something that only occupies a fraction of an individual's time require a faculty that is so expensive? Do hunter-gatherers (and we spent 98 per cent of our existence as a species as hunter-gatherers) need language (or, if it comes to that, big brains) to hunt? Much hunting by contemporary hunter-gatherers is a solitary business. A recent study of Inuit hunters produced a modal group size of just one person for 15 different kinds of hunts: indeed, only three of the hunt types had modal group sizes greater than two men. How much brain power is needed to coordinate the activities of two people? Even then, is more than a word or two necessary?

Even the manufacture of tools (another frequently touted suggestion) does not require language. You cannot tell someone how to produce a perfect arrowhead or end-scraper, you need to show them what to do - an activity that is as effective in the silent world of the deaf-and-dumb as it is in the world of language. "Do it like this!" is about as much linguistic competence as you need.

³ÉÈËVRÊÓƵ

To understand the evolution of language, we need to go back to the more fundamental question of why primates have larger brains than other species.

Primates in general differ from other species of animals in respect of their social skills and there is now considerable evidence to suggest that primates owe their large brains to the need to manage and manipulate large quantities of information about social partners. Some of the evidence for this is provided by the fact that, in primates, group size correlates directly with neocortex size: living in larger groups requires proportionally more computing power to keep track of what is going on. In contrast, purely ecological variables do not correlate with neocortex size once the effects of group size have been statistically removed. The primate brain is a social brain.

The real surprise, however, was the discovery that neocortex size also seems to predict group size in humans. The value of around 150 predicted by the size of the modern human neocortex turns up with monotonous frequency in human societies from all around the world. Roughly speaking, it is the number of people you know well enough to beg a favour of without undue embarrassment.

This raises an interesting question that brings us sharply back to language. Primates use social grooming to service the relationships that hold their social groups together and there is a roughly linear relationship between group size and the time devoted to social grooming among the more advanced species.

For the monkeys and apes, grooming is less of a hygienic activity (though it undoubtedly has a beneficial effect in this respect) and more of a social one. Among other things, it is a declaration of commitment that serves to reinforce friendships.

If humans used grooming to bond their groups as other primates do (and our ancestors must have done so at least when they were common or garden apes), how much time would we need to devote to grooming to keep our groups of 150 together? The equation based on the primate data predicts around 40 per cent of the day. This is twice as much time as any living primate species devotes to grooming and nearly double what we humans devote to social interaction. No species that has to wrest its living from the real world could afford to devote so much of its time to social intercourse, no matter how pleasurable such mutual mauling may be.

The evolutionary impasse here is this. If grooming is the principal mechanism that allows primates to hold their groups together, our ancestors would not have been able to evolve such large groups. But if ecological pressures demanded the evolution of larger groups, then some more efficient method of social bonding was needed. The stark alternative was extinction.

Does language function as a kind of vocal grooming, then? The answer would seem to be both yes and no. One thing grooming does for animals is it makes them feel good. If you watch what people do when they talk to each other, you soon find that much of our conversation time is devoted to making people laugh - in effect building up that same sense of warmth and intimacy grooming provides. Geoff Miller of the Max Planck Institute for Psychological Research at Munich has argued that humans' large brains were designed by sexual selection to maintain a constant stream of inventive entertaining chatter (and other activities like art and music) to entice and entertain prospective mates - what he calls the Scheherezade effect.

³ÉÈËVRÊÓƵ

³ÉÈËVRÊÓƵ

On the other hand, language does offer us a facility that grooming does not, namely the opportunity to exchange information during conversation. Whatever the entertainment value of conversation, the ability to acquire and disseminate social knowledge must surely have been crucial to the creation of large social groups. Monkeys and apes are limited to direct personal observation to learn who is a reliable ally and who should be avoided.

Language provides us with the opportunity to find out about more people more quickly because several individuals can pool their individual observations together. Gossip speeds up the process of building and monitoring our social relationships, as well as allowing us to control social backsliders.

If language evolved to allow us to gossip, we ought to see evidence of this in what people talk about in informal conversations with friends and acquaintances. And, indeed, our studies of natural conversations reveal that, for both sexes, around 70 per cent of all conversation time is taken up with matters directly related to personal experiences and social relationships. Work, philosophy, politics, culture, instructions, ethics, religion, even sport - all these are crammed into the remaining 30 per cent. Even highbrow newspapers devote up to half their column inches to what they loosely describe as "human interest" stories and features.

Sex differences emerge in respect of only two aspects of conversational behaviour. Although men and women devote the same amount of time to talking about social behaviour and relationships, it is their own experiences that tend to dominate men's conversations and other people's that are more common among women's. The only other difference is that technical and academic subjects have a much higher profile in men's conversations when women are present than when they are talking only with other men.

Men's conversations seem to function mainly as advertising. It is as though they are permanently on a kind of vocal lek. Leks are a fairly common form of mating system among birds and mammals in which males gather on areas frequented by females and vigorously advertise their qualities by displays or loud calls. The females wander among the displaying males and choose the one(s) they would prefer to mate with. The peacock is the classic case. Miller's work suggests that one of the key cues human females use in judging a male's quality is his ability to make them laugh. Look at advertisements in lonely hearts columns and see how often women's ads mention GSOH (good sense of humour) as a desirable trait in a prospective partner.

In contrast, women's conversations seem to be dominated by more mundane topics, in particular social gossip. It may look trivial, but in fact it is a crucial form of networking. Functioning effectively in the large (often dispersed) social groups so characteristic of humans requires a great deal of information as well as considerable skill. Keeping track of who is in and who is out, monitoring the activities of freeriders and seeking and giving advice on how to handle situations is what drives the social whirl.

As with all primates, managing this whirl effectively appears to be much more important for women than for men. Moreover, it requires a more subtle balancing of forces through the formation of alliances rather than the brute force and ignorance of the simple competitive strategies to which males of many species often resort.

There are some interesting implications to all this. One is that, since, in functional terms, group cohesion necessarily takes biological priority over mating, the pressures selecting for the evolution of language are likely to have been higher among the ancestral females than among the ancestral males. Gossip must have evolved before technical competence and symbolic language.

The other, as Miller has pointed out, is that politics, art, literature, music, perhaps even science, are really forms of male advertising and sexual display.

That does not, of course, make any of them less worth while in themselves - the global benefits of all these activities are surely beyond question. But it might explain why men seem more motivated than women to engage in - and especially succeed at - these activities. It would also explain why men and women so often complain conversations among the opposite sex are interminably boring. Of course they will be: they are designed to do different things.

³ÉÈËVRÊÓƵ

Robin Dunbar is professor of psychology, Liverpool University. His book, Gossip, Grooming and the Evolution of Language, is published in April by Faber and Faber.