Have you heard the latest?

Once we groomed each other, now we gossip. It's why we learnt to talk, says Robin Dunbar

GOSSIPING is trivial. Last week Stephen Dorrell condemned the gossip about BSE as a reckless substitute for proper scientific dialogue. The rest of us tend to see it as, at best, a harmless leisure activity which leavens the serious business of living, at worst the spreading of unkind rumours. We may secretly enjoy reading gossip columns but we probably profess to despise the columnists. Yet what if gossip was the reason our ancestors spoke in the first place? What if it were the key to being human?

If you listen in to your neighbours in a cafe you will find that around two-thirds of their conversation is taken up with "gossip": who is doing what to whom, and whether it's a good or bad thing; who is in and who is out, and why; how to treat a lover, a child, or a colleague. You may happen upon an intense exchange about work or a book just read. But listen on, and I'll wager that, within five minutes at the most, the conversation will drift away, back to the natural rhythms of social life.

This intense interest in each other's doings characterises the social lives of humans. We spend hours in each other's company, stroking, touching, talking, murmuring, being attentive to every detail of who is doing what with whom. You might think this marks us out as the most social of species, but you would be wrong. Monkeys and apes are just as social as we are. Without its friends and relatives, a monkey would no more be able to survive than a human being could. The social life of primates is intense and all- consuming. They spend a great deal of the day engaged in social grooming with their special friends. Their alliances are established and maintained by grooming.

Being groomed is very pleasant, addictive even. It causes the brain to flood the body with opiates. From the groomer's point of view it is an investment of time which could be spent on personal gain, and therefore a display of loyalty. A study of vervet monkeys showed that an animal that has been groomed in the last two hours is more likely to go to his groomer's defence. Grooming is a promise of future action in circumstances as yet unimagined.

But all this attention to friends and relatives is incredibly time-consuming. The usual size for groups of primates is 40 to 50 with a mean upper limit of 55 for chimpanzees. This is manageable for one-to-one grooming. Humans operate in larger networks, though they too have an upper limit. The optimum size for us (beyond which cohesion breaks down) seems to be about 150.

If the intensity of a relationship (and the willingness of one member to come to the aid of another) is related to the amount of grooming effort put in and hence to the quantities of opiates released, then our ancestors faced a serious problem when trying to push group sizes up beyond the levels observed in other primates. They needed to be able to keep "grooming" with a friend even while busy feeding at a distance from them.

Vocal "grooming" is one answer. The problem is that vocalisations are just, well, vocalisations. They don't have the same opiate-releasing properties as grooming and if opiates are a crucial part of the mechanism of bonding, vocal exchanges will only allow you to increase the size of the group by a limited amount. Sooner rather than later, you will hit a ceiling because the limited quantities of grooming cannot generate enough opiate production to maintain the bond.

However, suppose that as language developed, signals associated with language themselves began to stimulate opiate production. Smiling and, particularly, laughing do just this, and this may well explain why smiling and laughing are such important components of conversation. They may have begun as signals of submission. But it seems that, at some point, they were built into the business of bonding. We can now, quite literally, groom at a distance. Telling jokes allows us to stimulate opiate production in our grooming partners when we don't have time to sit and do it physically. We can get on with the other important activities of survival - travelling, hunting, gathering, preparing and eating food.

If we started to use language to facilitate the bonding of larger groups, then we should be able to show that it has design features that will achieve this. One is that conversation groups should be proportionately larger than conventional primate-grooming cliques. Another is that conversation time should be predominantly devoted to the exchange of social information, or gossip. This would be a strong test of the hypothesis, because conventional wisdom says that language exists to exchange information about the world in which we live - the "there's a herd of bison down by the lake" view of language.

And when my students and I sampled various conversation groups we found it to be so. We listened to conversations in university cafeterias, public receptions, during fire practices outside evacuated buildings, in trains and in bars.

The first thing we found was that conversation groups are not infinitely large. There seems to be an upper limit of about four. If you are at a party you may see two or three people talking to each other. Other individuals will join them one by one. As each does so, the speaker and listeners may try to involve them. However, when the group reaches five or above, things start to go wrong. The group becomes unstable: despite all efforts it proves impossible to retain the attention of everyone. Instead, two individuals will start talking to each other, setting up a rival conversation. Eventually, they will start a new conversation group. I guarantee that you will observe this if you spend a few minutes watching people in social settings.

The limit of four in a conversation group means that a talker has three listeners. A chimp grooms only one friend at a time. It may be an extraordinary coincidence but the predicted size of human groups, 150 individuals, is three times the mean group size of chimpanzees.

Gossip, then, far from being a trivial activity, is an essential part of being human. It allows us to use free time for social interaction more efficiently. It enables us to reach more individuals at the same time; to exchange information so that we can keep track of what's happening among the members of our network; to keep track of social cheats (perhaps keep them in check with malicious gossip); to engage in self-advertising in a way that monkeys cannot; and, last but not least, it enables us to produce the reinforcing effects of grooming (opiate release) from a distance.

It is true that our modern fragmented social networks make this more difficult. Traditional communities, both peasant and huntergatherer, are tightly integrated units. Everyone shares the same wider network of acquaintances, everyone knows everyone else. Two individuals may not have the same circle of immediate friends and relatives, but their wider networks of 150 friends, relatives and acquaintances overlap. In post-industrial societies, this is virtually never the case. A colleague and I may share one subset of acquaintances at work, but our spouses may be involved in other subsets by virtue of their work or leisure activities. Rather than having one large shared network, we have sets of sub-networks that only partially overlap. Each of us still has 150 people in our individual networks, but we may have only 15-20 members in common.

Our ties of common interest are weakened. By co-operating with you, I gain only from my immediate self-interest and by the benefits that you return to me in due course. In traditional communities, that benefit reverberates around the community in a series of overlapping waves as you pass on the benefit you received from me to your aunt, who passes it on to her cousin, who passes it on to his friend - who eventually passes it back to me. My momentary generosity to you is repaid to me not once but repeatedly in the round-robin of social life in small communities. Despite the inevitable petty frustrations of life in small communities, the benefits of social obligation and reciprocation are magnified over and over again.

It is easy to see the function of gossip in such a close-knit society, but though few of us enjoy such social cohesion we strive to replicate it in our behaviour. We reconstruct our overlapping groups of 150, acquire "friends" in common from the gallery of the royal family or soap star characters, and ... we gossip.

Stephen Dorrell might like to reflect on this. When we gossip about the dangers of BSE, we are acting out a basic human instinct - identifying the free-riders to maintain the cohesion of the group.

The author is professor of evolutionary psychology at the University of Liverpool. His book `Grooming, Gossip and the Evolution of Language' (Faber, pounds 15.99) is published tomorrow.