The magic number

If social networks become too large, communities fail to function effectively – and we’ve got evolution to prove it, claims Robin Dunbar

As a species, we have spent more than 95 percent of our evolutionary history in small-scale foraging societies in which no more than a few hundred people are dispersed over a wide geographical area. But ever since the Industrial Revolution, with its voracious demand for large labour forces, human communities have become increasingly concentrated in urban centres. In 1800, just 2 percent of the world’s population lived in cities, rising to 13 percent in 1900 and 60 percent in 2000. While we can obviously manage to live in very large communities, our psychology is not really designed to handle the kind of densities and social pressures associated with urban centres. Our social world is still built on a very small scale, and the relative cohesion of traditional small-scale societies has become difficult to replicate in the urban environment, in part explaining our civic dysfunctionality.

Among primates in general, there is a simple relationship between a species’ typical social group size and the size of its neocortex (very roughly, the thinking part of the brain). Humans fit nicely on to the end of this line, with a predicted group size based on our neocortex size of about 150 – the figure that is now known as ‘Dunbar’s number’. I remember being surprised at how small this figure was when I first predicted it. The apparent mismatch between this prediction and the fact that people now live in ‘mega-cities’ of 10 million or more prompted me to set about finding out how big natural human groups actually were. Initially, I searched the literature on ethnographic societies – hunter-gatherers such as the !Kung San of Namibia or many Amazonian Indian tribes – thinking that these would best reflect our natural ancestral state. Their typical community size turned out to be about 150. But in the industrialised world, perhaps we had managed to break through this particular glass ceiling and found ways to live in much larger communities?

To see if this was at all likely, one obvious place to look was personal social networks – the number of people you know and have relationships with. Our first attempt to do this exploited the old Christmas card trick. Writing cards costs money and time and, at the very least, implies that you think the recipients are worth the effort. We asked a number of people to tell us exactly whom they were sending cards to, and it turned out that, on average, there were about 148 recipients, taking into consideration all household members.

This looked sufficiently promising for us to begin sampling social networks in earnest. We have now asked about 500 people about their social networks. And by this, we don’t mean that they merely know the names of the people in their networks, but rather that they have a reciprocal relationship with a proper history. Informally, a good definition is the number of people you wouldn’t feel embarrassed about joining if you happened to find them at the bar in the transit lounge of Hong Kong airport at 3am. You know where they fit into your social world and they know where you fit into theirs. This figure also turns out to be about 150.

In fact, the number 150 starts to emerge from all kinds of unexpected places once you know what to look for. It is the average village size recorded in William the Conqueror’s Domesday Book of England in 1087, and it is the average size of the smallest standalone unit (the company) in all modern armies. In business organisations, it turns out to be the point at which businesses start to need formal management structures if they are not to fall apart as they grow in size. Intriguingly, it is also the typical size of communities among both the Hutterites and the Amish, both of which practise a communalistic form of fundamentalist Christianity in the USA. In sum, the number 150 seems to define the limit on the number of people with whom you can form personalised relationships that involve a sense of mutual obligation, trust and reciprocity.

Circles of acquaintanceship

That said, the 150 people in your social world do not form a homogenous group. Our research has revealed that social networks actually consist of a series of layers, or circles of acquaintanceship. The size of these layers tends to increase by a multiple of three – an inner layer of five intimates, then 15 good friends, 50 friends and 150 acquaintances, with each successive layer including those below it. As you go up through the layers, the average emotional intensity of the relationship declines, as does the frequency with which you see individuals. What seems to set the limit at 150 – the outer layer – is that you run out of time and psychological capital to give to more people.

Two things seem to be important in creating the relationships that give rise to this pattern. When social psychologists have looked at intimate relationships, they consistently come out with two key dimensions, labelled ‘being close’ and ‘feeling close’. ‘Being close’ is clearly related to spatial proximity – that is, time spent together. ‘Feeling close’ is an altogether more nebulous thing, but it has something to do with the emotional quality of our human interactions.

Part of that ‘feeling close’ component seems to relate to the fact that we are still a very ‘touchy-feely’ species. Monkeys and apes create a sense of feeling close through social grooming, an activity that has less to do with keeping their fur clean than with stimulating the brain to release endorphins by massaging the skin. Endorphins are part of the brain’s pain control system and are released whenever the body is under physical stress – which is what happens when monkeys rub, stroke, pinch and pull skin and fur during grooming. The critical side effect of endorphins is that they make you feel relaxed and at peace with the world, a state of mind that allows you to build a sense of trust and intimacy. Clearly, we still use grooming to create these effects, but it is usually confined to more intimate relationships, where we call it petting or cuddling. However, touch turns out to be far more important in conveying people’s intentions and shaping their relationships than we realise at first.

It’s a small world

One characteristic of small-scale communities is that everyone knows everyone else and, more importantly, there is almost complete overlap in people’s social networks. The trends towards urbanisation, economic migration and social transience that have come to dominate modern life have changed all that. We grow up in Huddersfield, go to university in Brighton, get our first job in London and move (or are moved by our employer) to Glasgow a few years later. At each step, we leave behind small groups of friends until time and distance eventually dim our relationships with them beyond the point of rescue.

The effect of all this is that our networks of 150 people become increasingly fragmented, consisting of small clusters of friends who are forever associated with a particular time and place. These clusters rarely overlap; indeed, our social network only partially overlaps with even that of our partner, despite the fact that we live in the same house and share a life together. The core clusters of best friends and family may overlap, but we tend to have separate friends for work, hobbies and so on.

In small-scale societies, the fact that the community is spatially and socially integrated means that its members can maintain social cohesion and social discipline. This doesn’t mean to say that they never fall out or quarrel, but it does mean that they will look out for one another. Peer pressure is usually sufficient to police everyone’s behaviour and prevent individuals from stepping too far out of line. What bonds the community together is a common sense of obligation, reciprocity and trust. Religion often plays a seminal role in these societies, providing a common signal of community membership through shared values and beliefs. Rituals such as trance-dancing and sweat-lodges are extremely effective at releasing the endorphins that seem to be so crucial for social bonding.

The fact that social networks in our urbanised world are fragmented and geographically distributed means that there is less to bind us into the fabric of our local community. We don’t even know most of the people we pass in the street. This fragmentation may be part – and perhaps a major part – of the reason why modern urban societies seem to be so dysfunctional. So we have a genuine problem – one to which we need to find a solution if we are not to be overwhelmed by the aspects of modern society that give rise to lack of social cohesion.

Might the internet offer us a way of creating virtual communities that reinstate small-scale society on a grand, global scale? There are several reasons for suspecting not. First, some exaggerated claims notwithstanding, most people have only their everyday, face-to-face friends on their Facebook page. The number of people with 500 or even 1,000 ‘Facebook friends’ is in fact very small. Second, those who do have very large numbers of virtual ‘friends’ do not know most of these people in any meaningful sense.

Third, social networking sites lack the everyday physical contact that seems to be so important in establishing real relationships of trust and obligation. They are very good for slowing down the rate at which relationships decay, but not for preventing that decay altogether. Sooner or later, you need to get to grips with people in real life. Nor do social networking sites really work well for establishing new relationships. In the end, a touch in the real world is worth a thousand words in the virtual world. Finally, analysis of community sizes in virtual worlds such as Second Life and World of Warcraft suggests that they mirror exactly what we see in the world of real-life communities. Dunbar’s number reigns supreme even here.

Cohesive communities

So, if the internet isn’t any help, how do we create large, integrated communities? There are probably only two serious options: the stick or the carrot. It is always possible to enforce integration by resorting to draconian punishment. But the problem, as everyone who breaks the speed limit knows only too well, is that it is not the size of the punishment that makes you obey the rules, but how likely you are to get caught. The temptation to cheat is always present. The punishment solution leads inevitably to a police state, which seems an unattractive option.

In fact, social cohesion is always more effective under ‘carrot’ regimes: if people sign up to the rules voluntarily, their commitment to the community is greater and requires much less (if any) policing. In the end, small-scale communities work precisely because everyone is signed up to the same grand project – survival and successful reproduction – and cannot afford to drop out or leave. We are all in it together, and we had better pull together or we all go down – even if we can’t stand one another.

Modern life allows us more independence. If that is the root of our problem, then might the answer lie in inventing a new grand project that will pull us all together? I have thought long and hard about this, but the only grand project that seems to work effectively in this kind of way seems, for better or worse, to be religion. Data on 19th-century American millenarian cults show rather dramatically that cults based on religious beliefs, however outlandish, survive for considerably longer than purely secular utopian cults. Religion has undoubtedly played a central role in community bonding throughout the course of recent human evolution, but the prospect of its becoming the centrepiece of modern life again fills me with deep angst, not least because it can (and has been) a very strong mechanism for social strife and for creating deeply divisive, ‘them-and-us’ polarisations.

Another answer might lie in some kind of Swiss model: smaller, semi-independent, old-fashioned ‘city states’ in which a sense of community can more easily be created at the local level. This would almost certainly require greater diversity than the current enthusiasm for central government control, from Westminster to Brussels, allows. As much as I like this idea, I suspect that, given the current level of job mobility, it may still be difficult to create that sense of local commitment in smaller civic units. This is because belonging invariably arises out of history – being part of a community for a lengthy period. 

Inevitably, most of the obvious solutions will either be difficult to achieve or have less than desirable consequences. But we cannot duck the problem, because it won’t simply go away. Our task in the century ahead must be to find a better alternative – before we get overwhelmed by the dysfunctionality of the urban state.

Robin Dunbar is professor of evolutionary anthropology at the University of Oxford

Who’s in your social network?

Make the most out of your social network with by visiting RSA Fellowship pages or emailing the Fellowship team.