A couple of months ago, about 80 people – some of whom knew each other and some of whom did not – gathered in a small lecture room at Nuffield College, Oxford, to hear a man give a lecture about how, if one of them suddenly got fat, the chances are that others would get fat, too. The same applied to happiness: if someone in the room spent the next week elated, that joy would probably become infectious. And the same for smoking: if a man in the room finally managed to quit, the chances were good that his friend sitting two rows in front of him would quit as well. And then, a short while later, a friend of his friend whom he didn't know would do the same thing.
The lecture was given by Dr Nicholas Christakis, a professor from Harvard who had flown over to expand on theories that he once thought of as "cockamamie". His talk examined the power of social networks to influence our behaviour, and suggested that our actions were only partly determined by our own free will. Increasingly, something he called "social contagion" seemed to be getting the upper hand.
Some of Dr Christakis's theories seemed obvious – the chances of becoming obese because we hang around with obese friends who like eating cake – but some are more surprising, including his findings that we may become obese just by knowing someone who knows someone who is fat.
One person at the lecture, who's not obese, was Dr Ben Goldacre, author of Bad Science, the bestselling exposure of quackery and lazy research. He kept his mouth shut during the talk, which was not necessarily a sign of approval, but probably indicated a certain level of intrigue in what was being delivered.
Dr Christakis, who is 47 and has grey hair and a gregarious outlook, describes himself "both as someone who craves solitude, but also as someone who is energised by the contact with friends and family; my wife thinks I am an extrovert, and I guess I am". Recently he has become a bit of a media star in the United States, not least upon the publication of his book Connected (written with his colleague James Fowler, a professor of political science at the University of California at San Diego). Connected is one of those popular social science books, like The Tipping Point or Freakonomics, that attempts to explain how we live and how the herding instinct of the crowd influences our actions. It has not yet sold quite as well as the others, although its contents seem to have featured on almost every talk show in the United States as well as the cover of the New York Times Magazine. Last year Time magazine named Dr Christakis in its list of the 100 most influential people in the world.
Connected, which is published in the UK next month, makes many claims, and many startling observations. Christakis sums them up thus: "People affect each other even in things [like body size or emotional states] that many readers would not necessarily expect." He calls these things "emotional stampedes" and "a social chain reaction". The authors believe that at least some part of this is deeply embedded in our genetic heritage, which is, as Christakis points out, "a non-trivial finding".
Dr Christakis is nothing if not an entertainer, and Connected is nothing if not diverting. It is not all about getting plump and happy; it is also about sex and making money. At Nuffield, Christakis tells the story of a friend of his, Brian Uzzi, who has used the impact of social networks to analyse the success or otherwise of Broadway musicals. "He finds that if the key players – the director, costume designer, sound person, producer, etc – all worked together before, and everyone knows everyone else, then the show is a flop. He also finds that if you put together a group of people, who have never worked together before, the show is also a flop. But if you put together a group of people some of whom have worked together and some who haven't, then the show is a runaway critical success with enormous financial rewards."
He had told me the same story when we had met a few hours before for a chat in the Nuffield College common room. He picked up his book repeatedly as he spoke, pointing to the coloured diagrams that show the distribution of obesity, happiness, sexual activity and smoking among groups of hundreds or thousands of people. Clusters of red, green and blue nodes spread out seemingly randomly towards the edges of their pages. But they are not random; they are connected and to some extent predetermined, and they are the cause of zealous excitement.
"We're not just social animals in the conventional way that people think," Christakis says. "It's not just a bunch of us who hang out together. We have a very specific pattern of ties, and they have a particular shape and structure that is encoded in our genes. It means that human beings have evolved to live their lives embedded in social networks."
These networks can harbour a flow of generally undesirable things – violence, germs, sexually transmitted diseases, suicide, unhappiness. But good things also flow – happiness, love, altruism, valuable information on how to find a job. "It is the spread of the good things that vindicates the whole reason we live our lives in networks," Christakis says. "If I was always violent to you or gave you germs, you would cut the ties to me and the network would disintegrate. In a deep and fundamental way, networks are connected to goodness, and goodness is required for networks to emerge and spread."
Christakis's work is new in its scope and ingenuity, but his interest in human interaction has many forebears, stretching back at least as far as Aristotle. Academic interest in the impact of social networks goes back at least a century, to the work of Georg Simmel, the German sociologist who became interested in triads, extending the study of relationships between two people to three. From the 1950s to the 70s this interest broadened, and people began to map networks of between 30 to 100 people, most famously studies of how a group of 30 monks in a monastery and 70 people in a karate club interacted with each other.
In the 1990s the study of large US networks took a huge leap forward when a group of physicists began borrowing modelling techniques from social scientists and mathematicians, and turning their gaze towards newly interesting things – neurons, genes, computers. An advance in conceptual ideas accompanied a revolution in statistical tools, not least the worldwide digital traces offered by email and other online activity.
Christakis's multidisciplinary background (he is a qualified clinical doctor as well as a social scientist and healthcare adviser) set him up as an ideal candidate to exploit these developments. For several years he had worked as a hospice doctor in London, examining how doctors made predictions about life expectancy. He was also looking at the health benefits of marriage, and the old problem known as the Widow Effect, first studied in England in the 19th century, asking why an elderly person's death was so frequently swiftly followed by the death of their spouse. This was a non-biological spread of disease from one person to another, and it was crucial in what was to follow. After Christakis had published a paper in the New England Journal of Medicine studying 1 million married couples, it dawned on him that this most basic of social networks could be agglomerated to form structures of greater complexity. "It was the simple idea that I'm connected to you, and you to others, on endlessly into the distance," he remembers. "And then these patterns could branch and form these incredibly complicated networks. I became very interested in this rippling effect – how a health phenomena could affect not just one person, but hundreds or thousands."
In one sense this was a grand extension of Six Degrees of Separation, the concept that we are all just six links away from being connected to everyone else in the world. But now the connections didn't come in straight and direct links but in fuzzy multiples, and Christakis and Fowler were trying to demonstrate something equally stunning: "three degrees of influence". But there was a problem with this thinking: how to prove anything. Christakis needed well-established research groups with hundreds and thousands of people in them. These people should not be lone individuals but somehow linked, and would have to be studied across time. Fortunately, such a group of people already existed.
In Framingham, Massachusetts, the Framingham Heart Study had tracked the cardiovascular health of more than 5,000 people since 1948. It had also kept extensive and interlinking personal records of its patients across three generations.
In 2001, Dr Christakis moved from the University of Chicago to Harvard, and it was here that he met his long-term collaborator James Fowler. Fowler was studying the classic problem of why people vote. (The classic conundrum: everyone knows that a single vote is highly unlikely to have any decisive bearing on an election, so why does anyone bother? The uniform conclusion had been that voting makes no rational sense. Fowler concluded that we vote because other people do – a clear influence of the effects of social connectivity.)
To prove their emerging network theories, the two professors hatched an ambitious plan for a long-term study involving thousands of people, at an estimated cost of $25m. The National Institute on Aging gave them an initial grant of $2m, too little to set up their own new research project but enough to start remodelling what was already out there. So they contacted Framingham, and found its basement records far more useful than its founders had ever anticipated. Every participant had logged their spouse, their children, their friends and their work details, and by doing so had also logged their ties with others in the study. "Previously they had only been examined as individuals," Christakis says, "but our insight was that you could take that population and reconstruct the network. I knew that [from these records] we could get 32 years of social network data if we just spent the money." In 2004 they began the digital reconstruction of about 50,000 social network ties involving 5,000 people.
The digitisation of the records began in 2004, and the reconstruction of the links among the participants began to take shape. The network pictures in Christakis and Fowler's book all originate from the Framingham Heart Study, and have since become famous.
"This is a social cluster of 22,000, mapping obesity," Dr Christakis says, pointing to an illustration. "Or this image here is one of my favourites – mapping happiness. We think of happiness being an individual phenomenon, but think about this: why do you show your emotions? Evolutionarily it would be to our advantage to hide fear or sadness. But we show those emotions on our faces, and surprise and joy, and not only do you read it on my face, but you copy it. An emotional contagion takes place. This suggests that emotion should have a collective existence." The picture shows clusters of happy and unhappy people within the network, and suggests that our happiness is connected with the happiness of people three degrees removed from us; whether we're happy or not depends in part on our friends' friends' friends.
The author's first paper on obesity was published in the New England Journal of Medicine in 2007 and was downloaded 120,000 times. No wonder: the research contained some remarkable figures. If someone on the Framingham study became clinically obese, their friends were 57% more likely also to become obese. A friend of a friend of that obese person was about 20% more likely to become fat, and this was the case even if the weight of the linking friend remained unaltered.
A year later came their paper on smoking, which contained similarly arresting ties. If a person began to smoke for the first time, the chances of their friend doing the same increased by 36%. "The impact surprised us," Christakis says. "We got very lucky."
Christakis and Fowler suggest that human beings in many ways behave like the flocking of birds and the schooling of fish – changing direction all of a sudden. "It's not always explicable in terms of individual actions," Christakis argues. "It would make as much sense to ask an individual smoker: 'Why did you quit?' as it would to ask a single buffalo in a stampeding herd: 'Why are you running to the left?'"
I suggested that a group of people who suddenly quit smoking might have other external causes – an advertising campaign, perhaps, or a plot in a popular soap opera. "Overall, smoking is in decline," Christakis concedes. "But it's patchy. We are often misunderstood in our work. Just because we say networks are important doesn't mean that networks explain everything. We're just adding additional information. Networks don't work like a match – they work like a magnifying glass."
Dr Christakis is keen to emphasise that not everything spreads in networks, and that not everything that spreads in networks spreads the same way. Germs spread differently to emotions, which spread differently to ideas. And online ties spread differently to those we encounter face to face. Christakis has done less conclusive work here, hampered partly by the fact that the definition of "friend" is very different online, and connections tend to be accumulated and then rarely abandoned. But he has produced one intriguing bit of research about Facebook. If a member lists a favourite musical artist on their page, it is unlikely to influence or reflect the choices of their friends. But there are two exceptions. List the Beatles or the Killers, and their popularity on your friends' lists will increase crazily overnight.
Recently, Christakis and Fowler have begun to wonder if we have purposely evolved to live in certain networks. Last year, the Proceedings of the National Academy of Sciences published its first study of the network genetics, suggesting that about 30% of our social ties to others may be embedded in our DNA. This is a profound claim, with a number of implications, and it has led to a search for possible genetic determinants. The authors suggest there may be three: sociability, transitivity and centrality.
The first concerns how many friends we have. "People are genetically programmed to vary in this," Christakis suggests. "Some people are shy, others outgoing. Historically it could be valuable to have more or less friends. Sometimes having more can be very useful – to your health, finding a job. But it can also be costly – they make demands on you, they might want you to lend them money, they can be a pain in the butt." Transitivity is concerned with whether one's friends know each other. Christakis and Fowler contend that people have a genetic predilection to introduce their friends or not, and this too has a basis in evolution. "If you want to hunt a mastodon," Christakis says, "it's really good if all your friends know each other because you can work closely together to kill it. But if you want to find a mastodon, it's much better if your friends don't know each other – because they'll all have the same information. If you don't know your friend's friend, the chances are he will be able to tap more distant regions of the network." Finally, the notion of centrality defines where you are in a network. If a germ enters a connected group, it is preferable to be a loner on the periphery; but if you want to be in on the gossip, its helps to be with more friends in the middle.
Critical reaction to Christakis and Fowler's work has been generally enthusiastic. Reviews of Connected have called it "obvious and brilliant" (New York Times) "unsettling" (FT) and "alluring… another way of seeing the world" (New Scientist). Online criticisms tend to focus on cause and effects, suggesting that their findings are predominantly caused by factors other than our social ties. "The social network wasn't needed to make people fat," one person observed on Wired.com. "The high fructose corn syrup did that for them."
Robin Dunbar, professor of evolutionary biology at Oxford, told me in an email that he found the work of Christakis and Fowler "absolutely fascinating". He believed that healthcare educators have "completely ignored the extraordinarily strong effect of very close relationships in influencing our actions. This effect must apply more widely – to, for example, antisocial behaviour as well as health patterns." Sanjeev Goyal, the professor of economics at Cambridge who has conducted his own pioneering work into networks, also told me that Connected raises timely and fundamental questions about public health policies. If further research proves the author's work to be true, Goyal suggests that our provision of healthcare may have to be reformulated.
At the end of his lecture in Oxford, Dr Christakis put up a final slide of the jacket of his book. The UK title of Connected has changed slightly from the US one, and will now appear with a deftly hyped-up sub-heading: "The Surprising Power of Our Social Networks" has become "The Amazing Power of Our Social Networks". But ultimately its success will depend not on hyperbole, but on something far more embedded in the reading public's natural habits: the amazing power of word-of-mouth. If one of your friends' friends reads the book, it may be only a matter of time before you read it, too – the social network effect that extends as far back as Gutenberg.★
Connected by Nicholas Christakis and James Fowler is published next month by Harper Press, £12.99. To order a copy for £11.99 with free UK p&p go to guardian.co.uk/bookshop or call 0330 333 6847