Social network theory presents a drastic new way for approaching personal responsibility as well as how we influence one another.

Joe is a typical American guy; a little pudgy in middle age, he definitely understands messages about eating right, exercising and taking care of himself. Yet there he goes again, grabbing a fast-food lunch for the third time this week. Why the disconnect? Who's responsible for the gap between his knowledge and his behavior?

Health experts typically see only two culprits. One is the individual himself: a hypothetical Joe has only himself to blame if, despite all he's heard about sedentary living and a high-fat diet, he winds up with diabetes or heart disease. The opposing outlook holds that eating, drinking and other "lifestyle" choices are driven by vast forces outside any individual's control. If we eat too many fries, blame fast-food advertising, agricultural policy, the effects of social class and culture, and even heredity.

Real people, of course, fall between these two stereotypes. They're neither impervious to influence, nor its slaves. To better represent that reality, some new ideas about responsibility have entered the conversation through the burgeoning field of social-network research. Its practitioners track how behaviors spread via the social ties that bind friend to friend, co-worker to co-worker, neighbor to neighbor. Not all ties are equal, but all can transmit influence from person to person. "The study of human social networks provides a way, in some sense, to bridge the gap between notions of pure individual choice and notions of collective determinacy," says Nicholas A. Christakis, a physician and medical sociologist at Harvard and one of the new field's inventors.

Though "social network theory" sounds as if it stems from the Internet or Facebook, it's rooted in Christakis's experience in Chicago in the early 1990s, when he worked as a hospice physician. There, he noticed that the illness of terminal patients had a huge impact on the health of other family members: a husband, for instance, might became ill due to the stress and exhaustion experienced by his wife as she cared for her demented mother. Christakis came to think of these effects as "the non-biological transmission of disease." With input from new work on networks in the social sciences, he was soon able to attach hard statistics to his intuitions. Many kinds of behavior, both good and bad, are influenced by the ties that people have to one another.

Christakis moved on to study much larger social networks, which offer a new and often better way to explain the spread of many kinds of health-related behavior, including smoking, sexually transmitted diseases, catching the flu, losing sleep and feeling depressed. The statistics are fascinating: If a friend of a friend of a friend of yours takes up smoking, you are 11 percent more likely to take up the habit; a teenager's friend who gets less than seven hours of sleep a night increases the risk by nearly 20 percent that another of that first teenager's friends will sleep less than seven hours -- even though those two don't directly know each other.

In some sense the numbers aren't surprising; smokers tend to know smokers, and your chances of being obese are greater if a lot of your friends and relatives are obese. The question, especially when it comes to assigning responsibility, is why. Some common explanations for such clustering of behavior depend on the idea that "like attracts like" (people are more comfortable with others who are similar). Other theories look to large-scale social forces (smoking advertisements, a bad economy) that touch many people at the same time. Both kinds of models allow, of course, for the fact that individuals influence one another. Again, it's not hard to imagine how a spouse or co-worker might exert a pull toward fatty dinners or social drinking.

Network theory is novel because it suggests that influence doesn't work that way. Instead, it holds, the characteristics of networks themselves explain how behaviors move through society. To understand why your odds of smoking increase based on the actions of a friend's friend's friend, Christakis doesn't look at individual characteristics nor at the effects of the recession on everyone. Instead, he argues, we must understand the social ties that bring strangers' influence to bear on your choices.

In the most famous example of this approach, in 2007, Christakis and longtime collaborator James Fowler, a political scientist at the University of California San Diego Medical School, used thousands of health records from the multi-decade Framingham Heart Study to demonstrate that obesity seems to spread through social ties. Since 1948, the study has collected detailed medical and lifestyle data on thousands of residents of Framingham, Massachusetts. (The study is one of the main sources of current knowledge about the links between diet, lifestyle and heart disease). Because the Framingham researchers kept detailed information about the social lives of their volunteers, Christakis and Fowler were able to map the social relationships of thousands of friends, relatives, neighbors and co-workers over 32 years.

Relating that map to medical data, the pair found that, on average, the chance of any particular person becoming obese was 20 percent higher if a friend of a friend (or friend of a co-worker, or co-worker of a relative) was obese. The connecting friend need not become fat himself, and he might have moved a thousand miles away. Most importantly, the two people connected through him need not know one another. Even more surprisingly, the pair found that the effect extends to three degrees of separation. Any individual's risk of being obese was 10 percent greater if someone at three removes -- say, the friend of a friend of a co-worker -- was obese.

The image of a social virus, like the sexually transmitted diseases and flu the pair has also studied, proved irresistible to the media. "Obesity Can Be Contagious," was one headline. "Are Your Friends Making You Fat?" asked a national magazine. In fact, the striking news in Christakis and Fowler's social-network studies is not that behaviors like unhealthy eating or careless sex can be influenced by friends and family. Rather, it's the impact of people you don't know -- your co-worker's sister's friend from college -- that matter more. "You may not know him personally," Fowler and Christakis wrote in Connected, their book about the work, "but your friend's husband's coworker can make you fat."

Other research has found similar effects. Because the Framingham participants were also asked about their mood and outlook, Christakis and Fowler were able to trace feelings of loneliness through social networks. Over the years, they found, individuals in the study were 52 percent more likely to say they felt lonely if someone they knew directly was lonely. A lonely person at two degrees of separation (a relative's friend, for example) boosted the person's odds of being lonely by 25 percent. And if even a friend of a friend of a friend was lonely, that made the subject 15 percent more likely to report loneliness. Similarly, an individual is 40 percent more likely to report being depressed if that person has a friend of a friend (or other second-order social tie) who is depressed.

Not all their studies have used Framingham data. In a paper on teen habits, for instance, they asked 90,000 American teenagers who had answered a health survey to name up to 10 friends, and then they mapped the connections between answers and social ties. As it turns out, a teenager is over 80 percent more likely to use marijuana if a friend of a friend uses the drug, and 38 percent more likely to do so if it's a friend of a friend of friend who's doing it.

Social ties carry influence in many ways, Christakis and Fowler believe. Sometimes, it's a slow and subtle alteration of what feels normal: As obesity spreads in a network, it becomes unexceptional to those in it. In other cases, a few people with many social ties may deliberately "change the rules," and the effect of their decision travels far along the ties that bind friend to friend.

The key point, Christakis says, is that people act like waves in a pond, and that our behaviors, like our diseases, ripple out as if from a fallen pebble, following the laws of physics whether we know it or not. In some respects this contradicts the social-atom model of decision-making, whereby each individual can and should be held responsible for each and every choice. In fact, "a smoker may have as much control over quitting as a bird has to stop a flock from flying in a particular direction," Christakis and Fowler write.

But, Christakis adds, any social link is a two-way street: The ties that affect you are also carrying your influence out to other people. The American media didn't pick up on that symmetry, but news outlets in less individual-oriented cultures did. So most United States newspaper headlines telegraphed the obesity study by saying, "‘Are You Packing it On? Blame Your Fat Friends,’" Christakis says. "But a lot of the British headlines had a different tone: 'Are Your Friends Gaining Weight? Perhaps You Are to Blame.'” Even as social-network theory seems to undermine the primacy of the individual, it also raises the individual's role, Christakis says. "We're saying that if you make a positive change in your life, you don't just benefit yourself; you also benefit your friends and their friends and so forth."

That expands the standard view on personal responsibility. The individual is not a social atom, with no impact on others; nor is it a social victim, helpless to resist their influence. Rather, Christakis believes, we each should realize that even our "private" decisions really do have an impact on others far removed from those we know directly, and we should take responsibility for the influences we transmit. If you decide to eat better and get more exercise, he says, "Maybe you don't care about all the strangers you'll help. But you're going to affect people you do care about--your family, your friends, and your neighbors."

In fact, Christakis notes, the new theory of social networks invokes ancient principles -- responsibility matters precisely because we care about what we do to and for others, not just ourselves.

David Berreby blogs about behavior at Bigthink.com and has written about science for The New Yorker, The New York Times Magazine, and many other publications. He is the author of Us and Them: The Science of Identity, published by Little, Brown.