Humans are social animals, and our species has evolved some unique ways of enforcing the bonds of friendship. Robin Dunbar, an evolutionary psychologist at the University of Oxford, studies the behavioral mechanisms behind the number and nature of such relationships. His work suggests social cohesion and long-term bonding among primates—Homo sapiens included—are the keys to their evolutionary success.

Primate societies are held together by unspoken contracts grounded in “social grooming,” whether in the form of physical affection or nonphysical activities such as storytelling. Primates possess limited time for this, but the more time individuals invest in such activities the more relationships they can maintain—which, in part, explains how big or small social groups are, Dunbar says.

His earlier research from 1992 revealed a correlation between brain size and social group size, suggesting humans can only comfortably maintain about 150 stable friendships whereas nonhuman primates can only sustain about 50.

These days he investigates the cognitive tricks humans have used to reach out to one another and expand our social groups. At a recent meeting of the American Association for the Advancement of Science, Scientific American  spoke with Dunbar about how these “rules” of friendship relate to modern armies, singing groups and classroom size.

[An edited transcript of the interview follows.]

What’s the most notable thing that you have learned in your recent studies of human relationships?
My colleagues and I have found that, since humans have much bigger social groups than nonhuman primates, our species must be using more sophisticated bonding mechanisms than simple physical grooming—for instance, laughter, singing and dancing. We have also concluded that the function of social grooming is to activate the brain’s endorphin receptors. These activities give us a mental and physical “social high.” We have also found that human social groups are structured in a series of layers that extend out beyond the 150 number, and these layers have a very specific relationship to one another.

How many of these layers are there?
Generally speaking, humans each have one to two special friends, five intimate friends, 15 best friends, 50 good friends, 150 “just” friends and 500 acquaintances. Our relationships form a series of expanding circles of increasing size and decreasing intensity and quality of the relationship. Not only do we see these circles in the structure of [modern, real-world] social networks and of hunter–gatherer communities—they’re also reflected in big data gleaned from Facebook posting and telephone call frequencies. It also turns out these layers are germane to the organizational structure of modern armies.

How so?
Company size in the British army is 120 troops, and it’s 180 in the American army. These sit nicely on either side of 150, the number of friends per person my research identified. There’s just something special about these numbers, something to do with the structure of relationships that makes them very stable. Even the number of “intimate friends” is reflected in the size of special forces units. For example, British Special Air Service squadrons have four men each.

Are these numbers also reflected in social media?
Yes. Stephen Wolfram, creator of Wolfram|Alpha and the founder and CEO of Wolfram Research, did a study that looked at one million Facebook pages and the most common number of friends was between 150 and 250. We did a smaller-scale study of 3,500 adults in the U.K. and got similar results. Our social world is extremely small-scale.

Now that research shows these group size numbers have stabilized—what does that information tell us?
Well, first about the nature of your friendships and your social world. It has become apparent in the last 10 years that the most important factor influencing your health, well-being, risk of falling ill, even your risk of dying and divorce is actually the size of your friend network. Particularly on the health side, your network seems to cushion you—your friends help you out when you’re in trouble—and more importantly, the endorphins that kick in from interacting with them seem to tune up your immune system. Laughing together, jogging together, dancing together, singing together, telling emotionally wrenching stories, going to see weepy films—these activities buffer the body biochemically and immunologically against the kinds of coughs and colds of everyday life.

How would you encourage someone to make friends?
One of the best ways is joining a singing club. We did a study comparing novice singing classes with novice hobby classes in terms of how much these activities produced feelings of social bonding. Singing produces a massive hit of endorphins, and that makes you feel very bonded to the people with whom you’re doing it. We call it the “icebreaker effect.” It seems when you go once a week to your glee club or barber’s quartet and sing together, it just ramps up this sense of belonging. It’s similar with dancing and jogging. I see people jogging with their ear buds, listening to music, and I think, “You’re doing it wrong. Take the freaking earpieces out and talk to the guy next to you.” It’ll ramp up the effect of the workout or the dance. You’re going to get an endorphin kick from any physical activity, but if it’s done in synchronicity with somebody else, the effect ramps up significantly, our studies of dancers and rowers have found. And that’s why dancing works, too—you can have large numbers of people doing the exact same thing in sync.

How have singing, dancing and laughter helped humans maximize our social bonding and group sizes?
The problem is that physical grooming is really inefficient. It is a one on one activity that’s very good at stimulating the release of endorphins, but you can only do it with one person at a time. For humans to increase group size, we somehow we had to break through that particular barrier of just using grooming to bolster community bonds. We had to find new ways of doing these things. We cannot invest more time, we just don’t have it. The only option we have, then, is to “groom” with more people at once. And laughter allows you to do that with several people simultaneously. What’s more, you get the benefit yourself—in grooming, only the person being groomed is getting the endorphin hit, but when you laugh, you get the hit, too. So it’s already twice as efficient. It turns out the limit of conversational laughter group size—say, at a bar—is about three people, which is slightly below the limit for conversation group size, which is four. These numbers are very consistent in my research. When laughter triggers these endorphins, it’s three times as effective as physical grooming, so it should allow you to increase group size by a comparable proportion. The same goes with storytelling.

What other areas might these stable social numbers might be informative for?
The other side has to do more with whether there are sizes of organizations that work more efficiently than others. One that is really glaringly obvious is schools. By and large, American high schools are pretty big affairs, which we’re endeavoring to copy in Britain. Our secondary schools are around 1,500 students in size, and I think that is completely dysfunctional in terms of creating a sense of community. I think you need much smaller groups, particularly for the younger kids. The average social network size for an 18-year-old is still only about 50 people. You don’t hit the 150 mark until your mid-20s. And if your natural social world is 50 people, being thrown into a school environment of several thousand is not only going to be stressful, it’s also going to be much more difficult to build a sense of community.