Robin Dunbar was not studying humans at all. In the early 1990s, the British anthropologist was puzzling over monkeys and apes: why do some primates live in small bands and others in large troops? His answer was that social life is computationally expensive. Keeping track of who is friends with whom, who owes a favour, who cannot be trusted — all of that bookkeeping takes brainpower. When Dunbar plotted the size of each primate species' neocortex, the brain's outer thinking layer, against the size of its typical social group, the two lined up remarkably well. Bigger social brains, bigger groups.
Then came the irresistible move. Humans are primates too. If you take the human neocortex and read our species off the same curve, out comes a prediction for our natural group size: roughly 148, which everyone promptly rounded to 150. Dunbar's number was born — a claimed cognitive ceiling on the number of stable, meaningful relationships one person can maintain. His informal definition is the set of people you would not feel awkward joining uninvited for a drink if you bumped into them.
The number that keeps turning up
What made the idea famous is how often something like 150 appears when humans organise themselves. Dunbar pointed to Neolithic villages, and to the average English village recorded in the Domesday Book of 1086, both in the same range. The Hutterites, communal Anabaptist farmers in North America, deliberately split their colonies once they grow to around 150 people, having learned that beyond that size, peer pressure alone no longer keeps order and formal rules become necessary. Military organisation tells a similar story: the company, the basic fighting unit of armies for centuries, typically musters somewhere between about 100 and 200 soldiers — big enough to fight, small enough that everyone knows everyone.
Business found the pattern independently. Bill Gore, founder of W. L. Gore & Associates — the company behind Gore-Tex — capped his factories at around 150 workers, building a new plant rather than expanding an old one, because he had noticed that beyond that point people stopped saying "we" and started saying "they". And when Dunbar and a colleague studied British Christmas card lists — in the pre-social-media era, a decent census of who someone bothered to maintain a connection with — the typical network of people reached came out at about 150.
Circles within circles
Dunbar's number is really the outer ring of a series. His research describes social life as a set of nested layers, each roughly three times the size of the last: about 5 intimate loved ones, around 15 close friends, roughly 50 good friends, then the famous 150 meaningful contacts. Beyond that lie about 500 acquaintances and something like 1,500 people whose faces you can put a name to. As you move outwards, both emotional closeness and time invested drop sharply. The layers are not just poetic: similar ratios show up in hunter-gatherer band structures and even in the organisation of armies, from squads up through platoons and companies.
The glue holding all this together, in Dunbar's account, is time. Monkeys maintain their alliances by grooming one another, one partner at a time — a method that simply cannot scale to groups of 150. Dunbar's proposed solution is one of his boldest claims: language evolved, at least in part, as a form of vocal grooming. Conversation, and especially gossip, lets humans "groom" several companions at once and keep tabs on absent third parties. Studies of casual conversation suggest a large share of it is indeed devoted to social topics — who did what with whom.
How seriously to take it
The number has critics. In 2021, researchers at Stockholm University reran the primate analysis with newer statistical methods and argued that the confidence intervals are so wide — spanning from a few dozen to many hundreds — that no single human number can honestly be extracted. Dunbar has defended the figure, pointing to the sheer range of independent human datasets that cluster near 150. The fairest reading may be that 150 is not a law of nature but a useful order-of-magnitude truth: human social capacity is limited, and the limit sits in the low hundreds, not the thousands.
Social media makes the point rather than breaking it. People may accumulate thousands of followers, but studies of interaction patterns online find that the number of people someone actually exchanges messages with, repeatedly and reciprocally, remains stuck near the old layers. The platforms changed; the brain running them did not. Whatever the exact figure, Dunbar's core insight has aged well: friendship is not free. It runs on time, memory and attention — and all three are finite.
Quiz nuggets
- Dunbar's number — about 150 — is the proposed limit on stable social relationships, derived from primate neocortex size.
- It was proposed by British anthropologist Robin Dunbar in the early 1990s.
- Hutterite communities traditionally split their colonies when they grow to around 150 people.
- Gore-Tex maker W. L. Gore capped factories at about 150 workers to keep everyone on first-name terms.
- Dunbar describes friendship in layers of roughly 5, 15, 50 and 150, each about three times the size of the last.