Dunbar's number

Dunbar's number is the proposed limit on how many stable social relationships one person can maintain at a time, usually given as about 150. It comes from the British anthropologist Robin Dunbar, who found a correlation between the size of a primate's neocortex and the size of the group it lives in, extrapolated a human figure from human brain size, and then checked it against the sizes of real human groups.

The number is really a series. Dunbar described nested layers, each roughly three times the one inside it: about 5 people in the intimate support group, 15 close friends, 50 good friends, 150 meaningful contacts, 500 acquaintances, and 1,500 faces you can put a name to. The layers matter more than the headline figure, because they say something the single number does not. Attention is not spread evenly, and a person keeps their layer only while the relationship is fed.

The supporting evidence is a mix: comparative neuroanatomy across primates, the average size of hunter-gatherer clans and Neolithic farming villages, the size of a Roman maniple and of a modern military company, and later work on how calls and messages distribute across a person's contacts, which does show a layered pattern. Christmas card studies put the average sending network close to 150 as well.

It is also contested, and worth reporting as contested. A 2021 replication at Stockholm University reran the neocortex analysis with newer data and better statistics and found the confidence interval so wide that no point estimate survives. Critics add that group sizes in the historical record were shaped by ecology and logistics as much as by cognition. What holds up under the criticism is the layered structure and the constraint underneath it, which is time. Relationships are maintained by contact, and contact is finite.

The ceiling is on active maintenance, not on acquaintance. You can hold thousands of contacts in a phone and recognize thousands of faces. What runs out is the capacity to track what is currently happening in each of those lives. That is why a network past the ceiling does not collapse loudly. It decays quietly, one missed follow-up at a time, and the people who drop out are rarely the ones you would have chosen to drop.

Because the binding constraint is memory rather than affection, an external record moves it. Written correspondence, address books and now software all do the same job: they hold the detail that would otherwise have to sit in your head, so the layer past 150 stops being anonymous. The ceiling on how many people you can genuinely care about does not move. The ceiling on how many you can keep track of does.