Robin Dunbar’s Number: The Brain Behind Social Circles
Robin Dunbar’s famous number did not begin with a survey of friendships, a count of Christmas cards, or a village census. It began with primate brains. In 1992, the British anthropologist reported that primate species living in larger social groups tended to have a larger neocortex relative to the rest of the brain. When he extended that relationship to humans, the calculation produced an expected group size of 147.8. The memorable version became 150.
Then the number seemed to turn up in the human world. Groupings close to it appeared in accounts of hunter-gatherer communities, traditional settlements, and military organisation. A later study of Christmas-card networks produced a maximum-network average of 153.5.
It is an unusually elegant story: the architecture of our brains quietly shaping the architecture of social life. But 150 is neither a hard biological wall nor a prescription for how many friends anyone should have. The evidence is more interesting when its limits remain visible.
The Original Prediction: An Extrapolation from Primate Brains
Dunbar’s 1992 paper in the Journal of Human Evolution examined the relationship between social-group size and the neocortex in primates. This foundational idea became widely known as the social-brain hypothesis.
Group living creates mental work. Animals must recognise individuals, remember past encounters, follow alliances and conflicts, and understand how two others relate even when not directly involved. Dunbar proposed that the computational demands of this social world helped shape primate brain evolution.
He fitted a statistical relationship to non-human primate data, then inserted the human neocortex ratio. The result was 147.8, rounded to the widely cited Dunbar’s number of 150.
This origin matters. Researchers did not first count representative samples of human friendship networks and discover that everyone stopped at 150. Rather, the number was a prediction generated by a cross-species regression, awaiting comparison with human social groups.
Human Comparisons: When 150 Looks Uncannily Familiar
In a 1993 analysis of human group size, language, and the neocortex, Dunbar compared his prediction with the scales at which people had repeatedly organised themselves.
Accounts of traditional societies included community groupings near the predicted range. Hunter-gatherer life was not organised at a single scale: small, shifting foraging bands could belong to wider communities containing roughly 150 people. Understanding the distinction between a camp and the broader community is important when comparing seemingly different estimates.
Military organisation offered another striking parallel. A company often contains somewhere around 100 to 200 people, while smaller platoons and squads sit inside it. This structural arrangement has existed in various forms across long periods, despite enormous changes in technology and administration.
These examples helped transform a brain-based estimate into a theory of human organisation. However, they are suggestive rather than definitive. A soldier, a neighbour, and a friend do not occupy equivalent relationships. Military units are deliberately designed institutions, while village records vary in whether they count residents, households, adults, or regular participants.
Finding a recurring scale across such settings is intriguing, but it is not the same as measuring a universal social capacity directly.
Christmas Cards: A Clever Real-World Test
In 2003, Russell Hill and Dunbar turned to a mundane seasonal ritual. Their study of Christmas-card exchanges treated cards as traces of active social relationships in contemporary Western life.
The method had a practical intelligence. Sending a card requires enough connection to remember someone, possess their address, and deliberately include them. At the same time, a card list can extend beyond the handful of relatives and friends a person might see regularly.
The maximum network size averaged 153.5 individuals, strikingly close to Dunbar’s prediction. When researchers counted only people explicitly contacted, rather than every person represented within a recipient household, the mean was 124.9. Kin made up a fairly stable 21% of networks.
The difference between 153.5 and 124.9 is revealing. A card addressed to a couple or family can represent several relationships, but not necessarily relationships of equal depth. The answer changes depending on whether researchers count households, individuals, direct contacts, or everyone socially represented by the exchange.
The Christmas-card results support the idea of a broad active network near this scale but do not demonstrate that every adult possesses exactly 150 stable bonds.
The Network Is Layered, Not Filled With 150 Close Friends
One of the most common misunderstandings is that Dunbar’s number means people can have 150 friends of equal emotional importance. The model is hierarchical.
A later analysis of social-group sizes found clusters near 4.6, 14.3, 42.6, and 132.5, with each layer about three times the size of the one inside it. These are statistical centres, not mandatory slots.
The innermost layer is the small support clique someone would turn to in a crisis. Around it sits a sympathy group of close friends and family, then a wider friendship layer, and finally the broader active network associated with approximately 150 individuals. Larger layers extend to familiar acquaintances and recognisable faces.
This hierarchical structure makes intuitive sense. A person may know roughly 150 people well enough to understand who they are and how they fit into the network, while trusting only a few with grief, fear, or a late-night phone call. Recognition is cognitively cheaper than emotional intimacy.
Time may be as important as memory. Close relationships require conversation, shared experiences, and repeated attention. Adding one bond does not simply consume another “name-sized” compartment in the brain; it competes for hours that are already divided among other people.
This view aligns with research on why some happier people maintain fewer, less draining friendships. Network size and relationship quality are different questions. A smaller circle can be socially rich, while a large contact list might contain little dependable closeness.
A Serious Statistical Challenge to the Precise Ceiling
In 2021, Patrik Lindenfors, Andreas Wartel, and Johan Lind revisited the primate-to-human inference using updated datasets and phylogenetic methods. Their paper, “Dunbar’s number deconstructed,” did not recover one stable value.
Depending on the dataset and method, their point estimates for average human group size ranged from 16 to 42 or from 69 to 109. More damaging to the idea of a precise number, the 95% confidence intervals stretched from 2 to 336 and from 4 to 520.
The authors concluded that a single cognitive limit could not be reliably derived from this kind of primate regression. The uncertainty was simply too large.
Defenders of Dunbar’s number place more weight on the convergence of human observations than on the original estimate alone. They also argue that researchers must compare the same social layer across species and societies. A temporary camp, an intimate support group, and a whole community answer different questions.
This disagreement leaves a sensible middle position. Human social networks may face real cognitive and time constraints without nature having engraved the integer 150 into every brain.
One Hundred and Fifty: A Scale, Not a Social Target
Dunbar’s number should not become another score people use against themselves. Having fewer than 150 relationships is not evidence of social failure. Having more than 150 contacts does not violate any biological law.
A person can recognise thousands of names, accumulate online followers, and remain loosely connected to former classmates or colleagues. The theory concerns reciprocal relationships that are stable enough to require maintenance. Even there, personality, age, culture, work, family structure, and communication technology can shift the boundary.
The more useful insight is that social networks have depth and attention has limits. Relationships move between layers. A new friendship can become central; an old one can soften into occasional affection. Parenthood, relocation, illness, and retirement can reorganise the entire pattern.
That is why the theory remains compelling even after its precise number becomes fuzzy. It asks us to see social life not as an unlimited collection of contacts, but as a living structure sustained by memory, time, and mutual recognition.
One hundred and fifty may be less like a wall than a recurring scale: a place where human groups often seem to settle, not a fixed line where every individual must stop.
Read more Here.
