Dunbar's Number: The Cognitive Limit on Human Social Relationships

150 Is Not a Rule — It Is a Neurological Ceiling

In 1992, British anthropologist and evolutionary psychologist Robin Dunbar published a paper in the Journal of Human Evolution presenting a regression line that has since become one of the most cited and contested findings in social science: the size of a primate's social group correlates with the size of its neocortex, and extrapolating this relationship to human brain size predicts a social group limit of approximately 150 individuals. This figure — now called Dunbar's number — is not a preference or a cultural rule. It is, Dunbar argues, a constraint imposed by the cognitive demands of tracking complex social relationships in a primate brain that has not fundamentally changed in 100,000 years.

The number has been tested, extended, replicated, and criticized across multiple disciplines, and it has influenced organizational design at companies including Gore-Tex (W.L. Gore and Associates), the Hutterite communities of North America, and military unit structures that appear to have independently converged on similar sizes.

The Social Brain Hypothesis

Dunbar's number rests on the Social Brain Hypothesis — the proposal that the evolutionary pressure driving the expansion of the primate neocortex was not environmental complexity or tool use but the demands of navigating complex social groups. Managing a network of relationships requires tracking individual identities, remembering past interactions, predicting future behavior, and modeling what other individuals know and intend (Theory of Mind). This cognitive load scales faster than linearly with group size.

The data supporting the neocortex-group size correlation spans 36 primate genera:

The neocortex ratio is the volume of the neocortex divided by the volume of the rest of the brain. The human neocortex is disproportionately large by primate standards — the prediction from the regression line is 148 ± some range, typically cited as 100–230.

The Layered Structure of Social Networks

Dunbar's research identified not a single number but a nested series of grouping layers, each roughly three times larger than the previous:

• Support clique (~5): The innermost circle — people whose death would be devastating and who provide emotional and practical support in a crisis. This layer requires the most maintenance — frequent, high-quality interaction.
• Sympathy group (~15): Close friends seen regularly; people you would confide in and whose welfare you follow actively.
• Affinity group (~50): Good friends; people you know well enough to invite to a gathering; regular but not constant contact.
• Active network (~150): Dunbar's number proper — the full social community within which you maintain stable, trust-based relationships. Beyond this, you know who people are but lack the relational depth needed for collaboration without formal rules.
• Acquaintances (~500) and recognition (~1,500): People you recognize by face or name; shallow connections.

Real-World Validation

Several independent observations have appeared to support the 150 figure:

• Hutterite communities: The Hutterites — an Anabaptist communal society — traditionally split their communities when they exceed 150 members. Members report that above this size, social control breaks down and formal organizational hierarchy becomes necessary. The Hutterite rule was observed to be approximately 150 before Dunbar published his work.
• Military unit size: The Roman legion's basic tactical unit (maniple, later cohort), the medieval English "hundred," and modern military companies all cluster around 100–200 individuals — the unit size within which officers can know all subordinates personally.
• W.L. Gore and Associates: The materials company (makers of Gore-Tex) deliberately limits factory units to approximately 150 employees, a practice reportedly adopted before awareness of Dunbar's research, based on founder Bill Gore's observation that personal relationships and intrinsic motivation break down in larger units.
• Christmas card studies: Dunbar's own empirical work on Christmas card sending in British households found a mean active network size of 153 people per household.

Criticisms and Limitations

Dunbar's number has attracted significant criticism:

• Wide confidence intervals: The regression line's extrapolation to humans has large confidence intervals — the predicted range in the original work was approximately 100–230. A number cited as "150" encompasses a broad statistical range.
• Online social networks: Studies of Facebook, Twitter, and other platforms show that active social interactions (comments, messages) cluster at smaller numbers than 150, even in accounts with thousands of followers. Others argue this supports Dunbar — most connections are superficial, and the active core remains around 100–200.
• Cultural variation: A 2021 meta-analysis by Lindenfors et al. in Biology Letters reanalyzed Dunbar's original dataset and found that when appropriate statistical corrections are applied, the data are less constraining — predicting a wider range for human group size than the iconic 150 suggests.
• Individual variation: Network size varies considerably between individuals based on personality (particularly extraversion), occupation (e.g., politicians, priests), and social context. The «150» describes a mean, not a universal limit.