Is Dunbar's number scientifically proven?

Partially. The correlation between primate neocortex size and group size is solid, and human networks do show layered structure. But the specific value 150 is contested: a 2021 Biology Letters reanalysis by Lindenfors, Wartel and Lind found that different statistical methods yield estimates from about 16 to 109, with confidence intervals spanning 4 to 520. Treat 150 as a useful rule of thumb, not a hard biological ceiling.

How many close friends can you actually have?

Dunbar's layer research puts the innermost 'support clique' at about 5 people and the 'sympathy group' at about 15. These inner layers absorb the bulk of your social time — roughly 40 percent of total effort goes to the closest 5 alone. You can know 150 people meaningfully, but genuine close friendship at the weekly-contact level rarely extends beyond a handful.

Does social media raise Dunbar's number?

Apparently not. Studies of Twitter and Facebook activity by Dunbar and others found that even users with thousands of followers actively interact with a layered network of the familiar sizes — the inner circles still top out around 5, 15 and 50. Platforms make it cheaper to keep weak ties on life support, but they don't expand the capacity for relationships that need real attention.

Network science

Dunbar's number

Dunbar's number is the proposed cognitive limit of about 150 stable relationships per person, derived by anthropologist Robin Dunbar from primate brain data.

Part of the Endearist relationship glossary →

Dunbar's number says that the human brain can sustain roughly 150 meaningful relationships at a time — people you know as individuals, whose connection to you has a history, and whom you would not feel awkward joining for a drink. The figure comes from Robin Dunbar's 1992 paper in the Journal of Human Evolution, which found that across primate species, the bigger the neocortex relative to the rest of the brain, the bigger the typical social group. Extrapolating that regression to the human neocortex ratio yields a predicted group size of about 148 — rounded, ever since, to 150.

The single number is the headline, but the more useful finding is the structure behind it. Your 150 is not a flat list: it is organized in nested layers of roughly 5 intimate confidants, 15 close friends, 50 good friends, and 150 meaningful contacts, with each layer about three times the size of the one inside it. Time and emotional investment are concentrated brutally in the inner layers — Dunbar's group found that about 40 percent of total social effort goes to just the innermost 5.

The number is also contested. A 2021 reanalysis by Lindenfors, Wartel and Lind showed that applying modern phylogenetic statistics to the same primate data produces wildly different estimates with enormous confidence intervals, so 150 should be read as a memorable order of magnitude, not a biological constant.

Where the 150 comes from: neocortex size in primates

Dunbar's 1992 study tested the 'social brain hypothesis': that primate brains grew large to handle social complexity, not ecology. Plotting mean group size against the neocortex ratio across primate genera produced a tight correlation — species with relatively larger neocortices live in larger groups, apparently because tracking who is allied with whom is cognitively expensive. Humans were never in the dataset; the famous 150 is an extrapolation of that primate regression line to the human neocortex ratio. Dunbar then looked for corroborating patterns in human life — Neolithic village sizes, Hutterite community splits, military company sizes, Christmas-card networks — and found many clustering near 150. Critics note that this is exactly where the argument gets soft: a 2021 Biology Letters paper by Lindenfors and colleagues reran the extrapolation with Bayesian and phylogenetic least-squares methods and got point estimates ranging from roughly 16 to 109, with 95 percent intervals as wide as 4 to 520. The cognitive-limit idea survives; the precise number does not.

The layers: 5, 15, 50, 150

Later work by Dunbar's group found that personal networks are organized in a discrete hierarchy rather than a smooth gradient. Zhou, Sornette, Hill and Dunbar (2005) analyzed grouping patterns and identified preferred layer sizes scaling by a ratio close to three: about 5, then 15, then 50, then 150 — sometimes extending outward to roughly 500 acquaintances and 1,500 faces you can put a name to. Each layer corresponds to a different contact frequency and emotional intensity: the inner 5 are typically contacted at least weekly, the 15 about monthly, the 50 and 150 progressively less often. The practical insight is that the layers are maintained by time. When you stop investing the hours a layer requires, a relationship does not vanish — it drifts outward to the next ring. That makes your calendar, not your contact list, the real constraint on your social world.

Using the layers deliberately (and where Endearist fits)

You cannot will your way past the limit, but you can choose who occupies each layer instead of letting proximity and habit choose for you. A useful exercise: list the people you actually contacted in the last three months, sort them into 5 / 15 / 50, and compare that revealed network with the one you would design on purpose. Most people discover that a few relationships they value deeply have drifted two layers further out than intended. Endearist is built around exactly this model — you assign each contact a priority tier that mirrors the Dunbar layers, and the app derives how often each tier should hear from you, so attention flows where you decided it should. The free Dunbar calculator on this site walks you through mapping your own layers in a few minutes, no account needed.

Try it yourself

dunbar calculator →

Frequently asked questions

Is Dunbar's number scientifically proven?: Partially. The correlation between primate neocortex size and group size is solid, and human networks do show layered structure. But the specific value 150 is contested: a 2021 Biology Letters reanalysis by Lindenfors, Wartel and Lind found that different statistical methods yield estimates from about 16 to 109, with confidence intervals spanning 4 to 520. Treat 150 as a useful rule of thumb, not a hard biological ceiling.
How many close friends can you actually have?: Dunbar's layer research puts the innermost 'support clique' at about 5 people and the 'sympathy group' at about 15. These inner layers absorb the bulk of your social time — roughly 40 percent of total effort goes to the closest 5 alone. You can know 150 people meaningfully, but genuine close friendship at the weekly-contact level rarely extends beyond a handful.
Does social media raise Dunbar's number?: Apparently not. Studies of Twitter and Facebook activity by Dunbar and others found that even users with thousands of followers actively interact with a layered network of the familiar sizes — the inner circles still top out around 5, 15 and 50. Platforms make it cheaper to keep weak ties on life support, but they don't expand the capacity for relationships that need real attention.

Sources

Last updated: 2026-06-10

Tend relationships, not records.

Endearist is a local-first personal CRM. Free up to 25 contacts.

Start free