150 — DUNBAR'S NUMBER

Robin Dunbar's argument is short. Across primate species, the size of the neocortex relative to the rest of the brain predicts the size of the social group the species lives in — plausibly because keeping track of who owes what to whom is the expensive part of being social. Dunbar fitted that relationship in 1992, put the human neocortex ratio into the fit, and got a group size of about 150. The wall is made of brain.

The wall is made of brain.

150 is not a limit on how many people you can recognise or name; that runs into the thousands. It is a limit on stable relationships — people whose place in your world you keep updated, and who would not be strange to sit down with. Dunbar's layers nest: roughly 5 intimates, 15 close friends, 50 friends, 150 meaningful contacts, then 500 and 1500 as the acquaintance and face-recognition tiers, each layer about three times the one inside it.

The corroboration is why the number stuck. Neolithic farming villages, Hutterite communities that split when they pass about 150, the company as the basic manoeuvre unit of an army, the size of a working academic sub-discipline, the bands of hunter-gatherer societies — all land in the same range. None of that was predicted by the regression. It was found afterwards, which makes it support rather than proof.

And the number is contested exactly where it is quoted most confidently: in its precision. A 2021 re-analysis in Biology Letters refitted the same primate data with modern phylogenetic methods and got point estimates between about 16 and about 109 depending on the model, with 95% confidence intervals running from roughly 2 to 520. The authors' conclusion was not that humans have no social ceiling — it was that the available data cannot locate it to within a factor of ten, and that stating a single figure to three digits claims a precision the regression never had. 150 is a good order of magnitude and a bad decimal.