What is Dunbar's number, and does it hold up?

Dunbar's number is the idea, proposed by the British anthropologist Robin Dunbar, that humans have a natural social group size of about 150 people. It is an average, not a hard cap. Dunbar and some other researchers say decades of data support it. Critics say the statistics behind it are too uncertain and that humans have no fixed cognitive limit. [1] [2]

Dunbar noticed a relationship between brain size and group size in primates, which became known as the social brain hypothesis. The figure of 150 was not observed first and then explained. According to a FactCheck piece in The Conversation, it is a prediction from an equation describing the statistical association between neocortex size and typical group size across primate species. The same piece stresses that it does not come from an ecological principle or evolutionary law. Dunbar's own account says the community level of organisation turned out to be almost exactly 150. [3] [4] [1]

Dunbar says 150 is an average and not a strict limit, and that it is one layer in a fractal series of friendship circles. A 2024 review he co-authored lists the layers as roughly 1.5, 5, 15, 50, 150, 500, 1500 and 5000. [2] [5]

In the 2024 review, Dunbar's group reports a robust statistical relationship between social group size and neocortex size across many animal groups. They say the predicted human figure of about 150 was confirmed by 23 studies of personal networks and ethnographic communities, drawn from many cultures and from the last 2000 years. They accept considerable variation between individuals because of gender, age, personality and circumstances. Dunbar told Live Science that there has been no change in the number of relationships. [5] [2]

Some independent researchers are partly supportive. Sarah Johns of the University of Kent told Live Science the number is broadly supported but not absolute in every circumstance, and that counting only close and meaningful relationships would give a much smaller figure. Samuel Roberts of Liverpool John Moores University pointed to wide variation in network sizes and noted that the theory has critics. [2]

Patrik Lindenfors, Andreas Wartel and Johan Lind reanalysed primate data in Biology Letters in 2021. Their estimates of human group size ranged from about 16 to 109 depending on the method, with 95% confidence intervals from roughly 2 to 520. They concluded that no single number can be derived this way and that the upper limits often far exceed 150. [6]

They also argue that primate group size is mainly explained by foraging, predation and reproductive factors, not brain size. They add that human culture and social structures can offset cognitive limits, and that other researchers have reported mean group sizes unlike 150. They call for the term to be dropped from science. Which side most specialists favor is not settled by these results, and the two camps rely on different data and methods. [6]