In 1992, the British anthropologist Robin Dunbar published a paper that suggested an unusual hypothesis. Looking across primate species, he noticed that the size of the neocortex correlated with the size of the typical social group those primates lived in. Bigger neocortex, bigger group. He extrapolated to humans and proposed that the natural size of a human social group — the number of people we can maintain stable, real relationships with — was somewhere around 150.
That number has had an unusual life. It's been embraced by management consultants, anthropologists, military planners, religious community designers, and tech founders. It's also been criticized by other researchers who think the number is softer than the popular framing suggests. Both groups have a point.
The more interesting and more practical thing isn't the precise number. It's the underlying observation, which holds up across multiple lines of evidence: human brains have a finite capacity for tracking social relationships, that capacity is meaningfully constrained, and modern social environments routinely exceed it. What that means for how to live is worth thinking through.
What Dunbar's research actually claimed
The original 150 number came from a regression of neocortex ratio against typical group size in primates. Dunbar's argument was that maintaining stable relationships requires cognitive resources — you have to remember people, track their relationships with each other, predict their behavior, manage your own social position relative to them. More cortex enables more relationships.
Subsequent research has refined this. Dunbar himself proposed a layered structure: roughly five intimate relationships (the people you'd reach for in a real crisis), fifteen close friends, fifty good friends, 150 meaningful contacts, 500 acquaintances, and 1,500 people whose faces you can recognize. The exact numbers vary by individual and method, but the layered pattern is robust.
What the research consistently finds is that these layers aren't symmetric. The amount of attention you give to each is roughly half what you give to the layer below. Most of your social cognition is spent on a small number of people. The 150 isn't 150 equal relationships; it's a tapering distribution where the top few get most of the attention.
The 150 number specifically has been challenged. Some researchers argue the real number is more variable, perhaps wider than 150 in some contexts, narrower in others. Studies looking at things like Christmas card lists, military unit sizes, or self-organized communities tend to find numbers in roughly that range, which lends some support to the original claim. But it's not a hard ceiling.
What's more solid is the underlying constraint: there's an upper limit, the limit isn't huge, and the layered structure where intimate relationships consume disproportionate cognitive resources holds up well.
What modern environments do to this system
The relevant question isn't whether the number is exactly 150. It's what happens when humans are placed in social environments much larger than the system was designed for.
Social media is the obvious case. The average Facebook user has somewhere between 200 and 350 friends. Active Twitter users may have thousands of followers. LinkedIn networks can run into tens of thousands. None of these come close to fitting within the cognitive system designed to track social relationships.
What appears to happen is something like overflow. The brain processes the network as if it's a normal social environment, but it can't actually maintain real relationships with that many people, so the relationships flatten. Studies on social media use and loneliness have found a counterintuitive pattern: more friends often correlates with more loneliness. People with thousands of online connections often report feeling less close to others than people with smaller, deeper networks.
The mechanism is roughly that the cognitive resources that would have gone into maintaining a small number of real relationships get distributed thinly across a large number of weak ones. The strong-tie benefits — the actual physiological and emotional rewards of close relationships — require concentrated attention that gets diluted in large networks.
Workplaces have their own version of this problem. Companies that grow past a few hundred employees tend to fragment into smaller clusters where real working relationships exist. Beyond a certain size, the same person can be in the same building with people they functionally don't know. The Dunbar limit shows up structurally even when no one is talking about it.
The cognitive cost of the wider network
There are a few specific costs that come from operating with a social network larger than the system can handle.
Decision fatigue. The brain is trying to track people it can't actually keep up with. Every additional weak connection adds a small ongoing cognitive load — checking notifications, parsing messages, formulating responses, monitoring social position. The aggregate weight is real even when each individual demand seems small.
Eroded depth. As attention spreads, the inner layers of the network often get less attention than they would in a smaller social environment. The intimate five and the close fifteen need active maintenance. When attention is being pulled by the outer layers — the 1,500 — the inner ones get less of you.
Distorted social comparison. Networks larger than the brain was designed for tend to produce comparison effects that are physiologically taxing. Tracking the wins of 500 people produces a different stress profile than tracking the wins of fifty. Most of the research on social media and well-being points to comparison as one of the main mechanisms.
Loss of granularity. Real relationships involve detail — knowing what someone is going through, what they care about, what's hard for them right now. Networks beyond a certain size start to lose this resolution. People become symbols of themselves rather than the actual people they are.
What seems to actually help
A few practical implications fall out of this research. They're not novel, but they're more grounded than the typical "delete social media" framing.
**Distinguish your layers explicitly.** Most people have a vague sense of who's close to them and who isn't, but rarely sort it consciously. Worth doing. Who are your five? Your fifteen? The clarity changes how you allocate attention.
**Protect the inner layers actively.** The five and the fifteen need regular contact, real conversation, and shared experience. They're the layer where the actual physiological benefits of relationships happen. They also atrophy faster than people realize. Friendships that aren't tended don't survive — closeness research suggests the strong-tie benefits drop off within months of contact lapsing.
**Be honest about the outer layers.** Most of your "network" isn't really doing relationship work. Acknowledging that is freeing. You don't have to maintain real connection with the 500. You can keep them as the loose ties they actually are.
**Reduce the surface area where it doesn't serve you.** This isn't about deleting social media; it's about asking whether the size of your network is producing the benefits you think it is. Most people who reduce the size of their active social network — closer friends, fewer of them — report higher satisfaction within months.
**Default to in-person and voice for the inner layers.** Strong-tie benefits depend partly on the social engagement system, which requires real-time, embodied contact. Text-only relationships don't deliver the same physiological benefits as voice or in-person. The inner layers should mostly happen in person where possible.
**Notice when you're operating beyond your cognitive capacity.** The signs are real: low-grade ambient anxiety about responses, difficulty remembering details about people, blurring of who's close to you, feeling lonely despite high contact volume. Those are signals that the system is overloaded.
The community framing
There's a related thread of research worth touching, on what kinds of communities humans actually thrive in.
Studies of religious communities, neighborhood associations, intentional communities, and small towns find that the well-being benefits of community plateau or peak at sizes roughly consistent with Dunbar's framework. Communities of 100-200 people produce most of the benefit. Communities of 1,000+ usually fragment into smaller working units. Communities of 10,000+ stop functioning as community in the meaningful sense and start functioning as institutions.
This suggests that "find your community" advice runs into a structural problem when the community is the wrong size. Online communities of millions of people don't deliver the benefits that the framing promises. The benefits come from the smaller-scale, deeper connections within those communities, not from the size itself.
The implication is that the work of building real community is mostly the work of finding or building sub-groups within whatever larger structure you're in. The church of 5,000 doesn't deliver community; the small group within it might. The company of 1,000 doesn't; the team of seven might. The online community of millions doesn't; the group chat of fifteen might.
Where this leaves us
Dunbar's research, taken with appropriate caveats, points at something useful. Human brains weren't designed for the social environments most of us now operate in. The mismatch isn't fatal, but it has real costs. The costs accumulate quietly and show up as things people don't always connect to network size: ambient stress, low-grade loneliness, eroded depth in close relationships, comparison-driven dissatisfaction.
The work isn't to retreat from the modern world. It's to be deliberate about which layers of your social system you're feeding and which ones you're letting run on autopilot. Most people, given the choice, want a real five and a real fifteen. The cognitive system is designed for that. The question is whether you've structured your life to actually have it.
Five is not lonely. Fifty isn't ambitious. The numbers work the way they work for reasons that go back hundreds of thousands of years.
---
*Pairs well with: "Reliable Friends Are Brain Health" and "Why You Have to Put Your Own Mask On First."*