How did people live together before writing? In groups, almost certainly, but nobody from that time left us a head count. What researchers have instead is indirect evidence. Put together, it shows that recent foragers live in groups nested inside bigger groups, and that many of the people they live with aren’t close relatives. Whether people lived that way 50,000 years ago is Plausible but can’t be checked directly. The famous claim that our brains cap our social world at 150 people is Disputed. And what belonging felt like to people before writing is Unknown.
Why can’t we just dig up a group?
Bones and stone tools show where people were and sometimes how they lived, but rarely who counted as “us.” This project reads prehistoric evidence like a ladder: a skeleton can show an injury and how long someone survived it, but feelings, motives and relationships sit on the top rung, where bone alone can’t reach (see our page on care for the injured). Loyalty doesn’t fossilize.
So researchers borrow three kinds of evidence. Each has a weak spot:
- Other primates. Comparing species can hint at what shaped human groups. But we aren’t chimpanzees, and they aren’t our ancestors.
- Recent foraging societies. Living peoples with their own long histories, not time capsules. Researchers who compile this work warn that cultures change, so present-day records can’t simply be projected onto the deep past (Ember et al., 2025, p. 4).
- Archaeology. House sizes and village layouts can show who had more, but mostly for later, settled communities.
Is there really a 150-person limit?
In papers from 1992 and 1993, the anthropologist Robin Dunbar, then at University College London, compared 36 primate genera (groups of closely related species). Species with a bigger neocortex (the brain’s wrinkly outer layer) relative to the rest of the brain tended to live in bigger groups. He extended that line to humans and got a predicted group size of 147.8. He noted that this stretched the line past its data, with a wide 95% range of about 100 to 231. He then pointed to groups of roughly 100 to 200 people in recent foraging and small-scale farming societies, Hutterite farm communities and army companies (Dunbar, 1993, §3.1). That’s “Dunbar’s number.”
Analogy: imagine measuring how far kids aged 3 to 10 can throw a ball, then extending the line to guess how far an adult can throw. You might land close. But the further you stretch the line, the shakier the guess.
In 2021, three Stockholm University researchers reran the comparison with larger datasets and newer statistics. Depending on the method, their estimates ran from about 16 to 109 people, with uncertainty ranges from about 2 to 520, so no single number can be pulled out this way. They also noted that food and predators shape primate group size, not brains alone (Lindenfors et al., 2021). Dunbar replied that they used the wrong kind of regression; his method gives about 158 (Dunbar, 2021). The Stockholm authors answered that his preferred method, run on their data, gives about 290, or 404 for monkeys and apes alone (Lind & Lindenfors, 2021). Disputed.
Groups inside groups
Here the ground is firmer. A 2007 study pooled 1,189 group-size estimates from 339 recent hunter-gatherer societies. Families averaged about 4.5 people. Then came groups of about 16, then about 54, then gatherings of about 165, and regional populations of about 840. Each level was close to four times the one below it (Hamilton et al., 2007). A separate 2005 study, which Dunbar co-authored, found preferred group sizes across many kinds of human groups in steps of about three: roughly 3–5, 9–15, 30–45, and so on (Zhou et al., 2005).
Analogy: think of nesting boxes. A household fits inside a camp, a camp inside a cluster of camps that meet now and then, and all of those inside a wider population.
Layering in recent human societies is Strongly supported: two teams using different data both found layers with a roughly steady ratio. Why the layers exist is Disputed. Dunbar points to brain limits; Hamilton’s team also lists foraging, population swings and group splitting. Projecting the pattern back 50,000 years is Plausible.
Were early bands made of relatives?
A 2011 comparison of 32 foraging societies challenged models that treat camps mainly as groups of close relatives (Hill et al., 2011, pp. 1286–1287).
Newer data complicate both pictures. A mobile camp usually holds about four families, roughly 16 to 20 people, and relatedness inside hunter-gatherer groups is generally low (Ember et al., 2025, p. 12). Among foragers, couples living with either spouse’s relatives is more common in very small communities (p. 15). A study of 205 Hadza adults in 17 camps in Tanzania mapped whom people wanted to live with next and whom they’d give honey to. People tended to be tied to others who gave similar amounts in a sharing game, and social ties explained that pattern about as well as family relatedness did (Apicella et al., 2012). For recent foragers, mixed groups of kin and non-kin are Strongly supported. For the deep past they’re Plausible.
How does a group keep things fair?
Keeping leaders humble. Our closest ape relatives live in clear pecking orders. Yet many small-scale human societies run with weak leaders. In 1993 the anthropologist Christopher Boehm searched ethnographies of about 200 politically independent societies. He found 48 with solid evidence that ordinary members deliberately kept would-be bosses in check. They used public opinion, criticism and ridicule, disobedience, and simply leaving. In 11 of the 48, the record includes killing an overbearing person. Boehm called this a “reverse dominance hierarchy” (Boehm, 1993, pp. 227–231). He called his own survey preliminary.
Punishing free riders. In a 2002 experiment, 240 university students in Zurich played a money game. Players chose how much to put into a shared pot, and groups were reshuffled every round. Cooperation grew when players could pay to punish free riders and fell apart when they couldn’t (Fehr & Gächter, 2002). A 2006 study of 1,762 adults in 15 populations on five continents found some willingness to pay to punish unfair play in every population tested, but the amount varied a lot (Henrich et al., 2006). Francesco Guala of the University of Milan argued in 2012 that there was no evidence that the small egalitarian societies anthropologists study enforce cooperation through costly punishment, and real communities tend to build institutions that make punishment cheaper (Guala, 2012). The lab findings are documented. Whether costly punishment is what held early groups together is Disputed.
Talking about each other. Dunbar proposed that language let people learn how others behaved faster than watching alone could (Dunbar, 1993). Plausible. Our language page covers this.
Why did human cooperation get so big?
The unusual human trait is how much we cooperate with people who aren’t relatives, often in large groups and with strangers (Fehr & Gächter, 2002, p. 137). One big explanation is cultural group selection. It says groups whose shared rules supported cooperation did better than groups without them, so those rules spread. Richerson and 13 co-authors laid out evidence for it in 2016 (Richerson et al., 2016). In the same journal issue, critics answered. Some called the evidence thin after 30 years (Krasnow & Delton, 2016). Some argued that minds built for small-scale life can scale up without group selection (Tooby & Cosmides, 2016). Disputed.
In Mutual Aid (1902), Peter Kropotkin argued that mutual aid is “as much a law of animal life as mutual struggle,” and probably the more important factor in evolution (Kropotkin, 1902, ch. I). Our page on violence and cooperation tells that story.
When did big differences in rank appear?
Recent foragers generally have little wealth difference and no specialized political officials, though some foraging societies are exceptions. Archaeologists also infer, from rich burials and prestige items, that some hunter-gatherers developed inequality before farming (Ember et al., 2025, pp. 5–6). Plausible. A 2017 study used house sizes as a measure of wealth at 63 archaeological sites or site groups, the oldest about 11,000 years old. Wealth gaps generally grew as people domesticated plants and animals and as societies got bigger. After the Neolithic, gaps grew larger in Eurasia than in North America and Mesoamerica (Kohler et al., 2017).
How we know
- What evidence exists: primate comparisons; field records of hundreds of recent societies; lab and field experiments; house sizes at dated sites.
- Whose voices are missing: everyone who lived before writing. Most of what’s known about recent foragers comes from anthropologists who lived with them (Ember et al., 2025, p. 4), not from the communities’ own accounts. Boehm wrote that women’s part in group decisions had surely been under-reported, especially in older ethnographies (p. 229). This page did not consult the Hadza’s own published accounts.
- How power shaped the record: In 1993 Dunbar himself warned that most surviving hunter-gatherer societies had been seriously disrupted by contact with colonial cultures (Dunbar, 1993, §3.1). Cross-cultural studies link flexible living arrangements partly to recent drastic population loss, so colonial-era epidemics may have reshaped some groups before they were measured (Ember et al., 2025, p. 15). Plausible. And as of 2006, most cooperation experiments had used students in industrialized societies (Henrich et al., 2006).
Then vs. now
- Then (1877): Lewis Henry Morgan sorted human societies into seven stages from “savagery” through “barbarism” to “civilization” (Morgan, 1877, ch. I). Even Kropotkin, arguing for cooperation, titled chapters “Mutual Aid among Savages” and “Mutual Aid among the Barbarians.” Now: Cross-cultural researchers ask what conditions predict differences between societies (Ember et al., 2025, p. 4), a different question from where a people sits on a ladder.
- Earlier models: camps were often treated as groups of related men (Hill et al., 2011, p. 1286). Now: recent foragers live in mixed groups; one estimate is that parents, siblings and children make up less than 10% of a camp (Apicella et al., 2012).
- Today’s evidence on belonging: in 2025, a World Health Organization commission reported that about 1 in 6 people worldwide is affected by loneliness. It discussed loneliness as a present-day research topic (WHO, 2025). That describes people now. Whether people before writing felt lonely or left out, and what their groups did about it, is Unknown.
Myths to retire
- “Your brain can only handle 150 people.” Disputed. Even Dunbar describes layers, from about 5 intimate friends to about 1,500 names you know (Dunbar, 2021).
- “Early humans lived in bands of blood relatives.” Recent foragers don’t. For the deep past, either picture is only Plausible.
- “Societies climb a ladder to civilization.” Morgan’s hierarchy is a historical argument, not a result established by the comparisons described here.
- “Humans are the most social animal on Earth.” We found no study that ranks species this way. What stands out is cooperation with non-relatives at large scale.
What we still don’t know
- Group sizes 100,000 or 50,000 years ago: Unknown.
- Why groups come in layers: Disputed.
- Whether competition between groups spread cooperative rules: Disputed.
- How people before writing experienced belonging, exclusion or loneliness: Unknown.