Probably, in a flexible way, but nobody saw it happen. [Plausible] Anthropologists describe marriage as found in every human society they know of, and in most societies most marriages, at any one time, join one partner to one partner (Schacht & Kramer, 2019). When our ancestors started pairing up, and why, is still argued over, because bones and genes can show who had children together but not who loved whom. And pair bonding isn’t a human invention. About 9 in 10 bird species live in pairs, and so do nearly 3 in 10 primate species.
How to read the labels on this page: [Documented] means the evidence shows it directly. [Strongly supported] means strong sources agree. [Plausible] means the evidence allows it but doesn’t prove it. [Disputed] means qualified experts disagree. [Unknown] means the evidence can’t answer the question.
What counts as a pair bond?
Social monogamy just means living as a pair, or “pair-living.” That’s different from sexual faithfulness, since pair-living animals sometimes have offspring with outsiders. A pair bond is the lasting preference two individuals show for each other. Lab researchers measure it in prairie voles with a “partner preference test”: a bonded vole chooses to stay close to its partner rather than a stranger (Walum & Young, 2018).
Marriage adds something no vole has: a community that defines the relationship, sets its rules and enforces them. Joseph Henrich and colleagues argue that this community role is what separates marriage from “pure pair-bonding” (Henrich et al., 2012).
Analogy: a pair bond is like two friends who always save each other a seat. Marriage is that friendship plus rules the whole school has agreed to.
Are humans the only animals that pair up?
Not even close. About 90% of bird species are considered socially monogamous, and in groups like geese, swans and eagles, lifelong pairs are common (Lukas & Clutton-Brock, 2013; Walum & Young, 2018). Mammals are different. A 2013 survey classified 229 of 2,545 mammal species, about 9%, as socially monogamous. Primates stand out, at 106 of 361 species, or 29% (Lukas & Clutton-Brock, 2013, p. 526). [Documented for these counts in this survey] Pair-living has evolved separately in every major branch of the primate family tree (Opie et al., 2013), from owl monkeys in the Americas to gibbons.
Why it evolves is a live argument. In 2013, two teams disagreed. Dieter Lukas and Tim Clutton-Brock concluded that mammal pairs formed where females lived spread out and a male couldn’t keep several to himself. Christopher Opie and colleagues, studying 230 primate species, concluded that the most compelling explanation in primates was protecting infants from males who kill young that aren’t their own. The teams traded published replies in 2014. [Disputed] Both agreed that fathers helping with babies seems to have come after pair-living, not before. [Plausible: two studies agree]
What about our closest relatives?
Chimpanzees and bonobos live in large communities of many males and many females, and both species mate with multiple partners. In bonobos, the strongest adult relationships researchers report are between mothers and grown sons, and among unrelated females, not between mates. Gorillas usually live in stable groups of several adult females with one, or occasionally more, adult males (Grueter, Chapais & Zinner, 2012). Gibbons are the apes that live in pairs, and their males and females are nearly the same size (Schacht & Kramer, 2019). [Strongly supported]
None of these apes is our ancestor; they’re cousins. The last shared ancestor of humans, chimpanzees and bonobos has been tentatively dated to 6–8 million years ago, with no clear fossil candidate yet (Grueter et al., 2012). [Plausible]
When did our ancestors start pairing up?
Nobody knows; the clues point different ways. [Disputed]
- Body size. Size differences between the sexes are expected to be largest where one male competes to control a group of females, as in gorillas. Human males today average about 15% heavier than females, chimpanzee males about 30%, and gorilla males nearly twice as heavy. Researchers agree the gap was bigger in our past but not on when it shrank. Some think Australopithecus males and females were fairly similar in size around 4 million years ago; others think they differed a lot (Schacht & Kramer, 2019).
- Testis size. Relative to body size, human testes are much smaller than chimpanzees’. That points away from a past where females routinely mated with many males. It can’t separate pair-living from one-male groups, though: human values are close to gorillas’ (same source).
- Which ape to compare? Researchers still debate whether early hominins lived in chimpanzee-like mixed groups, gorilla-like one-male groups, or clusters of one-male units like hamadryas baboons (same source, Box 2). Lukas and Clutton-Brock think the common ancestor was probably polygynous, meaning one male mated with several females (p. 529).
Schacht and Kramer’s review concludes that polygyny was likely present in the human past, but “the weight of evidence seems to support social monogamy.” That’s one review’s judgment. [Plausible]
What do human societies show?
In one widely used sample of 186 mostly pre-industrial societies, nearly 85% allowed a man more than one wife. Yet inside those societies, most marriages were one-to-one. In many societies, remarriage after a spouse’s death or divorce is common, so many people have several partners over a lifetime, a pattern called serial monogamy. Some societies also allow a woman more than one husband (Schacht & Kramer, 2019). [Strongly supported for the overall pattern]
Marriage also tends to tie families together. A 2011 paper compared recorded marriage practices using an earlier ethnographic dataset. Using a DNA-based tree with 15 matches between genetic and cultural datasets, the authors estimated that arranged marriage goes back at least 50,000 years (Walker et al., 2011). [Plausible: a model estimate] A 2012 review co-written by Bernard Chapais argues that once people recognized their in-laws, marriages could link whole groups into alliances (Grueter et al., 2012). [Plausible]
These records describe recent peoples living in the modern world, not frozen snapshots of the Stone Age. The authors treat them as clues, not proof.
Can ancient DNA see a relationship?
Partly. Hazleton North is a stone tomb in Gloucestershire, England, built between about 3700 and 3600 BCE. Researchers analyzed DNA from 35 people buried there. Twenty-seven belonged to one five-generation family, and all 15 generation-to-generation links used to trace the tomb’s family line ran through men. One man had children with four women. At the same time, or one after another? The team says it can’t tell, and it can’t rule out children from unions the community didn’t sanction. Eight people in the tomb weren’t close biological relatives of the family, which the authors say raises the possibility that kinship there included bonds not based on biology (Fowler et al., 2022). [Documented for the genetic relationships; Unknown for what they meant] This is one farming family in one place, not “prehistoric people” in general.
What’s happening in the brain?
Most of what scientists know about the biology of pair bonds comes from prairie voles, rodents in which males and females form long-lasting pairs and both care for the young. In voles, the hormones oxytocin and vasopressin, working with dopamine, help link signals from a partner with the brain’s reward system (Walum & Young, 2018). [Strongly supported in voles] Whether people work the same way is much less clear; the authors say human oxytocin studies “should be interpreted with some caution.” [Plausible for humans]
A longitudinal study also found associations between fatherhood and hormone levels. A study of 624 young men in Cebu, in the Philippines, between 2005 and 2009 found that men who became partnered fathers had large drops in testosterone. Fathers reporting three or more hours of childcare a day had lower levels than uninvolved fathers (Gettler et al., 2011). [Documented for this sample] The authors describe the link as running both ways: men with higher testosterone at the start were more likely to become partnered fathers, and their levels then fell.
How did earlier thinkers explain it?
In 1877, Lewis Henry Morgan’s Ancient Society promised to trace “human progress from savagery, through barbarism to civilization.” He described five family forms, rising from group marriage to “the monogamian family,” which he called “pre-eminently the family of civilized society.” [Documented for what he wrote]
In 1891, Edward Westermarck, a scholar from Finland, pushed back in The History of Human Marriage (we quote its 1901 third edition). He rejected the “promiscuity” stage that Morgan and others proposed and called marriage “an inheritance from some ape-like progenitor.” He also believed marriage occurred “as a rule” among monkeys and apes. Field studies since then show that’s wrong for chimpanzees and bonobos. [Documented for what he wrote]
Psychiatrist and psychoanalyst John Bowlby (1907–1990) set out attachment theory in papers from 1958 and in three volumes of Attachment and Loss (1969, 1973, 1980) (Bretherton, 1992; Wellcome catalogue). His focus was a young child’s tie to a mother figure, though he held that attachment keeps working in adult life (Bretherton, 1992). Research on attachment between adults began in the early 1970s with studies of grief and marital separation. In 1987, Cindy Hazan and Phillip Shaver published “Romantic love conceptualized as an attachment process,” and their work carried infant attachment patterns over to adult romance (Bretherton, 1992). Our page on the evolution of attachment goes deeper.
How we know
What evidence exists: animal surveys, ape field studies, fossil body sizes, descriptions of marriage in hundreds of recent societies, ancient DNA from a few tombs, vole experiments, and hormone studies in living people.
Whose voices are missing: no one from before writing left words about partners, love, jealousy or loss. At Hazleton North we know who had children with whom, not whether anyone chose, wanted or grieved those relationships. The comparative sources code marriage practices; this page does not include first-person accounts from each coded community. The sources used here say nothing about same-sex partnerships.
How power shaped the record: Morgan’s ladder ranked living peoples’ families below his own. And the worldwide spread of monogamous marriage as the norm happened “only in recent centuries,” according to Henrich and colleagues (Henrich et al., 2012).
Then vs. now
- Then (1877): Morgan argued that families climbed from group marriage up to monogamy. Now: one review finds that body and testis size seem to rule out a highly promiscuous past (Schacht & Kramer, 2019). [Plausible]
- Then (1891–1901): Westermarck thought apes generally “married.” Now: chimpanzees and bonobos mate with multiple partners in large communities; gibbons are the apes that pair up. [Strongly supported]
What we still don’t know
- When pair bonds became common in our lineage. [Unknown]
- Why pair-living evolved in primates: spread-out females, protection from infanticide, or a mix. [Disputed]
- Whether early hominins lived in chimpanzee-like, gorilla-like or other kinds of groups. [Disputed]
- What partnership meant or felt like to anyone before writing. [Unknown]