Why Are Human Brains So Big?

Your brain is about 2% of your body weight and consumes roughly 20% of your resting energy — some 300 to 500 calories a day, running constantly, whether you are solving equations or asleep. For a chimpanzee the figure is about 8%. For most mammals it is 3–5%.

In evolutionary terms that is an outrageous running cost. Natural selection does not fund expensive organs out of generosity. So what paid for ours?

The scale of the change

Two things in that table surprise people. First, brain size stayed flat for millions of years and then roughly tripled in the last two — this is a recent, rapid event. Second, Neanderthals had bigger brains than we do, on average. Absolute size is not a simple ranking of intelligence.

Theory 1: better food paid for it

The expensive tissue hypothesis, proposed by Leslie Aiello and Peter Wheeler in 1995, is the most influential account of the energetics. Brains and guts are both metabolically costly. Great apes eating bulky, low-quality plant food need a large gut with a big fermenting colon. If diet quality rises — more meat, more marrow, more energy-dense food — the gut can shrink, and the energy saved is available elsewhere.

The fossils fit. Homo erectus has a narrow waist and flat ribcage where australopiths had a flaring, gut-heavy trunk. Cut-marked animal bones appear in quantity in the same period. The correlation is real, though the strict trade-off has been challenged: some comparative studies across primates do not find the predicted inverse relationship between gut and brain size.

Theory 2: cooking

Richard Wrangham’s cooking hypothesis sharpens the same idea. Cooking gelatinises starch, denatures protein and physically softens food, raising the calories you actually absorb and slashing the time and muscular effort spent chewing. Chimpanzees spend around six hours a day chewing; humans spend under one.

The objection is chronology. Wrangham argues cooking began with Homo erectus around 1.8 million years ago, but the strong archaeological evidence for controlled fire is much later — Wonderwerk Cave at about 1 million years, and truly routine hearths only after 400,000. Either fire evidence is systematically under-preserved, or cooking arrived too late to have started the trend, though it may well have sustained it.

Theory 3: other people

Robin Dunbar’s social brain hypothesis shifts the question from fuel to function: what is the brain for? Across primates, the relative size of the neocortex correlates with typical social group size. Tracking who is allied with whom, who owes whom, who is deceiving whom, is computationally brutal — and it scales explosively with group size.

Extrapolating the primate relationship to human neocortex size yields a predicted group size of roughly 150 — “Dunbar’s number”. The specific figure is much argued over, and the correlation is disputed by researchers who find ecological variables predict brain size better. But the underlying point is hard to dismiss: for a highly social species, the most demanding part of the environment is the other members of your own group.

Theory 4: culture that accumulates

The cultural brain hypothesis is the modern synthesis of these ideas. Once a species can learn reliably from others, information accumulates across generations — better tools, better tracking, better plant knowledge. A bigger brain becomes worth its cost because there is more worth learning; and a bigger brain makes better learning possible. It is a feedback loop, sometimes called autocatalytic, and it explains a rapid acceleration far better than any single external pressure does.

Climate instability may have set the loop running. The African Pleistocene swung repeatedly between wet and arid, and flexible, learning-heavy generalists cope with unstable environments better than specialists do.

What actually changed in the genome

A handful of genetic changes are strong candidates:

• ARHGAP11B — a human-specific gene created by a partial duplication. Inserted into mouse and marmoset embryos, it expands the neocortex and can induce folding in a normally smooth brain.
• NOTCH2NL genes — human-specific duplications that delay neural progenitor cells from differentiating, letting them divide longer and produce more neurons.
• SRGAP2C — a human-specific duplicate that slows synapse maturation, extending the window of developmental plasticity.

The theme is timing. Human brains are not built from radically different parts; they are built by running the same developmental programme for longer.

The costs

Nothing this expensive is free.

• Childbirth. A wide birth canal helps deliver a big-headed infant; a narrow pelvis is better for efficient bipedal walking. The compromise makes human birth uniquely difficult and, historically, dangerous.
• Helpless infants. Human babies are born developmentally early — secondary altriciality — because a more mature brain simply would not fit through. A newborn’s brain consumes an estimated 60% or more of its total energy intake.
• A long, costly childhood. Human brain growth continues at a fetal-like rate for the first year and the brain is not structurally mature until the mid-twenties. That requires provisioning by adults for a decade or more — which in turn requires cooperative childrearing.

And it has been shrinking

An awkward footnote: average human brain size has decreased by roughly 10% since the Late Pleistocene, some 150 cc. Explanations range from overall body size reduction, to greater energetic efficiency and reorganisation, to the offloading of memory and computation onto external culture. A 2021 paper proposing a sharp drop coinciding with agriculture was rebutted the following year on the grounds that the dataset could not support it. The shrinkage is real; its cause is not settled.

Which points at the deeper caveat: size is a crude proxy. Suzana Herculano-Houzel’s neuron counts show the human cerebral cortex holds about 16 billion neurons, more than any other primate — more than would be expected even from an elephant’s much larger brain. What matters is the number, wiring and organisation of cells, not the volume of the box.