Two kinds of brain?

Paris, 1879

In 1879 the French physician Gustave Le Bon published a long paper on the weight of the brain in the Revue d’Anthropologie, one of the leading journals of its day. Working from autopsy data gathered in Paris hospitals, he argued that even among the most intelligent peoples there were many women whose brains were, in the words of Stephen Jay Gould’s translation, “closer in size to those of gorillas” than to the best male brains. Exceptional women, he added, were about as common as a gorilla with two heads, and could be ignored.

The scale readings behind this were real. Le Bon’s figures came from the anthropologist Paul Broca, who had weighed brains himself in four Paris hospitals: 292 men averaged 1,325 grams and 140 women 1,144 grams. What the weights meant was another matter. A century later Gould pointed out that Broca’s women were older than his men, and that his average man was almost half a foot taller. Allowing for such things shrank the gap from 181 grams to about 113 by Gould’s reckoning, and Broca’s own follower Léonce Manouvrier, correcting for body mass, found women slightly ahead. The numbers were measured. The story built on them was an argument, and the argument came first.

Others tested it. In 1901 the mathematician Alice Lee, working in Karl Pearson’s laboratory at University College London, estimated the skull capacities of women students, of male staff, and of 35 distinguished male anatomists, and reported that skull size did not track intellectual standing. In 1903 Helen Thompson (later Helen Thompson Woolley) at the University of Chicago put 25 men and 25 women through 15 to 20 hours each of tests of movement, the senses, memory, and reasoning. Men did better on most tests of movement; women did better on several of the senses, memory, and association. She argued that most of the differences owed more to upbringing and social circumstances than to nature. In 1914 Leta Hollingworth tested 23 women over three months, on tapping, steadiness, color naming, and typing, to see whether the monthly cycle made them mentally less able, as many physicians said. She found no sign of it.

This history is the method of the whole course. A measurement can be correct while the story told about it is not, and the story is where prejudice enters. Our job here is to look at what is measured today, how large it is, and what it can and cannot be made to say.

A bigger brain, a bigger body

The best single source on adult brain size is the UK Biobank, a British research project that has scanned the brains of tens of thousands of volunteers. In an analysis of 5,216 of them (2,750 women and 2,466 men, aged 44 to 77), the average brain volume was about 1,116 cubic centimeters for women and 1,234 for men: men’s brains were about 11 percent larger. In the units psychologists use, the gap is d = 1.41, which means that about 92 percent of the men had a larger brain than the average woman. That is a big difference, though a little smaller than the difference in adult height. Boys are born with slightly larger heads: in one scan study of 293 newborns, the intracranial volume of boys was about 6 percent larger.

Even a gap this size leaves the two groups overlapping: their curves share nearly half of their area. And how much of it is about the brain, and how much about the body? When the researchers allowed for height, the gap in total volume fell from 1.41 to 0.42. A smaller difference remains, but most of the raw gap went with body size.

There is also a surprise for anyone who has heard that women’s brains have the larger capacity. On average they do not: men’s brains are bigger, and women’s cerebral cortex is slightly thicker (d = 0.22, women higher, across most of the surface), while men’s has more surface area. Nor does size settle the question that the nineteenth century cared about. Across people, brain size and intelligence-test scores are correlated at about 0.24 in the largest review so far (88 studies, more than 8,000 people), which means size accounts for roughly 6 percent of the differences in scores, and the review found signs of publication bias that probably inflate even that. Men’s brains are about 11 percent larger, yet average scores on general reasoning tests are close. In a British study of 320,000 children aged 11 and 12, girls were a little ahead in verbal reasoning (about 0.15 standard deviations) and boys a little ahead in quantitative reasoning (about 0.05), with boys more spread out. A bigger brain is not a smarter one.

Which parts are bigger?

Popular accounts go further and name particular regions. Women’s hippocampus, a structure important for memory, is said to be larger, which would explain why wives remember everything. Men’s amygdala, a small structure involved in fear and emotion, is said to be larger, which would explain anger and aggression. Each claim sounds like a result, and each depends on a decision most readers never see: how to take head size into account.

The problem is that a bigger head has bigger parts, so a fair comparison has to allow for overall size, and the way it is done changes the answer. Three common methods are to compare the raw volumes, to divide each region by total brain size, and to use a statistical adjustment. For a region that grows more slowly than the whole brain, which is true of many deep structures, dividing by total size makes the owner of the bigger brain look as if their region were smaller. Try it below.

The hippocampus shows how this plays out. In the UK Biobank sample the raw difference is large (d about 0.55, men larger); after allowing for total brain volume it is below 0.03 in either direction. A meta-analysis of 68 samples found raw volumes about half a standard deviation larger in men and, after adjustment, no reliable difference: between 0.01 in men’s favor and 0.21 in women’s, depending on the method, and not statistically significant. A tiny difference in women’s favor shows up in some large, carefully scaled data, at the scale of 0.06 to 0.1 standard deviations; others find nothing. As for memory, large reviews do find an edge for women, but it depends on the task: across 617 studies and 1.2 million participants it was 0.19 overall, 0.28 for verbal material, and it reversed for routes and abstract pictures, where men did better. We found nothing that links that pattern to the size of the hippocampus.

The amygdala is harder, because careful adjustment does leave a small difference in men’s favor in large samples (about 0.18 in the UK Biobank), while one meta-analysis of smaller studies found the corrected difference was not significant. What is missing is the link to behavior. Aggression has many causes, from hormones to upbringing, and we found no study that ties the size of the amygdala to the difference between men and women in aggression. Likewise the anterior cingulate and the insula, which the popular accounts claim as female strengths, change direction depending on how the data are scaled. Overall, in the UK Biobank cortex data, of 68 regions, 44 showed no difference once total brain size was allowed for; 11 stayed larger in men and 13 became larger in women, by small amounts (about 0.14 and 0.15 standard deviations).

Before birth

The difference between a male and a female body begins early. A male fetus’s testes begin to make testosterone around weeks 8 to 11 of pregnancy, the level peaks around weeks 11 to 14, and it falls after about week 20, while a female fetus has far less. This is often told as a story in which the brain starts out female and is switched to male at six to eight weeks by a flood of hormone. That is a simplification. Six to eight weeks is about when the testes form and begin to work, and no human study has found a switch at that point. In 1959 the researchers Charles Phoenix, Robert Goy, Arnold Gerall, and William Young gave testosterone to pregnant guinea pigs and showed that the female offspring grew up behaving more like males, which suggested that early hormones “organize” the brain once and for all. The idea has since been refined. In rodents, much of the masculinizing is done by estradiol made from testosterone inside the brain, the chromosomes have effects of their own, and being female is an active process and not a blank default (McCarthy and Arnold, 2011).

In people, the best evidence comes from children whose bodies were exposed to extra male hormones before birth. Congenital adrenal hyperplasia (CAH) is a condition in which the adrenal glands produce excess androgens. Girls born with CAH, studied from the early 1990s on, play more with cars and construction toys and less with dolls than unaffected girls in their families. In a 2023 study of 43 girls with CAH and 41 comparison girls, the difference in choosing a car over a doll was large (d = 1.01), and the CAH girls fell between the comparison girls and the boys. Boys with CAH, who get the same medical treatment, do not differ from other boys, which makes it less likely that treatment is the explanation.

How large are sex differences in toy choice in general? A meta-analysis of 75 studies by Davis and Hines (2020) put each sex’s preference for its own typical toys at d of 1.6 or more, which makes it one of the largest differences in psychology, not far below the gap in height. It shows up in children as young as a year or two (in one study of 120 infants, girls looked longer at dolls and boys at cars) and grows with age, which is what you would expect if biology and learning both contribute. In monkeys, the picture is mixed. A 2008 study of rhesus monkeys found that males preferred wheeled toys to plush ones (d = 1.14) while females showed no preference; a 2023 study that tested 14 monkeys one at a time with 16 toys found no sex differences by type of toy at all. Prenatal hormones are one real input. They are not a switch, and they do not produce “a male brain” and “a female brain” as two types.

Hormones surge again after birth, in a phase called mini-puberty. In baby boys testosterone peaks at one to three months and returns to low levels by about six to nine months; in baby girls the ovaries show follicle cycles for about two years, with high and fluctuating levels of estradiol and FSH. A boy’s measure of early growth in this window (penile growth in the first three months) has been found to predict how male-typical his play is at three to four years. What the surges do to the brain is still unclear.

A famous 2000 experiment on 102 newborns, at an average of 36 hours old, found that girls looked longer at a live face and boys at a mobile. It has been criticized because the live face varied from baby to baby and the experimenter’s expectations might have played a part, and for many years it was not replicated. A 2026 attempt with 130 newborns and video stimuli found a difference in the same direction (d about 0.4), and the senior author of both was the same researcher. For now this is an interesting result and not a settled one.

Wired differently?

The corpus callosum is a thick band of about 200 million nerve fibers that connects the two halves of the brain. In 1982 a short report in Science described its rear part, the splenium, as larger and more bulb-shaped in women, in 14 brains obtained after death (nine men’s and five women’s, by one later account). It was a small study, but it became a story. By 20 January 1992 Time had put the question on its cover, “Why Are Men and Women Different?”, and told readers that parts of the callosum were up to 23 percent wider in women, which it linked to the idea of female intuition. Five years later Bishop and Wahlsten pooled 49 studies and found no reliable sex difference in the size or shape of the splenium once brain size was accounted for. A 2013 study of 316 scans found the callosum a little larger in women relative to brain size after adjusting for it, while a 2014 study that matched men and women on brain size found hardly any difference. As for the claim that a thicker callosum makes women better at doing several things at once, no study links the two, and one comparison of 96 people on ten multitasking measures found no significant difference between men and women.

In December 2013 the headlines announced a more detailed version. A study in PNAS led by Madhura Ingalhalikar scanned 949 young people aged 8 to 22 and mapped their brain connections. Within each hemisphere men’s connections were stronger on average, and between the hemispheres women’s were, a difference that grew through adolescence. A British newspaper summed it up as “Male and female brains wired differently, scans reveal,” and the university’s news release added that men are better at single tasks like cycling and women at multitasking. That gloss was the press office’s own and not a finding in the paper. Critics wrote that the paper’s figures showed only the connections that differed and not how many did not, and that it reported no effect sizes. Another group found in 138 brains that the balance of between-hemisphere and within-hemisphere connections shifts with brain size regardless of sex, which suggests that size explains much of the pattern.

Language has its own folklore: that women’s language is spread across both hemispheres and men’s sits in the left, which is why men are said to lose speech more often after a left-brain stroke. A 1995 imaging study of 38 people (19 men and 19 women) found that on a task of judging how words sound, men’s activity was concentrated in the left frontal region while women’s was more spread out. When 14 studies with 377 men and 442 women were later pooled, the difference in language lateralization was d = 0.21 with a range of −0.05 to 0.48, not statistically significant, and a 100-person study found no difference. On strokes, we could not trace the claim that men are three times as likely to lose language. A review of 25 studies of 48,362 stroke patients found aphasia slightly more common in women (a ratio of 1.14), a difference that age fully explained.

A mosaic

Suppose you measure many parts of the brain and each differs a little on average between the sexes. Does that mean there are two kinds of brain? In 2015 the Israeli psychologist Daphna Joel and her colleagues tested the idea on more than 1,400 brain scans from four data sets. They marked, for each feature, the third of men and the third of women most at each end: a “male end” and a “female end,” with a middle zone. They then asked how many people had features at only the male end or only the female end. In the first sample about 6 percent of brains were consistent that way, and about 35 percent were “substantially variable,” with both male-end and female-end features. Most people were a mosaic. A matching analysis of personality and behavior in more than 5,500 people gave the same picture.

The paper drew sharp replies, and they are worth hearing. Marco Del Giudice and colleagues showed that a statistical method could predict a person’s sex from Joel’s own data 69 to 77 percent of the time. They also applied Joel’s method to the faces of three monkey species, where only 1.1 to 5.1 percent of animals counted as typical of their species, to argue that a method that finds almost no consistent types even where differences are enormous cannot show there are none. A separate letter from Chekroud and colleagues reached 93 percent accuracy on 1,566 brain scans, though about 70 percent when head size was removed statistically. Joel replied that sex clearly affects the brain, that high accuracy does not show two distinct types, and that accuracy can drop sharply when a model trained on one data set is tested on another.

Both sides are right about something, because they are answering different questions. Try the toy model below, in which every feature differs a little between the sexes.

What the real data show depends on how head size is removed. Studies of the best algorithms run from about 60 to 70 percent when size is removed statistically up to 85 to 93 percent in samples matched for size, and we found none that was perfect. A widely repeated story says that in 2014 fourteen neuroscientists were shown 30 brains and identified the sex of every one, but we could not find any study that matches it.