A Scientist Dissents on Sex and Cognition

Suppose we want to ask how left handers and right handers are different, and relate those differences to brain organization and cognition? The project is accepted without objection by behavioral scientists and the public; it is clearly a special case of studying individual behavioral differences that we realize may be important. Yet exploring what may underlie the far more striking differences between the sexes—the concept of cognitive patterns associated with one’s gender—is fiercely resisted. In most cases this resistance arises from a misguided ideology that says social equality requires (or even means) individuals are identical in all important respects. This form of radical egalitarian feminism insists that men and women are self-evidently the same; the corresponding article of faith for behavioral studies is that social and cultural influences produce all variation among individuals—case closed.

Some feminists simply reject the study of sex differences, implying the field should be suppressed. The notion that any field of inquiry should be banned is inimical to science, of course, and to ideals of inquiry that virtually define—and help guarantee—free societies. Less extreme attacks have labeled research on gender and cognition flawed (usually asserted with no supporting argument) or as manifesting a male-dominated approach to science.

The facts are otherwise. Over the last two decades scientists have learned a great deal about the basic biology of sexual differentiation. It has become increasingly evident that humans, like other mammals, are affected throughout their lives by prenatal and current exposure to sex hormones. Most scientists conducting these investigations on human subjects have, in fact, been women who are trying to understand the basic mechanisms explaining individual differences in and across sexes.

We should keep in mind the fundamental knowledge that the brain mediates all behavior. Therefore if any two people differ in behavior—or in intellect, personality, or any other characteristic—their brains must differ in some way.

The questions I pose here are: Do systematic, meaningful, reliable differences exist in the cognitive or problem-solving abilities of men and women? Can these differences be convincingly related to parallel hormonal or brain differences? To the first question, the answer is an unequivocal yes. On the question of biological mechanisms I can make a convincing case for the contribution of sex hormones but must offer a far more tentative explanation of how the brain may mediate human cognitive sex differences.

Hunters and Homemakers

Apart from their obvious physical differences, men and women have a long history of dividing labor. Modern humans evolved from hunter-gatherer societies in which men were primarily responsible for hunting and scavenging, often taking them far from home. They were also responsible for defending the group and manufacturing weapons. Their evolution could be expected to favor long-distance navigation and accurate targeting. Women contributed by gathering food near home, manufacturing clothing, preparing food and, above all, tending and nursing children. We might, therefore, expect their navigational skills to be tied to more local information and their motor skills to emphasize small-amplitude movements required for domestic chores and child care. Heightened perceptual skills might reflect the importance of detecting small changes in the environment or in a child’s physical state.

How can we test these evolution-based theories? We do not have detailed information on how early hominids lived. Nor can we create a prehistoric environment in the modern world. We can confirm this scenario somewhat, however, by looking at preliterate societies that still exist or that were studied before they came under the influence of modern technology. We also know from analyzing skulls that general characteristics of the brain 50,000 years ago seem to differ little from those of modern people, suggesting that brain mechanisms are similar.

Alternative theories that sex differences in cognitive pattern result only or primarily from how boys and girls are raised today are less credible as we learn about the role of prenatal biological influences, about the early ages at which sex differences first appear, and about cross-cultural similarities.

He Hits the Target, She Remembers the Place

In controlled experiments, men outperform women, on average, on several spatial tasks, particularly those that require imaginary manipulation or rotation of an object. Men also do better on simple spatial tasks such as perceiving the slant of a line. Women excel on a different kind of spatial task—remembering the relative locations of several objects presented in an array, but apparently not when they must recall absolute locations.

Superior performance by women in remembering the relative locations of objects may relate to better verbal memory: that is, memory for material processed or presented in words. Women are clearly better than men at recalling words—abstract or concrete— and recalling drawings of objects as long as the objects can be readily labeled. Not only do women have better verbal memory than men, they have slightly better verbal fluency as shown in tests of the ability to generate words that start with a specific letter or under any other constraint on letters the words contain. Women can label common colors (and perhaps simple forms) more rapidly than men, but the common impression that women generally are better at verbal tasks—are more verbal than men—is incorrect. Men and women on average do not differ in vocabulary or verbal reasoning.

Males have the advantage on tests of mathematical reasoning where a problem must be solved; women are better at outright calculations. This male edge in math reasoning increases with school grade and level of difficulty. Girls often get better school grades in math as they do in other subjects; but on math aptitude tests, where the same abilities are tested on material not specifically taught, boys do better. On extremely difficult advanced math problems, such as those posed in a North American test known as the Putnam Competition, women very infrequently score among the top 500.

Although men are better able than women to identify or match the slope of a line, women are faster in matching stimuli like pictures or letters, an ability called perceptual speed. Besides these obviously cognitive abilities, there are interesting differences in motor tasks that are not reducible to differences in physical characteristics such as strength or size. For example, men are much more accurate, on average, in hitting a target with a dart, ball, or firearm. This was once attributed to men’s physical structure; instead it seems more related to the brain’s spatiomotor capacity to locate a target accurately and direct movements toward it. In contrast, women perform better on fine motor tasks, probably not because their fingers are smaller, as some have suggested, but because women are more dextrous than men.

Are Differences Significant?

All ability differences I just described are significant statistically, although on most tests the scores of men and women overlap a good deal. But on tests with less overlap, the sex difference involved may have more practical significance. To assess group differences across tests, scientists use the standard deviation (SD)—a measure of how much scores vary within a group. If scores are widely dispersed, the SD is large; if they’re narrowly dispersed, SD is small. We can measure the overlap or effect size between groups (in this case men and women) by dividing the mean difference between them by the combined SD groups. If effect size is large, overlap between scores is smaller.

Intellectual profiles or patterns of male and female cognitive abilities, and how those differ, seem strongly influenced by sex hormones.

First consider how basic sexual differentiation—the formation of a male or female with characteristic behaviors— occurs. Despite the gulf between the reproductive behaviors of rodents and human beings, laboratory rats are a good, if rough, model for human sexual differentiation. Rats and humans are mammals, and in mammals the basic or default form is female. This means that, absent masculinizing hormonal influences, a female will form regardless of genetic makeup.

Because we cannot experimentally manipulate human hormonal status, we have less information about details of human sexual differentiation. But hormonal anomalies in human beings provide useful information. For example, in human androgen-insensitivity syndrome, cells do not respond to the body’s endogenous androgens. Consequently, XY individuals with this syndrome look (and act) like females, despite having testes in the abdominal cavity and producing testosterone at levels typical of males. Only at puberty, when they do not menstruate, are they discovered to be males.

These XY individuals without functional androgen grow up considering themselves female. Contrast them with XY individuals who do experience androgenic influences before birth but are born with penile abnormalities. Until recently we assumed that if a baby boy lost his penis accidentally or was born with a malformed penis, he could safely be raised as a girl, because he did not yet have a gender identity. The usual procedure was for the testes to be removed and eventually a vagina to be constructed. As a result the castrated boys lost androgenic hormonal influences in their first year of life. Earlier reports suggested they grew up as normal girls, without obvious difficulty, although they could not become mothers. More recently we have learned that this sex reassignment does not always work as intended. These girls tend to retain strong male characteristics as they get older and most revert to living as males by adolescence, even without reconstructive surgery.² It seems that the effect of male hormones on their brains before birth, and perhaps in the first few months after birth, had an irreversible masculinizing influence that persisted despite their upbringing as girls who did not know their own early history.

Cognitive patterns of boys raised as girls have not been studied in depth, so we do not know if early androgens also shape a masculine pattern of abilities, despite the child being raised as a girl. This is a key question. But we do have information on the effect of androgen overproduction before birth in girls that suggests a lasting effect on cognitive abilities. This overproduction occurs in a condition called congenital adrenal hyperplasia (CAH), in which the adrenal glands produce excess androgens. In girls this can cause virilized genitals, a condition corrected surgically in the first few months of life. The flow of virilizing androgens is reduced right after birth by hormone therapy; the girls usually are raised without knowing their early history, although they know they must take hormone therapy (cortisol) for life.

The behavioral effect of early exposure to excess androgens on CAH girls is seen in characteristics more typical of boys than girls. CAH girls show more interest in traditionally masculine occupations and in rough-and-tumble play than do unaffected female relatives. They have less interest in infants and prefer boys’ toys to girls’ toys.³ They have better spatial manipulation and spatial rotation ability than unaffected girls, though not quite equal to unaffected boys. In contrast, boys with CAH who also are exposed to excess androgen do not show enhanced spatial ability; one study found them worse at this than unaffected boys, suggesting that abnormally high androgen levels may be disadvantageous.

Abnormally low androgen levels seem to have a negative effect on spatial ability. In a condition with the tongue-twisting name idiopathic hypogonadotrophic hypogonadism (IHH), testosterone levels beginning in early life are far below normal. In men with IHH, spatial ability is affected, although other cognitive skills seem normal.⁴ These discoveries together suggest there may be an optimal early level of androgens for certain spatial abilities, and that this optimal level is in the low-normal male range.

One confirmation of this hypothesis comes from studies of normal young adults whose hormonal background can be assumed to be unremarkable. We can measure testosterone levels in their saliva and relate this to spatial manipulation and spatial rotation abilities. Studies found that normal young men with lower levels of testosterone performed better on such spatial tasks than those with higher levels. In women the opposite happens. Those with higher testosterone levels perform better than those with lower levels. The highest scores are thus achieved by men with lower testosterone levels. We should note that men’s and women’s testosterone levels barely overlap, so the average level in low-testosterone men (the highest scoring group) is still much higher than that of high-testosterone women.

When Hormones Fluctuate by Day and Season

Hormones early in life organize the brain in ways that affect behavior for life, but natural hormonal fluctuations in adult life also cause variations in cognitive patterns. For example, men’s testosterone levels fluctuate from morning to evening; they also vary by season, with higher testosterone levels (and sperm counts) in the fall. Presumably this is related, in the northern hemisphere, to summer being an advantageous birthing time for our distant ancestors, because food is plentiful and the weather is warm.

What advantage could men gain by having better spatial ability in spring? One possibility is that, in hunter-gatherer times, this contributed to tracking and finding food after the lean winter. Or are seasonal changes in spatial ability merely a byproduct of hormonal changes that have other survival value? I will discuss this again when I consider cognitive variations in women.

Natural fluctuations in hormone levels during the female menstrual cycle are accompanied by changes in ability patterns. Periods of high estrogen (pre-ovulatory and mid-luteal phases) see increased performance relative to low-estrogen periods on tasks females are generally better at, like verbal fluency and fine motor skills.⁵ Performance on spatial tests seems affected in reverse: It is enhanced when estrogen levels are low rather than high. It seems reasonable to expect that feminine skills might be enhanced by higher estrogen, but why should spatial skills be depressed?

Perhaps masculine and feminine abilities are in a kind of trade-off relationship. In other words, for genital structure and reproductive behavior it is not reproductively advantageous for an individual to exhibit both kinds of traits (both a penis and a vagina), so one might speculate that the hormonal basis of masculinity and femininity operates reciprocally. The more masculine the hormonal milieu, the less feminine the traits. If so, cognitive fluctuation would be carried along by an inexorable system of hormonal changes that had other, non-cognitive bases. To date, however, we have not reliably seen depression of female-favoring tasks when testosterone levels are high.

Correlation and Cause

We know there is an association between sex hormone levels and cognitive function, but we are not sure about causes. Obviously one could infer that the hormones directly cause the cognitive pattern. On the other hand, factors we do not or cannot measure or detect that occur along with hormone changes may actually determine the cognitive pattern. We could speculate, for example, that women with high testosterone are more athletic, more likely to explore the external world, and so have more experience solving spatial problems.

A direct way to research this—one used in animals—is to manipulate hormone levels. Administering hormones to people and getting effects similar to those seen during natural fluctuations strengthens our inference that hormones are the cause. Studying people who receive various hormone therapies tends to confirm observations about natural hormonal fluctuations. Visual-constructional ability in older men improves after administering testosterone, cognitive performance is enhanced in post-menopausal women after estrogen-replacement therapy, and cognitive patterns show predicted effects when masculinizing or feminizing hormones are administered as part of transsexual treatment.

Sex Hormones Acting on the Brain

The bloodstream carries hormones to various parts of the body. Sex hormones that influence behavior directly must first affect brain cells in some way. Scientists think androgens and estrogens act by binding to specific receptors within cells. Systems of hormone receptors then promote DNA synthesis—a critical step in the process by which genes express themselves in body changes. Different hormones have different receptors, so one determinant of behavior is how many androgen or estrogen receptors are available; another is the location of the receptors in brain regions controlling particular cognitive functions. In animals, high concentrations of sex hormone receptors are found in the hypothalamus, the hippocampus, and parts of the cerebral cortex—areas that differ in structure in males and females.

The hypothalamus is important in determining whether male or female reproductive behavior will be forthcoming. In rats, a particular part of the preoptic area is larger in males than in females because of the early effects of sex hormones.⁶ The analogous region in humans may be the interstitial nuclei of the anterior hypothalamus. Certain parts of the region are larger in human males than in females, suggesting the regions may be involved in human sexuality.

The hippocampus is another structure that has different forms for males and females in some rodents (larger in males) and seems to support spatial memory and navigation. Size difference seems to result from the early influence of androgens. No sex-related size differences in the human hippocampal region are reported to date; and in humans this region may serve more general memory functions.

Sex Differences in the Human Brain

Scientists actively seek the precise brain mechanisms underlying cognitive differences between the sexes, but we are far from definitive answers. Many sex-related differences in human brains have been reported—anatomical and functional—but few reports are confirmed and several are contradictory. This makes it problematic to relate observed cognitive differences to specific brain differences. Instead, let us look at some areas being investigated.

The one sex difference about which there is no question is that men on average have larger brains than women. We used to see this as being determined by the difference in body size; the ratio of brain size to body size is the same in the sexes. However, Davison Ankney has pointed out that smaller people, male or female, have relatively higher brain-to-body ratios.⁷ When he compared groups of men and women matched for body size, the men’s brains were still found to be about 100 grams heavier than women’s. Men also have many more cortical neurons.⁸ Richard Lynn says such findings are consistent with his view that men on average are slightly more intelligent than women.⁹ This is a contentious issue, and one on which we as yet have no definitive answer.

Two other frequently claimed sex differences are that connections between the two cerebral hemispheres may be bigger in women, and that the dependence of speech functions on the left hemisphere and visual-spatial function on the right is more marked in men.

The implication is that larger connections between the hemispheres might mean that information transfer between hemispheres is better in women. Whether this would improve or impair a cognitive function might depend on the function. It seems unlikely that a highly coordinated motor skill like speech, for example, would be helped by readier access to the other hemisphere, which might make coordination looser. Although there are suggestions that some cognitive abilities relate to variations in splenium size in women, there is no evidence that cognitive differences between men and women depend on these variations.

The sometimes lower incidence of outright speech disorders (aphasia) in women with left-hemisphere damage is interpreted to mean that speech is less left-dependent in them than in men, presumably because women’s brains are more bilateral. A strong objection to this claim is that no study has ever reported a higher incidence of aphasia after right-hemisphere pathology in women than in men. My own research suggests that men and women differ less in the way speech functions are organized between hemispheres than within the left hemisphere itself. We found men to be more often rendered aphasic after posterior lesions, and women after anterior lesions, both in the left hemisphere.

This leaves us uncertain about whether men and women differ in left-hemisphere dependence for basic speech functions like naming and saying days of the week, which are routine in tests of aphasia. Still, there may be some sex differences in asymmetry for less fundamental speaking functions. For example, we found that the less rote vocabulary function (defining words) is affected by left- and right-hemisphere lesions in women, but only by left-hemisphere lesions in men, suggesting a more bilateral contribution to this ability in women. The same pattern holds for verbal fluency (saying as many words as possible that begin with a specified letter), but not for verbal memory. In contrast, recall of orally presented stories and word pairs is equivalently affected by left damage in men and women.

Scientists have not yet demonstrated a coherent relationship between a function that is more bilateral in one sex and the likelihood of that sex showing a cognitive advantage. Vocabulary score is bilaterally mediated (more dependent on both left and right hemispheres) in women, but there is no difference between adult men and women in vocabulary scores. Women perform better on tests of verbal fluency and verbal memory, yet their hemispheric organization for the tasks is different.

Some people claim men’s advantage on spatial tasks comes from their greater dependence for spatial functions on the right hemisphere. Yet when we compared the effects of left and right damage on a spatial rotation task (using mirror-image blocks), we found that right-hemisphere damage equally impaired male and female performance. These men, however, showed the same advantage over women in scores on the task as did men with no stroke damage. In a study¹¹ that compared men and women on line-slope identification after right-hemisphere strokes, women were at least as affected as men. In sum, it is premature to suggest that the sexes differ in performance on spatial tasks (as they do) because women depend less than men on their right hemisphere.

What Imaging Can Tell Us

What can brain imaging studies tell us about this question? In recent years, new techniques for assessing brain activity while people solve problems have promised to provide more information about how brain function may vary between individuals. Because these methods can be used in normal, healthy people, we do not have to depend solely on those with brain pathology or lesions. Functional Magnetic Resonance Imaging (fMRI) measures activity in particular brain areas by detecting the amount of oxygen the blood carries to those regions. Positron Emission Tomography (PET) works by showing how much glucose different brain areas absorb—another way to detect activity level.

Most of this neuroimaging research has asked if different brain regions in men and women light up during various language tasks. For example, it is reported that part of the inferior frontal region was more activated during a rhyme-judgment task than during other language tasks, including one requiring a decision about whether two words were related in meaning.¹² Men in that study, however, showed significantly more activation of the left frontal region than the right during rhyme judgment; in women there was no significant difference between sides. The researchers concluded that speech sound processes such as rhyme, but not necessarily all language processes, are more highly lateralized (one-sided) in men than in women. Similarly, a study using PET showed greater asymmetry of activation in men than in women during tasks in which they generated past tenses of verbs.

Other imaging studies, however, have failed to find differences in functional asymmetry between the sexes. One study using a PET scan found no difference in areas activated by two different speech production tasks. Another using fMRI found no difference in activation patterns on a task in which animal words were given aloud and the subject had to decide whether each animal fit a specified characteristic. This is considered a semantic or word-meaning task. Such apparently conflicting results may reflect only that different language tasks were studied. The sexes may differ in hemispheric organization for some language tasks but not others—as findings from brain lesion research indicate.

Imaging techniques have limitations when comparing brain events. Because the brain is always at least somewhat active, scientists must have a baseline (a measure of normal activity) against which to evaluate activation during the task they are investigating. Generally, they use a subtractive measure, comparing the activity occurring during the task with some background or control activity. For example, if we wanted to see what areas light up while subjects read aloud, we might subtract from it the activity seen when the subject reads the same words silently. Unfortunately, it is not always simple to choose the control or comparison conditions because not everyone agrees on the assumptions needed to describe what the subject is doing.

For the speech-production tasks described above, the comparison task was to stare at a central line without moving oral muscles. So the scan may have measured only the effects of oral movement, not speech-specific activation, in lighting up the brain. In the rhyme-judgment task, however, the procedure was more complex. Scientists inferred that specific brain areas related to speech sounds were lighting up after they subtracted activity observed during an upper case-lower case discrimination task from activity observed during a rhyming task. Also, the upper-lower case activity was measured by comparing it with a line-judgment task, creating three levels of subtraction.

From the Brain to Society

Scientists want to understand how cognitive patterns differ in men and women as they explore how individual cognitive patterns are shaped. Sex-related cognitive patterns and personality profiles are interesting to the public as well because they have ramifications for occupations and professions and their relative representation of the sexes. We cannot predict the occupational leanings of a particular man or woman with any accuracy, of course, but we can predict group representation more reliably.

Women on average have different cognitive profiles and have been found to be more nurturing than men and to have greater interest in people-oriented occupations. Men have been shown to be more theoretical and object-oriented. In line with these differences, women gravitate to education, nursing, and social work more often than do men. Men and women are about equally represented in other health-related professions, but men predominate in most scientific fields.

As I said at the outset, many people assume these differing preferences are driven by society and culture. For example, girls and boys engage in somewhat different activities; it seems reasonable to infer that these activities are important in determining their behavior as adults, including their cognitive patterns. But here we must remember the pitfalls of correlations. We do not know whether activities of the sexes as children determine their adult behavior. Child and adult behavior may arise from the same factors: prenatal or perinatal influences of sex hormones, for example. Determining the influence of socialization is difficult because we cannot often study the effects of child rearing independent of genetic or hormonal makeup. In contrast, we do have evidence, for example, that the effects of prenatal androgen-related masculinization can counteract socializing influences. Despite being raised as girls, CAH individuals and the unsuccessful sex reassignments described earlier show masculine behaviors lasting into adulthood.

Studies of gender representation in various professions show women in especially low numbers in the physical sciences and engineering. This seems to be a universal pattern, although proportions vary from country to country. This is commonly explained as the result of some form of negative discrimination, but where evidence exists it usually suggests the factors are differing talents and interests.

Spatial rotational ability, known to be greater in men, has been associated with better performance in organic chemistry,¹³ and in the physical sciences generally.¹⁴ Women are less often represented at the high end of mathematical reasoning tasks than are men, on standard tests like the SAT-Math and on specialized advanced exams like the Putnam Competition, where very few women excel. The difference in math reasoning between the sexes thus increases with increasing difficulty. Since math is a cornerstone of physics and engineering, it is not surprising that women are under represented in these fields.

Even mathematically talented girls still generally choose less theoretical, more people-oriented fields of study¹⁵ despite abundant parental and teacher support for their work in math-related fields. Women increasingly study medicine, agriculture, and basic biological sciences, where advanced math ability is less critical. Given the increased entry of women into these areas of science, interpreting the lower numbers of women in the physical sciences as a consequence of discrimination in science fields is illogical.

Today programs throughout North America promote the increased presence of women in these fields at the expense of better-qualified men. In Canada such programs include junior faculty awards available exclusively to women; in the United States they include post-doctoral awards for women and other programs especially for girls from the National Science Foundation. We have to wonder if these enterprises, in the name of egalitarianism, are enticing people into fields where they may neither excel nor enjoy themselves—and at the expense of abler individuals who might ultimately make more significant contributions.

We must guard against over-interpreting cognitive sex differences when dealing with individuals rather than with group averages. Every man or woman is unique and may defy the average profile for his or her sex. But why not face the probability that men and women will be disproportionately represented across a wide range of occupations and professions, without necessarily concluding that there are deliberate or systemic obstacles in the way of either sex? Undeniably women once encountered official discrimination in the form of quotas in many professions such as medicine. Now it seems likely that we can allow self-selection on the basis of talents and interests to determine career choices. Engaging in social engineering to balance sex ratios now threatens to become discrimination.

Sidebar: