Showdown at Sex Gap

Women's intrinsic math and science aptitude divides scientists

Here’s a good way to inflame the tempers of all those within earshot. Do as former Harvard University President Lawrence Summers did in 2005 and suggest that the relatively low number of high-achieving women in mathematics and science partly reflects a lack of an inherent aptitude for such pursuits. Summers lost his job in the campuswide tumult that followed his remarks. But in the ambiguous world of research on sex differences and their influence on math and science abilities, Summers’ saga prompted new attempts to clarify what scientists know and how their data apply to education and test taking.

At an Oct. 1 meeting at the American Enterprise Institute (AEI) in Washington, D.C., scientists tried to hash out why females lag behind males in math and science achievement. Participants focused considerable attention on a recent extensive review that found a place for sex-related biological disparities in explaining such achievement differences as well as on an earlier report that dismissed biology as a factor.

The latter report was issued in 2006 by the 18-member Committee on Maximizing the Potential of Women in Academic Science and Engineering, convened by the National Academy of Sciences (NAS). “It is not lack of talent,” the committee concluded, “but unintentional biases and outmoded institutional structures that are hindering the access and advancement of women” in technical fields.

GIRLS, INTERRUPTED Researchers disagree about why more boys than girls take up math and science careers. Photodisc

Controversy greeted the NAS report, notes psychologist Susan M. Barnett of the University of Cambridge, England. Some researchers suggested that committee members held their own biases against acknowledging any sex-related biological differences in math and science aptitude.

Enter a consensus statement, published in the August Psychological Science in the Public Interest, written by six researchers with varying takes on the reasons for sex differences. They conclude that “early experience, biological factors, educational policy, and cultural context affect the number of women and men who pursue advanced study in science and math” and that “these effects add and interact in complex ways.”

Psychologist Diane F. Halpern of Claremont (Calif.) McKenna College directed work on the consensus statement. She also spoke about sex differences to the NAS committee during its deliberations.

“Can we increase the number of women who enter careers in science and math? Yes,” Halpern asserts. “Is there evidence of a sex-related biological component to success in science and math? Yes.”

What difference?

At the AEI sessions, two psychologists challenged the assumption that biology in any way undermines women’s math and science proficiency.

Psychologist Rosalind C. Barnett of Brandeis University in Waltham, Mass., noted that several research reviews—including analyses conducted by Janet S. Hyde of the University of Wisconsin–Madison, a contributor to the consensus statement—find no or minimal sex differences in math and science aptitude.

Although more males than females earn extremely high scores on standardized math tests, such scores predict surprisingly little about who will succeed in math and science careers, Barnett says. Among college-educated men with math, science, or engineering jobs, less than one-third scored 650 or better out of 800 on the math portion of the SAT.

MEASURED SUCCESS Several lines of evidence suggest that sex-related biological factors contribute to math and science talent. Photodisc

Men’s monopoly on high-level math and science achievements derives largely from unfair social and institutional advantages, the Brandeis psychologist says. For the past several hundred years, social forces have limited women’s access to education and employment in the sciences, Barnett argues. Now, women receive the same education as men do but struggle against academic undercurrents of bias, she says. One study, cited in the NAS report, evaluated peer-reviewers’ ratings of applications for postdoctoral fellowships in Sweden. Researchers found that a woman had to be twice as productive as a man in publishing research and in other areas of scientific achievement in order to be judged equally competent.

Productivity aside, boys and girls possess the same three mental systems at the core of mathematical and scientific reasoning, according to Harvard University psychologist Elizabeth Spelke, a member of the NAS committee.

“Evidence to date does not favor the hypothesis of a male advantage in intrinsic aptitude for math and science,” Spelke says.

From infancy on, in her view, all typically developing children rely on one mental system that represents and reasons about objects, another that represents and reasons about numbers, and a third that does the same for geometric relations.

For instance, preschool-age boys and girls are equally adept at tracking items moving among distracting objects on a computer screen. Moreover, infants of both sexes recognize approximate quantities of items.

“Evidence to date does not favor the hypothesis of a male advantage in intrinsic aptitude for math and science.”

Elizabeth Spelke, Harvard University

Spelke and her coworkers have also tested 6- to 10-year-olds in the United States and in a remote Amazonian population for the ability to recognize relationships among simple visual forms and basic geometric concepts, such as distance and angle.

Overall, boys and girls performed comparably well on more than 40 geometric problems, Spelke says. Boys displayed superiority only on a mental-rotation problem, which requires a skill often incorrectly portrayed as the ultimate indicator of spatial ability, she argues.


Neuroscientist Richard J. Haier of the University of California, Irvine got plenty of flak 20 years ago when he talked about possible intelligence-related brain differences between men and women. Now he gets a friendlier reception from people who attend his public lectures, even if such work still makes many academics uneasy.

“The NAS committee prematurely dismissed biological research on sex differences,” Haier says. “The new consensus statement waffles a bit on the biological research as well.”

Men and women display comparable general-intelligence levels, on a measure derived from IQ scores. Yet the brain may foster intelligence differently in the two sexes, Haier suggests.

In the April Behavioral and Brain Sciences, he and Rex E. Jung of the University of New Mexico in Albuquerque reviewed 37 brain scan investigations of intelligence published to date. They concluded that parts of the parietal and frontal cortex, in concert with a few other neural regions, form a network that orchestrates individual differences in intelligence and reasoning.

Studies suggest that this network critically supports men’s intellect, including mathematical reasoning, Haier says. The same brain network modestly contributes to how well women do on various achievement tests.

Two large investigations conducted by separate teams indicate that girls call on a more distributed network of neural areas during reasoning tasks than boys do.

The new consensus statement suggests that male brains often rely on enhanced communication within each hemisphere, as indicated by measurements of large numbers of neural connections on each side of the brain in men. In contrast, female brains may specialize in communication across hemispheres, the consensus statement proposes, with extra assistance from language-related areas.

Neural clues coincide with what researchers know about the development of math and science expertise among mathematically talented youth. Researchers have tracked the accomplishments of more than 5,000 individuals from junior high school to middle age. As 12- to 13-year-olds, about 500 scored 700 or more on the SAT math test or 630 or more on the SAT verbal test, placing them within the top 1 percent of test takers.

More boys than girls received the highest math scores, although this gap has narrowed over the past decade to about four boys for every girl.

A larger proportion of mathematically talented males than females entered math and science careers, according to psychologist and study director Camilla P. Benbow of Vanderbilt University in Nashville. However, mathematically talented girls displayed a broader range of verbal strengths than boys did and often achieved advanced degrees in areas that required all of their skills, such as law, medicine, and the social sciences.

“Men choose more object-oriented jobs and women choose more people-oriented jobs, but so what?” remarks psychologist David Geary of the University of Missouri–Columbia. “Men and women report no differences in life satisfaction.”

Neuroscientist Simon Baron-Cohen of the University of Cambridge in England suspects that sex differences involved in intelligence and social interest emerge early in life, possibly as a result of fetal exposure to varying levels of the sex hormone testosterone.

“Men choose more object-oriented jobs and women choose more people-oriented jobs, but so what?”

David Geary, University of Missouri

In several studies conducted since 2002, his team has found that high testosterone concentrations—identified in amniotic fluid during pregnancy—predict a youngster’s tendency to avoid eye contact and to display a limited vocabulary at age 18 months as well as to show oral-communication difficulties at age 4.

These children typically show special interest in analyzing rule-based systems, involving computers, cars, and mathematics, for example. By age 8, they find it difficult to take another person’s perspective and to react appropriately in social situations.

Boys display elevated testosterone concentrations more often than girls do, although some girls show this pattern as well, Baron-Cohen says. He theorizes that autism develops in extreme cases of fetal-testosterone overload and often fosters mathematical talent, despite its other drawbacks.

Certain patterns of sex differences indicate that evolution sculpted male and female intellectual capacities along different lines, asserts Geary. For instance, in a 2005 study, school-age boys scored better on tests of spatial rotation and map reading than girls did, but only if the boys lived in middle- or upper-income families. Low-income children exhibited no such sex differences.

In other words, Geary says, males showed a special sensitivity to poverty that eliminated their superiority on spatial tasks.

This fits with the hypothesis that, by taking charge of long-distance journeys, big-game hunting, and warfare during the Stone Age, males evolved an affinity for spatial tasks. Early deprivation affects the brain in ways that undermine males’ evolved spatial aptitude, Geary proposes.

Still, the consensus statement, to which Geary contributed, notes that scientists sharply disagree about whether evolutionary forces lie behind the sex gap.


In 1995, psychologists Claude Steele of Stanford University and Joshua Aronson of New York University administered an achievement test to college students. One set of students was told that the test measured intelligence, whereas another group was told that the test was simply a research tool. Each group contained both black and white students.

Whites performed much better than blacks when taking a purported intelligence test, but the racial groups scored comparably well when they regarded the same test more neutrally. Steele and Aronson attributed this result to what they called stereotype threat, a sense of discomfort and distraction arising from subtle reminders of a group’s negative reputation in a situation. For example, anxiety about having read that blacks do poorly on IQ tests might have interfered with black students’ performance on the “intelligence test.”

Numerous laboratory studies have implicated stereotype threat as a drain on the intellectual abilities and test proficiency of women as well as blacks. At the AEI meeting, Aronson described a study in which female college students performed especially poorly compared to male students on a spatial rotation test after having been asked to indicate their sex. However, much of the sex-related disparity disappeared when male and female test-takers were first reminded that they all attended an elite college.

“Cultural ideas about group differences can exacerbate or lessen those differences,” Aronson says. “Intervention can boost performance and nurture intelligence so that biology need not mean destiny.”

However, stereotype-threat research draws fire for saying little about real-world, high-stakes testing situations. “The claims routinely made on behalf of stereotype threat are vastly exaggerated,” says neurologist and law professor Amy Wax of the University of Pennsylvania in Philadelphia. She uses social science research to examine public-policy issues.

Stereotype-threat studies often begin by statistically adjusting for the unequal prior test scores of men and women or of blacks and whites. Performance is then contrasted between groups either exposed to or protected from stereotype threat. Psychologist Paul R. Sackett of the University of Minnesota in Minneapolis and his coworkers argue that this approach indicates only that, absent stereotype threat, men and women, or blacks and whites, would display the same achievement gaps as they did before the experiment.

Moreover, stereotype-threat studies have not examined male and female samples that reflect the sex gap in math performance observed in the general population. As a result, it’s impossible to estimate whether stereotype threat accounts for 90 percent of that gap, 5 percent of it, or some other proportion, Wax says.

Psychologist Lawrence J. Stricker of the Education Testing Service in Princeton, N.J., doubts that stereotype threat sways scores on actual achievement tests. In a 2004 study, Stricker and a colleague collected information on race and sex either before or after 1,652 high school students took an advanced-placement test in calculus. The researchers did the same for 1,341 incoming community college students taking a battery of placement tests.

Stereotype threat, in the form of being reminded of one’s race and sex before the test, did not lower women’s calculus or math scores. Neither did it lower black students’ overall scores on the two tests.

However, a reanalysis of Stricker’s data using a looser statistical standard of success found that stereotype threat indeed lowered women’s calculus scores. Psychologists Kelly Danaher and Christian S. Crandall, both of the University of Kansas in Lawrence, estimate that 5.9 percent more women and 4.7 percent fewer men would receive passing scores if they indicated their sex after taking the test rather than before.

This “simple, small, and inexpensive change” would boost the number of U.S. women receiving advanced-placement calculus credit by more than 4,700 annually, Crandall says.

Stricker calls that conclusion “sensational but unwarranted.” Crandall’s projection is unreliable because the original advanced-placement sample was not chosen to represent the sex gap among all test takers, he says.

However this dispute shakes out, the rampant sexism that math- and science-oriented women struggled against in past decades shows signs of decay. As one young woman who recently earned an undergraduate degree in biochemistry and pharmacology put it at the AEI meeting, “Most of my professors didn’t look up from their podium long enough to realize that I was a woman, much less care. When I would fail an exam, I wouldn’t say, ‘Gosh, I wish my brain was more predisposed to science.’ It was ‘I really should have studied a hell of a lot harder than I did.'”

Bruce Bower has written about the behavioral sciences for Science News since 1984. He writes about psychology, anthropology, archaeology and mental health issues.