Girls do better than boys in school on average, and they know it. So why do girls tend to shy away from careers in physics, math, and other fields that are perceived as requiring brilliance?
Some of the blame may lie with stereotypes that take root as early as age 6, according to a new study.
By that age, girls are less likely than boys to believe that members of their gender are super-smart, the authors reported in Friday's issue of the journal Science.
The authors based this finding on several experiments, including one in which dozens of girls and boys were read a short story about a "really, really smart" person whose gender was not identified. The children then were shown pictures of four adults, two men and two women, and asked to guess which one was described in the story.
At age 5, girls and boys were equally likely to pick someone from their own sex, doing so about 70 percent of the time. But at age 6 and 7, girls picked women only about half the time, while boys of those ages continued to show a clear preference for men.
The drop-off among girls was unexpected, because past studies have shown that they have generally strong pro-girl attitudes at least through age 7, said senior author Andrei Cimpian, an associate professor of psychology at New York University.
"It's all the more surprising that girls are sort of dipping with respect to their ideas about who is really, really smart, given their overall positivity toward their gender," he said.
But there might be other explanations for the results, cautioned David C. Geary, a University of Missouri psychologist who was not involved with the study.
"Boys are simply overconfident about themselves and other boys and men generally," he said. "This is not always a good thing."
What is not in dispute is the gender disparity in the pursuit of careers.
In a previous study, Cimpian and others found that in 2011, women made up less than 30 percent of those who held Ph.D.'s in math, and less than 20 percent of the total for physics. The numbers of advanced degrees in biology and related life sciences were split more evenly among men and women.
As academics continue to untangle the thicket of nature and nurture to explain gender differences in the sciences, educators also are addressing the issue. Programs to interest girls in the "STEM" fields – science, technology, engineering, and math – have mushroomed, among them national nonprofit efforts such as TechGirlz and Girls Who Code.
The new study struck a chord with Steph Sessa, who teaches biology and chemistry at Science Leadership Academy in Philadelphia. A math fan from an early age, she started as a physics major at Boston University and found herself among a handful of women in that field.
"You kind of feel under a microscope a little bit, if you'll excuse the science pun," she said. "If you get anything wrong, you're kind of looked down upon. You're held to an even higher standard to be the female physicist."
She switched to biology in part because she saw more career options there, but now as a teacher she sees the power of role models. The academy, a partnership between the Philadelphia School District and the Franklin Institute, has a significant African American enrollment. So if she needs to show a video of a scientist, she is eager to showcase someone of color, such as astrophysicist Neil deGrasse Tyson.
The new study suggests that gender-branding starts early, say Cimpian and his coauthors, Lin Bian of the University of Illinois and Sarah-Jane Leslie of Princeton University.
In addition to the story-reading experiment, the authors presented boys and girls with two different games, and asked which one they would like to play.
One was described as being "for children who are really, really smart," while the other was said to be for those "who try really, really hard."
At ages 6 and 7, girls were less interested than boys in the game for smart children, while there was no difference in interest in the game for those who try hard.
Echoing Geary, the authors acknowledged that modesty was a possible explanation for these differences. But they noted that modesty would not explain the differences in the short-story experiment, as girls were not asked to judge whether they were smart, but whether others were smart.
And further analysis revealed that girls who were more likely to pick women as protagonists of the short story were also more likely to be interested in the "smart" game, the authors wrote.
So what's the answer? Role models? Mentoring? Training teachers to nourish girls' interest in the sciences? Cimpian said it may take more than one approach.
"It's important to know when some of these ideas start percolating in children's minds," Cimpian said. "That's the point where it's most effective to intervene."