In the first part of the study, girls and boys were told a story about a person who is "really, really smart," a child's idea of brilliance, and then asked to identify that person among the photos of two women and two men. The people in the photos were dressed professionally, looked the same age and appeared equally happy. At 5, both boys and girls tended to associate brilliance with their own gender, meaning that most girls chose women and most boys chose men.
But as they became older and began attending school, children apparently began endorsing gender stereotypes. At 6 and 7, girls were "significantly less likely" to pick women. The results were similar when the kids were shown photos of children.
Interestingly, when asked to select children who look like they do well in school, as opposed to being smart, girls tended to pick girls, which means that their perceptions of brilliance are not based on academic performance.