In experiments starting in 1939, American social psychologists Kenneth Clark and Mamie Clark discovered that black children preferred to play with dolls that were white. Their data helped to convince the Supreme Court in Brown v. Board of Education that separate education was inherently unequal.

Now civil-rights lawyers are turning once again to psychology as a way to reveal the powerful hidden barriers created by modern-day bias. In battles over issues from affirmative action to workplace discrimination, educators, political theorists and activists are relying on Stanford University social psychologist Claude Steele's studies on "stereotype threat" to argue for policies that might make access to jobs and education fair for everyone.

The roots of the concept trace back to a phenomenon Steele explored shortly after he began his academic career at the University of Utah in the 1970s. It turned on the potent consequences of being judged negatively, even when the only basis was one's membership in a group. Researchers telephoned residents and made negative characterizations, such as "people in your community are lousy drivers." In separate calls later, the subjects who had been insulted were more likely than others to agree to help a food co-op. They were trying to maintain a positive self-image, Steele concluded, and it didn't matter if the food co-op had nothing to do with driving skills. The results also demonstrated that threats to self-image in turn affect behavior.

Steele extended the concept to stereotypes after joining the faculty of the University of Michigan at Ann Arbor in 1987. While advising a program for minority students, he noticed that although these young people had enrolled with the same SAT scores as others, their grades trailed those of white students and they dropped out in much higher proportions. The trend still holds true across the country: In the 1990s proportionately one fifth more blacks left college than whites. When they did graduate, their grade-point averages were two thirds of a letter grade lower.

Steele wondered if the Michigan students suffered from a kind of self-image threat, so with colleagues Joshua Aronson and Steven Spencer, he designed a series of studies. They gave sophomores matched by SAT scores a frustrating section of the Graduate Record Examination. When first told that the test evaluated verbal ability, the black students scored a full standard deviation lower on average. But when the researchers described it as a study of problem-solving techniques unimportant to academic achievement, the scores for blacks leaped to the same level as those for whites.

Mathematically accomplished women react comparably, Steele has observed. When given a difficult set of problems, they assumed their math abilities were under fire and scored significantly lower than men. But when told that gender could not affect scores, the women did as well as equally skilled men. Steele developed the theory of stereotype threat--that is, when people are challenged in an area they care deeply about, such as intellectual ability, the fear of confirming negative stereotypes can hurt their performance.

Social psychologists rapidly accepted the idea and even identified stereotype threat in groups not typically associated with bias. When experimenters told white golfers that the quality of their game would reflect "natural athletic ability" instead of their strategic intellectual prowess, their performance was much worse than that of black players. White male students' performance was similarly depressed when they took a math test in which Asian-Americans were said to do better. Internalized self-doubt could be eliminated as the cause, Steele and his colleagues concluded, because white men are not typically susceptible to worries about collective inferiority.

But everyone feels marginalized at one time or another, so Steele hopes his ideas can foster more understanding and help to integrate society better. He has struggled with the problem since youth. His activist parents, an interracial couple who married in 1944, trained their children in nonviolent resistance and brought them to demonstrations against segregation in Chicago's parks. But even though Steele grew up in a fairly integrated setting, he could go to the skating rink or swimming pool only on certain days and was never allowed at his maternal grandparents' house. Such circumstances left the young Steele angry and mystified: "If my life had been a more exclusively black world, some of these issues may not have been as central."

Modest and reserved, the 59-year-old Steele isn't someone you might expect to find at the front of a social movement. As a scientist, he worries that his research will be misinterpreted to serve political aims. But like it or not, stereotype threat has become central to the debate on standardized tests in college admissions and, some predict, could take on an equally crucial role in criminal justice and other civil-rights issues.

In the closely watched University of Michigan affirmative action case, Steele's testimony underlined the need for admissions officers to look beyond test scores. "His research was part of the picture we were painting--that race matters for a variety of reasons in today's America," says Marvin Krislov, general counsel for the university. In June 2003 the Supreme Court decided that diversity was critical to effective higher education.

Even so, Steele would like to move the discussion of race away from affirmative action. "It distracts from some of the more profound inequalities in American society," he says. Disparities in academic tracking, discipline and resources limit intellectual achievement from a child's very first day in school, he argues. And silent, psychological cues impede success as well. When schools relegate ethnic studies to special programs, for instance, students' identities are marginalized in tandem, Steele asserts. Or when a teacher starts talking about "white males," those students begin to feel threatened and uncomfortable.

Even efforts to improve diversity may backfire if they are seen as "remedial." Instead schools and employers can remove the internal barriers that stereotypes produce by creating a setting that makes them unimportant. A critical factor is diverse numbers, Steele now says. "Critical mass gives people a sense of identity safety," he explains, and protects against the threat of assumptions based on race or gender.

Steele's work will most likely become a significant tool for understanding how race affects people's lives and opportunities, says legal theorist Charles J. Ogletree, Jr. Ogletree expects stereotype threat to become a central theme in studies at the new institute on race and justice he will be directing at Harvard University. "I think it's the cornerstone of a new field of expertise that will help us think through and resolve problems of race at a much grander scale than ever before," Ogletree asserts.

With a laugh, Steele predicts it will take a whole village of scientists to fully understand the process. In testing, for instance, other influences may swamp stereotype threat. "That's why I think it's exciting--it's important and powerful but not very well understood," Steele remarks. He believes the phenomenon, diffuse throughout society, affects everything from white self-segregation to minority self-limitation. "Stereotype threat is a first attempt to characterize these processes," he explains. "You hope people make some sense out of it and use it."