For more than 30 years, psychologists Mahzarin Banaji and Anthony Greenwald have been studying the unconscious biases that take root in our brains, coloring everything from hiring decisions to how doctors mete out medical care and judges rule on sentences. If you don't think you harbor any such mental stowaways, tugging you in favor of white over black, straight over gay, or male over female - yes, perhaps even if you're a proud black lesbian - then log onto Harvard University's Project Implicit and prepare to be disappointed in someone you never knew held such appalling views: you.

The news in Banaji and Greenwald's new book, "Blindspot: Hidden Biases of Good People," is not what we want to hear, either: The pair's research suggests that while children aren't born with racial preferences, they develop them incredibly early, so that a 3-year-old African American child is just as statistically likely as an adult to show preference for white over black. Across age groups, about 40 percent of African Americans show a pro-white bias on the Implicit Association Test developed by the authors and their colleague Brian Nosek, while 40 percent favor black and the rest are neutral. Some 14 million people have taken the test so far.

Similarly, women are even more likely than men to associate masculine names with career-related words and feminine names with home-related words. That last one stings a little less when Banaji explains that it's not so much that women find it hard to identify women with career. "But men with home? No."

To hold such hidden views is not necessarily to act on them, of course - but there is evidence that the stronger the bias, the more likely we are to do just that. Banaji's work has convinced her that one of the most common forms of acting on an unconscious bias is less likely to involve hurting "the other" than helping someone with whom we identify, or unconsciously favor as a member of the dominant group.

One example of this in the book is an incident in which a friend of Banaji's, a woman named Diane, cut her hand badly and her boyfriend rushed her to the emergency room, where he worriedly explained to the doctors that he feared the injury could keep her from doing the quilting that was her hobby. Not to worry, they said; stitching it up quickly was the key.

Just then, though, a volunteer came in, saw the quilter and cried, "Professor, what are you doing here?" Now identified as a Yale University professor, the doctors instantly reversed themselves and, instead of stitching her right up, called in a top hand surgeon. They weren't consciously, of course, valuing Diane the Ph.D. over Diane the humble quilter, and yet these are the kind of "favors" we might need to rethink if we want to be as fair and egalitarian as we say we want to be.

One of the strongest biases Banaji has studied favors slim folks over heavy, though unless you are morbidly obese or work as an athlete or a body double, metabolism is probably not the best measure of talent or competence in your field. So, to rise to the top, forget night school and lose 10 pounds? It's sad, Banaji said, but yes. We also favor extroverts, even when giving good meeting has little to do with the job.

Given her work on racial bias, was she surprised when Barack Obama was elected president? Yes, she said, but added, "Don't forget he was running against a woman and an elderly man who's short" - yes, the last is yet another shortchanged category.

The Implicit Association Test I tried - Franklin D. Roosevelt vs. Obama - is one of the more benign ones, but plenty telling, all the same. The exercise is to rapidly sort images of Obama in the same bucket with positive words, and images of FDR in the one with negative words, then switch; the president you can more quickly associate with "wonderful" than "miserable" and so on is the one you implicitly favor. Now, consciously, I would say that while I hold favorable views of both men, I surely give extra points to the one who got us out of the Great Depression, defeated the Nazis and dealt us the New Deal. Right?

Well, no. The result shows I favor Obama. When I asked Banaji to please tell me I could not have picked up this "mindbug" from that dreadful recent movie that showed FDR mistreating the women in his life - especially when I myself wrote about how inaccurate "Hyde Park on Hudson" was - she laughed and said that was actually a perfect example of how such views burrow in, even when our conscious minds know better.

So, are we stuck with these pernicious stowaways? They're incredibly hard to dislodge, she said, but being aware of them can help us work around them. And when we say a job candidate "just doesn't fit," those words ought to be a red flag. "Unless I can justify why I feel that way," Banaji says, "I don't say that anymore."

Melinda Henneberger writes for The Washington Post.