I have a moderate preference for white faces over black ones…at least according to Project Implicit, run by Harvard psychologists. And that’s not all. I have a preference for Islam over Hinduism, for young people over old ones, and for heterosexuality over homosexuality.

So, however, do most people, and you might, too, no matter how accepting and forward-thinking you think you are. The test, called the Implicit Association Test (IAT), which can be taken for free online, works by assessing our implicit negative or positive associations with various identity groups.

Here’s how it works: first, various words flash across the screen, and participants are asked to sort them into two categories—good and bad—by pressing down on the “I” or the “E” key. The words have either positive (good, superb, best, wonderful, laughter) or negative (bad, worst, horrible, tragic, miserable) connotations.

The tricky part comes when the researchers insert images or words that correspond to various identities: images of black or white faces, old- or young-associated words, words surrounding religious practice (Jew, Abraham, Torah or Christian, Jesus, Gospel).

Using race as an example, the test instructs you to first press down with your right finger for either “good” words OR white faces, and to press down with your left finger for either “bad” words OR black faces. Then it switches, asking you to associate “good” and black and “bad” and white. If it takes your brain a longer time to associate “good” and black than it does to associate “good” and white, and if your answers are less accurate, then you have your answer: your brain prefers white faces over black ones.

I visited the project’s website after reading about an explanation of the test in “Blindspot: Hidden Biases of Good People,” a recent book by psychologists Mahzarin Banaji and Anthony Greenwald. I began with the religion test, which measured my associations with “goodness” and “badness” for Judaism, Christianity, Islam, and Hinduism. In 5 minutes, I had my results: I have the best associations for Judaism, followed closely by Islam; Christianity lags a bit, and Hinduism is farther down the pole, resting closer to “bad” associations.

I was flabbergasted. HINDUISM? I know for sure that I have nothing against Hinduism; I don’t know a lot about it, sure, but I harbor no conscious prejudice towards its followers, who include, I don’t know, Mahatma Gandhi and Mindy Kaling. I’m prejudiced against HINDUISM? Could it be that I simply do not know many Hindus? Could it be that it feels more foreign to me than do Judaism, Christianity, and Islam?

Honestly, I have no idea. Before I lost my nerve, I swallowed my pride and clicked “race” on Project Implicit’s homepage. I wasn’t ready, exactly, but it was much like reaching into the garbage to find a lost ring. I had to take the plunge.

The test began the way the other one had: right hand for “good” words, left hand for “bad” ones. Once the test saw that I was capable of completing that simple task, it asked me to press down with my right hand for “good” words OR white faces and my left hand for “bad” words OR black faces. I clicked “begin” with trepidation. After the first round, the test changed the association: “good” words OR black faces, “bad” words OR white faces. The second part didn’t feel noticeably harder; maybe, I thought, I’m not a racist after all.

My relief was short-lived. I prefer white faces. The gap in preference isn’t huge, but it is there.

The most alarming part of the test, and of Banaji and Greenwald’s findings, is that these implicit associations are almost completely subconscious; Banaji and Greenwald assure their readers that it’s not our fault that we’re implicitly prejudiced. Many test takers who identify as black, such as Malcolm Gladwell, also prefer white faces to black ones. Elderly test takers associate better things with youth than they do with age. Overweight participants have bad associations with being heavy. Muslim participants’ brains sometimes prefer Christianity to Islam. Women—even high-powered, career-focused women—have a harder time associating “man” with “family” and “woman” with “career” than they do with the reverse association.

Looking consciously at our own “blind spots,” the parts of our brains that operate on a subconscious level, though, is not enough to combat implicit prejudice and discrimination. Banaji and Greenwald have both taken the test many times, and even conscious efforts to correct their own implicit associations—and to ensure that they treat colleagues and students of identity groups with which they have negative subconscious associations equitably—have not yielded different results. These “mindbugs,” as Banaji and Greenwald call them, “ingrained habits of thoughts that lead to errors in how we perceive, remember, reason, and make decisions,” are tough, if not impossible, to eradicate.

So what’s a racist, sexist, ageist, homophobe like me to do?

The resounding conclusion is that nobody knows. Banaji and Greenwald offer up one nugget of hope, however: one thing that has helped people with negative associations for, say, black faces, is increased exposure to highly respected black faces, along with exposure to terrible white people (serial killers, for example, or Rick Perry). As the media shines its light on powerful figures such as President Obama and Colin Powell, and on talented black artists, scientists, and economists, and as white people continue to do destructive things, we might be more likely to shift associations.

But even though conscious examinations of our own blind spots do not necessarily ameliorate the unjustified associations in our brains, the knowledge that we have these blind spots can help us adjust our conscious behavior. Professors who take the test and learn that they associate black faces with “bad” words can, for example, institute blind readings of tests and essays—reading them without their students’ names—to ensure that these biases do not result in inequitable treatment in grading. We, too, can learn to catch ourselves in the act of acting on our implicit associations, even though these associations might be in our brains for good.

But the first step is uncovering the associations themselves. Take the test. If you’re also racist, sexist, ageist, and homophobic, nobody but you has to know. But if you’re committed to aligning your actions with your beliefs, then it’s necessary that you face the music.