Where Bias Begins: The Truth About Stereotypes

By Annie Murphy Paul

Psychology Today May/June 1998

Psychologists once believed that only bigoted people used stereotypes. Now the study of unconscious bias is revealing the unsettling truth: We all use stereotypes, all the time, without knowing it. We have met the enemy of equality, and the enemy is us.

Mahazarin Banaji doesn't fit anybody's idea of a racist. A psychology professor at Yale University, she studies stereotypes for a living. And as a woman and a member of a minority ethnic group, she ahs felt firsthand the sting of discrimination. Yet when she took one of her own tests of unconscious bias, "I showed very strong prejudices," she says. It was truly a disconcerting experience." And an illuminating one. When Banaji was in graduate school in the early 1980s, theories about stereotypes were concerned only with their explicit expression: outright and unabashed racism, sexism, anti-Semitism. But in the years since, a new approach to stereotypes has shattered that simple notion. The bias Banaji and her colleagues are studying is something far more subtle, and more insidious: what's known as automatic or implicit stereotyping, which, they find, we do all the time without knowing it. Though out-and-out bigotry may be on the decline, says Banaji, "if anything, stereotyping is a bigger problem than we ever imagined. "Previously, researchers who studied stereotyping had simply asked people to record their feelings about minority groups and had used their answers as an index of their attitudes. Psychologists now understand that these conscious replies are only half the story. How progressive a person seems to be on the surface bears little or no relation to how prejudiced he or she is on an unconscious level-so that a bleeding-heart liberal might harbor just as many biases as a neo-Nazi skinhead. As surprising as these findings are, they confirmed the hunches of many students of human behavior. "Twenty years ago, we hypothesized that there were people who said they were not prejudiced but who really did have unconscious negative stereotypes and beliefs," says psychologist Jack Dovidio, Ph.D., of Colgate University. "It was like theorizing about the existence of a virus, and then one day seeing it under a microscope."The test that exposed Banaji's hidden biases-and that this writer took as well, with equally dismaying result-is typical of the ones used by automatic stereotype researchers. It presents the subject with a series of positive or negative adjectives, each paired with a characteristically "white" or "black" name. As the name and word appear together on a computer screen, the person taking the test presses a key, indicating whether the word is good or bad. Meanwhile, the computer records the speed of each response. A glance at subjects' response times reveals a startling phenomenon: Most people who participate in the experiment-even some African-Americans-respond more quickly when a positive word is paired with a white name or a negative word with a black name. Because our minds are more accustomed to making these associations, says Banaji, they process them more rapidly. Though the words and names aren't subliminal, they are presented so quickly that a subject's ability to make deliberate choices is diminished-allowing his or her underlying assumptions to show through. The same technique can be used to measure stereotypes about many different social groups, such as homosexuals, women, and the elderly.


From these tiny differences in reaction speed-a matter of a few hundred milliseconds-the study of automatic stereotyping was born. Its immediate ancestor was the cognitive revolution of the 1970s, an explosion of psychological research into the way people think. After decades dominated by the study of observable behavior, scientists wanted a closer look at the more mysterious operation of the human brain. And the development of computers-which enabled scientists to display information very quickly and to measure minute discrepancies in reaction time-permitted a peek into the unconscious. At the same time, the study of cognition was also illuminating the nature of stereotypes themselves. Research done after World War 11-mostly by European émigrés struggling to understand how the Holocaust had happened-concluded that stereotypes were used only by a particular type of person: rigid, repressed, authoritarian. Borrowing from the psychoanalytic perspective then in vogue, these theorists suggested that biased behavior emerged out of internal conflicts caused by inadequate parenting. The cognitive approach refused to let the rest of us off the hook. It made the simple but profound point that we all use categories-of people, places, things-to make sense of the world around us. "Our ability to categorize and evaluate is an important part of human intelligence," says Banaji. "Without it, we couldn't survive." But stereotypes are too much of a good thing. In the course of stereotyping, a useful category-say, women-becomes freighted with additional associations, usually negative. "Stereotypes are categories that have gone too far," says John Bargh, Ph.D., of New York University. "When we use stereotypes, we take in the gender, the age, the color of the skin of the person before us, and our minds respond with messages that say hostile, stupid, slow, weak. Those qualities aren't out there in the environment. They don't reflect reality."Bargh thinks that stereotypes may emerge from what social psychologists call in-group/out-group dynamics. Humans, like other species, need to feel that they are part of a group, and as villages, clans, and other traditional groupings have broken down, our identities have attached themselves to more ambiguous classifications, such as race and class. We want to feel good about the group we belong to-and one way of doing so is to denigrate all those who aren't in it. And while we tend to see members of our own group as individuals, we view those in out-groups as an undifferentiated-stereotyped-mass. The categories we use have changed, but it seems that stereotyping itself is bred in the bone.Though a small minority of scientists argues that stereotypes are usually accurate and can be relied upon without reservations, most disagree-and vehemently. "Even if there is a kernel of truth in the stereotype, you're still applying a generalization about a group to an individual, which is always incorrect," says Bargh. Accuracy aside, some believe that the use of stereotypes is simply unjust. "In a democratic society, people should be judged as individuals and not as members of a group," Banaji argues. "Stereotyping flies in the face of that ideal."


The problem, as Banaji's own research shows, is that people can't seem to help it. A recent experiment provides a good illustration. Banaji and her colleague, Anthony Greenwald, Ph.D., showed people a list of names-some famous, some not. The next day, the subjects returned to the lab and were shown a second list, which mixed names from the first list with new ones. Asked to identify which were famous, they picked out the Margaret Meads and the Miles Davises-but they also chose some of the names on the first list, which retained a lingering familiarity that they mistook for fame. (Psychologists call this the "famous over night-effect.") By a margin of two-to-one, these suddenly "famous" people were male. Participants weren't aware that they were preferring male names to female names. Banaji stresses. They were simply drawing on an unconscious stereotype of men as more important and influential than women. Something similar happened when she showed subjects a list of people who might be criminals: without knowing they were doing so, participants picked out an overwhelming number of African-American names. Banaji calls this kind of stereotyping implicit, because people know they are making a judgment-but just aren't aware of the basis upon which they are making it.

Even further below awareness is something that psychologists call automatic processing, in which stereotypes are triggered by the slightest interaction or encounter. An experiment conducted by Bargh required a group of white participants to perform a tedious computer task. While performing the task, some of the participants were subliminally exposed to pictures of African-Americans with neutral expressions. When the subjects were then asked to do the task over again, the ones who had been exposed to the faces reacted with more hostility to the request-because, Bargh believes, they were responding in kind to the hostility which is part of the African-American stereotype. Bargh calls this the "immediate hostile reaction," which he believes can have a real effect on race relations. When African-Americans accurately perceive the hostile expressions that their white counterparts are unaware of, they may respond with hostility of their own-thereby perpetuating the stereotype.

Of course, we aren't completely under the sway of our unconscious. Scientists think that the automatic activation of a stereotype is immediately followed by a conscious check on unacceptable thoughts-at least in people who think that they are not prejudiced. This internal censor successfully restrains overtly biased responses. But there's still the danger of leakage, which often shows up in non-verbal behavior: our expressions, our stance, how far away we stand, how much eye contact we make. The gap between what we say and what we do can lead African-Americans and whites to come away with very different impressions of the same encounter, says Jack Dovidio. "If I'm a white person talking to an African-American, I'm probably monitoring my conscious beliefs very carefully and making sure everything I say agrees with all the positive things I want to express," he says. "And I usually believe I'm pretty successful because I hear the right words coming out of my mouth." The listener who is paying attention to non-verbal behavior, however, may be getting quite the opposite message. An African-American student of Dovido's recently told him that when she was growing up, her mother had taught her to observe how white people moved to gauge their true feelings toward blacks. "Her mother was a very astute amateur psychologist-and about 20 years ahead of me," he remarks.


So where exactly do these stealth stereotypes come from? Though automatic-stereotyping researchers often refer to the unconscious, they don't mean the Freudian notion of a seething mass of thoughts and desires, only some of which are deemed presentable enough to be admitted to the conscious mind. In fact, the cognitive model holds that information flows in exactly the opposite direction: connections made often enough in the conscious mind eventually become unconscious. Says Bargh: "If conscious choice and decision making are not needed, they go away. Ideas recede from consciousness into the unconscious over time. "Much of what enters our consciousness, of course, comes from the culture around us. And like the culture, it seems that our minds are split on the subjects of race, gender, class, sexual orientation. "We not only mirror the ambivalence we see in society, but also mirror it in precisely the same way," says Dovidio. Our society talks out loud about justice, equality, and egalitarianism, and most Americans accept these values as their own. At the same time, such equality exists only as an ideal, and that fact is not lost on our unconscious. Images of women as sex objects, footage of African-American criminals on the six o'clock news,--"this is knowledge we cannot escape," explains Banaji. "We didn't choose to know it, but it still affects our behavior." We learn the subtext of our culture's messages early. By five years of age, says Margo Monteith, Ph.D., many children have definite and entrenched stereotypes about blacks, women, and other social groups. Adds Monteith, professor of psychology at the University of Kentucky: "Children don't have a choice about accepting or rejecting these conceptions, since they're acquired well before they have the cognitive abilities or experiences to form their own beliefs." And no matter how progressive the parents, they must compete with all the forces that would promote and perpetuate these stereotypes: peer pressure, mass media, the actual balance of power in society. In fact, prejudice may be as much a result as a cause of this imbalance. We create stereotypes-African-Americans are lazy, women are emotional-to explain why things are the way they are. As Dovidio notes, "Stereotypes don't have to be true to serve a purpose."


The idea of unconscious bias does clear up some nettlesome contradictions. "It accounts for a lot of people's ambivalence toward others who are different, a lot of their inconsistencies in behavior," says Dovidio. "It helps explain how good people can do bad things." But it also prompts some uncomfortable realizations. Because our conscious and unconscious beliefs may be very different-and because behavior often follows the lead of the latter-"good intentions aren't enough," as John Bargh puts it. In fact, he believes that they count for very little. "I don't think free will exists," he says, bluntly-because what feels like the exercise of free will may be only the application of unconscious assumptions. Not only may we be unable to control our biased responses, we may not even be aware that we have them. "We have to rely on our memories and our awareness of what we're doing to have a connection to reality," says Bargh. "But when it comes to automatic processing, those cues can be deceptive." Likewise, we can't always be sure how biased others are. "We all have this belief that the important thing about prejudice is the external expression of it," says Banaji. "That's going to be hard to give up." One thing is certain: We can't claim that we've eradicated prejudice just because its outright expression has waned. What's more, the strategies that were so effective in reducing that sort of bias won't work on unconscious beliefs. "What this research is saying is that we are going to have to change dramatically the way we think we can influence people's behavior," says Banaji, "It would be naive to think that exhortation is enough." Exhortation, education, political protest-all of these hammer away at our conscious beliefs while leaving the bedrock below untouched. Banaji notes, however, that one traditional remedy for discrimination-affirmative action-may still be effective since it bypasses our unconsciously compromised judgement.

But some stereotype researchers think that the solution to automatic stereotyping lies in the process itself. Through practice, they say, people can weaken the mental links that connect minorities to negative stereotypes and strengthen the ones that connect them to positive conscious beliefs. Margo Monteith explains how it might work. "Suppose you're at a party and someone tells a racist joke-and you laugh," she says. "Then you realize that you shouldn't have laughed at the joke. You feel guilty and become focused on your thought processes. Also, all sorts of cues become associated with laughing at the racist joke: the person who told the joke, the act of telling jokes, being at a party, drinking." The next time you encounter these cues, "a warning signal of sorts should go off-wait, didn't you mess up in this situation before?- and your responses will be slowed and executed with greater restraint." That slight pause in the processing of a stereotype gives conscious, unprejudiced beliefs a chance to take over. With time, the tendency to prevent automatic stereotyping may itself become automatic. Monteith's research suggests that, given enough motivation, people may be able to teach themselves to inhibit prejudice so well that even their tests of implicit bias come clean.The success of this process of "deautomatization" comes with a few caveats, however. First, even its proponents concede that it works only for people disturbed by the discrepancy between their conscious and unconscious beliefs, since unapologetic racists or sexists have no motivation to change. Second, some studies have shown that attempts to suppress stereotypes may actually cause them to return later, stronger than ever. And finally, the results that Monteith and other researchers have achieved in the laboratory may not stick in the real world, where people must struggle to maintain their commitment to equality under less-than-ideal conditions. Challenging though that task might be, it is not as daunting as the alternative researchers suggest: changing society itself. Bargh, who likens de-automatization to closing the barn door once the horses have escaped, says that "it's clear that the way to get rid of stereotypes is by the roots, by where they come from in the first place." The study of culture may someday tell us where the seeds of prejudice originated; for now, the study of the unconscious shows us just how deeply they're planted.