The psychology department at University of Washington offers an excellent service that is a very welcome by-product of the digital age—creating electronic documents from old books. Tony recently requested an electronic version of a chapter from a long-cherished forty-year-old book. When the electronic file arrived, he discovered that each of the even-numbered pages had lost the last letter or two from the right-hand side of every line of text. Tony drew the problem to the attention of the department administrator to request a correction. It came back later that day—same problem. He sent it back once again, expressing some consternation that the problem was still there. Surely this was something that could be easily fixed! When it returned the third time, all the pages had been properly reproduced.
Curious about what had happened, Tony stopped at the administrator’s office to find out what the problem had been. The answer: An “intelligent” feature of the copier had “thought” that the margins of all the pages of the chapter should have the same copiable area. However, because the electronic version was made from a book in which margins of even-numbered and odd-numbered pages were different, the text of all even-numbered pages was positioned more to the right than that of the odd-numbered pages. The eventual fix: Create a paper copy on which even-numbered and odd-numbered pages had similar margins and then use that as the basis for the electronic version.
The administrator who figured this out made a telling observation: “The trick is to outsmart the machine.” This idea of outsmarting the machine captures the essence of what we would like to be able to do to meet a much bigger challenge: enabling the human brain to outsmart the mindbugs that reside within it.1 Can we use our reflective, analytic minds to devise techniques that will allow us to override unintended results of our automatic, reflexive patterns of thought? In other words, can we outsmart the machinery of our own hidden biases?
The Musicians’ Guide to Outsmarting a Gender Mindbug
In 1970, fewer than 10 percent of the instrumentalists in America’s major symphony orchestras were women, and women made up less than 20 percent of new orchestra hires. Not many considered this a cause for concern. As most people understood it, nature had favored men more than women with whatever it is that makes for musical virtuosity. The evidence for this had to be immediately obvious to anyone who could reel off the names of the world’s top instrumental virtuosos—Sviatoslav Richter, David Oistrakh, Vladimir Horowitz, Pablo Casals, Mstislav Rostropovich, Yehudi Menuhin, Jascha Heifetz, Sergei Rachmaninoff, Glenn Gould, and Fritz Kreisler—all men.2
Or maybe not. Another way of looking at male dominance in instrumental virtuosity is that it is less a gift of nature to men than a gift of culture that recognizes, encourages, and promotes male talent. But this thought had occurred to very few until about forty years ago, when feminism became a cultural force in the United States and allowed such disparities to be noticed and questioned. Was the Boston Symphony really selecting the best talent when it sought to recruit the best available trombonist? If not, how could the symphony determine whether a blindspot was preventing its selection of the best musicians?
By long tradition, orchestra applicants competed for positions by performing before an audition committee—experts selected mostly from the orchestra’s musicians. Starting in the 1970s, several major American symphony orchestras experimented with a new procedure that involved interposing a screen between the auditioning instrumentalists and the committee, leaving the applicants audible but not visible to the judges. The adoption of this procedure was not prompted by any suspicion that women were being discriminated against. Rather, it was suspected that selections might be biased in favor of the students (almost all of whom happened to be male, of course) of a relatively small group of renowned teachers.
The next twenty years provided interesting evidence. After the adoption of blind auditions, the proportion of women hired by major symphony orchestras doubled—from 20 percent to 40 percent. In retrospect it is easy to see that a virtuoso = male stereotype was an invalid but potent mindbug, undermining the orchestra’s ability to select the most talented musicians. Two things stand out about the introduction of blind auditions for orchestra hiring. First is that they did the experiment at all; few experts are able to have sufficient distrust of their own abilities to actually put themselves to a test. Second, the fix was simple and cheap—a piece of cloth. Outsmarting this particular mindbug required awareness, a desire to improve, and a method for improving. It did not need to be complicated or costly.3
The Search for More Mindbug-Outsmarting Solutions
There is no doubt that the blind audition was an inspired and effective solution for outsmarting the virtuoso = male stereotype mindbug in orchestra hiring. So let’s consider similar strategies that could work in other situations. For example, high school and college teachers can often grade students’ essays without knowing the student author’s identity, and any kind of test that is done in written form can be graded without the grader knowing who is being graded.
Unfortunately, the blinding strategy is not possible in most work situations that involve hiring decisions or performance evaluations. Job applicants provide résumés from which names might be removed, but much other information that might reveal gender information would of necessity remain. For example, knowing that an applicant captained a women’s soccer team gives it away, as will the pronouns found repeatedly in letters of recommendation even after obscuring first names.
And, of course, it would be very difficult if not impossible to forgo personal interviews with applicants in the great majority of hiring situations. When it comes to evaluating the work of employees, there is typically no way to prevent supervisors from knowing the identities of those they must evaluate. Likewise, the blinding strategy can’t be extended to most other situations in which mindbugs can produce discrimination, such as in health care, criminal justice, and the search for housing, all of which generally involve face-to-face interactions. Doctors cannot treat patients without seeing them, nor can judges and juries be expected to decide about anonymous, invisible defendants. And although, in a digital world, one can learn a great deal about available apartments or homes without a personal visit, ultimately one cannot rent or purchase without being fully identified to the seller or renter. Technology has probably made obtaining demographic information about almost anyone easier now than ever before. Electronic searches now make it almost trivially easy to obtain personal information when one has a person’s name.
Can Mindbugs Be Exterminated?
Instead of trying to outsmart mindbugs, why not deal with them the way computer programmers deal with computer program bugs—just identify them and get rid of them? Unfortunately, effective methods for removing the mindbugs that contribute to hidden biases have yet to be convincingly established.
We’ve experienced a roller-coaster ride in our own understanding of the ease or difficulty of destroying mindbugs. When the IAT was new in the late 1990s, Mahzarin and Tony took some of the tests numerous times. As disappointing as it was to discover that the tests revealed associations that we preferred not to have, it was even more disappointing to observe that our results for these tests changed little over time, as we took them repeatedly. Awareness of the hidden biases did not seem to help us to eradicate them. At least, we could not see shifts in our continued taking of the test. Those early experiences with the IAT led us to think that the kinds of mindbugs measured by the IAT might be resistant to change.
At about the same time that we were becoming discouraged about eradicating our own hidden biases, other researchers were starting to experiment with procedures that they had devised with the aim of weakening, possibly even eliminating, IAT-measured mindbugs. One of the first of these researchers was Nilanjana (Buju) Dasgupta, who worked with both of us during the early stages of her career in the late 1990s.
Buju devised an innovative laboratory experiment using the face images of twenty well-known Americans. The experiment investigated whether results of the Race IAT would show reduced automatic White preference for subjects who started the experiment with a task that involved exposure to images of ten famous, highly esteemed Black Americans—among them Martin Luther King Jr., Colin Powell, Michael Jordan, Denzel Washington, and Bill Cosby. Given two alternative descriptions of each person, both quite positive but one correct and one incorrect, Buju’s subjects were asked to select the correct identification for each. For example, Colin Powell’s photo required a choice between “former chairman of the Joint Chiefs of Staff for the U.S. Department of Defense” (correct) and “U.S. ambassador to the United Nations.” Interspersed with the ten Black Americans’ pictures were those of ten infamous White Americans, including serial murderers Jeffrey Dahmer, Ted Bundy, Charles Manson, and Ted Kaczynski, and mass murderer Timothy McVeigh. The ten White images were also accompanied by a choice between two labels, both extremely negative. For Kaczynski, for example, the choice was between “the Unabomber who injured and killed using letter bombs” (correct) and “convicted pedophile.” After choosing a description for all twenty pictures twice—to ensure good exposure to each—subjects completed the Race IAT.
Having become skeptical about the possibility for change on the basis of observing the intractability of his own IAT scores, Tony was doubtful enough about the prospects for success of Buju’s procedure that he put $5 on the line—which he promptly lost, but with no regrets. The student research subjects who viewed the pictures of ten admirable Black Americans and ten despicable White Americans did indeed show weaker IAT-measured White = good associations than did those who had completed a comparison procedure involving an initial exposure to either admirable White Americans or pictures of flowers and insects.
Buju followed this experiment with a parallel one in which subjects viewed ten admirable elderly people, including Mother Teresa, Walter Cronkite, Eleanor Roosevelt, and Albert Einstein. The result was similar. Those who saw the admirable elders showed weaker IAT-measured young = good associations than did subjects who instead saw pictures of admirable young people. These results immediately changed our thinking about the malleability of IAT-measured hidden biases and about the prospects for exterminating mindbugs.4
At about the same time that Buju did her experiment, researchers in several other laboratories were similarly finding results that encouraged belief that implicit attitudes or stereotypes could be modified with relatively simple procedures. At the University of Colorado, for example, Irene Blair had college men and women do a brief imagination exercise, asking them to “take a few minutes to imagine what a strong woman is like, why she is considered strong, what she is capable of doing, and what kinds of hobbies and activities she enjoys.” That simple thought-generation task effectively weakened an IAT-measured male = strong stereotype. This was evident in comparison of the experimental subjects’ IAT measures of that association with those of comparison subjects who engaged in alternative mental exercises.5
Among the attractive aspects of being in our science at this time is the opportunity to have one’s mind routinely changed by research evidence. This was one such moment. Our initial belief that it might take long periods of hard work to change existing associations needed revision. The data were showing that hidden-bias mindbugs could be weakened by relatively minimal interventions. Perhaps repeated applications of the types of modest interventions that were used in these laboratory studies would provide a means of eradicating mindbugs.
We are often asked what steps we have taken to reduce our own hidden biases. Mahzarin came up with one. Inspired by the accumulation of results indicating malleability, and hoping that it would dislodge some of her own mindbugs, she created a screensaver for her computer that displays images of a diverse array of humanity. She assumes that these images may do little more than to keep her alerted to the actual range of diversity in the world, as opposed to that of the more limited set of humans she encounters in her daily experience. She also favored images that represent counterstereotypes. Short bald men who are senior executives is one of her favorite counterstereotyping images. Another is a drawing from a New Yorker magazine cover, of a construction worker with hard hat on, breast-feeding her baby. Her aim is to build up associations counter to the stereotypic ones that are strengthened in the rest of her daily life through observation and media exposures.
More than a decade after Buju’s study, experiments similar to hers continue to produce similar findings. As promising as these consistent results are, we must set a high standard for concluding that our science understands how to dismantle the hidden-bias machinery. Alas, there has not yet been a convincing (to us) demonstration that interventions of the types investigated in research of the last decade will produce durable changes. The experiment that has most successfully shown durable change in the strength of an IAT-measured association is a study of the effects of a three-week therapy program to reduce a spider phobia, which resulted in a phobia reduction measured one month after completion of the program.6
We now suspect that the changes observed in the studies by Dasgupta, Blair, and others were ones that we should understand as elastic changes. Like stretched rubber bands, the associations modified in these experiments likely soon return to their original configuration. Such elastic changes can be consequential, but they will require reapplication prior to each occasion on which one wishes them to be operative. Such suppression of mindbugs can be useful, but it is not the equivalent of eradication.
There are some other encouraging findings. Buju Dasgupta found a strengthening of female = leader and female = math associations in women college students after they had received sustained exposure via their college courses to women faculty members. Women who had taken more courses from women faculty showed the greatest weakening of these stereotypes. In an as yet unpublished study conducted at the University of Wisconsin, Patricia Devine tested a multiple-component training procedure that was aimed at reducing IAT-measured White racial preference. She found reductions that were observed up to six weeks after the training procedure. This promising work notwithstanding, we have come to regard mindbugs as dauntingly persistent.
At least for the time being, it appears necessary to find workarounds for mindbugs—so we come back to the strategy of trying to outsmart the machine.7 That possibility of outsmarting, rather than eradicating, has served well in many other places, especially in dealing with incurable diseases. For malaria, for example, outsmarting strategies include preventive strategies, such as the use of mosquito nets and the sterilization of disease-bearing mosquitoes; for HIV, outsmarting strategies include the use of condoms to prevent sexual transmission and, perhaps most ingeniously, a variety of antiretroviral drugs.
Outsmarting Mindbugs by Bypassing Them: The No-Brainer Solution
We met Joan A. in Chapter 6. Her physician, Dr. M., had thought that Joan, at age forty-eight, had no reason for concern about her cardiovascular health. His judgment was based partly on valid medical knowledge that middle-aged women are less at risk for heart disease than are middle-aged men and also on his knowledge of her body weight (low) and her level of athletic activity (high). When at Joan’s insistence he ordered a test of her blood cholesterol levels, Dr. M. discovered that—contrary to his probabilistic expectations—Joan’s cholesterol levels were high enough to call for some treatment.
The episode with Joan A. and Dr. M. took place about twenty-five years ago, at a time when there was less awareness of women’s risk for heart disease than there is today. It is now understood that, despite women’s risk level being generally lower than men’s, the risk is still significant enough to warrant screening, even if the woman in question is lean and athletic. As a result, the medical profession has adopted a different approach to cholesterol testing. Guidelines promulgated by the National Heart, Lung, and Blood Institute (NHLBI) now advise that everyone over age twenty have blood cholesterol levels checked at least once every five years. Of course, the NHLBI guideline is just that—a guideline, not a requirement. In day-to-day practice, some physicians no doubt still make recommendations for cholesterol testing that are based on the female = low cardiac risk stereotype. But doctors who choose to follow the latest guidelines no longer need to make any judgment about a given middle-aged woman’s likelihood of having a heart attack. The recommendation for periodic testing becomes a no-brainer: by using the guideline, the doctor does not need to take into account any personal knowledge of the relative risks for men and women, eliminating the possibility of inappropriately applying a stereotype.
It is easy to recognize that the female = low cardiac risk stereotype is a mindbug that one would want to outsmart. However, there are other situations in which we may have difficulty persuading ourselves that a stereotype should be outwitted. For example, recently the happy owner of a few American Staffordshire terriers sought Tony’s help in designing an information campaign to overcome a negative stereotype of his pets. You may be puzzled until you learn that these dogs are widely known as pit bulls.
In the United States, pit bulls account for perhaps 1 or 2 percent of all registered dogs but (according to various sources) are responsible for more than 25 percent of reported dog bites and for a similarly high proportion of fatalities that result from dog bites. Seattle, Washington, recently became a focus of attention among pit bull opponents after the launch of an anti-pit-bull website by a Seattle resident who had been attacked by one. Perhaps typical of other cities, in Seattle about half of the dogs who are euthanized in animal shelters are identified as pit bulls. It was to try to counter the stereotypes about these dogs and to prevent such widespread euthanasia that the Staffordshire terrier owner asked Tony for advice.8
Do you have a stereotype of pit bull = vicious? And if so, should you try to change your view of pit bulls and modify your behavior accordingly? We can help you to find your answers by replacing these questions with three others, the first two of which we can answer for you. (1) Are pit bulls, on average, more vicious than other types of dogs? This can get a yes response, meaning that the stereotype has some validity, even though this deserves to be qualified by the claims of pit bull advocates that aggressiveness in pit bulls is often the result either of deliberate training or of inhumane treatment. (2) Are all pit bulls vicious? Certainly not, so this gets a no. Many pit bulls are well-loved pets who play gently and safely with children as well as with other pets, including kittens. (3) The question you’ll have to answer for yourself is this: On first encounter with an unfamiliar pit bull, should you react to it by assuming that it will be vicious?
When you have your answer to that last question, consider the three parallel questions about middle-aged women in relation to the risk of heart disease. (1) Are middle-aged women less likely to have heart attacks than middle-aged men? This needs a yes response—the stereotype has some validity. (2) Are all middle-aged women unlikely to have heart attacks? Certainly not—large numbers of middle-aged women have heart attacks. (3) On encountering a middle-aged woman, should a doctor assume that she has no risk for heart disease? As we already know, the answer to the third question, now codified in NHLBI guidelines that call for periodic cholesterol testing for anyone over age twenty, is no.
We are dwelling on this pit bull example because we expect it to be challenging to many readers. Readers may have difficulty discovering any no-brainer solution (parallel to the NHLBI guideline for cholesterol testing) that can allow them to respond the same way to pit bulls as to all other breeds. We expect many readers to resist our suggestion that the two situations, of middle-aged women and pit bulls, are parallel. Many, we expect, will think that there are good reasons to treat pit bulls differently than other breeds.
We intend to leave the challenge of such parallels open for discussion. For our own part, we can argue it both ways. The case for parallel treatment of the two categories is straightforward—both cases reflect operations of stereotypes that are not justifiably treated as characterizing all individual members of the category. The case for nonparallel treatment of the two categories can be made by arguing that the pit bull = vicious association is actually more than a stereotype—rather, it may be an association of such great validity that viciousness should be treated as an essential property of pit bulls, much as poisonous is treated as an essential property of all rattlesnakes. The problem with this argument is that we know with certainty that viciousness is not an essential property of pit bulls—there is no evidence that viciousness characterizes more than some minority of pit bulls.
The best we can do in stating a defensible justification for treating pit bulls differently from other dog breeds is in terms of fear—“I want nothing to do with pit bulls for the very good reason that I’m scared to death when I see one.” Seen this way, one’s response to pit bulls has more the nature of a phobia than a stereotype. The significance of labeling this as a phobia rather than a stereotype is that the cause of aversion is seen to be a property of the person responding to the dog, rather than a property of the dog. However, we expect that pit bull advocates will find the phobia justification to be dubious.
In an alternate world there might be an easily administered test for canine aggressiveness that, when passed, would result in nonaggressive dogs wearing visible indicators of their test result, somewhat like the way doctors wear stethoscopes around their necks as status markers. This alternate world is not totally removed from present reality, because there do exist tests for dogs’ aggressiveness. However, the most valid of those tests require trained personnel and standardized test situations, and at present it is not practical to administer them routinely.9
Applying a guideline to eliminate a yes-or-no judgment from the decision to administer a test works to bypass a mindbug in the situation of Joan A., Dr. M., and cholesterol testing, but it is impractical for dealing with the abstractly parallel situation of pit bull stereotypes. More generally, the successful method of outwitting one mindbug has limited ability to deal with others. More methods are needed.
Outsmarting Mindbugs the Numerical Way
Testing, when done well—as is possible for blood cholesterol or canine aggressiveness—has the desirable property of producing numerical scores. It is easy to see testing as a cure-all. It is tempting to assume that judgments expressed in numbers guarantee objectivity and bias-free treatment. However, by themselves, numbers do not guarantee freedom either from hidden bias mindbugs or from more deliberate bias. This may never have been more apparent than in the figure-skating competition at the 2002 Winter Olympics in Salt Lake City, when numerical scores provided by the nine-judge panel provoked an international scandal.
International figure-skating competitions use a complex scoring system that requires each judge to make many numerical judgments of the components of a skating performance. These judgments are sorted into two basic categories: technical merit and presentation. Long regarded as being susceptible to bias, the scoring system for international figure skating fell into total disrepute after the judges in Salt Lake City awarded the gold medal in the pairs competition to a Russian pair who—as was thought by almost all observers at the skating venue and on television—had clearly been outperformed by a Canadian pair.
Shortly after the competition, a French member of the judging panel was reported to have admitted to colluding with another judge by agreeing to support the Russian pair in exchange for that judge’s support of a French skater in another event. Less than a week after the competition, officials of the International Olympic Committee and the International Skating Union (which presides over international ice skating competitions) decided that a second gold medal should be awarded to the Canadian pair—an action that acknowledged that the numerical system was compromised, but which was also seen by many as an insufficient response.
In the wake of the 2002 scandal, the International Skating Union overhauled its judging system, presumably aiming to achieve greater objectivity. However, after having been in operation for several years, the new system appears to offer no more than a partial improvement. By tying all judge-assigned numerical technical merit scores for performance elements (spins, jumps, etc.) to videotaped evidence of successful completion of the element, the new system came close to eliminating subjectivity—and consequent potential for bias—in the technical component of the score. However, the replacement for the presentation scoring, newly labeled “program components,” remained largely subjective. And, of course, the use of numbers did nothing to prevent vote trading and other deliberate forms of corruption.10
Hidden Biases of Good People: Understanding In-Group Favoritism
The “good people” of this book’s title are people who, along with their other good traits, have no conscious race preferences. But even though they regard themselves as egalitarian, they nevertheless obtain an “automatic White preference” result on the Race IAT. This, as we know, is no small group. Among the more than 1.5 million White Americans who have taken the Race IAT on the Internet, about 40 percent show this pattern of having explicit egalitarian beliefs accompanied by the automatic White preference result of the Race IAT.
Social psychologists Samuel Gaertner and Jack Dovidio have focused a sustained research program on a subset of Americans whom they describe as “aversive racists.” These are White Americans who earnestly describe themselves as egalitarian but, nevertheless, display subtle forms of race discrimination, such as by being more ready to offer help to Whites than to Blacks. The “aversive” piece of the “aversive racism” label captures Gaertner and Dovidio’s assumption that, for these White egalitarians, interracial interactions often provoke anxiety and discomfort, which can prompt avoidance or withdrawal rather than interracial engagement.
Even though we embrace Gaertner and Dovidio’s theoretical understanding of aversive racism, we may want to choose a different label. (Appendix 1 has a fuller treatment of the labeling issue in regard to race.) Our reason here is similar to our reason for concluding that it is unwarranted to attach a “racist” label to the many people who show an automatic White preference on the IAT. A possible alternative label for people who display Gaertner and Dovidio’s aversive racist syndrome is “uncomfortable egalitarians.”
We have some observations about these uncomfortable egalitarians. First, there are a lot of them, and we are entirely serious in including ourselves among them. Our present best estimate is that they comprise 40 percent or more of White Americans and Asian Americans, perhaps a bit smaller fraction of Latino Americans, and a substantially smaller but not at all negligible proportion of African Americans.
Second, their differential behavior toward White and Black Americans can well be responsible for a substantial portion of the disadvantage experienced by Black Americans (Appendix 2 has more detail on this). Third—and perhaps most needing explanation—is that uncomfortable egalitarians are extremely unlikely to notice that their differential behavior toward Whites and Blacks contributes in any way to the disadvantages experienced by Black Americans.
It may seem a long leap from the story of the doctor who took care of Carla’s hand in Chapter 7 to the understanding of uncomfortable egalitarians. However, the connection is actually close and provides insight into why uncomfortable egalitarians have no awareness that they are doing anything discriminatory. Uncomfortable egalitarians may be the prototypical “good people” who have hidden biases. They see themselves as helpful, but it turns out that their helpfulness is selective, caused in part by their discomfort in interracial interactions. Their discriminatory behavior consists of being selectively ready or able to help only or mostly those who are like them, those in their circle of friends and acquaintances—in other words, those in the groups for which they have automatic preferences.
In telling the story of Carla’s hand in Chapter 7, we mentioned that the doctor—whose behavior was altered when he discovered an in-group connection to Carla—could not possibly have regarded his subsequent extra-helpful behavior as contributing in any way to age, gender, race, or social class disparities in health care. He would have been aware only that he was going the “extra mile” to help a patient—clearly a good thing of the type that doctors do all the time. Like the doctor, uncomfortable egalitarians will remain unaware that their greater comfort and helpfulness in interactions with in-group members is not matched by similar levels of comfort and helpfulness toward out-group members.
Readers may wonder why we label the differentially directed helping of uncomfortable egalitarians as discrimination. How can we justify calling the act of helping an in-group member an act of discrimination? The answer requires distinguishing among a few different types of in-groups.
If you are a parent, no one (certainly not us) will accuse you of discriminating when you help your children by giving them meals, clothing, a secure home, and even some help with their schoolwork, even though you provide none of that help to the neighbors’ children. Likewise, no one will accuse you of discriminating if you donate a kidney to a sibling rather than to a stranger.
Now put yourself in the role of a hiring manager who gives a job to someone who is a college classmate or a friend from church. In doing this, you will likely reject other applications, possibly including that of a more qualified person whom you don’t know personally and who may be of a different nationality, religion, race, or ethnicity. Legally, that is discrimination. If you are a doctor serving on a hospital transplant committee, you would likewise be seen as discriminating if, on the list of candidate transplant recipients, you place someone of your own religion ahead of a more qualified recipient who happens to be of a different religion. If you are a school administrator, you would be seen as discriminating if you promote a teacher of your own race while not promoting a teacher of another race who has an equivalent or superior performance record.
The examples in the last two paragraphs were the easy ones. More challenging are situations falling between those that are obviously not discrimination (such as parental helpfulness toward children) and those that obviously are (such as nepotistic hiring). Consider the following example of a perfectly noble action that can have the unintended effect of increasing an existing societal advantage of your own group.
Suppose that you are a White American and contribute money to worthwhile charitable organizations that primarily serve needy people who happen to be primarily White Americans. Without your consciously planning it, your gifts are adding to the advantages of an already advantaged demographic category. Even though you are certainly not violating any civil rights law, your actions will contribute some to the relative advantages of White Americans and, thereby, to the relative disadvantages of others. To the extent that many others act in the same way that you do, the cumulative effects can be very substantial.
Another challenging example: Suppose that you are a bank manager and inform a fellow manager that one of the loan applications being received by the bank is from a personal friend, a relative, or a former school classmate, and that you would appreciate its receiving “careful attention.” Or suppose that you are a university faculty member and are asked to write a letter of recommendation to the university’s admissions office on behalf of a friend’s son or daughter. (We have been asked to do that.)
These situations are ones that divide the world into those who have connections—by virtue of money, social class, profession, or some other elite identification that is unequally distributed—and those who don’t. When you use those connections, your good actions can have the side effect of increasing the relative advantages of those to whom you are connected.
The transforming effect that our research has had on our own understanding of discrimination has gradually brought us to the point of believing that selective helping of the type we are now describing—helping that is nicely captured by a familiar term, in-group favoritism—may be the largest contributing factor to the relative disadvantages experienced by Black Americans. Yes, some people do actively intend harm to Black Americans, and their harmful actions do indeed contribute substantially to Black disadvantage. But we also know that these overtly prejudiced people are only a minority of Americans, quite likely a small minority.
In a society in which the ability to help falls more to Whites than to Blacks and in which in-group favoritism is the norm, ordinary acts of helping will necessarily contribute to White advantage. The cumulative effects of in-group favoritism are what sociologist Robert K. Merton had in mind when he described the “Matthew effect” as being the result of acts by which “the rich get richer at a rate that makes the poor become relatively poorer.”11
If you are an egalitarian who wishes to be helpful in a way that contributes to a level societal playing field, what can you do to avoid the Matthew effect? Here’s an unusual example of just such an action. After a good friend of ours heard us explain how acts of altruism may have unintended effects of increasing the existing advantage of those who are already relatively well off, she pondered the implications of a sizable monetary donation that she was about to make to her college. Although by long tradition her college was racially integrated and nondiscriminating, it was also true that it had substantially more White than Black students, so her gift ultimately would have the effect of increasing White advantage—which was not at all her intention.
At first she asked herself if, under the circumstances, she should make the gift at all. She resolved the dilemma imaginatively (even if expensively) by carrying through on her intention to donate to her college while also contributing an equal amount to the United Negro College Fund. Interestingly, she neither outsmarted nor eliminated the mindbug that prompted in-group favoritism—she merely neutralized it.
Can Self-Undermining Mindbugs Be Undone?
Most women do not endorse the stereotype that men are naturally better at having careers or that they have superior abilities in science, mathematics, or music. Because women do not endorse these stereotypes, the gender-career IAT (which we hope you tried in Chapter 6) has been an eye-opener for many women. Upon trying it, they may discover that they have automatic associations of exactly this type—strong female = family and male = career associations. One might think that a male = career association would affect women’s careers primarily because men in powerful positions may draw on that association to obstruct women’s progress in outside-the-home careers. But that’s not the only possible influence, and it may not be the most important influence.
Automatic gender stereotypes have recently been found to have the possibility of adversely affecting women’s careers when they are self-directed. Women are at risk of applying automatic gender stereotypes to themselves. Female = family and male = career mindbugs can constitute a subtle and sustained influence on women by becoming an unrecognized source of discomfort in their pursuit of full-time careers or in their nonpursuit of child rearing.
Other possibly self-undermining mindbugs include the old = infirm stereotype that has been implicated as a factor impairing the health of elderly people, and race stereotypes that associate Asians or Whites more than Blacks and Latinos with academic achievement. (We described these in Chapter 6.) The latter set of stereotypes is widely believed to be an influence that guides Blacks and Latinos away from academic pursuits.12
We do not yet know how to go about either eliminating or outsmarting self-directed mindbugs. However, they may prove modifiable by exposure to role models—this was found in Dasgupta’s study of women college students whose male = math stereotype was weakened when they took math courses taught by female faculty members. At the University of Washington, our faculty colleague Sapna Cheryan has demonstrated situational interventions that (at least temporarily) strengthen women’s associations of female gender with the possibility of computer science careers. Her method is surprisingly simple—it involves adding typically feminine decor to computer science classrooms.13
Mass media are potentially rich sources of counterstereotypic role models. Media exposures to attractive, healthy, and vital elders, for example, may help counter mindbugs triggered by automatic old = infirm associations. This reasoning could be the inspiration for the frequent appearance of attractive, sixtyish male and female movie stars on the cover of the American Association of Retired Persons magazine in recent years. A quarter century of national media exposure to Oprah Winfrey, followed by Barack Obama’s election as president of the United States, may have occupied enough American media space to be contributing to alterations of African Americans’ stereotypes of their own race.
That said, there is no reason to doubt that the mindbugs we direct toward ourselves are every bit as durable as those we direct toward others. The challenge of overcoming self-undermining stereotype mindbugs brought to mind an eighteenth-century poem by the Scottish writer Robert Burns, whose inspiration was a different kind of bug. After watching a head louse wandering through the folds of a bonnet atop the head of a well-to-do lady seated in front of him in church, Burns penned “To a Louse,” ending with the well-known verses in which he imagined what would happen if the fine lady could see herself as he was seeing her or, more generally, what would happen if we all could see ourselves as we appeared to others.
O wad some Pow’r the giftie gie us
To see oursels as ithers see us!
It wad frae mony a blunder free us,
An’ foolish notion!
What airs in dress an’ gait wad lea’e us,
An’ ev’n devotion!
As it happens, the mindbugs we have been writing about, when they are stereotypes of our own groups, do cause us to see ourselves as others see us—even if this seeing happens without our being aware of it. When it takes the form of applying societal stereotypes to ourselves, “seeing ourselves as others see us” is remarkable in that it shows no signs of bringing the benefits that Burns assumed should result (freedom from blunders, foolish notions, and airs in dress and gait).
Recognition of the potential damage from self-applying cultural stereotypes of our groups to ourselves provided us the inspiration to rewrite Burns’s lines. Our rewrite (in modern English rather than Burns’s Scottish) proclaims the value of not seeing ourselves as others see us.
To a Mindbug
Oh would some power deign to free us
From seeing ourselves as others see us!
What errors in dress and gait would leave us,
And limitation;
What aims and plans might come to please us,
And aspiration!
Mindbugs, Blindspots, Machines, and Good People
We began this book by introducing the idea of a mindbug, which we borrowed from computer scientist Kurt VanLehn’s name for mental arithmetic habits that could malfunction when applied in situations they weren’t intended for. We extended VanLehn’s notion far beyond arithmetic to include a wide array of mental automatisms that could cause undesired and unintended results. Although the automatisms that produce mindbugs function well in their most appropriate circumstances, they can fail when used in tasks that require thoughtful (conscious) attention.
In our preface, we used the retinal blind spot as a metaphor that likened hidden biases to a very ordinary gap in vision. An even better-known blind spot provides another useful metaphor. This is one that is well known to automobile drivers—the gap in side-view mirrors’ coverage of areas on either side of the car. A new device that may soon be widely available uses radar that can signal when approaching vehicles are in the side-view mirrors’ blind spots. When the sensor detects something, the driver is alerted to enter a state of heightened caution to avoid accidental collision.
As a means of sensing mental associations that reside in the blindspot that houses hidden biases, the IAT shares properties with the new automotive device. IAT measures likewise can provide warnings that prompt caution for those who wish to avoid unintended discrimination. Although it is not now possible, it is conceivable that a future descendant of the IAT might operate in real time, looking into the blindspot and generating warnings when an important judgment might be unintentionally contaminated by hidden bias.
At present, the strategies available to avoid unintended discrimination resulting from hidden biases are less easily described than are the methods for avoiding collisions with vehicles that enter side-view mirrors’ blind spots. But there are a few good weapons against hidden biases. The blinding method that worked to dramatically increase women’s success in symphony orchestra auditions is a known successful strategy that remains underused in many circumstances in which it can work. Another underutilized strategy is the “no-brainer” of developing evidence-based guidelines for to eliminate discretion from judgments that might otherwise afford opportunity for hidden-bias mindbugs to operate. When faithfully applied, intelligently developed guidelines will leave little room for hidden biases.
We expect the next several years to produce a steady accumulation of research on methods to eradicate or outsmart mindbugs. Although we (presently) lack optimism about eradicating mindbugs, we are not similarly pessimistic about prospects for research to develop and refine methods for outsmarting mindbugs.
Among the situations for which effective mindbug-outsmarting strategies do not yet exist, most troublesome are the many routine, daily social interactions in which automatic in-group preferences can trigger in-group favoritism that unintentionally puts others (out-group members) at a disadvantage. The “good people” of this book’s title likely include many who want to identify the situations in which hidden bias mindbugs operate and want to be prepared to outsmart them. It is not an easy task, but there is nothing to suggest that it cannot be done.