A properly functioning brain recognizes certain patterns and even makes generalizations about what it observes. But these same brain processes also can lead to overgeneralization and discrimination via “implicit bias,” which describes a prejudice, stereotype, or presumption made about certain groups or populations pre-reflexively, or without conscious knowledge of that bias.1 To better understand the effect of implicit bias in the courtroom, Judge Bernice Donald of the United States Court of Appeals for the Sixth Circuit talked with Professors Jeffrey Rachlinski and Andrew Wistrich of Cornell Law School. Rachlinski holds both a J.D. and a Ph.D. in psychology, and Wistrich previously served as a magistrate judge of the United States District Court for the Central District of California; the two professors have published several studies on judicial behavior. Their discussion follows.
Our research suggests that the influence of unconscious bias on judges is subtle. We know that judges harbor many of the same implicit associations as most adults. For example, in our study using the implicit association test, we found that 80 percent of white judges more strongly associated Black faces with negative words, and white faces with positive words.2 Black judges expressed a more complex pattern, with some judges showing the same white-good/Black-bad association as white judges, but an equal number showing the opposite preference. These results suggest that judges are no different than most adults in the United States.
What is more important for judges, however, is whether this bias affects their decision-making. Judges take an oath to be impartial and follow a code of ethics that demands that race does not play a role in their decisions. Commitments like these could motivate judges to avoid relying on implicit biases. In fact, we have some evidence that judges sometimes can avoid relying on their implicit associations when making judgments.
In one study, we asked judges to assess a series of hypothetical cases in which we manipulated the race of the defendant. One of the cases described an 18-year-old defendant charged with battery arising from a fight in a high-school locker room. The defendant alleged self-defense, although his claim was weak. We gave judges two versions of this case: one in which the defendant was white and the victim was Black and one in which the races were reversed. The problem was first used by Samuel R. Sommers and Phoebe C. Ellsworth, who found that white lay adults given the problem were more likely to convict the Black defendant than the white defendant (90 percent versus 70 percent).3 White judges in our study, however, expressed no difference in conviction rates.
The same judges, however, showed an influence of implicit bias in their judgment of juvenile offenders. In this part of the study, we used a subliminal priming technique and contextual cues to suggest that the defendants were either white or Black, rather than merely identifying their race outright. We found that implicit biases correlated with judgments. Judges who harbored a strong white-good/Black-bad association imposed harsher penalties on the defendant we had suggested was Black. In effect, when we overtly identified race, judges did not treat Black and white defendants differently. When we subtly suggested the defendant’s race, however, the judges’ implicit associations influenced their judgment. We concluded that judges were on guard when race was mentioned overtly. We encourage judges to maintain that kind of vigilance in their courtrooms as well. We believe that many already do.
Other studies have expanded on our initial findings. Justin D. Levinson, Mark W. Bennett, and Koichi Hioki showed that judges harbor invidious implicit associations concerning Asian Americans and Jews.4 As in our study, however, the authors found that judges did not act on these associations in making judgments when the ethnic background of the parties was salient. We also collected data on race in sentencing decisions in hypothetical cases and found that race had no effect. Judges harbor the same implicit associations as most adults, but do not seem to act on them when race is made explicit.
In other research, we have found that implicit gender associations affect judges as well.5 Most adults more easily associate males with career concepts and females with domestic concepts. Using a hypothetical case in which we varied the gender of the parties, we found that judges were more apt to grant a request from a mother to alter child custody arrangements than when the identical request came from a father. Judges, it seemed, felt more deferential to women’s preferences as parents, perhaps owing to the association most adults have with women and domesticity. We also found that judges sentenced a female criminal defendant less harshly than an identical male defendant. The latter finding is consistent with evidence that people more strongly associate men with violence than women. Although this association is obviously rooted in the reality that men commit far more violent crime than women, the result is still unequal treatment of identically situated defendants.
“Racist” is the wrong term for someone who harbors invidious implicit associations. Even Nelson Mandela reported having negative implicit associations with Black Africans as a result of decades of Apartheid. Individuals who embrace egalitarian norms can nevertheless harbor negative implicit associations that parallel invidious stereotypes about race and gender. “Implicit bias” is perhaps also a troubling term as it is increasingly used as a synonym for racism. The key findings in the social psychological literature concern how closely people associate races or genders with concepts and how that can influence rapid judgments. That is not the same as the kind of racial animus that leads people to join the Ku Klux Klan or refuse to associate with people because of their race. Bigotry, racial animus, invidious beliefs about women, and the like are not the same as implicit associations (though doubtless such people with such explicit beliefs also harbor the same implicit associations).
Researchers who study implicit bias often use the old example of the Stroop effect to illustrate their basic point. The Stroop effect refers to the delay in recognition created by the pairing of congruent and incongruent stimuli.6 For example, when the words are printed in the same color that they denote (e.g., the word “red” printed in the color red), reading the word is easy. When the words are printed in incongruent colors, however, (e.g., the word “red” printed in the color green) the task is difficult. The result obviously does not depend upon overt animosity towards various colors. In the same way, if a lifetime of exposure to stereotypic associations in the media has led a person to think of Asian Americans as reserved, mathematically oriented people, that person will have difficulty thinking of an Asian American as a talkative, extraverted litigator. That association is hardly the same as an explicit unwillingness to hire an Asian American as a litigator.
This is an important point. We observe enormous disparities in outcomes in the criminal justice system. Blacks comprise about 13 percent of the general population,7 but about 38 percent of the prison population.8 Hispanics are likewise overrepresented in prisons. If our research shows that judges are able to suppress implicit biases, how is it we continue to see such wide disparities?
We have to consider that our research might not be perfectly externally valid. Judges might be on guard in the educational settings in which we collect our data in ways that they are not in the courtroom. Although this is a possibility, we think that, if anything, judges are actually more apt to be on guard in a real case. Trials and hearings are public events in which judges are aware that they are being watched. We think the answer to this puzzle requires considering the other steps in the criminal justice process.
We believe a more plausible explanation for the gap between our research results and the actual disparities arise from other sources of bias. Implicit bias can play a role at every stage of the process, from the first encounter a suspect has with the police through criminal sentencing. Police might be more inclined to arrest Black suspects and prosecutors might be more apt to pursue cases against Black defendants. Furthermore, judges might be given different information about Black defendants than white defendants. With disparities at every stage, the effect of implicit bias can snowball.
We do not mean to exonerate judges completely. As we note below, some evidence suggests that they do impose disparate sentences by race, notwithstanding our research. Also, judges are responsible for monitoring prosecutors, police, probation officers, and others who might themselves be expressing implicit bias.
The adversary system creates the potential for lawyers to de-bias a judge or jury. Lawyers, however, also face significant practical obstacles that must be overcome in order for that to be an effective solution to implicit bias.
First, it is estimated that about one-third of federal cases have at least one pro se litigant. In state courts the percentage is even higher, with estimates ranging from 50 to 80 percent, depending on the court and type of case. Obviously, a lawyer can only help with de-biasing if he or she participates in a case.
Second, many proceedings are ex parte. Search warrant applications, applications for prejudgment remedies, and so on, only allow presentation of one side of the story. If one side’s presentation triggers reliance on implicit associations, no one is present to counteract it.
Third, even when both sides have lawyers, they are commonly not evenly matched. Although the risks posed by implicit bias have become better known, many lawyers are still not aware of it. And even if both lawyers are aware of it, one may lack the knowledge or skill to mitigate implicit bias.
Fourth, lawyers are partisans. Although most are committed to ensuring that the litigation process is orderly and fair, their principal objective is to win. This is not a criticism or a slur. Our adversary system is designed that way. Therefore, even if both sides have retained lawyers who are aware of implicit bias and understand how to combat it, only one side may be motivated to reduce it in any particular case. The other side might believe that implicit bias would improve its odds of success, so — without necessarily encouraging it or exacerbating it — the other side might not try to lessen it. Judges, however, can communicate to lawyers that they expect them to cooperate to reduce bias as part of their obligations as officers of the court.
Fifth, lawyers exhibit implicit bias themselves. One study showed that law firm partners gave an identical legal memorandum a lower rating, and found more of the errors embedded within it, when they thought the memorandum had been authored by a Black associate as opposed to a white associate.9 Another study revealed that even highly idealistic egalitarians, such as death penalty defense lawyers, also exhibited implicit bias.10 Yet another study revealed that Nevada lawyers responding to a judicial evaluation survey rated female judges lower than male judges, even after controlling for objective measures of their qualifications and performance on the bench.11 Therefore, lawyers might be part of the problem rather than part of the solution.
In sum, although the adversary system might allow lawyers to help minimize the risk of implicit bias in the courtroom, it is unclear how frequently or effectively they will do so. This means that judges may need to be proactive in attempting to develop courtroom de-biasing techniques.
Our research suggests that judges have a difficult time accepting their weaknesses, especially when it comes to implicit bias. In one of our studies, 88 percent of judges rated themselves as better than their median colleague at avoiding reversal, thereby expressing a notable sense of overconfidence.12 When it comes to avoiding racial bias in decision-making, 97 percent rated themselves as better than the median judge.13 Although judges recognize that they are human and therefore imperfect, some imperfections are difficult to detect and accept. Not only are judges typically successful mid-career professionals who have been selected for a prestigious position, but they also take seriously their commitment to fairness and impartiality. Many may find it hard to believe that they could be failing to live up to their oath and professional norms. The key is training judges actively rather than passively, so they receive feedback on their own performance that is hard to ignore.
There is plenty of evidence that judges are being influenced by litigant race and gender beyond just the experimental studies we have conducted with hypothetical questions. As an example, studies show that Black defendants receive longer sentences and female defendants receive shorter sentences.14 These results have persisted for decades. Of course, sentencing data can be noisy in the sense that others — probation officers, prosecutors, etc. — are involved in setting the stage for judges’ decisions. These results, however, dovetail with our experiments in which such factors are controlled.
Our research shows that judges can counteract racial bias, and, as we noted above, sometimes they do. Many judges are alert to the danger of bias in the courtroom and work to neutralize it. Some types of implicit bias, however, such as those based on age, skin tone, height, weight, citizenship, etc., also have an influence on judges.15 We worry that even judges who are sensitive to racial inequity might overlook some of these other sources of unfairness.
Of course, the suspicion that judges are influenced by race or gender bias is profoundly disillusioning and dispiriting for a society that rightly demands equality in the courtroom. Disparities in the administration of justice by a judge are particularly hurtful for racial or ethnic minorities and for women, perhaps particularly so when they turn to the courts for justice and redress for the effects of prejudice in the broader society. Acknowledging the imperfections of the judiciary can be painful for judges — especially those subject to reselection — and can give rise to public criticism and even cynicism. Nevertheless, Jerome Frank was right when he said that hiding the flaws of courts and judges is not the road to minimizing or eliminating them.16
Judges can, and should, do a great deal. Judges have a responsibility to educate themselves about anything that might contribute to unfair or biased outcomes in their cases, and to take action to ensure that their decisions are unbiased and just.
Such measures might include calling out lawyers who exhibit disrespect for others in the courtroom, striking remarks or arguments that display prejudice or rely upon unfair or misleading stereotypes, and making it clear that ethnic bias will not be tolerated, even in non-public settings like private caucusing during settlement conferences. Reminding lawyers in local rules or case management orders of their professional responsibility to avoid ethnic or gender bias or harassment pursuant to ABA Model Rule of Professional Conduct 8.4(g) or similar state bar legal ethics rules may help to set the appropriate tone. Judges also should assiduously police jury voir dire, a setting in which, even under the bias-conscious framework Batson v. Kentucky and its progeny, race and gender biases in jury selection have been repeatedly demonstrated. Judges should also take advantage of opportunities to strengthen the justice system by increasing diversity. As an example, a judge presiding over a class action could insist that lawyers appointed to serve as lead counsel for the class reflect the gender and ethnic diversity of the class members. Of course, judges should reject explicit invocation of bias and stereotypes in any setting.
Judges wear many hats, and one of these hats has always been serving as a mentor to lawyers. Judges can encourage, and participate in, bar association programs about implicit bias. This will raise awareness of the problem within the bar and help lawyers devise strategies to combat it. Judges also can encourage training for lawyers, probation officers, prosecutors, police, and others. These important professionals provide the facts, arguments, and recommendations that influence judges’ rulings. If the inputs judges receive are tainted by implicit bias, there is a heightened risk that the quality and fairness of their rulings may inadvertently be compromised. Garbage in, garbage out, in a sense.
To begin with, we need to make sure that our expectations are realistic. Implicit bias is the product of deep acculturation. It accumulates over the course of a lifetime, beginning as early as age three. It cannot be fixed by an afternoon of training.
Focusing on their own behavior, judges can take two categories of steps.17 The first is to reduce the role of stereotypes and other shortcuts in their decision-making. Reliance on intuition and “going with your gut” can be useful, but only if the technique matches the task. In general, the nature of judging dictates that judges are better off proceeding deliberatively, keeping the facts and the law — rather than impressions and feelings — in the driver’s seat whenever possible. Such measures include:
Avoid hurried rulings. Judges who are forced to rule quickly or feel rushed are more prone to make mistakes or to rely upon potentially misleading shortcuts and stereotypes, or upon the suggestions of others.
Take breaks, rest, and eat. Hunger and fatigue can produce reliance on intuitive judgment that is more easily influenced by implicit bias.
Use checklists and objective criteria. These tools promote structured thinking in which all cognitive and legal bases are considered.
Write opinions. Though time-consuming, the practice is worthwhile. Every judge has experienced the opinion that “won’t write” — a sure sign that a judge’s intuition has led him or her down a path that more mature reflection reveals is inappropriate or unworkable.
Seek feedback. The adversary system may not work perfectly, but that is no excuse for failing to take advantage of what it does offer. Issue tentative rulings and welcome motions for reconsideration. Both present opportunities to make sure our inclinations or rulings are on track, to reflect about them, and to correct them when input and our second look persuades us that we missed the boat.
The second set of steps consists of measures designed to combat implicit biases directly rather than indirectly. These include:
Obtain training about implicit bias. This can enhance awareness, heighten motivation, and suggest practical strategies for minimizing it. Awareness alone is not a complete solution, but it helps.
Remind yourself of your commitment to fairness and impartiality under law. Every judge has days when he or she is tired, frustrated by the shenanigans of lawyers, stressed by overwork, or pressured by the need to rule promptly. That is the time for judges to remember judicial ideals and to redouble their commitment to them.
Promote diversity in chambers and within the court as a whole. Exposure to different sorts of people helps to break down stereotypical thinking. Judges can learn a great deal from judicial colleagues, law clerks, assistants, and others who hail from different backgrounds and lived experiences.
Consider the opposite. When sentencing a defendant or assessing the credibility of a witness who is white, male, or sympathetic, judges should consider how the sentence or witness would seem if the defendant or witness were Black or Latinx, female, or unsympathetic. Research reveals that this technique successfully mitigates a variety of cognitive errors.
Audit judicial performance. Regrettably, most judges receive little systematic, meaningful feedback. Litigant reactions and lawyers’ praise or criticism can be tainted by pleasure or displeasure at the outcome. Appellate rulings often reflect other considerations — such as changes in law or other intervening events — and are usually too delayed to help us perform better day-to-day. But data concerning judicial performance is available. Computerized research services now compile statistics on judges’ rulings in various contexts. Judges can often examine such statistics. If readily available data is lacking, the clerk of the court, or the national or state administrative office, especially in this era of computerized dockets, can compile such data. Where that is not possible, some judges have compiled their own data by building spreadsheets of their sentencings to identify patterns that may not measure up to their ideals of impartiality.
Of course, all of these measures must be applied with common sense. Judges sometimes have to rule instantly, do not always have time to write opinions, and cannot limit their workloads. Sometimes one simply has to be satisfied with doing the best that circumstances permit. On the other hand, judges have more control over their environments and the timing of their decisions than most decision-makers. That advantage should not be overlooked.