Proceedings of a Workshop

THE SCIENCE OF IMPLICIT BIAS: IMPLICATIONS FOR LAW AND POLICY

Proceedings of a Workshop—in Brief

INTRODUCTION

On March 22–23, 2021, an ad hoc planning committee under the auspices of the National Academies of Sciences, Engineering, and Medicine’s Committee on Science, Technology, and Law (CSTL) hosted a virtual workshop titled The Science of Implicit Bias: Implications for Law and Policy. Implicit bias has been commonly defined as any unconscious or unacknowledged preferences that can affect a person’s beliefs or behaviors, and in particular, an unconscious favoritism toward or prejudice against people of a certain race, gender, or group that influences one’s own actions or perceptions. The methods for identifying the presence and degree of an individual’s implicit bias, the presence of implicit bias throughout society, and the successes or failures of attempts to mitigate implicit bias are topics of much scientific inquiry, with ramifications for law and policy as well as a multitude of organizational settings. The ways in which implicit bias reflects or contributes to structural and systemic racism in the U.S. remains an open and urgent question. The workshop, organized by the Committee on the Science of Implicit Bias: Implications for Law and Policy, was convened to better understand the state of the science on this topic in the context of critical and ongoing discussions about racism in the U.S. Funding for the workshop was provided by the Ford Foundation.

David S. Tatel, United States Court of Appeals for the District of Columbia Circuit and co-chair of CSTL, welcomed the workshop participants and attendees, emphasizing the importance of the topic of implicit bias in many arenas, including courtrooms. He quoted Michelle Obama, who wrote in her recent memoir, “Let’s invite one another in. Maybe then we can begin to fear less, to make fewer wrong assumptions, to let go of the biases and stereotypes that unnecessarily divide us.”¹

In introducing the workshop, planning committee co-chair Goodwin Liu (Supreme Court of California) noted that there have been great strides in psychology and neuroscience in recent decades, which have provided insight into human cognition, mental processes, visual processing, rational choice, and decision-making. Studies on implicit bias, he said, build on this foundation. He also spoke about his friend and mentor, the late Supreme Court Justice Ruth Bader Ginsberg, who faced explicit discrimination in her life and career, but believed that implicit bias is the next big challenge. Liu mentioned recent unrest, particularly following the killing of George Floyd, and said that we are not beyond explicit bias and old-fashioned racism: “bias is our inheritance of history.” He ended by invoking William Faulkner, who wrote that “the past is never dead. It is not even past.”

Planning committee co-chair Camara Jones (Morehouse School of Medicine) emphasized a need to understand implicit bias in the context of structural racism. She cautioned against invoking implicit bias in an effort to avoid naming and addressing racism. Jones indicated that while the focus of the workshop was to better understand the science underlying implicit bias and to discuss mitigation strategies and approaches to disrupting implicit bias, we cannot forget that implicit bias resides within the larger context of systemic

__________________

racism, which she defined as a system of structuring opportunity and assigning value based on race that unfairly disadvantages some communities and unfairly advantages some individuals in other communities.

OPENING REMARKS

Marcella Nunez-Smith (Yale School of Medicine and Chair of the Biden Administration’s COVID-19 Health Equity Task Force) spoke to workshop attendees in a recorded address that provided a perspective on how implicit bias and discrimination can be mitigated through policy. She discussed the Administration’s approach to ensuring that racial equity was front and center in its COVID-19 response, and said that the Administration uses a data-driven strategy to ensure that resources, including personal protective equipment, testing capacity, and vaccinations, are targeted toward communities most adversely affected by the pandemic. This strategy includes coordination with states as well as outreach to community and faith-based organizations.

Nunez-Smith identified trust as central to the success of public health measures. She noted that, although all groups are showing an increase in willingness to get vaccinated against COVID-19, serious distrust remains, particularly in the Black community. There are historical reasons for this lack of trust, she said, including the Tuskegee experiments² and the treatment of Henrietta Lacks;³ but personal experience also plays a critical role, as there is substantial evidence that Black people often receive sub-par care from medical professionals.⁴ To address some of this deep-seated mistrust, the Administration is partnering with trusted organizations and individuals, including community leaders, in their public health messaging.

Nunez-Smith said that a longer-term priority for the Administration’s COVID-19 response is to address deficiencies in public health infrastructure, including structural inequalities. She noted that the recently enacted American Rescue Plan provides support for community-based health centers and workers. She also emphasized the need for better demographic data about COVID-19, which will allow a better understanding of the impacts of nutrition, housing security, and other factors. She expressed concern that some people remain invisible in the data and believes that better data collection (combined with appropriate data privacy) will help in the identification of high-risk communities.⁵ To build resilience into the pandemic recovery, Nunez-Smith said that we must look at the underlying structural realities.

OVERVIEW OF SCIENCE AND LAW

Jennifer Eberhardt (Stanford University) began her remarks by observing that, when people think of “racial bias,” many think in terms of “old-fashioned racism” and acts such as burning crosses. However, bias is present, she said, in many contexts, including the criminal justice system, in education, and in the workplace. She noted that one of the strongest negative associations is between Black people and crime. Eberhardt described a study in which even subconscious exposure to Black faces made it easier for subjects to identify a blurry image of a weapon.⁶ This association is reinforced by the fact that Black people make up a disproportionate percentage of the prison population and that Black defendants are more likely to get the death penalty than their white counterparts. Even when young Black faces (i.e., of kindergarten age) are presented, they are more easily associated with weapons than white faces. In education, Eberhardt discussed how biases result in teachers and administrators treating Black middle school students more harshly than white students. Furthermore, she said that a Black student may be more harshly disciplined after the misbehavior of other Black students while a white student involved in a similar instance of misbehavior would be evaluated as an individual. In workplaces, Eberhardt described how Asian and Black students often edit their resumes to reduce indicators of their race, as individuals whose resumes appear to be “white” are more likely to get call-backs.⁷ Similarly, she said, for venture capital funds, publicity that includes photographs of Black partners are less likely to receive investment than those featuring white partners.

__________________

Eberhardt listed situational triggers for bias that included: time pressure, subjective standards, lack of accountability, lack of training, lack of positive contact, lack of trust, lack of empathy, threat, fatigue, and cultural norms. She argued that, to address bias, we should build friction into decision making processes to minimize these triggers. For example, the Nextdoor social networking site successfully reduced racial profiling by requiring users to go through a checklist to ensure that they were not reporting suspicious activity based solely on race. The Oakland Police Department decreased the number of vehicle stops by requiring police to answer simple questions about the purpose of the stop.

In looking for ways to build friction into police decisions and interactions, Eberhardt said it was important to understand how they take place. She described a study using footage from police body cameras where it was found that police spoke to Black drivers with different language and with less respect than white drivers.⁸ Such footage is valuable and should be used to increase accountability. In closing, Eberhardt described a conversation with a white man from South Africa who told her that he never knew or thought about Apartheid even though he lived through it. She wondered what types of things we miss about our own society every day.

Eric Holder (Covington & Burling), described some of his personal experiences as a Black man, including being pulled over and stopped and the conversations that take place between Black parents and their sons, across generations, about the police. He noted that implicit bias exists even without prejudice or “racist intent,” but that it can be devastating because judgement and discretion underpin policing and the legal profession. Holder emphasized that bias is important not only in the context of criminal justice, but also in the context of voting rights. He observed that there is a lot of discretion in decision making with regard to voting locations, the placement of new voting machines, voter registration procedures, and the placement of voters on inactive lists. In one study from University of Southern California, state legislators were less likely to respond to emails asking about voter identification requirements sent from a fictional Latino person than from a fictional white person.⁹ Holder noted that while politics and explicit bias certainly impact access to voting, implicit bias also plays a role.

On criminal justice, Holder spoke about his activities as U.S. Attorney General, including a Department of Justice review of criminal justice in America and the Smart on Crime Initiative. The review found that bias (including both explicit and implicit bias) played a key role in the disparate outcomes seen by Black Americans, and the Smart on Crime Initiative¹⁰ suggested measures to address them, including less mandatory sentencing, more alternatives to incarceration, and more discretion to prosecutors in some cases (e.g., to offer lighter sentences or alternatives). Holder also raised the issue of trust, pointing out that people who believe that a system is fair are more likely to respect decisions by authorities, obey the law, and accept outcomes. He closed his remarks by saying that increased national and local awareness and discussion of racism, bias, and racial disparities over the past year made him hopeful that we could turn this historical moment into a movement for increased racial justice.

DISCUSSION

Liu opened the session by quoting Senator Kennedy (R-LA) who asked a question about implicit bias at the recent confirmation hearing of Merrick Garland: “Does this mean that we’re all racists?” Holder responded that everyone has biases, and that we are all more race-conscious than we concede. Eberhardt added that, although we all have biases, we are not destined to act on them. There are ways to intervene. Planning committee member Thomas D. Albright (Salk Institute for Biological Studies) pointed out that biases (or unconscious expectations based on prior beliefs or experiences) are natural and essential for human survival; the challenge comes when expectations or decisions are made based on “false priors.” Jones brought up the issue of accountability, saying that claiming an implicit bias can’t excuse someone from a racist act or for supporting a racist system. Eberhardt agreed and argued that knowledge of implicit biases may increase the responsibility to work to address them. Planning committee member William “Sandy” Darity (Duke University) wondered whether intentionality lies at the heart of racism. Holder pointed out that a lot of recent political activity to suppress the Black vote is not driven by implicit bias, but by intentional acts.

__________________

In a discussion on the role of discretion in the criminal justice system, Eberhardt and Holder agreed that it would be impossible and harmful to remove discretion entirely from the system. Eberhardt identified a need to insert friction into the decision making process, saying that “bias lives in automaticity.” Planning committee member Tanya Hernandez (Fordham University School of Law) raised the idea of checklists (i.e., lists of criteria, principles, or other unchanging conditions) for judges, which can be used to ensure a pause in decision making and more consistent evaluations. Holder agreed that checklists can be helpful. He said that as a judge of the Superior Court of the District of Columbia, he used checklists before sentencing hearings to help ensure fairness. Planning committee member Rayid Ghani (Carnegie Mellon University) and Jones discussed the need for adding friction into the whole system.

Liu mentioned the March 16, 2021 mass shooting of Asian-American women in Atlanta and asked about racism against the Asian community.¹¹ Eberhardt said that there is very little research on the topic, but that hate crimes against Asian-Americans have risen dramatically in the past year. She also pointed to a study that found that health and economic anxiety can increase anti-immigrant sentiment.¹² Holder mentioned that hate crimes against Asian-Americans have generally been a small percentage of overall hate crimes, but that there is reason to believe that such crimes are under-reported. He emphasized the need for community outreach and dialogue.

In response to audience questions, Eberhardt said that in her research, study participants and police officers of all races show similar biases. In terms of future research, she said that it would be useful to look at the intersectionality of bias, for example, bias that arises at the intersection of race and gender.

PANEL 1: THE PREVALENCE AND PERVASIVENESS OF IMPLICIT BIAS – EVIDENCE AND IMPLICATIONS

The session’s first panelist, Kate A. Ratliff (University of Florida), spoke about the Implicit Association Test (IAT),¹³ a widely used tool to measure the strength of mental associations, including implicit racial bias. When used to measure implicit bias, the test measures how quickly test subjects evaluate Black or white faces and make associations (such as “good” or “bad”). A faster response indicates a stronger association. IAT scores reflect a range of implicit biases, including toward people that are white, thin, abled, straight, have lighter skin, and are cisgendered.¹⁴ IAT results indicate that a higher percentage of white test-takers than Black test-takers show implicit bias against Black faces. Ratliff also discussed ways that people talk about implicit bias, including as a latent construct (i.e., a hidden force in people’s minds that can only be inferred by indirect measures),¹⁵ as concept accessibility (i.e., what comes to mind easily),¹⁶ or as a behavioral phenomenon.¹⁷

S. Michael Gaddis (University of California, Los Angeles) spoke about the audit method¹⁸ for measuring implicit bias. Unlike the IAT, this method looks at behaviors and differential treatment and can be applied in different contexts. He contrasted discriminatory behavior in the 1950s and 1960s (when housing developments may have posted explicit “Whites Only” signs) with the more hidden discriminatory practices of today. Gaddis described how audits of real estate purchases and rental applications demonstrate discrimination against Black people whose applications are otherwise equivalent to those of white applicants. He said that housing audits, including large-scale audits conducted by the Department of Housing and Urban Development¹⁹ (as well as similar employment audits)²⁰ have shown little progress since the 1970s. Similar audits can be conducted in a wide array of societal contexts, and have shown bias in

__________________

higher education admissions and other areas.

Patrick Mason (Florida State University), the session’s final panelist, discussed racial discrimination in the labor market. He noted that bias plays a role not only in hiring, but in interactions between managers, workers, and among peers. Small biases, he said, can have large effects, particularly when compounded through multiple agents and interactions, including recruiting, interviewing, hiring, training, supervising, evaluation, starting salary rate, promotion, and layoff decisions. He described a study that found that bias in managers leads to lower productivity in minority workers because they spend less time with them.²¹ This can, in turn, reinforce biases in hiring because the hiring manager only sees lower productivity in the minority group. Mason also referred to studies that seek to reduce discriminatory behavior. One study found that giving teachers the IAT and alerting them to their biases against immigrants reduced discrepancies in their grading between immigrant and non-immigrant students for similar performance, at least in the short term.²² Another study found that an organization’s use of the Rooney Rule, a simple form of affirmative action,²³ increased the quality of the applicant pool and decreased bias.²⁴

DISCUSSION

Session moderator and planning committee member Sheryl Heron (Emory University School of Medicine) asked about the relationship between the IAT and discriminatory behavior. Jones asked a similar question: If we eliminated implicit bias, would we get rid of discriminatory behavior? Ratliff responded by saying that people have implicit biases against Black people because our society is racist, not the other way around. To reduce those implicit biases, she said, people need to be exposed to a different reality, and structural changes are needed rather than individual interventions. Gaddis added that, while our society still wrestles with explicit bias, a better understanding is needed of how and why implicit bias arises. Planning committee member Deena Hayes-Greene (Racial Equity Institute) said that it may be wrong to talk about implicit bias and racism separately, as they are clearly linked. Jones expressed concern that talking about implicit bias rather than racism allows people to remain in a state of inaction that prevents outrage where outrage is needed.

PANEL 2: THE FOUNDATIONS OF IMPLICIT BIAS

Andrew Meltzoff (University of Washington) a developmental psychologist who studies observational social learning in children, opened the session by emphasizing that biases are learned by observing others rather than through explicit lessons. Biases, he said, are “caught, not taught.” He described his studies in which pre-school aged children observe how adults treat each other and then show preference for adults that were treated more favorably.²⁵ These preferences were generalizable to whole groups of people that shared a feature (e.g., the same shirt color). Meltzoff said that biases are generated rapidly, that their formation only requires observation, and that this may explain how biases are transmitted from generation to generation. Meltzoff noted that children are very good social pattern detectors, and suggested that parents be mindful of the fact that, “Our children are watching us.”

Jennifer Kubota (University of Delaware) spoke about what neuroscience tells us about implicit bias. She noted that humans naturally and spontaneously use cues to categorize people. In some of her studies, she used electroencephalogram imaging to show that human brains can process an image of a face after a 200 millisecond exposure

__________________

and process information related to emotion, age, gender, and race.²⁶ Other imaging studies using magnetic resonance imaging showed that Black faces elicit a stronger response in the amygdala, a part of the brain that processes emotion and fear.²⁷ Kubota then spoke about the importance of interracial contact in reducing these differential responses, and showed data that suggested that such contact decreased activation in the amygdala when viewing Black faces.

The session’s final panelist, Keith Payne (University of North Carolina, Chapel Hill), spoke about implicit bias as the cognitive expression of structural racism. He presented data that suggests that individual differences in implicit bias are only weakly correlated with discriminatory behavior,²⁸ and that individual IAT scores vary widely over time.²⁹ In IAT scores from different geographic locations, Payne said, there are stronger correlations that are stable over time: Countries with higher levels of gender bias also show disparities in standardized test results;³⁰ U.S. states with the largest number of Google searches for racial slurs show implicit racial biases as measured by IATs;³¹ and cities that show implicit associations between Black people and guns have a larger disparity in the use of force by police.³² He also described a study that found that counties and states in the U.S. that were the most dependent on slavery in 1860 showed the highest levels of implicit bias today, as measured by IAT scores of individuals in those places.³³ Payne argued that implicit biases reveal more about a person’s context than it does about the individual. In order to mitigate implicit bias, he said, the focus should be on changing systems, structures, and situations.

DISCUSSION

Moderator and planning committee member Stacey Sinclair (Princeton University) asked about the relative importance of individual implicit bias and more systemic, structural factors. All session panelists agreed that structural factors and societal context were important. Sinclair asked, “If we ended structural racism, would we also end implicit bias?” Meltzoff suggested that the biases that children learn are not about race per se, but about power. In the absence of race, they would pick up other indicators. Albright reminded the group that the capacity to develop rapid ways to categorize and evaluate others has a biological basis, and we will not be able to eliminate it. Payne noted that biases change based upon context and over time, and that if the environment changes, the brain will adapt.

Liu asked what people mean when they talk about structural racism. Payne suggested that structural racism is the fact of pre-existing disparities combined with the intergenerational transfer of resources. Planning committee member Diana Dunn (People’s Institute for Survival and Beyond) emphasized that American racism is not simply about the repression of Black people, but the elevation of white people. Jones spoke about the need to dismantle the system of structural racism, not just for Black people but because systemic racism saps energy and resources from the society as a whole. Hernandez reiterated that close, meaningful contacts between races is key to reducing bias and pointed out that systems influence opportunities for meaningful contact.

Meltzoff said that many white parents don’t know how to talk about race. Eberhardt described a new study wherein she found that Black parents talked with their children much more often about race, and that this difference increased after the death of George Floyd. CSTL co-chair David Baltimore asked if schools might present opportunities

__________________

to mitigate implicit bias. Meltzoff responded that, while teachers can play an important role for students in kindergarten through 12th grade, earlier interventions may be needed. Teachers receive very little training on how to have such a conversation, Meltzoff added, and there are very few resources about how to talk about bias in this context.

In response to audience questions, Eberhardt said that artificial intelligence systems, sometimes intended to level the playing field, can soak up information and biases the same way humans do. In response to a question about dealing with backlash against progress on racism, Payne struck an optimistic tone, saying that reforms move the system in the right direction and that progress against structural racism has been made despite backlash.

SECOND DAY WELCOME AND OVERVIEW

David Baltimore provided welcoming remarks in which he emphasized the role of young children, noting that they provide a clear reflection of societal attitudes and behaviors and may hold the key to understanding and mitigating implicit bias. Liu provided a summary of the previous day’s sessions, highlighting that implicit bias is detected in individuals, but its origins and its consequences are societal. Implicit bias represents, he said, the accretion of our history and it is transmitted to our children. He emphasized that implicit bias is not the last frontier of racism, but is more like a canary in a coal mine. It expresses our tacit assumptions and expectations and can become a self-perpetuating reality. Jones placed the previous day’s discussions about implicit bias and structural racism in the context of recent mass shootings in the United States. She suggested that it is misleading to focus on individual motivations and interventions as it is most important to address structural issues that reside at the societal level.

OPENING TALK: CRIMINALIZATION OF BLACK GIRLS IN SCHOOL

Monique W. Morris (Grantmakers for Girls of Color [G4GC]) opened the proceedings by examining the criminalization of Black girls, particularly in the context of schools where Black girls are disproportionately subjected to disciplinary action. A key factor is the adultification³⁴ or age compression of Black girls. She stated that Black girls ages 10-14 are seen as much more adult than their white counterparts, and that this has serious implications for how they are treated and for assumptions about their knowledge and abilities.³⁵ Morris argued that Black girls are considered to be less in need of nurturing and more familiar with adult topics like sex. She said that hyper-sexualization of Black bodies contributes to disciplinary consequences, particularly with regard to the enforcement of school dress codes. Morris linked these biases to historical narratives, tropes, and stereotypes that support these attitudes about Black girls.

Morris spoke about the need to locate and center Black girls in the conversation and to be intentionally disruptive of structures that perpetuate these harmful narratives about Black girls. She said that new structures and spaces based on engagement and participation of Black girls are needed to allow them to bring their full selves to school. Schools, she said, need to be part of the “tapestry of healing” for Black girls.

DISCUSSION

Morris said that oppression of Black girls happens at many levels, including internally as they absorb these harmful narratives. She noted that biases extend even to pre-school aged children, where teachers are more likely to choose disciplinary action over nurturing responses for Black girls than for white girls.

Liu asked if perceptions of Black girls vary based on the level of segregation in a school or by the race of the teachers. Morris responded that the discipline disparity is nearly universal across the U.S. She said that adequate diversity in school and a commitment to addressing racism can increase the capacity to respond to problem behaviors with more empathy and less exclusionary discipline. She said that educators need to develop a “safety” curriculum focused on Black girls’ lived experiences to ensure that they feel safe.

__________________

PANEL 3: THE EFFECTS OF IMPLICIT BIAS IN VARIOUS ENVIRONMENTS

Panel moderator and planning committee member Ivan Fong (3M Company) stated that the purpose of the panel was to better understand manifestations of implicit bias and why it matters.

Rachel Godsil (Rutgers School of Law) spoke about the degree to which implicit bias has consequences, including in the legal system and in workplaces. She emphasized the need to understand in granular detail the many decisions that prosecutors and others in the criminal justice system make, from pre-filing decisions to trial strategy. She pointed out that prosecutors have many interactions with victims, families, witnesses, jurors, and others and that, although bias in these interactions may not directly impact the outcome for the defendant, such interactions impact the level of trust in the system and should be examined. Godsil also discussed the need for detailed examinations of workplace decisions and interactions so as to better understand where implicit bias may have its biggest impacts.

Ian Ayres (Yale Law School) spoke about a study on racial disparity in the rate at which Los Angeles Police Department officers arrested those stopped during stop and frisk encounters. The study found that police were more likely to arrest Black people, even after controlling for neighborhood racial demographics and criminality.³⁶ Although individual police officers varied widely in how likely they were to arrest Black people than white people, Ayres postulated that it might be useful to determine if these behaviors correlated with IAT scores. Ayres also mentioned research into whether teachers in New York show disparities in how they teach Black students relative to white students (as measured by value-added modeling).³⁷ Ayres noted that different types of association mistakes on the IAT can yield the same overall score, with a fast but wrong association scored the same as a correct association that is too slow. He said that these differences may be meaningful.

Jerry Kang (University of California, Los Angeles), the session’s final panelist, emphasized that small associations can have a large impact. While researchers have found very small correlations between IAT scores and biased behavior in meta-analyses of IAT scores, Kang argued that these are meaningful. He noted that these small correlations are statistically similar to correlations found between lead exposure and decreased IQ, taking aspirin and decreasing the risk of heart attack, or smoking and lung cancer. He presented a simple simulation of a law firm wherein lawyers had a small chance of being fired each month. By giving some lawyers (e.g., white lawyers) just a small decrease in the probability of firing and others (e.g., Black lawyers) a small increase, the differential outcome becomes very large over time. Kang also spoke about the need to incorporate behavioral realism into the law by better understanding human behavior and having judges incorporate this understanding into notions of common sense that are already embedded in the law. Kang said that the decision in Washington State vs. Gregory (2018)³⁸ (which abolished the death penalty in Washington) is an example of a court successfully incorporating behavioral realism by looking at how the death penalty had been applied in practice rather than depending on the principles alone.

DISCUSSION

Session moderator Fong asked panelists to expand on the idea of the compounding nature of small biases and associations. The panelists agreed that this concept is important, with Godsil pointing to the cumulative harm that could result from many small, biased interactions that prosecutors may have. Ayres pointed out the effect of small biases when aggregated over groups, such as an entire police force. Both Ayres and Kang cautioned that weaker correlations should be considered carefully to ensure that they are meaningful.

Sinclair asked the panelists what should be prioritized in addressing implicit bias when there are so many small decisions. Godsil responded that people should work to identify where biases matter most in their context and establish guardrails in those places. Kang added that a focus on IAT scores is not helpful, and that we should change policies and procedures where biases might have impacts.

Albright asked about the incorporation of behavioral realism into prosecutorial decisions. In the face of such complexity in these interactions, Liu asked whether we should focus on outcomes when making decisions on interven-

__________________

tions? Godsil responded that it is difficult to achieve desired outcomes if we do not do the work to understand where to intervene. She said that judges and prosecutors are willing to do this work. Kang said that we can directly inject better practices in some areas, and that many law firms are willing to attempt this. He stressed the importance of engaging the legal community by providing evidence for implicit bias and by drawing on values that they already espouse, such as fairness and rigor. Ayres suggested that, to help people see the need for interventions and better practices, the IAT should be described as evidence of disparate impacts rather than personal failings.

In response to a question from the audience about the use of the IAT in hiring, Kang cautioned that it is not a reliable test for that purpose. He said that the IAT is better used in aggregate to act as an early warning system, for example, in police departments.

PANEL 4: DISRUPTING THE IMPACTS OF IMPLICIT BIAS

The session’s first panelist, Calvin K. Lai (Washington University in St. Louis), spoke about efforts to combat discrimination. He described a study in which investigators used a variety of interventions to try to decrease individual IAT scores.³⁹ Only a few interventions worked, including those that used counter-stereotypical individuals (e.g., showing well-liked Black people and much-disliked white people before the test) and those that provided tools for individuals to change their habits of mind (e.g., simple instructions on how to focus on responding with “good” when seeing a Black face). Importantly, positive effects were found in testing immediately following the intervention but not 1–3 days later. Lai said that while more intensive training about implicit bias, including multi-day workshops, were effective in motivating change, there were inconsistent changes in behavior. To decrease the effects of bias, Lai stressed the importance of adding guardrails to decision making. In hiring practices, for example, Lai said employers should pre-commit to decision making criteria, structure interviews, remove biasing information, and evaluate candidates side-by-side.

Robert W. Livingston (Harvard Kennedy School) highlighted the difference between prejudice (feelings) and discrimination (behaviors), pointing out that there is modest correlation between the two. He described the PRESS framework for addressing discrimination (Problem awareness, Root causes, Empathy, Strategy, Sacrifice), stating that discussions about implicit bias are necessary for problem awareness, but are not, in and of themselves, sufficient.⁴⁰ He said that concern about discrimination is critical, but that some have self-interested and competing value systems and will not prioritize or support social justice. To reach a wider range of people with a wider range of value sets, Livingston suggested making three cases for addressing discrimination: a moral case, a business case, and a collective interest case. He suggested that behavioral interventions can be made at the individual level, the cultural or social level (i.e., people prioritize behaviors that are consistent with what their neighbors are doing), and the institutional level (e.g., the Massachusetts Port Authority changed evaluation criteria for contractors to include diversity and successfully increased diversity). In Livingston’s opinion, change can only be achieved if there is a willingness to sacrifice time, energy, and resources.

Jack Glaser (University of California, Berkeley), the final session panelist, focused his remarks on the role of implicit bias and discretion in policing. He recounted key points from the workshop that resonated in his own work: implicit bias effects are small, but cumulative; biases are caught, not taught; associations between Black people and crime are especially pernicious; changing biases at the individual level is not enough; and inter-group contact is the best hope for a meaningful reduction in bias. He then described a quasi-randomized study that assessed the value of implicit bias training in the New York Police Department and found no effect of the training on police behaviors.⁴¹ Glaser described studies on the role of discretion in racially disparate outcomes in policing, including a study in California that showed that police searches conducted with high levels of discretion showed increased disparities between Black and white subjects.⁴² He also pointed to studies that showed that decreasing the level of discretion decreased racial disparities, including a study where a policy change for U.S. customs that reduced behavioral criteria for vehicle searches (thereby

__________________

decreasing the amount of discretion given to customs agents) resulted in fewer searches and less disparity (with no decrease in the amount of contraband seized).

DISCUSSION

Hernandez moderated this session, beginning with a question of how to present the idea of implicit bias to those who may not be inclined to believe it. Jones said that some people see racial disparities in society, but develop explanations that act as cultural barriers to action, including a narrow focus on the individual, a lack of historical context, the myth of meritocracy, the myth of a zero-sum game, the idea of American exceptionalism, and white supremacy. She said that there are some people who may never acknowledge racism, but that we must move forward without them. Darity added that not everyone wants others to be better off, pointing to tribalism and explicit bias. Livingston raised the idea of social value orientation and how individuals believe resources should be allocated, recalling a meta-analysis that showed that 14 percent of people favor inequality and social dominance. Lai said that evidence suggests that training on implicit bias has the greatest impact on people most resistant to it. Albright suggested that one way to reach people was to highlight the fundamental, biological nature of implicit bias and discuss the challenge as a problem of developing “false priors.”

Darity stated that studies have shown that inter-group contact can improve affective relationships (i.e., friendship or fondness), but not necessarily disrupt cognitive bias (as might be measured by the IAT). Glaser said that intergroup friendships are a valuable pre-condition that reduces negative feelings in a generalizable way. Lai added that the small effects of inter-group contact accumulate over time and, with increased incidences of such contact, the reduction of biases can be meaningful.

KEYNOTE

The remarks of the workshop’s keynote speaker, Bryan Stevenson (Equal Justice Initiative), drew on lessons from his personal and professional life and touched on themes of justice, the need for proximity, and the need to get uncomfortable to address bias. Stevenson began by providing perspectives and statistics on racial disparities in prisons and the criminal justice system. He noted that the prison population has exploded since the 1970s, when politicians began to compete to see who could be the toughest on crime. Black people and communities have been disproportionately affected, and he lamented the indifference to this problem in much of America.

For solutions, Stevenson said, we need proximity to the problem. He described how scientists make progress by getting so close to what they are studying that they see what others can’t. He argued that we should follow the same approach to achieve justice and to address implicit bias: In both cases, he said, we need to get close to the poor and marginalized to see and understand the details. When we get closer to those who are poor and marginalized, we can see things that others can’t. He described the power of proximity in stories about his grandmother and about the lawyers who worked to integrate his school in the 1960s. Their proximity gave them lasting impacts.

Stevenson argued that distance allows biases to grow. He described how criminologists created a narrative in the U.S. that some children are not children and should be tried as adults when they have committed crimes. Others fostered a narrative that people suffering from drug addiction should be charged with crimes rather than treated medically. Such narratives, he said, shape how we think and are often driven by fear and anger. In arguing against trying children as adults in front of the Supreme Court, Stevenson said that he relied on science and understandings of brain development to combat biases and harmful narratives.

Stevenson said that America has been contaminated by its history of injustice, which is like smog in the air. To address this contamination, Stevenson said that narratives must change. He called upon the audience to acknowledge that we live in a post-genocide country, after the mass killings and displacement of indigenous people. The true evil of slavery, Stevenson said, was the narrative we used to justify it; that Black people were inferior and not deserving of the same consideration as white people. Over the years, the narrative has persisted despite the 13th Amendment, which freed the slaves, and the Civil Rights and Voting Rights Acts, which took down some of the legal infrastructure that institutionalized discrimination. He pointed to the examples of South Africa, which established a Truth and Reconciliation Commission after apartheid, and Germany, which has numerous markers and monuments that acknowledge its role in the Holocaust. In contrast, the United States continues to celebrate the Confederacy. Stevenson highlighted the efforts of the National Memorial for Peace and Justice to change the historical narrative. Reckoning with important issues, he said, requires that we get uncomfortable.

DISCUSSION

Sinclair asked when we might reach critical mass for meaningful change. In response, Stevenson said that there is a knowledge gap about biases, narratives, and history that still needs to be filled. He pointed out that there were few laws in the 1970s and 1980s against drunk driving, but Mothers Against Drunk Driving successfully raised awareness of the issue to create change. He expressed hope that efforts like those that resulted in the National Memorial for Peace and Justice will generate useful conversations. He suggested that discussions should be extended beyond individuals to institutions, and that many institutions will need to be pushed to understand the history and legacy of racism. Jones asked about the importance of naming racism. Stevenson said that it is essential to name racism and white supremacy, and that we should not pretend that it is inappropriate to do so.

ROUNDTABLE DISCUSSION WITH COMMITTEE MEMBERS: KEY TAKEAWAYS

Each member of the planning committee provided their perspectives on the important messages that they had taken from the workshop.

Co-chair Camara Jones said that the key messages of the workshop are that racism exists, that it saps strength from all of society, and that we can act to dismantle it. Jones provided a definition of racism: “Racism is a system of structuring opportunity and assigning value based on the social interpretation of how one looks, which is what we call race, that unfairly disadvantages some communities and unfairly advantages some individuals in other communities.” Jones then shared an allegory which compared racism to cement dust in the air. In this instance, individuals can reduce their exposure to the dust or clear their lungs, but that such interventions will not solve the problem. What is needed, she said, is to organize and shut down the factory that is producing the dust.

Thomas D. Albright said that, as a scientist, his inclination is to address many of the topics raised in the abstract, even though they cannot be divorced from their consequences or their history. In an abstract sense, the solutions seem to be either palliative or curative. However, he said, the problem is not like rocket science or brain surgery where progress can be made through the focused attention of small groups of people; instead, it will require listening to a wide range of perspectives to understand and address the problem. This workshop provided an opportunity to tap into such diversity in the pursuit of finding the bigger truth.

William “Sandy” Darity said that we need to be cautious and realistic in approaching racism. He observed that, as the plantation economy showed, you can have both discrimination and economic growth. It will be important, he said, to reckon with the economic and functional role of discrimination. White parents will ensure exclusionary benefits for their children, he argued, even though they say they are not racist. Darity said that we must deal with systems and should look at stratification economics (the study of how social classes or social identity drive economic disparities).

Diana Dunn provided a definition of race: “Race is a specious classification of humans created by Europeans or whites to assign human worth and social status, using himself as the model of humanity and the height of human achievement, for the purpose of establishing and maintaining privilege and power.” Dunn noted that when white people struggle, we change structures, but when Black people struggle, we give them programs. She suggested that empowering those most impacted by racism can create movements and lead to meaningful change.

Ivan Fong said that the business community is at the forefront of thinking about implicit bias and looking for solutions to achieve social justice and racial equality. He said that companies are embracing these issues, particularly since the death of George Floyd. He pointed to listening sessions, diversity initiatives, corporate philanthropy, work to improve culture, and increased transparency about salaries and promotions to increase accountability as examples of corporate efforts to address racial inequity. He observed that three things can be true about the world simultaneously: The world is much better than it used to be, the world is awful, and the world can be much better.

Rayid Ghani reflected on his personal story as an immigrant and said that it took him a long time and more proximity with the problem to feel comfortable talking about American racism. He highlighted artificial intelligence, which can develop or perpetuate biases, but can also be used as a tool to understand bias, to intervene, and to reduce the impact of a biased decision or interaction. He noted that it is challenging for policy makers to address racial disparities without explicit racial criteria, which can get them into legal trouble.

Deena Hayes-Greene said that racism is a groundwater issue: If one fish dies, you may wonder what was wrong with the fish, but if all the fish in the lake die, you will wonder what is wrong with the water. She described a collaboration between the University of North Carolina and the Moses Cohn Hospital that aimed to close disparities in treatment completion between Black and white patients with breast and lung cancer. After many years, the gap was

closed and both Black and white completion rates were increased to 97–98 percent.⁴³ Hayes-Greene said that it was crucial to consider the racial aspects of the disparity even though some did not want to acknowledge it.

Tanya Hernandez spoke about implicit bias from the perspective of a lawyer and law professor. She was pleased that the workshop not only explained implicit bias, but also linked it to systemic problems. She said that, while there has been a lot of previous work on trying to change individuals, we need to focus on systemic interventions. Concrete interventions focused on systemic and structural issues made the difference between good and bad implicit bias training, she said, and consumers of such training should push implicit bias training programs to speak to these systemic and structural issues.

Sheryl Heron referred to her professional career both as an emergency physician, where she worked to stop the bleeding, and as a public health practitioner, where she focuses on prevention. She said that while we have been trying to address implicit bias issues at the individual level, the problem is in the prevention of racism at the systemic level. Heron also reflected on her background as an immigrant from Jamaica and a Black woman, and said that we need to consider the role of caste in this country. She said we must uncover compassion and empathy. Little things, she said, can make a big difference.

Stacey Sinclair said that many of the insights discussed at the workshop were built on basic behavioral research done over decades and said that funders shouldn’t shy away from such research. She suggested that structural racism and implicit biases are not separate, but are linked and based on historical underpinnings. Sinclair said that work at all levels is needed to make a difference: from individual to systemic, from basic research to applied.

Co-chair Goodwin Liu closed the workshop by thanking the National Academies for taking on the issue of implicit bias and for speaking about the power of science in the context of important social issues. He said that, as a judge working to address bias in his courtroom, the workshop discussions were highly relevant. As an Asian-American in a time of increasing racism against that community, he also found the discussions to be personally meaningful. He called for other opportunities to get broader consensus and to broaden participation and knowledge of the issues raised.

__________________

DISCLAIMER: This Proceedings of a Workshop—in Brief has been prepared by Sarah Carter, Vern Dunn, Steven Kendall, and Anne-Marie Mazza, as a factual summary of what occurred at the meeting. The committee’s role was limited to planning the event. The statements made are those of the individual workshop participants and do not necessarily represent the views of all participants, the planning committee, the Committee on Science, Technology, and Law, or the National Academies.

REVIEWERS: To ensure that it meets institutional standards for quality and objectivity, this Proceedings of a Workshop—in Brief was reviewed by Jordan Axt, McGill University; Susan Fiske, Princeton University; Tanya Hernandez, Fordham University; and David Tatel, U.S. Court of Appeals for the District of Columbia Circuit. Marilyn Baker, National Academies of Sciences, Engineering, and Medicine, served as the review coordinator.

Planning Committee for The Science of Implicit Bias: Implications for Law and Policy: A Workshop: CAMARA P. JONES (Co-Chair) (Morehouse School of Medicine); GOODWIN LIU (Co-Chair) (Supreme Court of California); THOMAS D. ALBRIGHT (NAS) (Salk Institute for Biological Studies); WILLIAM "SANDY" DARITY (Duke University); DIANA DUNN (People’s Institute for Survival and Beyond); IVAN FONG (3M Company); RAYID GHANI (Carnegie Mellon University); DEENA HAYES-GREENE (Racial Equity Institute, LLC); TANYA HERNANDEZ (Fordham University School of Law); SHERYL HERON (Emory University School of Medicine); and STACEY SINCLAIR (Princeton University).

National Academies of Sciences, Engineering, and Medicine Staff: Anne-Marie Mazza, Senior Director; Vernon Dunn, Program Officer; Steven Kendall, Program Officer; Dominic LoBuglio, Senior Program Assistant.

SPONSORS: This activity was sponsored by the Ford Foundation. For additional information about the Planning Committee for The Science of Implicit Bias: Implications for Law and Policy, visit https://www.nationalacademies.org/our-work/the-science-of-implicit-bias-implications-for-law-and-policy-a-workshop. For additional information about the Committee on Science, Technology, and Law (CSTL), visit https://www.nationalacademies.org/cstl/committee-on-science-technology-and-law.

Suggested citation: National Academies of Sciences, Engineering, and Medicine. 2021. The Science of Implicit Bias: Implications for Law and Policy: Proceedings of a Workshop—in Brief. Washington, DC: The National Academies Press. https://doi.org/10.17226/26191.

Policy and Global Affairs

Copyright 2021 by the National Academy of Sciences. All rights reserved.