Internal Affairs

In an effort to combat police mistreatment of minorities, law enforcement agencies have embraced the science of implicit bias. But what do such studies—and related training programs—really reveal?

By Dwyer Gunn, published on March 2, 2017 - last reviewed on March 17, 2017

When Michael Brown, an unarmed black teenager, was killed by Darren Wilson, a white police officer, in Ferguson, Missouri, in August 2014, protests and unrest consumed the city for weeks. The reaction was quickly understood as a response not to a single isolated incident but to something much bigger—corrosive mistrust of the police and long-festering patterns of abuse

Illustration by Paul Sahre 

Seven months later, Attorney General Eric Holder announced the findings of two investigations of the Ferguson Police Department by the U.S. Department of Justice. The picture he painted was bleak. Voluminous data confirmed uneven, often unconstitutional treatment of African-Americans by law enforcement. Kafkaesque details appeared in reports of seemingly spurious charges, like "manner of walking along roadway." Holder described a black population that felt perpetually "under assault and under siege by those charged to serve and protect them." 

In this context, he said, it wasn't difficult to comprehend why Ferguson exploded "like a powder keg" after Brown's death. There was no other explanation for the stark disparity in treatment, Holder said, besides both explicit and implicit bias.

While the role of race in police misconduct is a hard subject to study empirically, numerous studies have shown that people of color are more likely than whites to be targeted, stopped, and arrested by police and are more likely to be subject to force, even when variables like demographics are controlled for. It's impossible to determine the precise motivations of any individual police officer, but in the past few years, with the proliferation of phone-captured videos of apparently unprovoked police violence, a growing chorus of analysts has offered an array of explanations for the disparity in treatment. One factor increasingly makes all the lists: implicit bias, or the evaluations that the mind makes automatically, below the level of consciousness.

Social scientists have known for decades that human beings have a fundamental tendency to create stereotypes—an essential mental shortcut related to the mind's need to assess situations and make decisions quickly. From the earliest moments of childhood, "we gather associations from our surroundings—the media, the stories we hear, the people we interact with on a day-to-day basis," says Rachel Godsil, the cofounder and research director of the Perception Institute, which studies bias and develops intervention measures. "For example, when you say 'engineer,' the example that pops into my head is male. That's not consciously what I think an engineer should be. That's the automatic association that I grew up with."

Nearly two decades ago, the study of racial stereotyping was upended by a new test said to measure it. In 1998, the psychologists Anthony Greenwald, Debbie McGee, and Jordan L.K. Schwartz unveiled the Implicit Association Test, or IAT, a computer-based psychological tool that assesses the strength of participants' automatic association of positive or negative characteristics with concepts. The earliest test concepts included flowers, insects, musical instruments,  weapons, and typically black and white American names. Greenwald, along with psychologists Mahzarin Banaji and Brian Nosek, went on to use the IAT to reveal what they called the "roots of unconscious prejudice," and reported implicit antiblack bias in 70 percent of those studied. Millions of people have since taken a self-administered IAT on the website Project Implicit, hosted by Harvard University. "The concept isn't new, but the Implicit Association Test provided a means to understand it in a very visceral way," Godsil says.

Among police officers, experts say, implicit racial bias isn't necessarily stronger than it is in the general population. It's just that it may influence behavior that can have unusually powerful repercussions because of the nature of their job. "Police officers are human and have the same set of biases the rest of us do, but the consequences can be worse," says Phillip Atiba Goff, a criminologist at John Jay University and the president of the Center for Policing Equity. "If you or I make a mistake based on implicit bias, the consequences are relatively minor. If a police officer makes a mistake, someone might go to jail, be separated from his or her kids, or, God forbid, get killed."

The concept even popped up in last year's presidential debates—when Hillary Clinton was asked if she thinks police are implicitly biased against black people. She responded that "implicit bias is a problem for everyone, not just police" but "when it comes to policing, it can have literally fatal consequences." 

Embracing this idea, police departments in dozens of cities, including Baltimore, Detroit, Dallas, and Miami, have rushed to provide officers with implicit-bias training. A coterie of social scientists now crisscross the country teaching police about the psychological construct in the belief that such instruction can perhaps help mitigate its effects. The New York Police Department is developing a bias-awareness curriculum for its police force, the nation's largest. And last summer the DOJ announced that all of its law enforcement agents and personnel—more than 23,000 people—will be required to undergo implicit-bias training.

The nascent science of implicit bias is still evolving, and psychologists continue to debate the extent to which it drives behavior. As training  becomes commonplace, key questions remain unanswered: Is bias-free policing possible? And what can implicit-bias training really accomplish?

Illustration by Paul Sahre 

To Serve and Protect

On an overcast day in November, I traveled to Tacoma, Washington, to observe a two-day program on the science of implicit bias for command-level police officers. Tacoma is a quiet city of about 200,000 situated 25 miles south of Seattle on picturesque Puget Sound. Its population is two-thirds white and a third nonwhite, and it's not known for hostile race relations or police violence. Nevertheless, Tacoma is not without racial tension and mistrust, and like many cities in the wake of Ferguson, the police department embarked on an initiative to better its relationship with the community. Implicit-bias training was one piece of that effort.

The conference room at the department's headquarters was filled with uniformed officers, along with a handful of city council members and community leaders—people who had been involved in the conversations that prompted the training. Police Chief Don Ramsdell opened the session by introducing Lorie Fridell, a professor of criminology at the University of South Florida and the director of Fair and Impartial Policing, an organization that teaches law enforcement agents around the country about implicit bias and how it figures into police work.

A slim woman with cropped honey-blond hair, a lean face, and sharp blue eyes, Fridell buzzed about the room, her bright red jacket a stark contrast to the officers' navy blue. She seemed to have boundless enthusiasm for her topic and sensitivity for how to broach it. After warming up the crowd with a few jokes about college sports, she drew them in by sharing her personal "aha!" moment.

In the early aughts, Fridell said, as the research director of an organization that studies policing practices and policy, she spent much of her time speaking with law enforcement and community groups about racial profiling. Her sense was that there were essentially two kinds of cops: those who were overtly ill-intentioned and racist and those who were well-meaning and free of bias. "I thought bias was explicit, in your face, out on the table," she said.

Fridell found the truth to be more complicated. "I came to believe the problem was more than just a few bad apples," she said. "But I also came to know that most law enforcement officers are well-intentioned individuals who want to help and serve their communities."

Fridell learned about the burgeoning study of implicit bias around the same time that revelatory research about the subconscious association of African-Americans with crime was capturing headlines. In 2004, Goff, along with Stanford University psychologist Jennifer Eberhardt and their colleagues, published a landmark study in which they primed participants with pictures of black or white faces and then showed them images of a variety of crime-related or noncrime-related objects at different levels of blurriness. They found that exposure to black faces results in dramatically faster identification of the crime-related objects. Exposure to white male faces has the opposite effect, slowing the rate at which people identify crime-related objects.

The study came on the heels of another suggesting that such associations could, in fact, have life-or-death consequences. In a 2002 paper, University of Colorado psychologist Joshua Correll had participants play a computer game in which pictures of black or white men carrying either a weapon or a nonthreatening object flashed in front of their faces; students had to either "shoot" if the man held a gun or not "shoot" if he carried something harmless. Correll found that the students were slower to identify a weapon when the man in the picture was white and slower to identify a harmless item when the man was black. They were also more likely to incorrectly "shoot" a black man with no gun and to fail to "shoot" a white man with a gun. When police officers were the participants, Correll found that they held the same set of implicit biases as the general population but were actually less prone to making errors related to stereotypes, a finding that he attributed to their use-of-force training.

In Fridell's framing, the theory of implicit bias offers a way to confront the problem of unequal treatment by police without hostility or accusations of racism—it creates what she calls a "blame-free environment." In her Tacoma presentation, she went to great lengths to illustrate how implicit biases might hamper police officers' ability to do their jobs safely—not just by prompting erroneous judgments about black men, for instance, but also by leading police to be less vigilant with women or white men. And she peppered her lessons with disarming examples of her own unconscious biases, emphasizing the central message over and over again: Everyone is biased.

"Because they are human, even the best officers might practice biased policing," she said. "Because they hire humans, even the best agencies will have biased policing."

Much of the day was spent in this way, with Fridell explaining the mechanics of implicit bias, its prevalence, and its potential problems. By the afternoon, with the room warmed up to the idea, she facilitated discussions about how it personally affected—or could affect—the participants. Black and Hispanic officers talked about occasions when they had been subject to prejudice. Ramsdell, the police chief, suggested that fear of their own bias may prevent cops from doing their job. "I think they're asking themselves, 'Why should I stop that car if I might end up shooting someone?'" he said.

Sitting beside Ramsdell was Eric Jackson, the pastor of an African-American church who has worked with the police chief on community outreach. Jackson expressed disappointment that most of his young parishioners were opposed to a career in law enforcement, even as the Tacoma department tried to recruit them, and he relayed their general anxiety around police. He also described having "the talk" with one of his sons.

"I told him that if he's ever in a situation with a cop, to lie face down and extend his arms," Jackson said. "The very next day, my son saw on the news that the cops had shot a man in that exact position. My older son lives in Atlanta, and every time the phone rings, I'm scared." Chief Ramsdell, a parent himself, flinched and shook his head.

Illustration by Paul Sahre 

Science at Work

No researcher disputes the existence of implicit bias, also known as implicit social cognition—its roots lie in the fact that most of human cognition takes place outside of conscious awareness.Yet the research on whether it can be changed is mixed. A 2012 study by University of Wisconsin psychologist Patricia Devine framed implicit racial bias as a habit and reported "dramatic reductions" among those who gained awareness of it, were concerned about its effects, and practiced strategies to lessen it, such as thinking concertedly about people who don't conform to stereotypes. Another study, published last year, suggested that biases can be overridden with the chance to reflect before making a decision (something that police often don't have the opportunity to do.)

Even if bias can be changed, however, questions persist about how well it predicts behavior. Last year, a meta-analysis led by University of Wisconsin psychologist Patrick Forscher looked at 426 studies and found that while implicit bias as measured by the IAT is malleable, there's scant evidence that changing it relates to a change in behavior. Another meta-analysis, published in 2013 and led by Rice University psychologist Frederick Oswald, found that the IAT is a poor predictor of real-life discriminatory behavior. Over the past several years, a group of research psychologists, including Philip Tetlock at the University of Pennsylvania and Hart Blanton at the University of Connecticut, have forcefully questioned the IAT's validity as a psychometric instrument altogether and asserted that it doesn't really measure what it purports to.

Some have jumped on academic debates like these to imply that implicit-bias training is ineffective and, in some cases, that the whole concept of implicit bias is invalid. But Robert Mather, a professor of experimental psychology at the University of Central Oklahoma who studies social cognition, says that popular discounting of implicit bias comes from "the fact that people don't understand how science works."

"Science is slow, and when the public steps into the middle of an unfolding scientific process, it can look messy," Mather says. "In the case of bias, it does and doesn't link to behavior. It affects us sometimes, and at other times it doesn't. And the IAT is only one measure of implicit bias. I think the most important thing to understand is that implicit bias is a broad concept and it will require much more experimental testing to determine what type of biases affect what types of behavior, and under what types of circumstance."

Discounting the training programs may also misunderstand their goal. Experts, including Fridell and Eberhardt, say that the intent isn't to scrub people of bias but rather to raise awareness so police officers and departments can craft policies and cultures to mitigate its possible effects. They note that implicit-bias training is just one initiative of many that can work in concert to improve policing. These include building relationships with the community; recruiting more diverse candidates; setting up systems of accountability for police misconduct; requiring officers to wear body cameras, which research suggests decreases the use of force; and mandating high-quality use-of-force training—simulations that train officers to assess rapidly and accurately whether a suspect is a threat.

Eberhardt, who codesigned an implicit-bias training program that's been used by 28 California law enforcement agencies, also points out that bias doesn't act in a vacuum: It's contingent on situations and more likely to be triggered when commingled with stress, fear, and anxiety. Departments can take steps to address those triggers, she says, like frequently rotating officers with stressful posts or changing foot-pursuit policies so police don't follow suspects into places where they might get trapped and become more likely to react out of fear. "To the extent that bias is about the situation that we find ourselves in, we need to think about those situations," she says.

Goff agrees. "We use police in the exact situations where implicit bias can be a problem," he says. "You're tired, you've got too many things to do, and you're more likely to rely on implicit biases just by virtue of the job you're doing. It's a perfect storm. So it might be that changing policies and culture is far more powerful in changing police behavior than an individual training program. The training isn't unimportant, but it's not a silver bullet. It needs to be part of a broader justice gumbo."

Still, to the question of what such training accomplishes, whether alone or in concert with other reforms, the truth is, nobody knows. Fridell's program, supported by more than $1 million in grants from the Justice Department, has not been rigorously evaluated. In 2014, the Justice Department gave $4.75 million to Goff and others to launch the National Initiative for Building Community Trust and Justice, part of which would go to research and evaluate implicit-bias interventions. In assessments of Eberhardt's training, officers have reported hazily positive outcomes, like feeling hopeful about change and desiring more connection with community members, but she readily acknowledges that the work demands more empirical evaluation and is now working on a project that will use body camera footage to assess police behavior before and after undergoing her training. "You don't want a situation where the community is calling for something and the police do the training, but it's still not clear whether that actually moves the needle," Eberhardt says.

Illustration by Paul Sahre 

A Bitter Truth

After the first day of training ended, I went back to my hotel room and reviewed my notes. At around 8:30 p.m., I received an email from Fridell. Day two of the training was canceled, she wrote. A few hours earlier, a Tacoma police officer had been shot and killed. It was the first time a Tacoma officer had been killed in the line of duty in 19 years.

I followed the tragedy unfolding across town on a local news site. The officer, Jake Gutierrez, and his partner had responded to a domestic disturbance call in the city's Eastside neighborhood in the late afternoon. Gutierrez, a 17-year veteran of the police department and a father of three, had entered the house with his gun still holstered and was shot by the suspect, Bruce Johnson. He was transported to Tacoma General Hospital, where he died on the operating table. Meanwhile, a standoff ensued at the house, where Johnson was using his children, ages 6 and 8, as human shields. At around 3 a.m., a SWAT team sniper positioned outside killed Johnson with a single bullet. The children were rescued unharmed.

Days later, Gutierrez's public memorial service was attended by thousands, including police officers from around the country. Ramsdell gave a eulogy. He wore a black mourning band over his badge and had purple shadows under his eyes. With a heavy voice, he described Gutierrez as an exemplary police officer and summarized a letter of commendation that a mother had sent after he'd responded to a family dispute. "The mother related that even though her son was arrested, Jake treated everyone involved with dignity and respect, including her son," Ramsdell said. "He thoroughly explained the steps that were being followed along the way. The mother shared that his taking the time to explain the circumstances to the entire family helped ease the tension and stress that they were experiencing."

Sixty-three police officers were killed by intentional gunfire in the line of duty last year, a number slightly higher than in the few preceding years, in part because of antipolice ambushes like the one in Dallas that killed five officers. The bitter contrast of a Tacoma officer—one who was highly respected by and involved in the community he served—being killed at the same moment that his colleagues were wrapping up a day-long effort to build a bridge with the community was striking.

I called up Eric Jackson, the Tacoma pastor, a few days later. He had reached out to Chief Ramsdell to offer his condolences after the killing and had attended Gutierrez's memorial. He told me how shaken he'd been by the reminder that violence against police officers can happen anywhere and at any time.

I asked him how he felt about Fridell's program even though it was cut short. "I don't know that in two days you can change people," he said. "I'm still an advocate for trying to make change. But I find myself lately, with the current climate, thinking, My goodness, I don't have a solution."

Then he reflected on the broader social forces that had led to Fridell's arrival in Tacoma for the training—the acts of aggression recorded on cell phones and spread over social media, the movement among police to repair and strengthen their community relations, the advancement of psychological science and the efforts to translate its findings to daily life. As he saw it, implicit-bias training is perhaps one small piece of a solution for police in Tacoma and everywhere. It isn't an end, he said. It's just a beginning.

Submit your response to this story to letters@psychologytoday.com If you would like us to consider your letter for publication, please include your name, city, and state. Letters may be edited for length and clarity. 

Pick up a copy of Psychology Today on newsstands now or subscribe to read the the rest of the latest issue.

Facebook image: a katz/Shutterstock