Police implicit bias training originates with a flawed interpretation of antiblack police violence. It’s worth revisiting that origin. Early on February 4, 1999, four white New York Police Department (NYPD) officers patrolling a neighborhood in the Bronx stopped Amadou Diallo, a young Black man “acting suspiciously”  who matched the description of a suspected rapist. They pulled over their unmarked car, identified themselves as police officers, and ordered the man to stand still. When Diallo reached into his pocket for his wallet, the lead officer, per his testimony, misidentified it as a gun, triggering the group to shoot and kill 23-year-old Diallo. The officers were acquitted, their actions judged to be reasonable self-defense under the circumstances as they allegedly perceived them. This is the story told by the police in their trial — we do not know what Diallo would say.
Diallo’s murder brought mainstream attention to the aggressively racist stop-and-frisk policy that was a hallmark of the NYPD under Mayor Rudy Giuliani. This history is well known. But the way Diallo’s murder would echo through the linoleum-lined halls and windowless cubicles of psychological research labs has yet to be reckoned with. It inspired the lucrative research paradigm of implicit bias, which appeals to researchers, funders, and the general public alike because it offers an easy way to address racism without casting blame or leveling inconvenient institutional critiques.
In 2001, psychologist Keith Payne published the first psychology paper arguing that implicit bias drives racist police violence. Inspired by Diallo’s death, Payne’s experiments showed that simply looking at a Black man’s face primed subjects to misperceive harmless objects (like a wallet) as a weapon, through no conscious fault of their own. For Payne, this was a key insight: brutality was a problem of honest misperception. Diallo’s murderers, taken at their word, were acquitted: “[I]n this ambiguous situation, [the officers] acted on the information available, sincerely believing that they were in danger.” A year later, Joshua Correll and his colleagues introduced a similar set of experiments on police violence, also based on the NYPD’s account of Diallo’s death. A new paradigm was born, with implicit bias at its center.
In psychologizing Diallo’s murder, Payne and Correll divorced Diallo’s death from the unconstitutional and racist practices that defined Guiliani-era NYPD. In their repeated retellings, these researchers uphold the NYPD officers’ self-exonerating account over the experience and incredulity of activists and observers. Instead, they might have attended to contemporaneous legal commentary on the case, which argued that “[n]o one who knows police practices believes that the four officers involved that night were actually looking for a rapist in a year-old rape case.” They could have pointed out that the officers didn’t just happen upon Diallo in a neutral, everyday situation, a common fantasy of social psychology. Rather, the officers were part of the infamous NYPD Street Crimes Unit, which carried out “broken windows” policing until it was rebranded. Harassing communities of color in the hopes of turning up guns or drugs was already at the time known to be the aggressive “modus operandi of the [Street Crimes U]nit.”
Psychologists stressed the moment immediately before police first fired at Diallo: “This case is interesting to psychologists,” Payne writes, “not because of its legal or ethical implications, but because of the psychological processes that it dramatically highlights” (emphasis added). Their writing construes police violence as natural human error. Two decades of implicit bias research have followed in their footsteps, repeatedly casting police officers as apolitical agents in a system whose potential fairness was being undercut by ambiguity and unconscious prejudices. NYPD’s incredible alibi of an innocent murder thus became enshrined as universal psychological vulnerability.
Correll and colleagues’ motivating question — “Would the police have responded differently if Diallo had been [w]hite?” — ignores other factors in presupposing the need for new psychological evidence. It wasn’t by chance that police were patrolling Soundview, a neighborhood that became majority-Black after white flight, economic decline, and redlining. Officers would have been unlikely to encounter, let alone approach, a white person. It was equally unlikely that Soundview would have become the type of neighborhood targeted by the NYPD were it not for decades of financial disinvestment. This naturalization of police violence as cognitive bias effaces the complex social dynamics that did give rise to the murder of Diallo, and others like Diallo.
Acknowledging the ugly antiblack history built into Diallo’s murder demands radical interventions aimed at undoing racist policing and housing discrimination, solutions that would require psychologists to dramatically alter their methods and theories. Implicit bias research appeals to law enforcement because, rather than addressing these fundamental issues, it provides neat explanations for complex structural problems and humanizes police violence as the tragic outcome of “hidden bias.” This narrative captivates audiences and spins out into convenient policy reform.
Two initiatives illuminate how psychologists package implicit bias and take it on the road. Stanford University’s Social Psychological Answers to Real-World Questions (SPARQ) and the Center for Policing Equity (CPE) reside at the nexus of academia, private industry, nonprofits, and local government, partnering with law enforcement to implement the most promising translations from bias research to applied intervention.
Their efforts in police reform have three prongs: training courses for police, new studies using policing data, and policy recommendations. The training courses developed by SPARQ and CPE promote the familiar blame-averting lessons of implicit bias, going to great length to withhold critiques of the police. CPE materials state, “This Session Is Not About Your Character […] [D]ismiss the notion that this training is in anyway targeting the character of officers.” Trainings, they say, help police achieve their personal goals, which include “Get promoted,” “Don’t get sued,” and “Have fun every day at work.” SPARQ claims their training is successful based on surveys of the officers who have taken them. In actuality, such reports show nothing more than the alarming potential of bias trainings to promote the illusion of equity in the face of ongoing brutality.
SPARQ and CPE courses also teach what’s known as procedural justice, which trains police officers to act fairly and impartially when enforcing laws. Like implicit bias, procedural justice is predicated on the fantasy that police are fair agents of bureaucracy objectively carrying out apolitical social functions. Yet the concept of procedural justice ignores the degree to which criminalization and law enforcement are inherently political, regardless of officers’ comportment. For example, when police conduct sweeps of unhoused people to enforce anti-encampment ordinances, they are targeting a group that is disproportionately Black, LGBTQ, and disabled, often in the service of gentrification.
Both implicit bias and procedural justice trainings peddle the idea that we can purge the police of human error through data-driven tools. SPARQ and CPE are proponents of this approach, where equity benchmarks and empirical “precision” in policing are considered evidence of improvements. The problem is that the data are suspect: police generally, and SPARQ’s partner, the Oakland Police Department specifically, are unreliable reporters of their own conduct. And even with trustworthy data, psychologists and their police collaborators remain in charge of defining what “equity” is by choosing which outcomes are measured. This methodological latitude allows them to market their reforms as making progress, even as the underlying causes of police violence remain untouched.
For example, SPARQ celebrates a pilot program developed with Oakland PD that shrank the total number of police stops in 2019. However, racial disparities in police stops remained stark. Any success in scaling back policing is a step in the right direction, but SPARQ’s data-driven analysis illuminates how what looks like an amelioration can actually be an intensification of policing. Behind SPARQ’s “success” is precision policing, a popular reform championed by law enforcement, including William Bratton, the former NYC police commissioner and principal architect, overseer, and salesman of the broken windows policing that resulted in Diallo’s murder. Precision policing involves targeting individuals or neighborhoods for enforcement based on computer models derived from past crime data — but that data reflect and reproduce existing racist hierarchies. Thus, the number of police stops may indeed shrink but disparities are reinforced as policing is made ever more “efficient,” and the burden of police racism confined to an ever narrower geographic “hot spot.”
As with any research field, psychologists do not merely describe their objects, they construct them using narratives composed of motivating anecdotes, data, and theoretical assumptions. These narratives have political stakes. Payne’s and Correll’s research in the early 2000s turned a suspicious NYPD narrative into psychological fact. Ever since, mainstream psychology has pedaled a reform without evidence of its success.  In spite of its empirical failure to produce substantive reductions in racist policing, each year a new spate of articles continue to modify and elaborate on Correll’s and Payne’s 20-year-old bias tests. These hair-splitting refinements only preserve the failed reformist promises of implicit bias training. Further research into trainings will never make them as effective as they need to be precisely because the problems these trainings seek to address are rooted in structural and historical crises. They cannot be solved by training individuals to act more fairly.
The failed promises of implicit bias are symptomatic of psychology’s longer-standing failure to grapple with racism as a structural problem. Gordon Allport’s The Nature of Prejudice (1954) is the foundational text of modern intergroup psychology; it is also the origin point of psychology’s conceptualization of racism as located in individual psychology. The book opens with a series of racist encounters in countries like Rhodesia, the West Indies, South Africa, and the United States — regions marked with legacies of colonialism and enslavement. Allport elides this history and instead frames racism as a feature of universal tribalism (we are too “fettered to our respective cultures”) and cognitive mishandling (racism as “an antipathy based upon a faulty and inflexible generalization”). Payne’s and Correll’s psychological reframing of Diallo’s murder is indebted to Allport, which they acknowledge in their recent review. In the same way that Allport reduces colonial and plantation legacies to cognitive bias, bias researchers reduce police violence — erupting in our contemporary carceral system with its plantation origins — to a problem of “labor-saving heuristics.”
Instead of turning to Allport, psychologists might have turned to Frantz Fanon’s 1952 Black Skin, White Masks and a now-jettisoned lineage of psychology that foregrounds the role of colonialism and enslavement in shaping social life. An anticipatory corrective to Allport, Fanon overtly rejects the psychological impulse to conflate — and decontextualize — all forms of discrimination into a flat, universal category of bias. Instead, for Fanon, the psychology of antiblack racism is inextricable from the ideologies and material conditions imposed by colonial domination — it can only be redressed through anticolonial politics: “There will be an authentic disalienation [of Black people] only to the degree to which things, in the most materialistic meaning of the word, will have been restored to their proper places. […] It implies a restructuring of the world.”
Black Skin, White Masks remains almost entirely excluded from the canon of psychological theory, even as it has served as an indispensable touchstone in adjacent fields. Psychologists must reengage with this radical tradition of psychology to dispense with the Allportian liberal individualist conceptualization of racism as bias. Black studies, literary studies, history, and anthropology — disciplines that psychology has increasingly eschewed in favor of quantification and neural reductionism — recenter what’s structural about racism. Geographer and activist Ruth Wilson Gilmore provides an exemplary definition of racism as “the state-sanctioned or extralegal production and exploitation of group-differentiated vulnerability to premature death.” Similarly, abolitionist historian Dylan Rodríguez defines racism as “a logic of social organization that produces regimented, institutionalized, and militarized conceptions of hierarchized ‘human’ difference.” These structural formulations of racism acknowledge its roots in economic and political systems of which it’s a central feature, not a cognitive accident. “Put simply,” says Gilmore, “capitalism requires inequality and racism enshrines it.”
Crucial here is the insight that race and racism are not merely isolated issues of identity, bias, and split-second perceptions, but must instead be thought of in relation to racialization: the assemblage of institutions, ideologies, and economic forces that make some group differences and not others the basis for ongoing and tolerated subordination and exploitation. An abolitionist psychology grounds its work in analyzing and undoing these forces. Contra bias, it conceives of racism as an essential feature, not a bug, of the system. Migrating from the laboratory to policy and business ventures, psychology’s bias paradigm ends up collaborating with the police in upholding the prison-industrial complex. An abolitionist psychology by contrast challenges the logics by which prisons and policing produce themselves as common sense.
Certainly, structural racism is mediated by individual bias. But a structural approach to racism leaves in doubt the tacit assumption in much psychological work: that racism is principally or necessarily propagated through individual or collective prejudice, and that correcting such bias will undo longstanding structural racism. Instead, structures must be dismantled before the biases that pervade individual and collective psyches can be undone.
Lee Jasperse is a PhD student at the University of Chicago. His research and writing examine how scientific racism extends into the popular imaginary and lived experience.
Benjamin Stillerman received his PhD in Social Psychology from NYU studying prejudice and dehumanization.
Banner image: "Party: Over" by Derek Simeone is licensed under CC BY 2.0. Image has been cropped.
 The standards for establishing reasonable suspicion are easy to meet if the police claim they witnessed a combination of “‘furtive glances’ [and] ‘evasive behavior’ in a ‘high crime’ area. Not only are these factors subjective, they are easily manufactured.”
 This didn’t need to be the case. Implicit bias research exploded because psychologists studying prejudice faced a problem following the Civil Rights movement when their college-aged study participants stopped owning up to racism in surveys and questionnaires. But implicit bias was not the only innovation available. Their research could have traced the post–Civil Rights innovations to prejudice itself, when subjugation of devalued groups evolved into new forms of institutional and administrative violence.