Why “implicit bias” training is not the solution to police violence

Last week, U.S. Attorney General Loretta Lynch announced that some 28,000 federal agents would be required to undergo implict bias training. Lynch’s announcement came during a brief respite between police killings of unarmed Black, Latino, and Indigenous people. But as the week progressed, and the media reported on of the killings of Alton Sterling, Philando Castile, Anthony Nuñez, Raul Saavedra-Vargas, Melissa Ventura, and Pedro Villanueva, implicit bias training re-emerged, albeit quietly, as a talking point. Hillary Clinton referred to it as a “common sense” police reform. And, indeed, where implicit bias programs have been instituted at the local level, they have been greeted with approval on all sides. They are inexpensive, and do not pose any challenge to policing at a structural level. Yet I worry that focusing on unconscious racism is a misrepresentation of the problem and may lead to complacency (which, I suppose, is one meaning of common sense).

What is implicit bias training?

The keystone of implicit bias training is the Implicit Association Test (IAT), which was developed by a group of social psychologists in the 1990s.^[1]Greenwald, Anthony G., Debbie E. McGhee, and Jordan L. K. Schwartz. “Measuring Individual Differences in Implicit Cognition: The Implicit Association Test.” Journal of Personality and Social … Continue reading The premise is that it’s easier to form a conscious association if it already exists implicitly, in the pre-conceptual area of memory sometimes called “know-how.”^[2]The existence of which is disputed–although I have no dog in that particular fight. In the IAT, words or pictures flash across a computer monitor, which the subject has to categorize by pressing one key or another. There are usually two domains of categorization: that is, someone might be asked to classify items as either safe/dangerous or fruit/vegetable using the same pair of keys. The subject has about half a second to make those judgments, so it’s expected that she’ll hit the wrong key a lot of the time. But (the theory is) the keystrokes are faster and more accurate when they reflect what the subject intuitively believes.

It’s in the realm of implicit memory where prejudice is said to lie: hearing “nurse” and visualizing a woman, for instance. There are IATs designed to measure associations surrounding gender, sexuality, weight, age, works of art, and current events, as well as several dealing with race and the perception of danger. In one variation, the subject has to determine if the object on screen is a handgun or something else, like a cell phone or a wrench. The subjects–mostly civilians, mostly white–were quicker to identify the object as a gun after being “primed” with a black man’s face.^[3]CHECK: Payne, B. K. “Prejudice and perception: The role of automatic and controlled processes in misperceiving a weapon.” Journal of Personality and Social Psychology 81.2 (2001). Cited … Continue reading In another variation, this one administered to police officers, the subject has to click “shoot” or “don’t shoot” after glimpsing a photo of a man, armed or not, black or white. We can easily guess the results.^[4]Correll, Joshua, et al. “The Police Officer’s Dilemma: A Decade Of Research On Racial Bias In The Decision To Shoot.” Social & Personality Psychology Compass 8.5 (2014): … Continue reading

Rhetorically, these tests have succeeded in many corners at something very difficult–proving to skeptics that racism exists. (The need for “objective” scientific evidence of such things is symptomatic, as anti-racist and feminist activists have long known.) But they are rooted in a characteristically American understanding of racism as prejudice–a psychological failing, not a legal or economic one. Embraced after World War II as part of U.S. cultural diplomacy efforts, “prejudice” research was inoffensive and ostensibly neutral.^[5]Further, as the historians such as Jamie Cohen-Cole and Fred Turner have shown, the tolerant personality was partly a Cold War ideal–the person who valorized flexibility over ideological purity … Continue reading Using the IAT, a straight person who shows evidence of anti-gay bias and a gay person who displays a slight bias in favour of homosexuality (as I do, apparently) are seen as manifestations of the same phenomenon. Call it “All Lives Matter” liberalism.

What do we gain by calling racism unconscious?

The test’s advantage over traditional sensitivity training seems clear. Since officers are forced to make snap judgments in the field, it’s not enough for them to consciously disavow racial prejudice. That premise, however, rests on several unquestioned assumptions about the way policing works.

The more we convince ourselves that snap judgments are what matter most, the more our institutions reflect that assumption.^[6]And by the way, fuck you too, Malcolm Gladwell. Many police cadets receive a mere eight hours of de-escalation training. A small minority of cops receive any training in mental health crisis intervention, even though a high number of shooting victims have a mental illness or an intellectual disability. I think of Andrew Loku, a Somalian immigrant who had a history of mental illness but no criminal record, banging on a stairwell with a baseball bat. Or Philando Castile, driving with a broken headlight with his girlfriend and her four-year-old daughter in the backseat. Or Anthony Hill. Or Kevin Davis, who was shot after a 911 call he made himself, after another man had attacked his girlfriend. These did not have to be fraction-of-a-second decisions; perhaps they were only declared so in retrospect.

The forced decision (we have to act now!), which the IAT treats as paradigmatic of thinking in general, is part of the logic of terror.^[7]See, for example, David Simpson, 9/11: The Culture of Commemoration. At the same time, the conditions for racial violence are not whims or accidents. Will targeting implicit bias do anything to change the stop-and-frisk or carding policies targeting young people of colour, the influx of military weaponry into North American cities, or the assumption that police officers need deadly weapons at all times (even on traffic patrols or college campuses)? And what about the jury system, where deliberation is expected to prevail over intuitions? Can implicit bias account for a system that rarely indicts police officers for extrajudicial murder or manslaughter, and even more rarely convicts them?

In a 2010 study, university students were asked to read a scenario about “John,” a white manager who sometimes denies promotions to deserving African American employees. However, in some of the scenarios, John’s prejudice is described as unconscious–he is unaware of it–and in others, it is described as automatic, a “gut feeling” of dislike which John tries to suppress “because he sincerely believes in equality.” The difference between implicit and automatic may sound like semantic quibbling, but it had a real effect on the way the test subjects interpreted the case. Those who were exposed to the “implicit” theory of racial bias were less likely to hold John responsible for his actions. (Choose your cliche: “My brain made me do it” or “Everyone’s a little bit racist.”) I hardly want to reify the notion of personal responsibility, nor the lie of “a few bad apples” that often comes with it. But the implicit-bias protocol at once individualizes the problem and puts it beyond the scope of individual liability.

What, exactly, do IATs tell us about racialized violence? We know, unambiguously, that people of colour are murdered by police officers in far higher numbers than white people, regardless of the crime rate in the area. We know with almost as much certainty that implicit biases exist. Yet the correlation between unconscious racism, as revealed in the test, and on-the-ground use of force is imperfect. (Other researchers have noted a similar discrepancy in hospital doctors’ treatment decisions.) Even the Association for Psychological Science, which supports the use of IAT tools in police training, seems to admit that the tools may work best to improve the legitimacy of police in the public eye, at least as much as actually reducing the number of extrajudicial killings. (A populace that trusts the police, they claim, is more likely to “comply with officers’ directives and to co-police neighborhoods.”) The proliferation of algorithmic sentencing tools attests to the ways racial disparity is explicit and objective, untethered to individual psychology. Perhaps it’s easier to talk about implicit racism than to acknowledge all the ways racism is conscious, enshrined in our founding documents and cherished institutions.

From associationism to racial liberalism

In 1879, Francis Galton, the English psychologist, statistician, and eugenicist conducted the first word-association test on himself. He cut 75 slips of paper and wrote a different word on each one. Then, drawing a word at random, he recorded the first words that came to his mind. After running several trials, he noticed that his most frequent associations seemed to stem from earliest in his life. It was Carl Jung who added the stopwatch. Jung and the American psychologists who came in his wake argued that a longer interval between the stimulus word and the response could be an indication of repression, or subterfuge, on the subject’s part. Over time, it came to be assumed that, rather than idiosyncratic memory traces, a person’s responses to word- and picture-association tests revealed something fundamental about their personality.^[8]The strongest argument against this position is that individuals’ associations are remarkably quick to change.

I don’t know who first used an association test to study racial prejudice. But the most famous such experiments took place as part of The Authoritarian Personality (published 1950), a large study conducted under the auspices of the American Jewish Committee and led by the Frankfurt School theorists Theodor Adorno and Max Horkheimer. The AJC created a modified version of the Thematic Apperception Test (in short: look at a picture and make up a story about it) designed to attract racially charged projections—a pair of “zootsuiters or young jitterbugs,” a man and woman in a city slum, a foreign-looking man in a jail cell, and a black child with his mother or grandmother. Betty Aron, one of the AJC scientists, wrote that fascistic individuals “project onto members of minority groups” the “primitive, unsublimated forms of expression so often found in their fantasies.”^[9]Authoritarian Personality, 494-95, 526. At the same time, the lead authors of The Authoritarian Personality worried that the culture of testing posed a threat to democracy. Adorno and Horkheimer were especially dismayed by the erasure of conscious reflection from psychology, what 1947’s Dialectic of Enlightenment described as a “regression to judgment without judging.” And in Minima Moralia, Adorno argued that opinion polls, behaviourist lab work, and the psychoanalytic technique of free association all opened the door to a dangerous form of irrationalism:

The way into the patient’s unconscious is laid open by persuading him to forgo the responsibility of reflection…. Thought, in allowing itself to be reminded of its unconscious origins by the administrative structure of analysis, forgets to be thought. From true judgement it becomes neutral stuff. Instead of mastering itself by performing the task of conceptualization, it entrusts itself impotently to processing by the doctor, who in any case knows everything beforehand. (§44)

The Authoritarian Personality tests, in contrast, were designed to provoke reflection. Adorno wanted the participants to feel somewhat alienated from their answers, the better to see the alienated character of all interaction and avoid the traps of anti-Semitic and reactionary thinking. These Brechtian ambitions were not shared with all of his colleagues, and in any case, they were sorely disappointed. Many test subjects did not recognize the irony behind cartoon characters like “Mr. Biggott.”^[10]David Jenemann, Adorno in America. And so I have to regard The Authoritarian Personality as one more chapter in a history of depoliticizing social conflict, from which the IAT seems like a further regression.

The IAT, as we have seen, provides no time for reflection. That is its claim to scientific truth. Today, “judgment without judging” has been redescribed as a positive–not just in behavioural-economics bestsellers, but also in some areas of left social theory. Political theorist William Connolly describes the cultivation of ethics as brain work: “working tactically upon a dense sensibility whose layered composition is partly receptive to direct argument and deliberation, partly receptive to tactics that extend beyond the reach of argument, and partly resistant to both.”^[11]William Connolly, Neuropolitics: Thinking, Culture, Speed, 106 We are increasingly aware of just how partial the “reach” of argument and rationality is, how irrational their underpinnings–and yet I cannot fathom giving up on deliberation, cannot bear the thought of a neuropolitics without a politics.

I do not object to implicit bias training–especially not if it works. It doesn’t have to be at odds with de-escalation or mental health training initiatives. But the emphasis on unconscious processes and instantaneous decisions bears closer examination. Most of all, one must be wary of the self-congratulatory glow which comes from tackling systemic problems with individual solutions.^[12]Many activists and educators have been working on issues of policing and racial violence longer, more intimately, and more systematically than I have. Here are just a few: Black Lives Matter; Audre … Continue reading