Farid Alsabeh: What neuroscience can tell us about implicit bias

Saturday, June 16, 2018 - 9:46am

The April arrest of two black men at a Philadelphia Starbucks has reignited public discourse on the topic of racial discrimination. Arguments proposed by the most vocal critics of the incident, who have charged the coffee chain with racism, portray it as the latest scene of a systemic white supremacy that has terrorized minorities for decades. In a post-Jim Crow America, the question must be asked: How could such a widespread manifestation of racism persist in our society? The consensus has been, as far as Starbucks’ perception is concerned, that unconscious profiling is the culprit.

The idea that motivations can exist beneath the level of conscious knowledge is hardly a new idea; this was precisely the scandalous observation that set a prude Viennese aristocracy on fire in the days of Sigmund Freud. Marked as it is by racial tension, our current era takes as the object of scandal not implicit sexual attitudes but racial ones. As it applies to the Starbucks arrests, the argument states that even if the barista harbored no explicit prejudice against African-Americans, she was nonetheless motivated by an unconscious bias to call the police when the men declined to order anything. The assumption here is that race played a decisive role in the events that unfolded, that the arrest would not have been made if the men were white.

The Implicit Association Test has emerged as the most prominent assessment for the kind of implicit bias that has been implicated in the Starbucks incident. For those unfamiliar with the test, I encourage you to try it — it doesn’t take more than 20 minutes, and the firsthand experience will be a far better explanation than what follows. In the relevant version of the test, there are two parts. one is when there are two buttons: “white and good adjectives” and “Black and bad adjectives,” and the other is when the two buttons are “Black and good adjectives” and “white and bad adjectives.”. The finding of this simple test is similar to others in psychometry — the practice of timing mental processes — in that the speed of our categorization changes significantly based on another parameter. In this case, that other parameter is the face we see prior to the word: It consistently takes people longer to identify positive words, and shorter to identify negative words, when Black and good are the same button, because of implicit bias.

Perusing other versions of the assessment, it was startling to consider just how many identity groups may be affected by the phenomenon of unconscious bias. Age, weight, and physical disability among athletes were some of the most striking examples outside the typical race or sex considerations, all three of which have been associated with the classic IAT finding of delayed response time. The diversity of subgroups present in the assessment is a testament not only to the potential ubiquity of implicit bias, but to the rising sociocultural movement that is bringing this issue to the forefront. The floodgates holding back the uncomfortable topic of unconscious marginalization have been opened.

As far as strictly scientific evidence is concerned, the IAT is still as close as we can get to quantifying the phenomenon of implicit bias. But even this has been scrutinized by scientists wary of the test, a skepticism that has slowly but surely been extending into the field of social psychology at large. Some psychologists maintain that the results of this assessment suffer from low replicability, and that the test itself fails to live up to quality-control standards. Moreover, the IAT seems completely inept at doing what it’s designed to do: actually predict potential acts of bias. Despite these criticisms — if only for the sake of argument — I’ll assume that these results are valid and consistent. Where do we go from there?

It’s my belief that the psychometric approach outlined by the IAT is not a rigorous method of demonstrating that unconscious racism exists per se. This position was largely influenced by my own experience taking the test. Within a few rounds of the assessment, I found myself using a simple verbal strategy to my advantage: Whenever the Black face and negative word prompted the same key, I identified both with the verbal association Black/bad, an alliteration which I felt made my responses faster and more accurate. If the similar phonology of these words contributed to the increased response rate, this seems to cast doubt on the idea that negative emotional responses are necessarily at play in the IAT. Now, it’s certainly the case that not everyone was using this makeshift strategy, so I can’t argue empirically that this explains the results. But I think this demonstrates on theoretical grounds that a delay in response time is not necessarily caused by negative emotional biases against a particular race.

Still, influential studies have shown that preferences for names can result in a detectable and unconscious bias against minority groups. So the question is, if not the IAT, how else might we go about quantifying the phenomenon of implicit bias? In order to arrive at a more rigorous demonstration, I think the neurobiological findings by University of Southern California neuroscientist Antonio Damasio should be considered. According to his somatic marker hypothesis, our rational deliberation of options is not a purely indifferent process, devoid of any emotional content. Instead, certain outcomes are “marked” by physiological responses according to our own mindset and experiences.

Under this framework, a demonstration of implicit bias becomes clear. Imagine a manager who, in two separate cases, considers whether to call police on someone who has entered their establishment without ordering anything. If the physiological responses of such a decision, or something like it, could be measured (say, by using galvanic skin detectors, which measure our stress response), we might have objective evidence that negative emotions underlie our seemingly bias-free decisions when race is involved. Unlike the psychometric IAT test, which relies on the circumstantial fact of delayed response, I think such a physiological result could provide a definitive measure of implicit bias.

That said, I encourage everyone to take the IAT for themselves and grapple with the question of what scientific result constitutes direct evidence of negative racial bias. To reiterate, I feel that we haven’t yet arrived at an objective standard to definitively say that a certain individual is “unconsciously racist.” A scientific, and strictly neurobiological, approach might someday take us there. The history of America has been fraught with racial strife and injustice for a majority of its history; let’s all contribute to progress by tackling these difficult questions with an open and deliberative mind.

 

More like this