The Flaw in Unconscious Bias Training

Do you have a job that makes you take a compulsory training course on unconscious bias—maybe once a year or upon taking up a new position? I remember being confused by the term the first time I heard it in a human resources context and wondered what the difference was between it and implicit bias. It turns out, there's no difference, so why not stick with the established implicit bias?

I wonder if the naming strategy is simply a means of trying to simplify the concept for the masses. To any readers au fait with the rationale, please get in touch and let me know. With that, the concept is relatively simple: All human beings are biased in light of previous experiences, beliefs, thoughts, and emotions—and oftentimes, they’re not aware of it, hence its implicit nature. Together, these help form cognitive structures, known as schemas, which act as mental frameworks that help us interpret and guide us through our (social) worlds.

Schemas are a good thing; they are evolutionarily advantageous. They permit us to make quick, split-second decisions without effort or cognitive burden. However, as I often point out in relation to reflective judgment, though automatic, intuitive judgment is quick and easy, but one shouldn’t rely on it when we’re making important decisions. When an intuition (i.e., reliance solely on a schema) is wrong, it’s going to be very wrong. Given that we want employees to engage sound decision-making for things that matter, we need them to be aware of their intuitive, implicit bias and more strongly evaluate scenarios in advance of making decisions—hence the impetus for "unconscious" bias training.

I haven’t had to take such a training course in a good while and, truth be told, they used to get my back up bit—for a number of reasons. A couple of friends of mine have coincidentally taken separate unconscious bias training quite recently and spoken with me about it. Their grievances were similar to the problems I recall having with such courses, so I figured this would be a good platform to discuss.

First, I use the term course liberally here in reference to unconscious bias modules, as for many of these, it’s e-learning training that can be finished in a half hour—generally amounting to a couple of videos, a short exercise or two, and a quiz at the end—certainly not enough to bestow any meaningful instruction. It’s a box-ticking exercise from which little to nothing is actually gained, which is a shame. If you’re going to do it, do it right.

As I teach and research bias, I instinctively compare my own approaches with the material covered and how it’s delivered, typically leaving me to second-guess the efficacy of common approaches to training discussed here; "doing it right" isn’t always accomplished. Of course, I realize the potential for me to be suffering from a self-serving, false uniqueness bias here (e.g., they’re not doing it as well as I would), but this is a reasonable conjecture given the brevity of this training. If you’re an institution concerned that your employee population could be contaminated by unconscious bias in important decision-making scenarios, dedicate time and effort to such training: A half hour is not enough.

Second, the way in which key terminology is presented isn’t always entirely accurate. This may be a result of not providing enough clarity or context, given the limited time dedicated to the training, or, worse, may actually reflect the biased thinking of course developers—in which case, the whole situation is an ironic form of "meta-bias," a biased course on bias. Arguably, the two of these issues can be seen to converge in the following example:

During one course, I was asked a question following a short video about a fictitious company, regarding a topic that had not been addressed by the video: "Who is the CEO of this [fictitious] company?" Three options were given—as I recall, they were names like: John Taylor, Mary Smith, and Aisha Mahmood. I selected John and got the answer wrong; the system told me it was Aisha, and that my unconscious bias should not have excluded her because she was a woman or had a "non-Western-sounding" name.

I was annoyed, because, first, that was not my rationale; and second, no context was given. Thus, other than guessing at the sexes and ethnicities of the hypothetical people, all I had to go on was statistics. Statistically speaking, it was likely to be the man. Without context, objectivity is the only way to go; so, I went with the stats. Nevertheless, my answer was determined to be wrong—I thought maybe this was a clever mechanic of the software in that whomever I chose would be wrong and that it would be subsequently explained to me that without context, I shouldn’t make such guesses (that actually would have been a nice touch). I did the question over and chose Aisha. I was "correct" this time, and the program progressed. So, it wasn’t a clever tactic to show me how unconscious bias works; rather, they wanted to make it clear that I should be thinking of Aisha as the CEO, for whatever reason—and I say "for whatever reason," because it’d be wrong to assume without context.

Bias Essential Reads

Women Without Children Still Face a "Motherhood Penalty"

Are You Blind to Subtle Forms of Workplace Discrimination?

The problem with a scenario like this is that, for unconscious bias to be adequately detected and overcome, we need context. In fact, every decision we care about needs context. As I tell my social psychology students, whenever you are faced with a question to which you don’t know the answer, "it depends" is a perfectly adequate response. Instead of being forced to pick the name of a hypothetical person in this unconscious bias training exercise, "It depends" would have been a much better fit.

The problem with unconscious bias training is not the message it is trying to convey—it's actually a very important message, with which I completely agree. However, oversimplifying the concepts (e.g., stereotyping and prejudice) has the potential to facilitate misinterpretation of the core messages—perhaps as an arbitrary venture into value signaling and moral grandstanding, thus yielding a backfire effect of sorts (e.g., see Lewandowsky et al., 2020). So, with respect to developing unconscious bias training, if you're going to do it, do it right.