The Problem With Unconscious Bias Training Is....

There are three general views that I take on the subject of the unconscious and biases, but let’s imagine that the opposite of each holds much better ground, which would mean: 1) we have valid and reliable tools to show that unconscious processes fundamentally guide our thinking and behaviour, 2) we are inherently biased, and 3) unconscious biases can be eliminated if we find the appropriate tools to do so. In the following thought experiment, the aim is to see where some of the views take us.

So, the first thing that we might be interested to know is what proportion of the decisions we make in any given day fall into each of the categories listed in Table 1. Heuristics are broad rules of thumb that have some validity to them (e.g., using clarity of a perceived object as an indicator of its distance), and biases are the consequence of the overuse of heuristics, where essentially they lead to the overweighting of some information at the expense of other potentially relevant information with which to make a judgement or decision (e.g., distances are often overestimated when visibility is poor because the contours of objects are blurred).

Table 1

Source: Author

Assuming that we can trust that our tools can accurately measure a person’s decisions and the processes that generate them, and also given that a popular view is that we are inherently biased, and that unconscious processes drive what we do, we already have some idea of which of the cells we want to focus on populating in Table 1.

So, the only other matter to consider is whether the decisions are consequential or not. It is difficult to dichotomise decisions neatly into whether a decision is consequential, but for the sake of simplicity, we want to consider decisions of consequence as those that specifically impact protected values (e.g., protected by existing equality legislation related to age, disability, gender reassignment, race, religion or belief, sex, sexual orientation, marriage and civil partnership and pregnancy and maternity).

When it comes to doing this we need to go a step further because we have to decide which of two general assumptions to accept, which is the basis on which we determine the prevalence of consequential decisions, which results in something like Figure A or Figure B.

Figures A and B

Source: Author

One way to make the determination could be based on what profession an individual has, and to examine the frequency of decisions that impact multiple people (e.g., medical, legal, educational). But this might seem too restrictive since it only considers decisions taken in the workplace and means that not all of us are making consequential decisions frequently.

A much broader way to make the determination is based on the view that if you accumulate all decisions then overall they are of consequence because they shape our beliefs and attitudes, and collectively signal preferences at a population level. So this means that personal and professional decisions are consequential, including the food we choose to eat, the books we read, the films we watch, who we spend the most time with, what social media accounts we have, and what news media we most attend to, as well as our workplace decisions.

The basis on which we decide (professional only vs. personal and professional) impacts the way we end up classifying daily decisions. If we go for the former, then we further classify people according to whether they are in positions that mean they make frequent consequential decisions or not (some will be of Figure A, and some of Figure B). If we go for the latter, then everyone is considered as making consequential decisions (Figure B).

If we opt for the broadest assumption, and go with the popular claim that nearly all our decisions are based on the unconscious, and that we are inherently biased, then we get something like the results presented in Table 2.

Table 2

Source: Author

What this suggests is that no one escapes unconscious bias, and everyone makes consequential decisions that are biased. Irrespective of whether you are within any one of the protected values, you will be making decisions (inadvertently as a result of unconscious bias) that will negatively affect any one of a number of protected values all the time. In turn, this means that we are all regularly the victims of unconscious biases because cumulatively the decisions people have made are of consequence and are likely to impact our day-to-day lives. This also means that we predominately behave in ways that are beyond our conscious control, and that our sense of control and sense of agency are somewhat illusionary.

If unconscious bias training is a route out of all this, then we would need to know whether the methods (of which there are many) are valid and reliable and lead to sustainable effects. If they are, as yet, invalid and unreliable and do not achieve sustainable effects, then there are one of two views that could be taken to explain this. The first is that there is room for improvement, and we are still having to find out the best way to target unconscious biases effectively. The second is that the theoretical and empirical basis of the concept of unconscious processes, unconscious biases, and training out of unconscious biases is unsound.