In this post, Katharina Berndt Rasmussen (Stockholm University & Institute for Futures Studies) discusses her recently published article in the Journal of Applied Philosophy (co-authored by Nicolas Olsson Yaouzis) exploring the roles that implicit bias and social norms play in discriminating hiring practices.

The US, like many other countries, is marked by pervasive racial inequalities, not least in the job market. Yet many US Americans, when asked directly, uphold egalitarian “colour-blind” norms: one’s race shouldn’t matter for one’s chances to get hired. Sure enough, there is substantial disagreement about whether it (still) does matter, but most agree that it shouldn’t. Given such egalitarian attitudes, one would expect there to be very little hiring discrimination. The puzzle is how then to explain the racial inequalities in hiring outcomes.

A second puzzle is the frequent occurrence of complaints about “reverse discrimination” in contexts such as the US. “You only got the job because you’re black” is a reaction familiar to many who do get a prestigious job while being black, as it were. Why are people so suspicious when racial minorities are hired?

In a recent paper, my colleague Nicolas Olsson Yaouzis and I offer an explanation for both puzzles: we model the workings of implicit racial bias in a population of egalitarian norm followers. Implicit biases have been shown to affect basically all of us. They are, roughly, automatically activated stereotypes about social groups. They are often unnoticed and unendorsed by their bearers. And they correlate with social inequalities on population levels. But how, exactly, should we understand the underlying mechanism? Here’s our model:

Imagine a big firm consisting of a large number of subsections, each headed by a manager. From time to time these managers hire new people. Assume that the firm was initially all-white (think of the early sixties Mad Men era). Managers knew that there was some great competence among the black applicants, which would benefit the firm overall. Still, each preferred to head a racially homogenous subsection, because it saved them trouble. They were thus all trapped in a prisoner’s dilemma: each doing what was (supposedly) better for them, while the firm missed out on competence.

Social norms solve prisoner’s dilemmas. Suppose that (in the early seventies) the managers become aware of an egalitarian social norm: “When hiring, hire the most competent candidate, regardless of their race”. This norm changes their incentives: as long as they believe that enough others will both comply with it and expect them in turn to comply, they want to comply with it themselves. Imagine now that each manager comes to believe this about the other managers. Each then complies with the egalitarian norm and hires black applicants whenever they are the most competent. Slowly, the racial composition of the firm will change.

However, if there are implicit racial biases among the managers, this will sometimes distort their actions. They want to comply with the norm but sometimes make mistakes. Due to the nature of implicit bias these mistakes are asymmetrical. That is, they sometimes occur when the most qualified candidate is black (such that a white candidate is hired instead), but never occur when the most qualified candidate is white (such that a black candidate would be hired).

Now, assume that there are many decisions and many managers, so mistakes add up. This could, on the whole, explain large scale hiring inequalities. But this would mean that each of them, observing that the egalitarian norm is violated time and again, would cease to believe that enough others complied with it. Moreover, observing that such frequent norm violations are not met with protests by the others, each manager would cease to believe that enough others expect them in turn to comply. Then, each would no longer want to comply and the norm would break down.

Yet this is not what seems to happen. Rather, the norm stays in place (people uphold the norm when asked) and large-scale hiring discrimination persists (causing the pervasive racial inequalities). Our model can account for this, by illuminating the intricate interplay between implicit bias and job competence. To see this, consider a specific recruitment case, where the most competent candidate is black. Their race is a clearly observable feature – their competence typically is not. Suppose that the hiring manager makes an implicit bias-mistake and hires a less competent white candidate. The other managers likely cannot directly observe that a norm violation has taken place. They can, however, observe the successful candidate’s race. If they (like most of us) hold implicit racial biases, they may perceive the white candidate as more competent than they actually are, and (falsely) infer that the egalitarian norm was followed. Thus no one protests, and no one changes their belief that enough others comply with the norm and expect them in turn to comply. The norm may be repeatedly violated, but does not break down. This explains the first puzzle.

Now consider the same case, but suppose the manager doesn’t make a mistake. The most competent black candidate is hired. Again, the others can observe the candidate’s race but not their competence. Again, if they hold implicit racial biases, they may perceive the black candidate as less competent than they actually are – and (falsely) infer that the egalitarian norm was violated. If this happens repeatedly, observers might eventually (falsely) conclude that the egalitarian norm has come to be replaced with a norm of “political correctness”: “When hiring, hire the most qualified minority candidate (to increase firm diversity)”. This explains the second puzzle.

In sum, we propose a toy model of hiring decisions. Those are, of course, much more complex in real life. Still, the model helps us see the double role of implicit bias: in the hiring decisions themselves, and in bystander evaluations of these decisions. It solves the two puzzles by explaining how hiring discrimination can be invisible in seemingly egalitarian social contexts — and why instead non-discrimination may appear suspicious.

A final word: in a world where explicit racism is on the rise, why worry about implicit bias? Our analysis does not imply that explicit racism doesn’t matter (and we certainly think it does). It just shows that focusing narrowly on eradicating explicit racism will not be enough.

Katharina Berndt Rasmussen

The Journal of Applied Philosophy is a unique forum for philosophical research that seeks to make a constructive contribution to problems of practical concern. Open to the expression of diverse viewpoints, it brings the identification, justification, and discussion of values to bear on a broad spectrum of issues in environment, medicine, science, policy, law, politics, economics and education. The journal publishes in all areas of applied philosophy, and posts accessible summaries of its recent articles on Justice Everywhere.