Stereotype threat is a phenomenon described in Steele and Aronson (1995), in which invoking a negative stereotype about a group to which an individual belongs can cause that individual to perform below his or her actual ability. Calling attention to the fact that an individual belongs to a group stereotypically less skilled at a particular task can cause that person to perform more poorly at it. The original study considers African-Americans’ performance on standard aptitude tests, but many other studies have been conducted since then. Other studies have considered negative stereotypes about women’s mathematical aptitude and related them to women and girls’ performance on math tests (Spencer et al. 1999; Ambady et al. 2001). In general, what such studies have found is that negative stereotypes, particularly when highlighted, can become self-confirming.

However, several points about stereotype threat ought to be highlighted. First, one does not have to endorse the negative stereotype to be affected by it, so long as it is a recognizable stereotype in one’s culture. Women who do not believe that gender affects mathematical ability can nevertheless underperform as a result of stereotype threat. It is less a matter of our own self-conception than of our perception of the way in which others might see us. Being labeled as an individual who is rationally inferior, or less skilled in argumentation, can become self-confirming for an individual previously confident in her own abilities.

Second, many people belong to several different groups to which stereotypes are associated, some of which may conflict with each other. What makes a study such as Ambady et al. (2001) particularly interesting is their investigation into exactly this phenomenon, which looks at mathematical performance among Asian-American girls. Asians are typically stereotyped as being good at math, while girls are typically stereotyped as being bad at it. The girls in the study were first asked to color a randomly selected picture before taking a standardized math test. The three pictures girls could have received to color were intended to activate their female identity (a girl holding a doll), their Asian identity (two Asian children eating from rice bowls), or neither (a landscape). For most age groups, the best performances were among the girls whose ethnic identity was activated, and the lowest among girls whose gender identity was activated, with the control group intermediate between the two.¹ So due to the complex nature of many people’s identity, it is possible to affect performance on certain tasks by activating one stereotype or another about a group to which they belong.

The second psychological concept we will discuss is implicit bias. Where stereotype threat has largely to do with a person’s views about how she will be perceived, implicit bias has to do with the way in which others actually do see her. We will look primarily at ways in which implicit bias can affect others’ judgments of credibility about an individual. Now, judgments of credibility are not always conscious, and especially when unconscious, may be affected by negative stereotypes having to do with a person’s identity. For example, studies of implicit bias have shown that factors such as race and gender can affect even well-meaning individuals’ assessment of job candidates. One study found that fictitious resumes of identical quality sent out to employers were much more likely to receive callbacks if they were attached to a traditionally white name than if they were attached to a traditionally African-American name. This was even the case among employers who explicitly state that they are equal opportunity (Bertrand and Mullainathan 2004). Another similar study looked at the effects of gender, by sending identical CVs to various academic psychologists for evaluation, but varying the name. Some were given typically male names and others, typically female names. In general, the finding was that the CVs with male names were evaluated more highly than their identical counterparts with female names attached. Also important is that there was no significant difference between men and women’s evaluations of the CVs – both had a tendency to rank the male candidates more highly (Steinpreis et al. 1999). So judgements made by members of marginalized groups may still be affected by negative stereotypes, even when those are stereotypes about a group to which they belong.

This last fact may seem counterintuitive, but an important aspect of this type of bias is that it tends to manifest itself in ways that the biased person is typically unaware of. What makes it a particularly difficult thing to combat, or even mitigate, is that people who do have implicit biases do not generally see themselves as being influenced by bias. For instance, the bias blind spot is the commonly held (but mistaken) belief that one’s own judgments are less susceptible to bias than the judgments of others (Pronin et al. 2002; Ehrlinger et al. 2005). This means that even under very good conditions, in which we have a well-meaning person who does not harbor conscious prejudice, and is even aware of biases to which she might be susceptible, we still see the effects of implicit bias. Now, in order to articulate some of the harms that can result from these psychological phenomena, we will turn to the concept of epistemic injustice.

Epistemic injustice, as discussed in Fricker (2007) in particular, is a kind of epistemic wrong done to an individual, in her capacity as a knower, as a result of systemic injustice. Her main focus is on testimonial injustice, which stems from our often unconscious assessments of a speaker’s credibility. Literature on the epistemology of testimony does not give a uniform account of the manner in which we come to accept testimonial evidence, but it is acknowledged that some judgment on our part, whether explicit or implicit, of the testifier’s credibility plays a role. Put simply, we are less likely to accept a claim if we do not see the person making it as credible. And prejudice can result in a person’s being assigned a lower degree of credibility solely on the basis of a negative stereotype about a group to which she belongs. There are many ways in which this can actually happen – probably as many as there are factors involved the assessment of credibility. However, trustworthiness and competence can be singled out as important dimensions of credibility assessment, and both can be negatively impacted by prejudice. African-American males in North America are often unfairly criminalized, and this negative stereotype can affect assessments of trustworthiness. In different situations, members of racial groups who have negative stereotypes assorted with business practices might be assessed as less trustworthy than people who are not members of those groups. Similarly, there are negative stereotypes about competence at particular tasks. We have already mentioned negative stereotypes about women and math performance, but we will shortly discuss problematic stereotypes about gender and rationality that can cause women to be negatively evaluated. Fricker, in her book, makes use of an example from the screenplay from the film The Talented Mr Ripley, in which a woman is told “Marge, there’s female intuition, and then there are facts.” (Fricker 2007, p. 88) Fricker discusses this instance of testimonial injustice in more detail than will be covered here, but it illustrates the point, at least, that identity prejudice can seriously affect the reception of a person’s claims. In this case, Marge, despite intimate knowledge of the subject in question, is dismissed because of her gender, and her views are discounted.

There has been some work done in adapting this idea to arguments, and developing a concept of argumentative injustice (Bondy 2010). This is an analogous concept to Fricker’s in that it involves harm done to an individual due to false identity-prejudicial stereotypes. However, instead of harming someone as a knower, it harms her as a reasoner, or someone capable of drawing conclusions from premises. While Bondy does cite some disanalogies between his concept and Fricker’s, the issues under discussion in this paper could easily be discussed in terms of either or both.² Particularly when we consider the close relationship between ad hominem fallacies and testimony, it ought not matter too much which term we use to talk about the injustice being done – whether it is a wrong to the person as an arguer or as a source of good information. So at least in this particular case, testimonial injustice and argumentative injustice intersect. As such, we will continue to use the original term “epistemic injustice,” recognizing that our examples fall into both categories.

Some treatments of fallacies do consider their effect on dialogues in general. One particularly good treatment is Woods and Walton (1982), in which we see disagreements between two agents in a romantic relationship: Bill and Sue. Woods and Walton use these characters as part of a running example in order to illustrate different ways in which agents might disagree. For instance, they provide examples of their disagreeing about the facts, such as what Bill might have said on a particular occasion. They also provide examples of their drawing different conclusions from the same facts. The former is called premissory instability, and the latter, conclusional instability. These concepts are used to show when an argument becomes a quarrel, which is often what the word “argument” is taken to mean in ordinary language contexts. Quarrels, however, are typically unproductive and unpleasant.

If, as in the case of premissory instability, we cannot even get started on the road to agreement, then frustration, accusation, and hurt feelings are bound to occur. References will tend to become personal and disagreeable. Sue might eventually complain that if Bill can’t recall what he said last Friday, then he is a simpleton; Bill might retort that Sue is a hysterical shrew. Before you know it, things will have taken another nasty turn. Similarly, having got the discussion nicely under way with some basic premissory agreement, things might come grinding to a halt owing to a lack of common conclusions. Then the same personal disruptions could occur. Bill might contend that Sue shows herself to be a “typical woman” in having no capacity to reason beyond her nose, or to perceive what follows from what. And Sue may earnestly offer to slap Bill’s moronic face (and perhaps be forgiven for it.) (Woods and Walton 1982, p. 4)