Does Race Best Explain Racial Discrimination?

  • Keshav Singh orcid logo (University of Alabama at Birmingham)
  • Daniel Wodak orcid logo (University of Pennsylvania)


Our concern in this paper lies with a common argument from racial discrimination to realism about races: some people are discriminated against for being members of a particular race (i.e., racial discrimination exists), so some people must be members of that race (i.e., races exist). Error theorists have long responded that we can explain racial discrimination in terms of racial attitudes alone, so we need not explain it in terms of race itself. But to date there has been little detailed discussion of whether it is better to explain racial discrimination in terms of race or in terms of racial attitudes alone. Our goal is to offer a novel and detailed argument in defense of explaining racial discrimination in terms of racial attitudes alone, by attending to the neglected phenomenon of misperception discrimination, which involves differential treatment due to misperceived race. We argue that the discriminatory action in misperception cases must be explained in the same way as cases where (according to the realist) the victim’s race is accurately perceived. Thus, the victim’s actual race cannot provide the best explanation. The main upshot of our argument is that explanatory arguments from racial discrimination to realism about race fail.

Keywords: race, racial discrimination, racial error theory, social construction

How to Cite:

Singh, K. & Wodak, D., (2024) “Does Race Best Explain Racial Discrimination?”, Philosophers' Imprint 23: 24. doi:



Published on
13 Feb 2024
Peer Reviewed

I – Introduction

A central debate in contemporary philosophy of race is between realists and error theorists. Realists say races exist; error theorists deny this. Many realists — especially social constructionists — have argued from racial discrimination to realism. Say that A fires B for being Black. The common argument is that because B’s being Black explains A’s firing B, B must be Black, so individuals must have races. We contend that this argument fails once we attend to the neglected phenomenon of misperception discrimination. Say C fires D for being Black, but D is not Black. Our core argument is simple. D’s actual race cannot explain why C fired D. Instead, C’s perception that D is Black is the reason why C fired D. But if that is true, there is considerable pressure to offer this same explanation in the case of A’s firing B. In other words, misperception discrimination pushes us to explain instances of racial discrimination in terms of the discriminator’s racial attitudes, not the discriminatee’s actual race. The upshot is that explanatory arguments from racial discrimination to realism fail. For all we say, realism may be the correct theory of the metaphysics of race, but racial discrimination provides no good argument for it.

II – The Target

Let us begin by clarifying the target. The argument that racial discrimination requires us to be realists is often explored or advanced in overviews of the realism–error theory debate, such as that of Alissa Ney and Alan Hazlett:

[C]onsider the ongoing history of racism and racial oppression, which is constituted by real events. For example, suppose Sarah, a bank manager, suffers from an unconscious racist bias against Samoans. Maria, her employee, descended from ancestors from Samoa, has applied for a promotion at the bank. Sarah rejects Maria’s application. We might ask: Why was Maria’s application rejected? In such a case, it seems perfectly possible that among the causes of the rejection was the fact that Maria is Samoan. This, together with Sarah’s implicit bias against Samoan people resulted in Maria’s application being rejected. But if the fact that Maria is Samoan caused her application to be rejected, then races have causal powers. And if races have causal powers, then we should count them as existing (2014: 276, italics in original).1

There are interesting differences in how realists advance this argument. Sally Haslanger, for example, is more concerned with groups’ races explaining discriminatory patterns. “Race is used to explain a broad range of differences between social groups, including educational attainment, patterns of arrest and incarceration, health outcomes, social history, etc.,” Haslanger argues (2019: 18). Because of this, “an error theory about race has substantial costs… We would need to give up the idea that race explains certain group differences” (20). Similarly, Ásta’s chapter on race in Categories We Live By begins by suggesting that because race explains social phenomena such as discrimination, we should be realists about race, and social constructionists in particular (2018: 93):

Race is not a scientifically respectable explanatory category, as no natural phenomena can be explained by appealing to race (as opposed to some notion of a population with a shared history or a kindred notion). A host of social phenomena can, on the other hand, be explained by reference to a person’s race, ranging from the statistical likelihood of becoming a prison inmate to completing a college degree or being shot by police. Races are social categories.

There are other examples of this kind of argument,2 not all of which are about race.3 But notably, while they are commonly cited as supporting the reality of race, such arguments are rarely defended at length.

Error theorists offer a simple response to this argument:

Anti-realism about race can capture every moral phenomenon that racial realism can. The difference is in how they frame the phenomena. In particular, Lawrence Blum (2002), Tommie Shelby (2005), and others have drawn attention to this solution:

We can talk about real racialized groups even if there are no races.

We can say that anti-black discrimination is discrimination that targets people who have been racialized as black (Glasgow 2019: 137, italics and citations in original).4

What is it to be racialized as Black?5 This is a function of racial attitudes. As Khalifa and Lauer note, “racialization typically includes misrepresentations and false claims. This is why it is important that racialization only entails that groups are represented as races, since X’s being represented as Y does not entail that X actually is Y” (2021: 5, italics in original). Error theorists propose that we do not need to posit races to explain racial discrimination; racial attitudes can do the requisite explanatory work.

Ney and Hazlett offer a similar account of the error theorist’s response:

[A] correct causal explanation of the fact that Maria’s application was rejected would cite the fact that Sarah thought Maria was Samoan — and that this is enough to explain what we were inclined to explain by appeal to Maria’s actually being Samoan (2014: 276–77, italics in original).

In other words, the error theorist holds that Maria can be subject to racial discrimination without having a race, because Sarah’s (the discriminator’s) racial attitudes are sufficient to explain the discrimination.

Unfortunately, this error-theoretic response is also more often asserted than defended at length. As a result, while this disagreement in the realism–error theory debate is long-standing, it has not progressed far. Realists tend to appeal to the explanatory power of races almost in passing,6 and error theorists tend to reject the explanatory indispensability of races in passing too.7 Neither engages much with the relative virtues of these rival explanations of racial discrimination. Grant that we could explain instances of racial discrimination in terms of the discriminatee’s race or the discriminator’s racial attitudes. Which is the best explanation of racial discrimination? That is our central question.

To date, surprisingly little has been said in answer to this question. In their lengthy recent discussion of race and racial discrimination, Khalifa and Lauer note that anti-realists tend to appeal to the hypothesis that “some groups are racialized, but races do not exist” and hold that a racialization-only explanation “has no difficulty accounting for racial discrimination” (2021: 9).8 However, they also explicitly hold that the realist’s hypothesis is “one of many equally good explanations” and take no stand on whether the error theorist’s hypothesis is a better explanation (2021: 8–9). For related reasons, Charles Mills argues that “despite initial appearances, the ontology [of race] matters less than one might think for the moral issue” of racial justice (2018: 73). Why? Because on most realist views, we can take race to explain “invidious discriminatory treatment,” whereas on most antirealist views we can say that there are “groups wrongly believed to be races, [who are] historically subject to discrimination on that basis.” The two explanations of racial discrimination are presented by Mills as being on a par.

It is, of course, possible that the realist and error theorist have advanced equally good explanations of racial discrimination. But we think this verdict is premature. One very significant factor in determining which explanation (if either) is better has been unfortunately neglected in the debate to date.

III – Misperception Discrimination

Our goal in this paper is to defend at length the view that we should explain racial discrimination in terms of racial attitudes, not in terms of actual races. We will defend this answer by drawing attention to the neglected phenomenon of misperception discrimination. Imagine that an employer decides to reject job applicants who have excellent CVs but whose names are stereotypically Black, such as Lakisha and Jamal. Now imagine that some of these job applicants happen to be white (or Asian, Native American, etc.). Here the employer has engaged in misperception discrimination: some applicants were differentially treated by the employer on the basis of their misperceived race.9

This example is hypothetical, but the phenomenon of misperception discrimination is real, common, and important. It divides courts tasked with applying anti-discrimination law. A paradigmatic example is Burrage v. FedEx Freight.10 After being subject to anti-Mexican harassment while working at FedEx, Nathanial Burrage quit and sued under Title VII. But the harassment was based on a misperception. Burrage was not Mexican. The court sided with FedEx, taking the view that “Title VII contains no provision for those wrongly perceived to be of a certain national origin.”11 Here, the court followed related precedents where plaintiffs misperceived to be members of other protected classes (such as religious groups) were denied protection under United States antidiscrimination laws.12 As one court stated, “the protections of Title VII do not extend to persons who are merely ‘perceived’ to belong to a protected class.”13 By contrast, other courts have extended the protection of antidiscrimination laws to those who are misperceived to be members of protected classes.14

We will contrast misperception discrimination with veridical discrimination. If Burrage had been Mexican, Burrage would have been accurately perceived by other FedEx employees to be of a certain national origin; all else equal, it would have been a case of veridical discrimination.

To make the contrast sharp, we will focus on the following pair of cases:

Veridical: After seeing Jamal’s stereotypically Black name, an employer rejects Jamal’s application. Jamal is Black.

Misperceived: Exactly as above, but Jamal is white.

We use this pair of cases because they are as similar as possible, but for the veridicality of the employer’s perception. The distinguishing feature of the explanation of racial discrimination that has been offered by error theorists is that it appeals to fallible racial attitudes (as we noted above). So, a good way to test which explanation is better is to compare otherwise identical cases where such attitudes are true and false. Hence, focusing on veridical and misperception cases in general, and Veridical and Misperceived in particular, can help us make progress in the long-standing debate between realists and error theorists about discrimination.

In light of this methodological point, it is unfortunate that misperception discrimination has received scant attention in philosophy. Misperception cases are mostly discussed by legal commentators focusing on antidiscrimination law.15 To our knowledge, they have received no attention at all from those concerned with advancing or rejecting the argument from racial discrimination to realism about race.

This attention can be directed at both camps in the debate. One can also ask what error theorists should say about veridical discrimination. After all, on their view, all cases of racial discrimination seem to be misperception cases. But that will not be our focus, as we find it plausible enough that error theorists can explain the difference between the cases by drawing a distinction within racialized groups. (For example, in veridical cases, the discriminator’s perception of the target’s group membership lines up with how the target is racialized, whereas in misperception cases there would be a mismatch.16) Our focus will instead be on the more fruitful challenge: what can realists say about cases of misperception discrimination, and what does it reveal about arguments from racial discrimination to realism?

A final point about terminology before we move on. The term “misperception discrimination” should not suggest a fixation on (visual) perception in particular. As we noted, error theorists have appealed to fallible racial attitudes in general, which is a broader category that includes beliefs and suspicions. Likewise, the term “veridical discrimination” should not suggest that all of the discriminator’s attitudes are veridical. Cases of veridical discrimination will often involve false views about the specific and general target of discrimination (i.e., the relevant individual and group). All that matters for the purposes of the distinction is the veridicality of the discriminator’s classification of the individual(s) in the relevant group.

IV – Our Central Argument

Having outlined our central question and the phenomenon of misperception discrimination, we can now outline our argument by reductio. We will first say what the realist’s explanation is committed to concerning cases like Veridical, then argue that this leaves the realist with no good options when we attend to cases like Misperceived.

For now, let us say that the discriminatory action to be explained is the employee’s rejection of Jamal’s application. Call this the rejection. We assume that the realist’s explanation is committed to saying that in Veridical Jamal’s being Black partly explains the rejection. While some realists focus on group- or type-level (rather than individual- or token-level) explanations of discrimination, our assumption is warranted. For one thing, it would be at least surprising (and arguably incoherent) to hold that at a type level, race explains racial discrimination, but at a token level, B’s race never explains A’s discrimination against B.17 For another, the realist ultimately needs to attribute races to individuals — this debate does not concern whether races exist in Plato’s heaven, but whether individuals are members of racial groups. So it is germane to assume that if racial groups explain racial discrimination, then individual membership in a racial group at least sometimes explains actions like the rejection. If you still deny this assumption, bear with us for now; we’ll return to group-level explanations of discrimination later (especially in §VI).

If one assumes that the rejection in Veridical is explained by Jamal’s being Black, what can one say explains the rejection in Misperceived? Note that whether the rejection counts as discrimination (and in particular, whether it counts as racial discrimination) depends on how it is explained. In light of this, there are three possible views that one could adopt:

  1. Misperceived is not a case of discrimination at all.

  2. Misperceived is a case of discrimination and is explained the same way as Veridical (i.e., in terms of Jamal’s actual race).

  3. Misperceived is a case of discrimination but is not explained the same way as Veridical (i.e., not in terms of Jamal’s actual race).

As we will argue, all three options fail. The upshot is that we should reject the starting assumption. That is, we should deny that Jamal’s actual race is part of the best explanation of the rejection in cases like Veridical.

IV.I Is Misperception Discrimination Really Discrimination?

The first possibility aligns with the court’s decision in Burrage: there is no discrimination in Misperceived, because in this case Jamal is not Black. More generally, on this view the phenomenon called ‘misperception discrimination’ turns out not to be a kind of discrimination at all.

Why take this view? The main motivation seems to be that Jamal cannot be a victim of racial discrimination in Misperceived, as only members of protected classes can be victims of discrimination and white Americans are not marginalized.18 This fits with a common position in moral philosophy. Scanlon holds that the only cases of discrimination are “actions that disadvantage members of a group that has been subject to widespread denigrations and exclusion” (2008: 73–74). Relatedly, many prefer anti-subordination theories of anti-discrimination law, wherein the wrong of discrimination lies in enforcing “the inferior social status of historically oppressed groups” (Siegel 2004: 1472–1473). If the rejection in Misperceived only disadvantages Jamal, who is not a member of a historically oppressed group, it is not a case of discrimination.

There are three problems with this view. First, it has objectionable moral implications, as it often requires us to deny that people have been seriously wronged. Plausibly, there are contexts where we have significant moral latitude to engage in certain conduct unless what we do is discriminatory: it would be permissible for you to hire the white candidate or the Black candidate, but not permissible to hire the former because of their race. (If you disagree with this example, switch to another, like selecting a tenant.) So what if you hire the white candidate because you mistakenly think the other is Black? If this is a context where you have moral latitude and what you do is not discrimination, how can this be impermissible, and how has anyone been seriously wronged? Some legal commentators have echoed this concern: an implication of the position underlying decisions like Burrage is that employees may be subject to differential treatment “based on an assumed protected characteristic, like race, religion, or national origin, so long as the harassers or the employer are wrong about these assumptions” (Maril & Gill 2018: 2). Even if the courts could legally justify that position as a matter of statutory interpretation, we don’t think it’s a plausible moral view. But unless one denies that in the relevant contexts we have a lot of moral latitude to engage in certain conduct unless what we do is discriminatory, it is hard to see how one can capture the verdict that Jamal is seriously wronged in Misperceived.

More can be said on this point, as there are several possible ways of maintaining that Jamal is not discriminated against in Misperceived but is wronged nonetheless; we just do not think they turn out to be plausible. To illustrate, some may hold that Jamal is wronged in Misperceived because Jamal’s application was rejected on an irrational or irrelevant basis. While this response is tempting, its problems are instructive. For one, to hold that the rejection is immoral for this reason requires giving up on the idea that we have the relevant kind of moral latitude; it requires positing a general moral duty to make decisions only on rational or relevant bases, which is implausible and subject to serious counterexamples.19 For another, even if the rejection in Misperceived is somewhat immoral because it is irrational, this cannot explain why Jamal is seriously wronged.20 Similar problems will arise for other responses to this first argument.21 But we will not dwell on the issue any further. If you are unconvinced, the two arguments that follow provide independent reasons to reject the view that there is no discrimination in Misperceived.

The second problem for this view is that it does not follow from its main motivation (that only members of marginalized groups can be victims of discrimination and white Americans are not marginalized). Someone who is a member of one protected class may be misperceived by bigots to be a member of a different protected class. Burrage illustrates this well. Burrage was Black, but was not subject to anti-Black harassment:

At best, the references to Burrage as “the Mexican” and “cheap labor,” and the use of the Spanish terms “andale” and “ariba,” represent the very unfortunate employment of offensive stereotypes of Hispanics, and can be said to arise out of a misperception that Burrage was of Hispanic descent; or at worst, they amount to incomprehensible name calling. They cannot reasonably be considered to have referred to the fact that Burrage’s race was African-American.22

Similarly, if Jamal were Native American but misperceived to be Black and rejected on that basis, we would have a misperception case despite the victim being a member of a group that is marginalized in the United States. So, if only actual members of protected classes (i.e., marginalized groups) can be victims of discrimination, it does not follow that misperception discrimination is not really discrimination.

Finally, misperception cases provide interesting ways to challenge the premise that only members of marginalized groups can be victims of discrimination. The reason Scanlon takes this to be true is that “the prejudicial judgments it [wrongful discrimination] involves are not just the idiosyncratic attitudes of a particular agent but are widely shared in the society in question and commonly expressed and acted on in ways that have serious consequences” (2008: 73–74). This would explain why an employer who rejected a white applicant because of idiosyncratic anti-white prejudice is not engaged in discrimination. But it does not explain why there is no discrimination in Misperceived. In Misperceived and Veridical, the employer’s prejudicial attitudes are identical and so equally systematic. (We will return to systematicity in §V and §VII.)

This last point speaks to a general difficulty for the view that misperception cases are never instances of discrimination. Put simply, cases like Misperceived and Veridical have far too much in common. This makes it hard to deny that misperception cases are at least sometimes instances of discrimination, which is sufficient for our purposes, as it allows us to raise the crucial question: what explains the discriminatory act in cases like Misperceived?

IV.II Is Misperception Discrimination Explained by Race?

This leads us to the second possibility. On this view, there is discrimination in both Veridical and in Misperceived, and it is explained the same way in both cases. Because it is (ex hypothesi) explained by Jamal’s actual race in Veridical, the same must be true in Misperceived.

The first way one might aim to vindicate this view would be to argue that Jamal’s being white explains the discriminatory act in Misperceived. But what causal chain plausibly links the two? Aside from this obvious issue, there is a more general, philosophically important reason why no position like this is tenable, which turns on the significance of needing an action explanation of discrimination — an explanation of the discriminatory act in terms of the agent’s motivating reasons.

To tease out this issue, imagine a case where Mary is a Black Mormon, and Mary’s being Black somehow caused Mary to be a Mormon. (Insofar as realists ascribe races causal powers, this is surely possible.) Now imagine an employer rejects Mary’s application after learning Mary’s religion but is ignorant of Mary’s race and is driven by anti-Mormon prejudice but not by any form of anti-Black prejudice. The following two claims are plausible: Mary’s being Black is a cause of Mary’s application being rejected, but the employer does not reject Mary’s application on the basis of or for the reason that Mary is Black. This is a case of religious discrimination. It is not a case of racial discrimination.

This case illustrates the importance of the fact that when explaining discrimination, we are explaining an action, not an outcome. We are asking whether the agent acted because of someone’s race, in the sense of acting on the grounds or basis of their race.23 As Eidelson states, “the grounds on which a person discriminates are marked by the links on a chain of explanation that go through her mind” (2015: 19).24 Mary’s being Black may have caused Mary to be fired, but that is insufficient to show that the employer fired Mary because Mary is Black (as an action explanation), as this link on the causal chain did not go through the discriminator’s mind.25 (This does not mean that the link in the causal chain must be explicit or conscious in the discriminator’s mind.26)

Why does this matter? Consider Misperceived once more. Even if Jamal’s being white is somehow a link on the causal chain of Jamal’s application being rejected, it cannot be a link that goes through the employer’s mind, because Jamal’s race is misperceived. So, misperception cases cannot be cases where discrimination is explained by the victim’s actual race.

However, there is a second way in which one might aim to vindicate the claim that Jamal’s race explains the discriminatory act in Misperceived. One might claim that Jamal is Black in Veridical and in Misperceived. There may be ways of supporting such a stance on some social constructionist accounts of the metaphysics of race. We will discuss this more in §V and note that there are general reasons why such accounts cannot offer better explanations of racial discrimination than their error-theoretic rivals. But for now we will restrict ourselves to a simpler point. It is a datum that Jamal’s race is misperceived in Misperceived. To deny this datum has a cost. And it is a cost with consequences. Say we take the same view across the board. There are many cases of religious misperception discrimination. When Sikhs are systematically perceived to be Muslims and subject to Islamophobic discrimination, this does not mean they are Muslims (or “function as” Muslims). There are similar costs to denying the existence of racial misperception discrimination. For instance, if on one single occasion Rachel Dolezal or Jessica Krug passed as Black and were subject to anti-Black discrimination, we doubt it would follow from this alone that they were or “functioned as” Black in that context.

IV.III Is Misperception Discrimination Explained by Something Else?

We have assumed for the sake of argument that, in Veridical, Jamal’s actual race partly explains the discriminatory act. And we have argued that Jamal is a victim of discrimination in Misperceived, but this cannot be explained in terms of Jamal’s actual race. This leaves a final possibility: the discrimination must be explained in terms of something else. Despite their similarities, Veridical and Misperceived must be explained differently.

Anyone who takes this stance must identify what explains the discrimination in Misperceived. The obvious answer is that it is the discriminator’s perception that Jamal is Black. This suggests a disjunctive view: to be a victim of discrimination is to be differentially treated because of your actual race (in veridical cases) or because of your perceived race (in misperception cases). Notably, when definitions of discrimination can include misperception cases, they often involve disjunctive formulations (e.g., actual or perceived race).27 We will call this view disjunctivism about racial discrimination.

Adverting to disjunctivism raises two challenges for the argument from racial discrimination to racial realism. First, if the target’s perceived race is sufficient to explain why they were discriminated against, then the target’s actual race is redundant in explanations of discriminatory acts. Disjunctivism violates a general superfluity constraint on explanations.28

One way to cash out that constraint is that explanations must be potential difference-makers: if A explains B, then A must be capable of making a difference to whether B is the case. Many have endorsed a “necessary condition for explanatory relevance” of this form (Strevens 2004: 159).29 The constraint can also be cashed out in terms of minimality: for a set of factors, {P}, to explain Q, nothing could be subtracted from {P} without thereby losing the ability to explain Q (Audi 2012: 699). If when we subtract A from {P} the remaining set would still be able to explain Q, A is superfluous; in that sense, A is no part of the explanation of Q.30

Why does disjunctivism violate this superfluity constraint? Because all the facts that are explanatorily sufficient in the misperception case are also present in the veridical case. In Misperceived, Jamal’s being perceived as Black is sufficient to explain the discriminatory act. In Veridical, everything — including this racial perception — is held fixed, except we add that Jamal is actually Black. Because the facts of Misperceived are sufficient to explain the discriminatory act, the fact that Jamal is actually Black in Veridical is not a potential difference-maker. Put otherwise, if we start with the facts of Veridical and subtract the fact that Jamal is Black, the remaining set of facts about how the discriminator perceived Jamal does no worse at explaining the discriminator’s action. So, Jamal’s actual race is explanatorily superfluous, and therefore is no part of the explanation of the rejection in Veridical.31

This brings us to our second challenge to disjunctivism. As mentioned before, explanations of discrimination are action explanations. These are not just any explanations where the explanandum is an action, but specifically explanations of actions in terms of the agent’s motivating reasons — the reasons for which an action was done. And the discriminator’s state of mind is the same across pairs of cases like Veridical and Misperceived. So why should the action explanation differ? To hold that it does violates a constraint on action explanations in particular: that the motivating reasons that explain an agent’s actions supervene on internal facts about their mental states. Call this the supervenience constraint.32

Why accept the supervenience constraint? Here is the abstract argument. Someone’s motivating reasons for acting explain their actions by playing a certain motivational role in their psychology — that is why philosophers call them motivating reasons. Whether some reason plays a motivational role in one’s psychology depends on facts that are internal to one’s psychology. So differences in facts external to one’s psychology cannot by themselves make any difference to one’s motivating reasons.

To make things more concrete, we can consider cases where the relevant external features of the world change unbeknownst to us:

Airport: Rodrigo’s Auntie is flying into town. The flight is scheduled to arrive at 5pm. For this reason, at 4pm Rodrigo begins preparing to drive to pick Auntie up from the airport. At 4:05, Rodrigo is preoccupied and does not see a notification that Auntie’s flight has just been delayed by an hour.

At 4pm, what is the reason for which Rodrigo is preparing to drive to the airport? Suppose we say it is the fact that Auntie’s flight is scheduled to arrive at 5 pm. Here is the problem. At 4:10, this is no longer a fact, even though Rodrigo is still in the process of performing the same action. So what is Rodrigo’s motivating reason at 4:10? If we want to hold onto the first answer, we must hold that Rodrigo’s motivating reason changes (it becomes, e.g., Rodrigo’s belief that the flight is scheduled to arrive at 5) as a result of a change in the external world that makes no contact with Rodrigo’s psychology. This seems bizarre. Surely Rodrigo is acting on the basis of the same reason throughout the process of preparing to leave for the airport. Thus, cases like Airport support the supervenience constraint.

Notice that the flight delay in Airport plays a similar role to the two versions of Jamal’s case, Veridical and Misperceived: it introduces a difference in the external facts with no corresponding difference in the internal facts about agents’ mental states. In Veridical and Misperceived, the employers have the same motivating thought: that Jamal is Black. Their motivating reasons are psychologically indistinguishable. But disjunctivism says that in Veridical the fact that Jamal is Black explains the rejection, whereas in Misperceived something else (e.g., the perception that Jamal is Black) explains the same action. These disjunctivist verdicts violate the supervenience constraint in the same way as in Airport.

Of course, the Jamal cases may not be analogous to Airport in every respect. Perhaps facts about a flight arrival are mutable in a way that facts about one’s race are not. But we do not think that this undermines our point. As Boxill (1992: 16) has argued, even if race were mutable,33 this would not change the moral landscape with respect to discrimination. Regardless, we can construct more closely analogous cases of discrimination on the basis of an identity that can easily change, like religion. Compare this pair of cases:

Mary 1: Mary applies for a job and does not get an interview. Mary later learns this is because the employer assumed they were Mormon, since their CV says they went to Brigham Young University, which is operated by the Church of Jesus Christ and the Latter-day Saints. Mary is Catholic.34

Mary 2: As above, except that after submitting the application and while it is being considered, Mary converts to Mormonism.

In Mary 2, as in Airport, the relevant external fact (Mary’s religion) changes over the course of the agent’s action. But this change cannot make a difference to the relevant agent’s motivating reasons. If so, then in Mary 1 and Mary 2, the discriminator acts for the same reason. Thus, the discriminatory act should be explained in the same way. Whether across time or across minimal case pairs, providing two different explanations of the discriminatory act violates the supervenience constraint.

At this point, some might ask: Did we not just raise a general problem with disjunctive explanations of corresponding ‘good’ and ‘bad’ cases in philosophy of action more generally? And if it were this easy to argue against such explanations, would our arguments not also refute other similar forms of disjunctivism, such as disjunctivism about perception?

Our response is threefold. First, it is true that we have connected the problems with disjunctive explanations of veridical and misperception discrimination to more general issues with disjunctive explanations in the philosophy of action. But we think drawing this connection is both fruitful and independently motivated. Consider what Eidelson calls the “basic ‘internalist’ intuition” about discrimination: “If we know everything about how X treats Y and Z and why he treats them each as he does, we should know whether he is discriminating on any given basis in so doing” (2015: 28). Why is this an internalist intuition? Because the kind of ‘why’ question Eidelson references is an explanation of X’s actions in terms of their motivating reasons, which are internal to their psychology. Thus, Eidelson’s internalist intuition dovetails strikingly with the supervenience constraint. We think this connection between general action-theoretic considerations and specific considerations about discrimination is grist for our mill and renders disjunctivism about discrimination even more difficult to defend than its more general action-theoretic counterpart.

However, we do not think our arguments imply that all forms of disjunctivism should be rejected on the same grounds. This is our second response, and it points to a difference between disjunctivism about discrimination and disjunctivism about phenomena like perception: the former is distinctively undermotivated. A primary motivation for disjunctive explanations of perception is that there is some positive status present in the ‘good’ case that is absent in the ‘bad’ case. With perception, the idea is that you can gain knowledge of the world when you see a dog, but not when you have a vivid hallucination of a dog. By contrast, with discrimination there is no positive status to be explained in the veridical case — that is why we called it the veridical case, not the good case. So, disjunctivism about discrimination comes with similar costs to disjunctivism about perception and the like, but without similar benefits.

Finally, this response at most has traction against the argument from supervenience. It does not clearly affect the argument from superfluity. That is, it is not clear that disjunctivists about other phenomena are similarly committed to explanations of perceptions (etc.) in terms of facts about the external world that are always superfluous to the explanation. (And if they are, surely that is a serious cost for their view!)

IV.IV The Argument So Far

We have assumed for the sake of argument that Jamal’s being Black partly explains the rejection in Veridical. As we have shown, this assumption leaves no good options for explaining the rejection in Misperceived. So we should reject the starting assumption. That has been the argument so far. But we can also recast this argument to better emphasize its important upshot. We have seen that Misperceived is a case of discrimination (IV. I), that we should not explain it in terms of the victim’s actual race (IV. II), and that we should explain it the same way as we explain Veridical (IV. III). It follows that we should not explain the discriminatory action in Veridical in terms of the victim’s actual race.

Crucially, this argument does not show that realism about race is false. Neither the premises nor the conclusion concern whether race exists; they concern whether it plays a certain explanatory role. Our goal is to show why we should prefer explanations of racial discrimination in terms of racial attitudes (which are the same in veridical and misperception cases) to explanations in terms of the victim’s actual race. So, responding to this argument by citing other theoretical advantages of realism would miss the point. The issue realists need to face in light of our argument is different: what is the theoretical advantage of positing that race plays this explanatory role? To answer this question, realists would need to engage in precisely what has been missing from this debate so far: a comparative analysis of the virtues of explaining discrimination in terms of race rather than racial attitudes.

V – Does the Realist’s Metaphysics Make a Difference?

We have framed our discussion so far in terms of realism about race. But while realists agree that race is real, they disagree about its nature. Do those disagreements make a difference to the challenge we just posed? In other words, does the nature of race make a difference to whether race best explains racial discrimination?

We do not think so. What race is may make a difference to why race does not best explain racial discrimination, but it does not make a difference to whether race best explains racial discrimination. The best way to see why is to consider the following. If race best explains racial discrimination, what, for the realist, best explains race?

Start with views according to which race is a social kind, explained by social facts. Race’s putative role in explaining social phenomena like discrimination is often framed as supporting social constructionism. (As Ásta put it in the passage above, “social phenomena … can be explained by a person’s race,” so “[r]aces are social categories.”) As long as such views grant that there can be cases of misperception discrimination, the argument we offered in §IV can get off the ground. If so, appealing to such views will not undermine our arguments about explaining racial discrimination.

What if social constructionists try to block the arguments in §IV by denying that there can be racial misperception cases in the first place?35 One option for social constructionists is to hold that being perceived to be a member of race R is sufficient for being an R, such that misperceived race is impossible. The other option is to hold that being discriminated against as an R is sufficient for being an R, such that racial misperception discrimination is impossible. Neither of these options is promising. For good reasons, leading social constructionist views allow for racial misperception and misperception discrimination.36 Moreover, neither of these options generates a plausible way for social constructionists to show that someone’s actual race best explains racial discrimination.

Say they take the first option: racial perception fully explains race. But race is also meant to explain racial discrimination. By transitivity, it follows that racial perception explains racial discrimination. This looks now like the error theorist’s explanation, with an additional middle step positing race. If they take this option, it is unclear how the social constructionist can contend that we should not explain racial discrimination in terms of racial attitudes (like perception). Indeed, it is unclear how they could contend that race is explanatorily indispensable. Their view now entails that racial attitudes are always sufficient to explain racial discrimination.

What about if they take the second option? On this view, racial discrimination explains race. So how can race explain racial discrimination? Say we grant that in Misperceived Jamal is Black because Jamal is subject to anti-Black discrimination. If so, then Jamal cannot be subject to anti-Black discrimination because Jamal is Black. That would be explanatorily circular. The problem generalizes (indeed, it also arises in Veridical). As Khalifa and Lauer note, if we explain racial segregation partly in terms of race, we cannot explain race partly in terms of racial segregation (2021: 6). Perhaps there are creative ways for realists to coherently take race to be both the explanans and explanandum of racial injustice. But if so, we do not see why such explanations would be better than vastly simpler rival hypotheses that do not raise any specter of a vicious explanatory circle. This social constructionist explanation hardly seems like a promising candidate for being the best explanation of racial discrimination and racial injustice.

Instead of taking race to be a social kind, the realist could take race to be a biological kind — i.e., explained by biological facts like phenotypical traits or genetic ancestry.37 Biological racial realists do not typically make the argument from racial discrimination to realism. But that route might actually seem more promising. If visually indistinguishable biological racial differences are impossible,38 perhaps we can rule out misperception cases, and do so without taking race to be explained by racial perceptions. It is a tempting thought, but it does not pan out. As we noted, the term “misperception discrimination” should not suggest a fixation on (visual) perception in particular; it picks out a constellation of fallible racial attitudes that play a certain explanatory role in discriminatory acts. On plausible versions of biological racial realism, racial misperception in this broader sense must be possible, and is likely to be ubiquitous and systematic. Individuals’ and groups’ relevant racial attitudes can easily fail to line up with biological racial categories.

The implications of this point for discrimination are significant. One way to bring them into focus is to consider the position of Sikhs in the United States. Sikhs are not Muslims, but in the United States, one of the central ways in which Sikhs have been marginalized is tied to the frequency with which they are misperceived to be Muslims in a kind of erasure, resulting in their being subjected to Islamophobic violence and harassment.39 If we adopt forms of biological racial realism, many marginalized social groups are in a similar position to Sikhs: one of the central ways in which they are marginalized is by being subject to systematic racial misperception discrimination. So whether one adopts social or biological forms of racial realism, one cannot deny the reality of misperception discrimination.

VI – What About Indirect Discrimination?

So far, our discussion has focused on direct discrimination: A’s differential treatment of B on the basis of B’s being a member of a race. We have not yet discussed the phenomenon of indirect discrimination: cases where A’s actions have a disparate impact on some Bs who are members of a race (where A need not act for the reason that Bs are members of a race). As many of our realist interlocutors are interested in explaining group disparities,40 perhaps this gives them a way out. They can grant that race does not best explain direct racial discrimination, while insisting that it does best explain indirect racial discrimination.

It is not obvious to us how indirect racial discrimination can be explained by race without being ultimately explained by racial attitudes.41 But however it is cashed out, we think such a view faces a dilemma. Does the realist hold that race always explains indirect racial discrimination? If so, their view is very contentious.42 To see why, imagine that a collegiate sports group implements a policy that favors some athletes (e.g., swimmers) and disfavors others (e.g., basketball players), where the first group is predominantly white and the second predominantly Black.43 On standard approaches, something like this is sufficient to ground a complaint of indirect racial discrimination.44 The policy must disfavor Black athletes; it need not disfavor them because they are Black.

To evade the first horn of the dilemma, realists should grant that there can be cases of indirect discrimination like this. This means they can at most posit that indirect racial discrimination is sometimes, but not always, explained by race. Now we can bring back arguments from §IV. Take a case where there is a disparate impact on members of a racial group that suffices for indirect racial discrimination. If we add that the disparate impact on those members obtains because of their race, then ex hypothesi this cannot make a difference to whether it is a case of indirect discrimination. So to take race to sometimes explain indirect racial discrimination would violate the superfluity constraint on explanations. That is the second horn.

The distinction between direct and indirect discrimination is complex and fraught, so there is more to say here. But this dilemma is enough to illustrate how realists face similar problems in holding that race is the best explanation of direct or indirect discrimination.45 It does not seem that they would be on firm footing in granting that race does not best explain direct discrimination while insisting that it does best explain indirect discrimination.

VII – Does Race Best Explain Wrongful Discrimination?

Suppose the realist concedes that race does not best explain direct or indirect racial discrimination. There is one last route they can take — argue that it best explains the wrongfulness of racial discrimination. This would make race part of the best moral explanations, rather than the best action explanations. This strategy is underexplored as an explanatory argument for realism. It warrants consideration.

Is this strategy ultimately any more promising? We do not think so. First, we will show that it faces the same kind of objection as its predecessors by generalizing our overarching argument from §IV. Second, we will consider an important realist objection, which turns on whether systematicity makes veridical racial discrimination more seriously wrong. We will then respond by granting that the systematicity of discrimination is morally important while denying that veridical racial discrimination is necessarily more systematic.

VII.I The Three Options

We will begin, as before, by assuming for the sake of argument that, in Veridical, Jamal’s being Black partly explains the wrongfulness of the discriminatory act. This leaves three main possibilities for Misperceived:

  1. 1. Misperceived is not a case of wrongful discrimination at all.

  2. 2. Misperceived is a case of wrongful discrimination and the wrongfulness is explained in terms of Jamal’s actual race.

  3. 3. Misperceived is a case of wrongful discrimination but the wrongfulness is not explained in terms of Jamal’s actual race.

As before, we will show that all three options fail. The upshot is that we should reject the assumption that Jamal’s actual race is part of the explanation of the wrongfulness of the discriminatory act in cases like Veridical.

On the first option, when Jamal is discriminated against because he is misperceived as Black, this action is not (seriously) wrong. We’ve already noted that this view is implausible. As we discussed in §IV.I, one of the reasons it is implausible to deny that Misperceived is a case of discrimination in the first place is that it would imply, counterintuitively, that Jamal was not seriously wronged. But there is a more important point here, which starts from an observation we made in §IV.I: Misperceived and Veridical are identical but for the victim’s actual race. This implies that they share commonly posited wrong-making features of discrimination.

For one, in both cases the employer is motivated by the same objectionable attitudes: anti-Black prejudices. More generally, on agent-centered views of the wrongfulness of discrimination (those which focus on the discriminator, rather than the discriminatee), Misperceived and Veridical should be regarded as wrong for exactly the same reasons.

For another, it is plausible that in both cases the employer’s act expresses the same wrongful objective meaning.46 The expressive significance of rejecting job applications because the applicant’s name is racially coded is similar in both cases — it demeans the individual and Black people. More generally, insofar as such expressive views focus on the relation between the discriminator and discriminatee, we think they should also regard at least many cases of misperception discrimination as wrong.47 To return to an earlier example, Sikhs who are systematically subject to Islamophobic harassment and abuse are thereby demeaned. Part of why this is the case is that the features that discriminators conceive of Muslims having are features that Sikhs have. So insofar as discriminatory acts demean people with those features, they demean Sikhs. (We do not think this is the full explanation — as we will note below.) Both misperception and veridical discrimination, then, involve wrongful, unjust relations.48

Finally, in both cases, the act of discrimination is harmful. The harms of discrimination can be material, experiential, or consist in the deprivation of certain liberties.49 Appealing to such harms gives us patient-centered accounts of the wrongfulness of discrimination. But misperception discrimination can still be harmful in all of these ways, and hence be wrong on patient-centered views. When people who are perceived to be Muslim are subjected to harassment, it does not cease to be harmful if they happen to be Sikh. Likewise, when Burrage suffered humiliation and lost employment, this did not cease to be harmful because Burrage happened not to be Mexican. Again, veridical and misperception discrimination share this commonly posited wrong-making feature of discrimination.

Say the realist grants that misperception discrimination is also wrongful. They could now take the second option, insisting that in Veridical and in Misperceived the victim’s actual race explains the wrongfulness of the discrimination. It is not clear how one should defend this idea. In Misperceived, Jamal’s being white does nothing to explain why the discriminatory act is wrong. And for the reasons we discussed (§IV.II and §V), we do not think Jamal is Black in Misperceived or that Jamal’s being Black in this case could help explain why the racial discrimination that Jamal suffered was wrong. So this second option looks like a dead end.

This leaves the final option, which is to hold that misperception discrimination is wrongful, but its wrongfulness is explained by something other than the victim’s race. Call this view moral disjunctivism: the wrongfulness of discrimination is explained either by actual race (in cases like Veridical) or by something else (in cases like Misperceived). As before, we are owed an account of what, other than race, plays the explanatory role in cases like Misperceived, and the obvious answer will be perceived race. But whatever the answer, this view faces an identical challenge to the disjunctivist view we considered in §IV.III.

Recall the superfluity constraint. If actual race is part of the explanation, it must be capable of making a difference. If we can always subtract actual race from a set of factors and the remaining set would just as well explain the wrongfulness of discrimination just as well, then actual race is superfluous and not part of the explanation of wrongfulness. The fact that Jamal is perceived to be Black, along with the background moral facts, is sufficient to explain why Misperceived is a case of wrongful discrimination. All of these facts are present in Veridical. So they are sufficient to explain why Veridical is a case of wrongful discrimination. Jamal’s actual race is superfluous to any explanation of why veridical is a case of wrongful discrimination. The superfluity constraint is a fully general constraint on explanation, but moral disjunctivism violates it.

VII.II Systematicity and Gradeability

At this point, the moral disjunctivist may respond that we are being too quick. When it came to disjunctivism previously, the superfluity constraint applied because the explanandum was exactly the same in Veridical and Misperception: the discriminatory act. But when we explain the wrongfulness of the acts in Veridical and Misperception, we’re explaining something gradable. It could be that the victim’s actual race makes no difference to whether racial discrimination is wrong but does make a difference to how wrong it is. Plausibly, anti-Black discrimination against someone who is Black is more seriously wrong than anti-Black discrimination against someone who is white.

The most plausible explanation for this is patient-centered. Cases like Veridical involve harms to the victim that are far more systematic. A white person named Jamal may face adverse treatment here and there because of their name, but a Black person named Jamal faces a much more widespread and far-reaching constellation of anti-Black prejudice across the course of their life. As this differential treatment aggregates over a lifetime, it may bring a host of additional harms. These include internalization, a heightened sense of insecurity, and adaptation to oppressive social scripts.50

This is an important point. And while we think Veridical and Misperception both involve wrongful discrimination, we agree that the discrimination in Veridical is morally worse. However, we still do not think this tells in favor of the view that race is part of the best explanation of the wrongfulness of racial discrimination. Here is why.

We grant that systematicity matters for the seriousness of discrimination. But the systematicity and veridicality of discrimination can come apart. Sikhs are systematically subject to anti-Muslim harassment and violence. In fact, Sikhs may even be subject to higher rates of some forms of Islamophobic violence than actual Muslims, because the visible features that make them targets (turbans and beards) are features that Sikhs, not Muslims, tend to have. Systematicity and veridicality can also come apart in the other direction: sometimes actual members of marginalized groups can conceal a group identity and thereby be less systematically targeted.51 Of course, comparing how systematically real-world groups are discriminated against is messy. Our point is just that it is theoretically possible (and indeed easy to imagine) that systematicity and veridicality come far apart. One lesson from this is that if discrimination is more seriously wrong when it is systematic, it does not follow that veridical discrimination is more seriously wrong than misperception discrimination. In some cases, misperception discrimination may be no less systematic. Another lesson is that systematicity itself is better explained in terms of patterns of racial attitudes (independent of actual race membership), which illustrates another way that the complex wrong-making features of discrimination may be best explained in terms of racial attitudes (rather than actual race).

As we alluded to in §V, this point is especially important if we grant that entire groups can be subject to systematic racial misperception. For in such cases, as with Sikhs in the United States, there is an additional harm that comes with systematic misperception discrimination: a kind of erasure. This kind of erasure is felt by many non-Chinese East and Southeast Asian Americans who have been systematically subject to Sinophobic abuse. When one is systematically abused for being an X when one is actually a Y, part of what can be internalized is the social insignificance of what makes one’s actual group distinct in the first place. To be clear, we are not claiming that this makes misperception discrimination generally more seriously wrong than veridical discrimination. Like Moreau, we think that “practices that wrongfully discriminate vary enormously,” in terms of “the severity of their impact on the discriminatee,” “the motivation of the discriminator,” and “the discriminator’s awareness of the impact on the discriminatee and others who share the relevant [perceived] protected trait” (2020: 161). Our point is that, in the face of this variegation, the realist has an uphill battle to find a persuasive argument that veridical discrimination is more seriously wrong just in virtue of being veridical, such that the victim’s actual race is part of the best explanation of the gradable wrongness of discrimination.

Overall, then, the move to moral explanations does not save explanatory arguments for racial realism. We can provide better explanations of the wrongfulness of discrimination in terms of racial attitudes than in terms of actual races. This remains true even when we turn to more complex moral factors, such as the systematicity of discrimination.

VIII – Conclusion

Racial realists have long argued that because race explains racial discrimination, we need to be ontologically committed to races. Error theorists have long responded that we can explain racial discrimination without race by appealing to racial attitudes alone. Because neither side has offered much argument about which of these explanations we should favor, the debate on this issue has been at an impasse.

We have aimed to break this impasse by arguing that explanations of discrimination in terms of racial attitudes are superior. We have argued that this is true for explaining individual actions (direct racial discrimination), certain group inequalities (indirect racial discrimination), and the wrongfulness of discrimination. If we are right about this, a common explanatory argument for racial realism fails. This is a significant step in the dialectic between realists and error theorists.

Even if you are not convinced that we have established the superiority of explanations of discrimination in terms of racial attitudes, our discussion has at least two important upshots. First, our arguments at least push the realist to say far more about why we should explain discrimination in terms of race, instead of racial attitudes alone.52 We have also pushed the realist to say more about whether their argument shows that race explains direct discrimination, indirect discrimination, or the wrongfulness of either. And if realists cannot establish that similar explanatory arguments hold across the board, that result would be theoretically significant in its own right: for example, despite its common presentation, perhaps the realists’ argument is best understood as concerning moral explanations, not action explanations.

Second, our arguments have a methodological upshot. Discussions of discrimination should pay far closer attention to misperception cases. It is surprising that such cases are rarely mentioned. This is not only due to the legal controversies they generate but also because they provide an obvious way to tease apart rival explanations. An important feature of explanations in terms of racial attitudes is that these attitudes are fallible. Hence Khalifa and Lauer’s point that “racialization only entails that groups are represented as races, since X’s being represented as Y does not entail that X actually is Y” (2021: 5, italics in original). Misperception cases are precisely where we should expect explanations in terms of actual race and explanations in terms of racial attitudes to diverge. This is why focusing on cases of misperception discrimination, where reality does not match representation, is fruitful.

The importance of this methodological lesson stretches beyond race and racial discrimination. If we are right about race and racial discrimination, it does not automatically follow that analogous arguments to ours would work for race and racial oppression, for example.53 Nor does it follow automatically that analogous arguments would work for gender and gender discrimination or disability and disability discrimination.54 But we think more attention to the distinction between veridical and misperception cases would be fruitful in all of these contexts. And if disanalogies between different social hierarchies were to emerge, this would have important implications for social and moral philosophy more generally. It would reveal that for some social hierarchies, actual social groups play a role that cannot by captured by appeal to any constellation of social attitudes. In other words, paying more attention to misperception discrimination can help us discover whether, when, and why we need to appeal not just to social attitudes, but to social reality.55


  1. There are other examples. Ritchie offers this as the first argument for realism about racial and other groups: “Anything efficacious exists, so […] groups exist. Being part of a racial group can substantially affect one’s experiences, what one can do, and how one is treated,” where in context this includes discriminatory treatment (2015: 311). In a similar vein, Mason discusses a view according to which social categories such as race are natural kinds, in part because “these categories are explanatorily fruitful” (2016: 843). Both list authors who they say offer such arguments (see Ritchie’s fn. 10 and Mason’s fn. 11).
  2. A particularly interesting case is Root’s “How We Divide the World,” which defends a form of realism about race partly on the basis of how races feature in generalizations, which are explained by discrimination: “much of the variance between the races in socioeconomic standing, as well as health and disease, is explained by past or present acts of discrimination based on race” (2000: 629).
  3. That is, some use it to defend realism about other groups: e.g., Piper (1993: 24) suggests that we need the property being a homosexual to explain homophobic discrimination.
  4. Appiah (1992: 45) is the first error theorist we know to have offered this response.
  5. Hochman argues that racialized groups are “groups misunderstood to be biological races” (2020: 2). And like other error theorists, Hochman holds that “we should be careful not to infer the existence of race from the existence of racialization and racism, which do not require races, but only the belief in races” (2021: 34). See also Hochman (2017).
  6. We do not mean to suggest that nothing at all has been said on this front. Haslanger briefly argues against the error-theoretic response as follows: “Although false beliefs about racial groups may be the best explanation of early forms of racial hierarchy (though I find even that questionable, given the economic and other forces at work), it is implausible that such beliefs are the best explanation of ongoing racial injustice, including the perpetuation of economic and political injustice, social segregation, and cultural stigma” (2019: 20–21). This response is underdeveloped, but it laudably addresses whether attitudes (beliefs) best explain racial discrimination.
  7. Again, we do not mean to suggest that nothing at all has been said on this front. Hochman draws on Fields and Fields’ idea of the “race–racism evasion” in sentences like ‘black Southerners were segregated because of their skin color’: this “transforms racism, something an aggressor does, into race, something that the target is” (2012: 95, 17, italics in original). As Hochman writes (2021b: 38, italics in original), “Black Southerners were segregated because of anti-Black racism on the part of Whites, not because of their skin color.” Again, we think this argument needs development (can realists respond that Black Southerners were segregated because of their race and because of anti-Black racism?). This issue is too complex to resolve here, but the argument is plausibly better understood as a moral criticism of views about how we should talk about discrimination (whether we should, as Hochman (2021b: 38) writes, “name racism instead of euphemistically referring to ‘race’”), rather than what explains discrimination. As such, it would bear more on the debate between eliminativists and conservationists than between error theorists and realists (on this distinction, see Mallon 2006 and Wodak 2022). That said, other arguments for why racism is best explained without appealing to the reality of race (see, e.g., Hochman 2021a) may carry a similar lesson for explaining racial discrimination. Thanks to an anonymous reviewer for raising these points.
  8. Their racialization-only explanation is somewhat schematic, but its basic point is that “groups are represented as races and that discrimination arises (in part) because of false folk theories” (Khalifa and Lauer 2021: 10).
  9. This example is adapted from Eidelson (2015: 17). (These names were used in Bertrand and Mullainathan’s (2004) classic experiment on employment discrimination.) A similar case is used by Lippert-Rasmussen (2013: 20) to make the same point.
  10. Burrage v. FedEx Freight, Inc., 2012 WL 1068794 (N.D. Ohio Mar. 29, 2012).
  11. Burrage v. FedEx Freight, *9.
  12. See El v. Max Daetwyler Corp., No. 3:09CV415, 2011 U.S. Dist. LEXIS 49645, (W.D.N.C. May 9, 2011); Adler v. Evanston Nw. Healthcare Corp., No. 07CV4203, 2008 U.S. Dist. LEXIS 101744, (N.D. Ill. Dec. 16, 2008); Lopez-Galvan v. Mens Wearhouse, No. 3:06 CV 537, 2008 U.S. Dist. LEXIS 53456, at *23 (W.D.N.C. July 10, 2008).
  13. Lewis v. N. Gen. Hosp., 502 F. Supp. 2d 390, 401 (S.D.N.Y. 2007).
  14. See, e.g., Arsham v. Mayor & City Council of Baltimore 85 F. Supp. 3d 841 (D. Md. 2015).
  15. The term “misperception discrimination” was coined by Greene (2013). Other legal scholarship on the phenomenon includes Williams (2008), Senn (2009), Aronson (2016), and Flake (2016). Explicit philosophical discussions are limited and brief; see Lippert-Rasmussen (2013: 20), Eidelson (2015: 17), and Thomsen (2017: 24). But see Hellman (ms) for an illuminating recent discussion of the puzzles posed by misperception cases.
  16. Thanks to an anonymous reviewer for helpful discussion of this explanation.
  17. See, e.g., discussion in Gallow (2022: §2.1).
  18. This may need to be amended if we allow that white members of interracial couples (like Richard Loving) can be victims of racial discrimination via anti-miscegenation laws.
  19. Gardner argues, quite plausibly, that making a decision on an irrelevant basis is irrational but not in itself immoral, as we have no “across-the-board-duty to be rational, so our irrationality as such wrongs no one” (1998: 168). Even in the context of hiring, there are clear counterexamples to any such duty (see, especially, Shin 2009: 153).
  20. As Hellman (2016: 935) argues, being adversely treated on an irrational or irrelevant basis in the absence of distinguishing features of discrimination (“animus or differential sympathy,” or “denigration”) is “mere irrationality, pure stupidity.” Even if mere irrationality in hiring is somewhat immoral, why does it seriously wrong Jamal?
  21. For instance, some might say that misperception discrimination is attempted discrimination and is, therefore, an inchoate offense. We think this position also faces serious obstacles (including some related to the next problem), but we leave them aside here for brevity.
  22. Burrage v. FedEx Freight, *10.
  23. Gardner (1998: 179) notes that a variety of such locutions are used to the same end.
  24. This stance is fairly orthodox. See discussion in Moreau (2020: 19–20).
  25. For an extended discussion of this issue in relation to recent anti-discrimination jurisprudence, see Berman and Krishnamurthi (2021).
  26. We assume, like many others, that an agent’s motivating reasons may be opaque to them. Those who use the term “motivating reasons” differently can substitute terms accordingly.
  27. See, e.g., Lippert-Rasmussen (2013) and Altman (2020).
  28. Such a constraint plausibly undergirds many famous arguments, such as Harman’s (1977) argument for the superfluity of moral facts in explaining actions. Harman argues that because agents’ moral attitudes are always sufficient to explain their actions, moral facts cannot be part of the explanation of those actions. Our argument is similar in spirit.
  29. For more recent discussions, see Krämer and Roski (2017) and Wodak (2020).
  30. A plausible superfluity constraint will be compatible with familiar forms of causal overdetermination. Say that {A, B} cause C, where A and B are independent and sufficient causes for C. If we subtract A, B is still sufficient to explain C. But in the possible absence of B, A would make a difference to whether C obtains. That is what we mean by potential difference-making. It is worth emphasizing that in cases like Veridical we cannot treat actual and perceived race as independent, sufficient causes of discrimination. The presence of perceived race is sufficient to explain the discrimination, but in the absence of perceived race there is no racial discrimination left to explain.
  31. It could be that in some version of Veridical, Jamal’s actual race causes the perception that Jamal is Black. Perhaps the fact that Jamal is Black is part of what caused him to be named Jamal, which in turn caused the employer to perceive him as Black, which in turn caused the employer to discriminate against him. However, that will not always be true of veridical perception. And moreover, as we discussed above, the relevant kind of explanation of the discrimination is not just any old causal explanation. We want an explanation of the discriminator’s actions in terms of the reasons for which they were done. Regardless of their causal history, the reasons for which the discriminator acts are encoded in their racial attitudes in exactly the same way whether or not Jamal is Black.
  32. For a defense of this constraint and the origin of the case below, see Singh (2019).
  33. Mills (1997: ch. 3) also considers several thought experiments where race is arguably mutable. That one’s race can change (at least between contexts) is a theoretical possibility on many views on the metaphysics of race and an actuality on some of those views.
  34. Mary 1 is based on a case from Flake (2016: 88). The variant, Mary 2, is original.
  35. Thanks to Kate Ritchie for pushing us on this.
  36. To foreclose the possibility of racial misperception discrimination, the views we just described need to say that anytime X is perceived to be an R (or discriminated against as an R), X is an R. This is not a very sophisticated form of social constructionism. By contrast, Haslanger’s view might seem to block misperception cases, as Haslanger says that those who “pass” as an R “function as” an R regardless of their actual ancestry (2012: 237, fn. 18). But Haslanger holds that “membership in a racial/ethnic group” is understood “in terms of how one is viewed and treated regularly and for the most part” (2012: 238, fn. 18, emphasis added). So there can be misperception discrimination on Haslanger’s view.
  37. For a prominent example of biological racial realism, see Spencer (2014, 2019).
  38. There is some debate as to whether this is the case: cf. Hardimon (2003: 442) and Spencer (2019: 98). We do not intend to take a stand on it here.
  39. For example, Balbir Singh Sodhi, a Sikh American, was the first murder victim in a rash of Islamophobic hate crimes following 9/11.
  40. Khalifa and Lauer (2021: 3) also note this, citing Haslanger (2012: 226–227; 2019: 20), Jeffers (2019: 66–67), and Mills (1997: 5). They also note that many “social-scientific realists about race” similarly take race to explain group disparities, citing Kincaid (2018), Mallon (2017, 2018), Mallon and Kelly (2012), Root (2000), Sundstrom (2002a, 2002b, 2003), and Wiegman and Mallon (2017).
  41. As Eidelson notes, many hold that the relevant group disparities are explained by the interests of racial minorities being unduly “ignored” or “prejudiced” (2015: 50). Moreau also seems to appeal to the explanatory role of racial attitudes (like group stereotypes) in producing unequal outcomes for groups (2020: 16). Moreover, given our argument in §V, it will follow that whenever indirect discrimination is explained by prior acts of direct discrimination, it is (via transitivity) explained by racial attitudes.
  42. Eidelson notes the standard view is that it is not a “necessary element of indirect discrimination” that there be “a causal connection between the protected trait and the adverse effect,” as such a requirement is inconsistent with widely accepted features of how indirect discrimination is understood (2015: 54). That said, Moreau (2020: 20) seems to accept a requirement for such a causal connection, so the view is not unprecedented.
  43. See, e.g., the recent Title VI complaint against Clemson University (Murphy 2021).
  44. We say “something like this” as there are plausibly other conditions that are irrelevant for present purposes, such as that the policy cannot be justified as a business necessity.
  45. Many of the points we made in relation to this view about direct discrimination in §IV.IIII have analogs in relation to a similar view about indirect discrimination.
  46. For a prominent example of an objective meaning view, see Hellman (2008, 2017).
  47. Hellman’s account arguably falls into this category, as the discriminator must both express a demeaning meaning and have the social power to demean the discriminatee (2008: 35–38). And it is notable that expressive accounts take their cue from Anderson and Pildes (2000), which is partly motivated by Anderson’s relational egalitarianism (1999).
  48. You might think Hellman’s account would reach a different verdict, given its emphasis on the relative power of the discriminator and discriminatee. Perhaps that relative power is different in Veridical and Misperceived, such that in the former case the employers have the power to put Jamal down but lack it in the latter case. As Moreau notes (2020: 46), however, Hellman’s account tends to emphasize differences in institutional power. Insofar as the discriminatee being in a marginalized social group generates such a power imbalance, as Moreau argues it should (2020: 46), this feature would also be present in many paradigmatic misperception cases like Burrage — Burrage was Black, after all.
  49. See, e.g., Moreau (2017).
  50. Stoljar (2015: 118) seems to endorse the view that being an actual member of the target group affects whether one must adapt to oppressive scripts, relying on Iris Marion Young’s view that this adaptation occurs among members of oppressed groups “because they are forced to react to the behavior of others influenced by those images” (1990: 55).
  51. Gardner writes that “gay men, lesbians, and bisexual people need suffer no discrimination on grounds of their sexuality, even in the face of primitive prejudice and superstition, so long as they succeed in concealing their sexuality” (1998: 176). Of course, “passing as privileged” is morally complicated — see Silvermint (2018).
  52. As we noted (see above, fn. 6–7), surprisingly little has been said on this front.
  53. Some, such as Iris Marion Young, have held that we should carefully distinguish discrimination from other social injustices, such as subordination (1990: 196).
  54. Notably, the Americans with Disabilities Act is the only federal anti-discrimination law that expressly includes misperception cases. The “regarded as” language used by Eidelson in their definition of discrimination intentionally echoes the language used by the ADA (2015: 22, fn *).
  55. The authors are equally responsible for the content of this paper and are listed alphabetically. We are grateful to Deborah Hellman, Benjamin Eidelson, Daniel Fogal, Adam Hochman, Chris Howard, Zoe Johnson King, Alex King, Stephanie Leary, Katherine Ritchie, Alex Worsnip, audiences at the Rutgers/ Penn Reading Group on Philosophy of Race and the Southampton Ethics Center, and anonymous referees for their helpful comments on previous versions of this article.


Adler v. Evanston Nw. Healthcare Corp., No. 07CV4203, 2008 U. S. Dist. LEXIS 101744, (N. D. Ill. Dec. 16, 2008).

Altman, Andrew. “Discrimination.” The Stanford Encyclopedia of Philosophy (Winter 2020 Edition), Edward N. Zalta (ed.),

Anderson, Elizabeth. “What is the point of equality?” Ethics 109, no. 2 (1999): 287–337.

Anderson, Elizabeth, and Richard Pildes. “Expressive Theories of Law: A General Restatement.” University of Pennsylvania Law Review 148 (2000): 1503–1575.

Appiah, Kwame Anthony. In My Father’s House: Africa in the Philosophy of Culture. Oxford: Oxford University Press (1992).

Aronson, Elizabeth. “Perceived-As Plaintiffs: Expanding Title VII Coverage to Discrimination Based on Erroneous Perception.” Case Western Reserve Law Review 67, no. 2 (2016): 235–271.

Arsham v. Mayor & City Council of Balt., 85 F. Supp. 3d 841 (D. Md. 2015).

Ásta. Categories We Live By: The Construction of Sex, Gender, Race, and Other Social Categories. Oxford: Oxford University Press (2018).

Audi, Paul. “Grounding: Toward a Theory of the “In-Virtue-Of” Relation.” The Journal of Philosophy 109, no. 12 (2012):

Berman, Mitchell N. and Krishnamurthi, Guha. “Bostock was Bogus: Textualism, Pluralism, and Title VII.” 97 Notre Dame L. Rev. 67 (2021), U of Penn Law School, Public Law Research Paper No. 21–31.

Bertrand, Marianne, and Sendhil Mullainathan. “Are Emily and Greg More Employable Than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination.” American Economic Review, 94, no.4 (2004): 991–1013.

Blum, Lawrence. “Racism: What It Is and What It Isn’t.” Studies in Philosophy and Education 21, no. 3 (2002): 203–218.

Boxill, Bernard. Blacks and Social Justice. Maryland: Rowman & Little-field, (1992).

Burrage v. Fedex Freight, Inc., Case No. 4:10CV2755 (N. D. Ohio Mar. 29, 2012).

Eidelson, Benjamin. Discrimination and Disrespect. New York: Oxford University Press (2015).

El v. Max Daetwyler Corp., No. 3:09CV415, 2011 U. S. Dist. LEXIS 49645, (W. D.N. C. May 9, 2011).

Fields, K. E., and Fields, B. J. Racecraft: The soul of inequality in American life. New York: Verso (2012).

Flake, Dallan. “Religious Discrimination Based on Employer Misperception.” Wisconsin Law Review, no. 1 (2016): 87–132.

Gallow, J. Dmitri. “The Metaphysics of Causation.” The Stanford Encyclopedia of Philosophy (Fall 2022 Edition), Edward N. Zalta & Uri Nodelman (eds.), forthcoming,

Gardner, John. “On the Ground of Her Sex(uality).” Oxford Journal of Legal Studies 18, no. 1 (1998): 167–187.

Glasgow, J. “Is Race an Illusion or a (Very) Basic Reality?” In What Is Race?: Four Philosophical Views, Joshua Glasgow, Sally Haslanger, Chike Jeffers, and Quayshawn Spencer (eds.), 111–149. Oxford: Oxford University Press (2019).

Greene, D. “Categorically Black, White, or Wrong: Misperception Discrimination and the State of Title VII Protection.” University of Michigan Journal of Law Reform 47, no. 1 (2013): 87–166.

Hardimon, M. “The Ordinary Concept of Race.” The Journal of Philosophy 100, no. 9 (2003): 437–455.

Harman, Gilbert. The Nature of Morality: An Introduction to Ethics. Oxford: Oxford University Press (1977).

Haslanger, Sally. Resisting Reality: Social Construction and Social Critique. New York: Oxford University Press (2012).

Haslanger, Sally.“Tracing the Sociopolitical Reality of Race.” In What Is Race?: Four Philosophical Views, Joshua Glasgow, Sally Haslanger, Chike Jeffers, and Quayshawn Spencer (eds.), 4–37. Oxford: Oxford University Press (2019).

Hazlett, Allan, and Alyssa Ney. “The Metaphysics of Race.” In Metaphysics: An Introduction, Alyssa Ney (ed.), 259–279. London: Routledge (2014).

Hellman, Deborah. When Is Discrimination Wrong? Cambridge: Harvard University Press (2008).

Hellman, Deborah. “Two Concepts of Discrimination.” Virginia Law Review, Vol. 102, no. 4 (2016): 895–952.

Hellman, Deborah. “Discrimination and Social Meaning.” In The Routledge Handbook Of The Ethics Of Discrimination, Kasper Lippert-Rasmussen (ed), 95–106. London: Routledge (2018).

Hellman, Deborah. “Defining Disparate Treatment” (manuscript).

Hochman, Adam. “In defense of the metaphysics of race.” Philosophical Studies 174, no. 11 (2017): 2709–2729.

Hochman, Adam. “Janus-faced race: Is race biological, social, or mythical?” American Journal of Physical Anthropology 1 (2020).

Hochman, Adam. “Against the reification of race in bioethics: anti-racism without racial realism.” The American Journal of Bioethics 21 no. 2 (2021a), 88–90.

Hochman, Adam. “Further defense of the racialization concept: A Reply to Uyan.” Du Bois Review. 18, no. 1 (2021b), 31–48.

Jeffers, Chike. “Cultural Constructionism.” In What Is Race? Four Philosophical Views, Joshua Glasgow, Sally Haslanger, Chike Jeffers, and Quayshawn Spencer (eds.), 38–72. Oxford: Oxford University Press (2019).

Krämer, Stephan, and Stefan Roski. “Difference-making Grounds.” Philosophical Studies 174, no. 5 (2017): 1191–1215.

Kincaid, Harold. “Debating the Reality of Race, Caste, and Ethnicity.” Philosophy of the Social Sciences 48 no. 2 (2018):139–167.

Khalifa, Kareem & Lauer, Richard. “Do the Social Sciences Vindicate Race’s Reality?” Philosophers’ Imprint 21, no. 21 (2021).

Lewis v. N. Gen. Hosp., 502 F. Supp. 2d 390, 401 (S. D.N. Y. 2007).

Lippert-Rasmussen, Kasper. Born Free and Equal?: A Philosophical Inquiry into the Nature of Discrimination. New York: Oxford University Press (2013).

Lopez-Galvan v. Mens Wearhouse, No. 3:06 CV 537, 2008 U. S. Dist. LEXIS 53456, at *23 (W. D.N. C. July 10, 2008).

Maril, R. K., and A. M. Gill. Discrimination Based on Perceived Characteristics. Washington, DC: Human Rights Campaign Foundation (2018).

Mallon, Ron. “‘Race’: Normative, Not Metaphysical or Semantic.” Ethics 116, no. 3 (2006): 525–551.

Mallon, Ron. “Social Construction and Achieving Reference.” Noûs 51, no. 1 (2017): 113–131.

Mallon, Ron. “Constructing Race: Racialization, Causal Effects, or Both?” Philosophical Studies 175, no. 5 (2018): 1039–1056.

Mallon, Ron, and Daniel Kelly. “Making Race Out of Nothing: Psychologically Constrained Social Roles.” In The Oxford Handbook of Philosophy of Social Science, Harold Kincaid (ed.). Oxford University Press (2012).

Mason, Rebecca. “The Metaphysics of Social Kinds.” Philosophy Compass 11, no. 12 (2016): 841–850.

Mills, Charles W. The Racial Contract. Ithaca: Cornell University Press (1997).

Mason, Rebecca. “I — Racial Justice.” Aristotelian Society Supplementary Volume, 92, no. 1 (2018): 69–89.

Moreau, Sophia. “Discrimination and Freedom.” In The Routledge Handbook of the Ethics of Discrimination, edited by Kasper Lippert-Rasmussen, 164–173. London: Routledge (2017).

Moreau, Sophia. Faces of Inequality: A Theory of Wrongful Discrimination. New York: Oxford University Press (2020).

Murphy, Dan. Complaint alleges Clemson’s removal of men’s track is racial discrimination. ESPN (2021).

Piper, Adrian. “Two Kinds of Discrimination,” Yale Journal of Criticism 6, no. 1 (1993): 25–74.

Ritchie, Katherine. “The Metaphysics of Social Groups.” Philosophy Compass 10, no. 5 (2015): 310–321.

Root, Michael. “How We Divide the World.” Philosophy of Science 67, no. 3 (2000): 628–639.

Scanlon, Thomas. Moral Dimensions. Cambridge, MA: Harvard University Press (2008).

Senn, Craig. “Perception over Reality: Extending the ADA’s Concept of Regarded as Protection under Federal Employment Discrimination Law.” Florida State University Law Review 36, no. 4 (2009): 827–864.

Shin, Patrick. “The Substantive Principle of Equal Treatment.” Legal Theory, 15 (2009): 149–172.

Siegel, Reva B. “Equality Talk: Antisubordination and Anticlassification Values in Constitutional Struggles over Brown.” Harvard Law Review, 117, no. 5 (Mar., 2004): 1470–1547.

Silvermint, Daniel. “Passing as Privileged.” Ergo 5, no. 1 (2018): 1–43.

Singh, Keshav. “Acting and Believing Under the Guise of Normative Reasons.” Philosophy and Phenomenological Research 99, no. 2 (2019): 409–430.

Spencer, Quayshawn. “A Radical Solution to the Race Problem.” Philosophy of Science 81 no. 5 (2014): 1025–1038.

Spencer, Quayshawn. “How to Be a Biological Racial Realist.” In What Is Race? Four Philosophical Views, Joshua Glasgow, Sally Haslanger, Chike Jeffers, and Quayshawn Spencer (eds.), 73–110. Oxford: Oxford University Press (2019).

Stoljar, Natalie. “Essence, Identity, and the Concept of a Woman,” Philosophical Topics 23 (1995): 261–292.

Stoljar, Natalie. “‘Living Constantly at Tiptoe Stance’: Social Scripts, Psychological Freedom and Autonomy.” In Personal Autonomy and Social Oppression, Marina Oshana (ed.), 105–123. London: Routledge (2015).

Strevens, Michael. “The Causal and Unification Approaches to Explanation Unified — Causally,” Noûs 38, no. 1 (2004): 154–176.

Sundstrom, Ronald R. “Race as a Human Kind.” Philosophy & Social Criticism 28, no. 1 (2002a): 91–115.

Sundstrom, Ronald R. “‘Racial Nominalism.” Journal of Social Philosophy 33, no. 2 (2002b):193–210.

Sundstrom, Ronald R. “Race and Place: Social Space in the Production of Human Kinds.” Philosophy and Geography 6, no.1 (2003): 83–95.

Thomsen, Frej Klem. “Direct Discrimination.” In The Routledge Handbook of the Ethics of Discrimination, Kasper Lippert-Rasmussen (ed.), 17–28. Abingdon: Routledge (2017).

Wodak, Daniel. “Redundant Reasons.” The Australasian Journal of Philosophy 98, no. 2 (2020): 266–278.

Wodak, Daniel. “Of Witches and White Folks.” Philosophy and Phenomenological Research 104, no. 3 (2022): 587–605.

Wiegman, Isaac, and Ron Mallon. “Applied Philosophy of Social Science: The Social Construction of Race.” In A Companion to Applied Philosophy, Kasper Lippert-Rasmussen, Kimberley Brownlee and David Coady (eds.), 441–454. West Sussex: Wiley Blackwell (2017).

Williams, Charity. “Misperceptions Matter: Title VII of the Civil Rights Act of 1964 Protects Employees from Discrimination Based on Misperceived Religious Status,” Utah Law Review, no. 1 (2008): 357–376.

Young, Iris Marion. Justice and the Politics of Difference. Princeton: Princeton University Press (1990).