2. Two Competing Conceptions of Discrimination

As mentioned, I contend that the account of discrimination presented here allows us to understand discrimination in a more comprehensive way than other philosophical accounts on offer. As an in-depth survey of all alternative proposals would far exceed the limits of this article, I focus on the views presented by Kasper Lippert-Rasmussen and Deborah Hellman. Lippert-Rasmussen’s definition of discrimination in particular serves as a blueprint for most of the ongoing debate.Footnote ³ Showing that his account cannot do justice to the all-of-us intuition therefore is tantamount to showing that most of the contemporary debate about discrimination, too, fails to do so. In contrast, Hellman’s view is useful because, while it ultimately cannot accommodate the all-of-us intuition either, the way it falls short contains lessons for taking the intuition into account successfully.

Let us start with Lippert-Rasmussen’s conception of discrimination.

The general idea behind this conception is that discriminatory actions are instances of differential treatment that come about because of their targets’ (perceived) membership in some social group. Let us take a look at each condition. (1) stipulates that the target must belong to a specific social group or must be taken by the perpetrator to belong to a specific social group. Lippert-Rasmussen thereby wants to make room for what we might call misdirected discrimination, that is, instances of discrimination in which the perpetrator targets a person because they mistakenly believe them to belong to some social group. Intuitively, it should still count as discrimination if a person is not hired in virtue of being taken to be gay, for example, even if they are not. (4) spells out the social groups Lippert-Rasmussen is interested in, which he calls “socially salient.” According to him, a group is socially salient “if perceived membership of it is important to the structure of social interactions across a wide range of social contexts” (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 30). On this conception, dentists do not count as a socially salient group because there are few social interactions structured around being a dentist aside from those related to oral health care and perhaps some professional perks (invitations to annual conventions and the like). In contrast, women constitute a socially salient group because being perceived as a woman matters on the job market, in one’s dating life, for one’s health care and safety, etc. (3) states that discriminatory actions occur because of their targets’ perceived group membership. According to Lippert-Rasmussen this means that the target’s group membership is part of the agent’s motivating reason for x-ing, for instance, when racist employers refuse to hire Black people because they falsely believe Black people to be incompetent or because they are afraid of them (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 72–73). (5) is best understood as Lippert-Rasmussen’s attempt to tweak his definition to accommodate two recalcitrant examples. The first one concerns “a man who, despite falling in love with another man who reciprocates his love, rejects the latter’s advances because he is a man” (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 26). According to Lippert-Rasmussen, “it seems very strained to say that he discriminates against such a person” (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 27), just as it would seem strained to say that gay people discriminate against straight people and vice versa. But since being male or being gay or straight are socially salient properties and since being romantically rejected in favor of someone else certainly counts as a way of being treated worse than somebody else, both cases would qualify as discrimination on clauses (1)–(4). To prevent this, Lippert-Rasmussen adds the first part of (5) as he holds that even if gay people are regularly romantically rejected by straight people and vice versa, this alone does not make these two groups worse off relatively to each other. On its own, the first part of (5) would be too demanding, though, as it would exclude one-off cases of morally troublesome differential treatment caused by group-related animus. For instance, imagine that “just one Catholic employer in a Protestant, fiercely anti-Catholic society refuses to hire Protestant applicants simply as a result of animosity toward Protestants” (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 27). Since this employer is just one person, his refusal to hire Protestant applicants would not result in Protestants being significantly worse off. Lippert-Rasmussen, however, insists that cases like this should count as instances of discrimination. He therefore has to add the second part of (5). Finally, (2) captures Lippert-Rasmussen’s belief that discrimination involves the differential treatment of the person targeted. On his view, there can be cases of discrimination that, while not morally neutral, do not rise to the level of a moral wrong. According to Lippert-Rasmussen, an act of discrimination meets this level when it runs counter to what he calls desert-accommodating prioritarianism, that is, when by x-ing A has either not maximized the general well-being or done so in a way that did not prioritize the well-being of those worse-off or more deserving (cf. Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 166).Footnote ⁴

Let us now turn to Hellman’s “meaning-based view of wrongful discrimination” (Hellman, Reference Hellman and Lippert-Rasmussen2018, 97).Footnote ⁵ On the most general, nonmoralized level, A, by x-ing, discriminates against B vis-à-vis another person C if A treats B differently from C on the basis of B having or lacking some trait or property P constitutive of some social group G (Hellman, Reference Hellman and Lippert-Rasmussen2018, 98).Footnote ⁶ Hellman does not state explicitly how we are to understand the notion that a person is treated differently on the basis of having or lacking some group-related property, but it seems fair to assume that she adopts an understanding similar to Lippert-Rasmussen’s condition (3).Footnote ⁷ If we rely on this assumption, A, by x-ing, treats B differently from C on the basis of B having or lacking some property P if B’s having or lacking P is part of A’s motivating reason for x-ing.

Hellman claims that discriminatory actions of this general variety only turn wrongful in their distinctive way once they become demeaning. According to her, demeaningly discriminatory actions necessarily involve “an expressive dimension and a power dimension” (Hellman, Reference Hellman and Lippert-Rasmussen2018, 102): they must express that their target has lower status, that is to say, that the target can legitimately be treated worse than what the “bedrock moral principle” of “equal moral worth” (Hellman, Reference Hellman2008, 6) prescribes. Even though Hellman elaborates neither on the metaethical assumptions behind this principle nor on its specific content, it is clear that, on her understanding, a person’s equal moral worth constitutes an immutable, objective moral standard for how people are to be treated. In addition to this symbolic dimension, there must be “a certain efficacy” (Hellman, Reference Hellman2008, 35) to these expressions, so that demeaningly discriminatory actions end up lowering their targets’ status and thus subordinate them. At first glance, Hellman’s insistence on efficacy might seem incongruous: if the principle of equal moral worth immutably and objectively prescribes how people are to be treated, how could A possibly have sufficient social power negatively to affect B’s (immutable, objectively given) status? The most plausible solution to this puzzle is that, for Hellman, lowering a person’s status does not imply a change to how a person can legitimately be treated (i.e., a change to the topography of their normative landscape) but to how they are de facto treated (i.e., to how other people navigate this topography). Hellman insists on this power dimension because, according to her, discrimination involves a putdown (see Hellman, Reference Hellman2008, 35), which I take to mean a tangible worsening of the target’s situation. Hellman assumes that, on their own, none of the other elements of wrongful discrimination—that is, A treating B differently from C and A expressing that B can legitimately be treated worse than objectively required—constitutes such a putdown. By adding the power dimension, Hellman wants to make sure that wrongful discrimination packs the necessary punch. Hellman has recently expanded her account of wrongful discrimination by allowing for the possibility that its demeaning element pertains not only to its direct target B but to the whole social group G constituted by having P. Bearing this expansion in mind, her account can be summarized as follows:

The last issue I want to address regarding Hellman’s account is the notion of expression stipulated by (3). How can an instance of differential treatment—such as not giving someone a job they deserve, not letting them enter a facility otherwise open to the general public, etc.—acquire any meaning, let alone the very specific meaning of expressing that the target can legitimately be treated worse than objectively required? Hellman does not spell this out (I will do so in Section 3), but she is adamant that it does not depend on “whether or not the people who enacted” this treatment “intend to express this meaning and whether or not those affected by it interpret it in this way” (Hellman, Reference Hellman and Lippert-Rasmussen2018, 100); rather, “culture and context play a significant role in determining” it (Hellman, Reference Hellman2008, 38). According to her, the meaning of any action is determined by society’s common ground, that is, by the explicit as well as the implicit background beliefs most members of said society share. Occasionally, this will make the correct assessment of an action’s meaning complex or even impossible, for instance, when there are no society-wide but only highly localized interpretations of the action (Hellman, Reference Hellman and Lippert-Rasmussen2018, 101). But often enough, we will be able to determine an action’s meaning through rigorous interpretive work (see Hellman, Reference Hellman2008, 41). For instance, against the common ground of white supremacy, “the meaning of segregation in the Jim Crow South was clear” (Hellman, Reference Hellman and Lippert-Rasmussen2018, 101) as degrading Black people. Hellman’s account is therefore not just a meaning based view, but an objective meaning based view of wrongful discrimination (Hellman, Reference Hellman and Lippert-Rasmussen2018, 100) as only those instances of differential treatment become instances of wrongful discrimination that express, given a society’s common ground, that their target has lower status.

With these clarifications in place, let us now turn to the question whether Lippert-Rasmussen’s and Hellman’s conceptions can do justice to the all-of-us intuition. To answer it, I first need to establish what doing justice to the all-of-us intuition amounts to. I contend that it implies three things: (i) First, to do justice to the all-of-us intuition, a conception of discrimination must make sense of the idea that a discriminatory action wrongs its direct target, that is, the person it is perpetrated against. Otherwise, it cannot explain why, in cases like “Committee,” we feel outrage on behalf of B. (ii) Second, and involving more complications, it must accommodate the idea that a discriminatory action also wrongs every member of the social group G to which B is perceived to belong. This is necessary to bear out the observation that, as fellow women in “Committee,” we would also feel offended on our own behalf and entitled to an apology. (iii) Third, such a conception of discrimination must capture the idea that the wrong incurred by the direct target is somehow more troublesome than the one incurred by each of their fellow group members.Footnote ⁸ This is crucial to make sense of the observation that while, as fellow women in “Committee,” we would feel some outrage on our own behalf, our moral opprobrium would center on how candidate B was treated. As I see it, there are two ways to achieve (iii): either by arguing that the wrong incurred by B is worse than the one suffered by each of their fellow group members or by claiming that B is wronged in more ways than their fellow group members.

If (i)–(iii) correctly state what it takes for a conception of discrimination to do justice to the all-of-us intuition, then it is clear that Lippert-Rasmussen’s view will not necessarily do so. According to him, an instance of group-related differential treatment x amounts to wrongful discrimination if x leads to a reduction in well-being incompatible with desert prioritarianism. But while Lippert-Rasmussen is adamant that x is an instance of wrongful discrimination only if it harms either those who are already worst off or those undeserving of such harm, he does not say anything over and above this about how the harm caused by x has to be distributed. The structure of his definition suggests that some of this harm will likely befall x’s direct target B as, per Lippert-Rasmussen’s condition (2), B is to be treated worse by x than some other person C, thus making it likely that x will fulfill implication (i) of the all-of-us intuition. But nothing in his definition necessitates that x will also fulfill (ii), that is, that x will also harm every other member of B’s (presumed) social group G.Footnote ⁹ Admittedly, nothing in Lippert-Rasmussen’s definition precludes this, either, and there may well be instances of differential treatment that meet his definition and that result in a distribution of harm which bears out (ii).Footnote ¹⁰ But since his definition remains silent on this issue, it remains a matter of chance whether an instance of wrongful discrimination according to Lippert-Rasmussen lives up to the all-of-us intuition or not.

Let us see if Hellman’s conception fares any better. At first glance, her view seems more promising as its conditions (3) and (4) explicitly allow for instances of discrimination in which the differential treatment constitutive of discrimination is demeaning to all members of G. Let us call these cases instances of group-demeaning discrimination. Assuming that demeaning constitutes a moral wrong, it follows that in group-demeaning instances of discrimination, the discriminatory action in question wrongs all members of G. In such cases, Hellman’s conception therefore fulfills (ii). And her conception also fulfills (i), the demand that a discriminatory action wrongs its direct target B, because as a member of G, B too is demeaned by x and therefore wronged by it. In fact, in such cases, B is wronged at all only in virtue of social group G being demeaned. This is because, according to Hellman, the differential treatment constitutive of discrimination does not as such contribute to discrimination’s moral valence—discrimination becomes morally problematic only insofar as it is demeaning. Put differently, in cases of group-demeaning discrimination, the only morally troublesome thing that happens to B is that their social group is being demeaned and they along with it. But while instances of group-demeaning differential treatment enable Hellman’s conception to kill two birds with one stone, so to speak, the way they allow it to fulfill both (ii) and (i) entails that it necessarily fails to live up to (iii). Because in group-demeaning instances of discrimination, the only wrong suffered by B is the same wrong as the one suffered by their fellow group members, that is, the demeaning of their shared social group. Thus, on Hellman’s view, there is no way of claiming that B is wronged worse or more often than their fellow group members.Footnote ¹¹ So while it successfully makes sense of the idea that an individual instance of discrimination wrongs more people than its direct target alone, Hellman’s view cannot explain how an individual instance of group-demeaning discrimination, while wronging all members of the social group targeted, wrongs its direct target first and foremost.

3. “Discrimination” Ameliorated

In the previous section, I argued that, each in its own way, both Lippert-Rasmussen’s and Hellman’s conceptions of (wrongful) discrimination fail to reliably do justice to the all-of-us intuition. Nevertheless, there is much to learn from both views. Hellman’s account provides us with a solution to the puzzle posed by (ii), namely, the idea that the all-of-us intuition implies that an individual discriminatory action wrongs every member of the targeted social group. No discriminatory action can bear out this idea simply in virtue of the differential treatment it involves. In “Committee,” for instance, B is certainly wronged by the search committee passing her over in favor of C, yet it is simply not the case that in passing B over, the search committee has also passed over every other woman. But as we can glean from Hellman’s conception, there is also a morally dubious expressive element to discrimination, and this expressive element can very well wrong more than one person and possibly even a whole social group. In contrast, Lippert-Rasmussen’s conception provides us with a way to address (iii), that is, the idea that every discriminatory action wrongs its direct target first and foremost because for him discrimination necessarily includes the mistreatment of the specific person at whom it is directed. My contention is that by putting Hellman’s and Lippert-Rasmussen’s insights together, we can create a conception of discrimination that succeeds in doing justice to all aspects of the all-of-us intuition. Specifically, I claim that we should adopt the following conception:

The influence of both Lippert-Rasmussen and Hellman should be obvious: Conditions (1) and (2) take their cue from Lippert-Rasmussen: (1) takes up his idea that discrimination focuses on social groups, which he articulates in his conditions (1) and (4), where he also makes room for cases of misdirected discrimination; (2) radicalizes his condition (2) as it stipulates that the differential treatment involved in discrimination has to amount to a moral wrong. In contrast, condition (3) is inspired by Hellman’s condition (3), though only by its group-related aspects.

These similarities notwithstanding, my conception differs from theirs in some important respects, which I outline now before I turn to the question of how my conception fares with regard to the all-of-us intuition. The first difference is that, on my view, we do not need to know anything about the agent’s motivating reasons—specifically, whether they were guided by any beliefs about their target’s group-membership or by group-related animus—to determine whether their action was an instance of discrimination or not. This is so because, on my conception, the agent’s de facto motivating reasons or intentions are not constitutive of discrimination at all. In the next section, I explain why I believe this omission to be advantageous for my view; for now, I merely seek to explain how I can afford to make it.

A critic might remark that, by disregarding the agent’s motivating reasons, I run the risk of missing the most important point about discrimination, which is that discrimination is a group-related phenomenon, that discrimination is about social groups (see Altman, Reference Altman and Zalta2020). So far, all accounts of discrimination have captured this group relatedness by insisting that the target’s group membership figure prominently in the agent’s motivating reasons. If we leave this part out, as I propose we do, do we not rob ourselves of the ability to distinguish between instances of differential treatment of a personal variety and instances that are group related and therefore rightly called discrimination? I would reply that my account does not do away with the group relatedness of discrimination: it just spells it out in a different way. Specifically, on my understanding, discrimination is about social groups insofar as discriminatory actions express something negative about the status of the social group they target.

Another difference is that I omit Hellman’s power dimension. While I very much agree with her idea that discriminatory actions express something negative about how their targets can legitimately be treated, I do not believe that these expressions must be accompanied by sufficient social power to then make it the case that the targets are de facto treated worse than what is objectively required. Others have already argued that it seems counterintuitive to rule out that people with little or no social power are able to discriminate simply on conceptual grounds (see Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 138). I thus consider this omission an asset of my view. I am able to make this omission because, in contrast to Hellman, I stipulate that the differential treatment involved in discrimination has to amount, by itself, to a moral wrong. As I argued earlier, Hellman only insists on the power dimension to ensure that discriminatory actions are indeed wronging their targets. Since condition (2) of my conception already takes care of this (as does, as will become clear presently, my condition (3)), there is no need for me to accept Hellman’s controversial condition (4).

My stipulation that the differential treatment involved in discrimination has to amount to a moral wrong is my third divergence from Lippert-Rasmussen’s and Hellman’s conceptions. This stipulation is necessary not, as one might expect, to strengthen condition (2), but to get condition (3) and, more specifically, its idea of expressive meaning off the ground. Like Hellman, I contend that those instances of differential treatment that will amount to discrimination express that members of the social group targeted can legitimately be treated worse than objectively required. But how are we to understand that an action acquires this specific expressive meaning? Unfortunately, Hellman herself has little to say on this matter. All she suggests is that certain types of action, for example, spitting at someone, can acquire this meaning by convention. While true, this is not very helpful as there are many and varied ways of differential treatment that can constitute discrimination (e.g., passing someone over for a job, not letting them enter a facility which is otherwise open to the public, paying them less for their work than other people, etc.) and it seems unlikely that all of them have acquired the necessary degrading meaning by convention. Therefore, I want to suggest that we rely on a different theory of expressive meaning according to which “an agent’s action expresses a principle when it would be objectively reasonable for observers to attribute to the agent an attitude of endorsement regarding that principle” (Shin, Reference Shin2009, 166; see also Anderson & Pildes, Reference Anderson and Pildes2000, Bratu, Reference Bratu2023). Given this theory of expressive meaning, A’s action x can express that members of the targeted social group can legitimately be treated worse than they objectively can only if, on the basis of A’s x-ing, it is reasonable to assume that A believes that members of the targeted social group can legitimately be treated worse than they objectively can. But it is only reasonable to assume that A believes that members of the targeted social group can legitimately be treated worse than they objectively can on basis of A’s x-ing, if x is in fact an action by which A treats a member of the targeted social group worse than they should objectively be treated, that is, if by x-ing A wrongs B. Otherwise, A’s action and the belief that (presumably) accompanies it would not be adequately connected.Footnote ¹² If discriminatory actions are supposed to express that their targets can legitimately be wronged, as I claim in condition (3) of my definition (or, for that matter, Hellman in condition (4) of hers), the actions that are supposed to do this expressing need to be some kind of wrong. Condition (2) of my definition reflects this necessity. With one exception (to which I will return presently), I remain open about the specific nature of the wrong at play in (2). I can do so because, as will become clear shortly, the precise way in which the differential treatment called for by (2) wrongs its target does not matter for my conception; also, this openness allows my conception of discrimination to be compatible with different background theories of morality. Thus, the differential treatment constituted by x could be wrong because it is unfair (as is B’s being passed over for the job she is qualified for in “Committee”), because it fails to maximize overall well-being in a desert-accommodating way (see Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 166), because it illegitimately curtails B’s deliberative freedom or sustains conditions of social subordination or denies B access to basic goods (see Moreau, Reference Moreau2020, 153) or whichever background theory of morality turns out to be correct. All that matters is that the wrong called for in condition (2) occurs in such a way that it acquires the expressive meaning called for in condition (3), namely that members of G can legitimately be subjected to this kind of wrong and thus be treated worse than objectively required.

Elsewhere I argued that we should call the expression of this specific content—that a person can legitimately be treated worse than they objectively can and thus has less moral status than they in fact do (cf. Bratu, Reference Bratu2023)—disrespectful. Footnote ¹³ I also argued that we should consider disrespect a moral wrong because people have the right not to have their moral status publicly called into question as this exposes them to the risk of future mistreatment. If we accept this, the following picture of discrimination emerges: every discriminatory action packs not just one, put two punches as it combines two distinct, but interrelated moral wrongs—first, an instance of differential treatment which, given some underlying theory of morality, amounts to a moral wrong and, second, an instance of wrongful disrespect. Put differently, discrimination is a compound wrong that consists of two distinct wrongs: discriminatory actions wrong their targets not only in some tangible way (for instance, by not hiring them even though they are the most qualified candidate, by restricting their access to areas otherwise open to the public, etc.), they also wrong them by trying to legitimize this wrongful behavior by expressing that it is permissible (when in fact it is not). Discrimination is therefore never only an injury, it always adds an insult. And in fact, the insult discrimination involves is particularly pernicious because, on my understanding, discriminatory actions do not just disrespect their direct target, they also disrespect the whole social group the target is reasonably perceived to belong to, that is, every member of their social group. Another way to spell out this idea is to draw on Kate Manne’s distinction between those actions that function as the “legislative branch” of some hierarchical and oppressive world view and those that function as its “law enforcement branch.” According to Manne, the former try to justify said world view, while the latter enforce its norms (cf. Manne, Reference Manne2018, 78f.). If what I argued is correct, discriminatory actions blur this distinction as they simultaneously serve in both branches.

On this picture, there is nothing truly special about discriminatory actions since other kinds of action also instantiate the wrongs constitutive of discrimination. There are many instances of differential treatment that wrong their targets in some way or other, just as there are other types of harmful action, for instance, hate speech, that wrong their targets by disrespecting them.Footnote ¹⁴ What’s the point then of having a philosophical concept specifically dedicated to discrimination, or so a critic might ask. I contend that having such a concept is useful because, even though discriminatory actions are not the only ones that can be disrespectful, the way they express their disrespect is particularly pernicious. As I defined it earlier, (group-related) disrespect always consists in an expression, specifically in the expression that we can legitimately treat members of some social group worse than we in fact can. Discriminatory actions express this content, but in addition they also treat a member of the targeted social group worse than they can legitimately be treated and thus as if their disrespectful statement was true. Put differently, discriminatory actions do not just proclaim “look, we can do xyz to people like them,” but in doing xyz to them, they also show that xyz can (descriptively speaking) be done to them. By combining their disrespectful statement with a matching action, discriminatory actions thus ostensibly provide those who witness them with evidence that backs up their pernicious message. Obviously, this evidence is deeply flawed as normative claims can never be justified by simply pointing to the status quo (not to mention that, given the correct moral theory, no social group can legitimately be treated worse than othersFootnote ¹⁵); but as social beings, we all are social cognizers and thus tend to treat what other people do as at least indicative for how to navigate the world. In short, discriminatory actions do not only express disrespect, they also make it seem like their disrespectful content was true which in turn raises the risk of their content being taken seriously. For this alone, disrespectful actions deserve a philosophical concept tailor-made to identify them.

But my account of discrimination does not only help to pick out actions that exhibit the pernicious combination of wrongs just described; in addition and in contrast to Lippert-Rasmussen’s and Hellman’s, it can also make sense of the all-of-us intuition. On my view, every discriminatory action x involves two wrongs: a wrongful instance of differential treatment and a wrongful instance of group-related disrespect. The former only wrongs x’s direct target B; the latter, in contrast, wrongs every member of the social group G B is reasonably perceived to be a member of, including B. Since B is wronged, both by being treated worse than C and by being disrespected as a member of G, (i) holds. Since every member of G is wronged, given that expressing x’s belittling meaning violates their right not to have their rightful moral status publicly called into question, (ii) holds as well. Finally, since B is wronged twice while their fellow social group members are wronged only once, there is a sense in which the wrong incurred by B is more troublesome than that incurred by every other member of G; thus, (iii) holds as well. Against the backdrop of my view, the all-of-us intuition elicited by “Committee” therefore makes perfect sense: as B is passed over unfairly in favor of C, it is reasonable for us to feel outrage on B’s behalf. It is also reasonable for us to feel outrage on behalf of all other women, including ourselves (provided we identify as women), given that the blatantly sexist way B is passed over is disrespectful to all women. But even though all women are disrespected in “Committee,” B is also cheated out of a job she was best qualified for, which is why our outrage should center on the wrongs perpetrated against B rather than the one perpetrated against us.

The one way my conception of discrimination could fail to live up to the all-of-us intuition is if the differential treatment wronging target B (which condition (2) calls for) was identical to the wrong vis-à-vis all members of the social group targeted (which condition (3) calls for), that is, if the differential treatment wronging B was itself an instance of disrespect toward the social group B is reasonably perceived to be a member of. In such a case, in which the wrong suffered by B was the same one as the one suffered by every other member of B’s presumed social group, my conception of discrimination would meet the same fate as Hellman’s and fall short of fulfilling implication (iii) of the all-of-us intuition as it would be impossible to claim that B had been wronged worse or in more ways by the discriminatory action than their fellow group members. Therefore, I am forced to make an exception to my general parsimony regarding the kinds of wrong we can insert in condition (2) and insist that it cannot be an instance of disrespect toward B’s presumed social group. But, as I will argue in section 5, this exception will allow us to make an important distinction.

Let us sum up. I argued that the all-of-us intuition helps us understand that individual discriminatory acts are even worse than hitherto acknowledged as they wrong not only their direct target but everybody in the targeted social group. Conceptions like Lippert-Rasmussen’s and Hellman’s that focus on just one aspect of the compound wrong discrimination consists in are thus not just in breach of some random conceptual intuition—they fail to give us a comprehensive account of the moral wrong intrinsic to discrimination and therefore risk trivializing it. This is problematic in terms of advocacy. As emancipatory-minded philosophers, we want to present the strongest possible case against discrimination and, arguably, we are better able to do so if we draw on a conception of discrimination that shows how every individual instance of discrimination attacks not just its direct target but every member of their (presumed) social group. But more importantly, as philosophers, we strive to paint a comprehensive picture of moral reality and in the case of discrimination we cannot do this without taking the all-of-us intuition into account.

4. What Else? Other Considerations in Favor of a Comprehensive Account

As noted, the main advantage of my understanding of discrimination is that it allows us to see and to advocate against all that is intrinsically wrong with it. But there are two additional advantages to my account.

First, my conception makes it easier to identify individual instances of discrimination than the intention-based (to use Mari Mikkola’s helpful term; see Mikkola, Reference Mikkola2016, 192) accounts of Lippert-Rasmussen and Hellman. For them, an action only turns discriminatory if the agent’s motivating reasons for it include some belief about or desire concerning the target’s social group membership (such as a false belief about women being meek or a desire to put them in their “proper place”). Such intention-based accounts make it difficult to identify discriminatory actions, since to do so, we need reliable information about what is in its agent’s head, so to speak, which is notoriously hard to come by. Interestingly, this difficulty has led some feminist philosophers to call into question the very usefulness of discrimination as a conceptual tool or to at least think of it as ideal theorizing. For instance, Iris Marion Young argues that we should think about women’s situation in terms of oppression rather than of discrimination, since the latter focuses on “perpetrators and intents” and therefore “becomes very difficult to prove” (Young, Reference Young1990, 196).Footnote ¹⁶

In contrast, my conception of discrimination requires no mindreading to assess whether an action is discriminatory or not. All we need is information about whether said action wrongs its target and whether it does so in a way that is disrespectful toward their (presumed) social group. Admittedly, accessing this type of information will not always be easy either. It requires considerable knowledge concerning how a specific action will likely be interpreted given society’s common ground. And this, in turn, requires considerable knowledge of a society’s history, culture, public discourse, and so forth. Still, I contend that acquiring this type of information is easier than divining another person’s motivating reasons, since it can be found in the public domain, that is, in history books, newspaper articles, blogs, etc., rather than stored away in people’s minds.

A critic might be willing to grant the point about my account’s greater accessibility, but wonder whether it might be just as much a bug as it is a positive feature. Specifically, they might wonder whether getting rid of the intention requirement renders my account too comprehensive.Footnote ¹⁷ A similar dynamic can be found in the debate about racism, where some want to reserve the term “racist” to actions that are caused by “particularly hateful beliefs and attitudes toward members of a racial group” (Anderson, Reference Anderson2010, 47) in order to avoid the proliferation of racist actions that would presumably result from a nonintention-based understanding of the term. In reply to this, I would point out two things. First, whether my account does in fact entail that many more actions would have to be considered discriminatory is an open question rather than a foregone conclusion. The answer depends on how many wrongful actions also acquire the necessary disrespectful meaning. Second and more importantly, even if it turns out that on my comprehensive conception there is more discrimination than on an intention-based one, this increase is not per se problematic because, as the debate about indirect discrimination shows (see the next point), we are already willing to accept that discrimination can occur without its perpetrators being motivated by bigoted beliefs or motives concerning the targeted social group. The fact that we are willing to accept indirect discrimination as a kind of discrimination proves that we are not wedded to an intention-based understanding of discrimination.

This brings me to the second additional advantage of my comprehensive conception, which concerns, precisely, indirect discrimination. I contend that my conception does a better job of making sense of and addressing what Lippert-Rasmussen calls the “indirect discrimination definition challenge” (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 55). To show this, I first have to clarify what I mean by “indirect discrimination” and the challenge raised by it.

Consider “Ambivalent,” the case of an employer who genuinely does not harbor any prejudices or animosity toward women but who, for purely financial reasons, does not want to hire people who are likely to request time off due to childcare responsibilities. Given the prevailing gendered division of labor, this policy will in all likelihood result in women not being hired, which intuitively should count as an instance of gender-based discrimination. Intention-based accounts of (direct) discrimination like Lippert-Rasmussen’s or Hellman’s cannot reflect this intuition since, by stipulation, the agent’s motivating reasons are devoid of any bigoted beliefs or desires. This assessment is problematic not only because it is counterintuitive, but also because it clashes with existing law. Assuming that in societies marred by prejudices and social exclusion, discrimination “often takes a form where it exists independently of discriminatory intentions” (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 55), lawmakers around the world have tried to accommodate cases like “Ambivalent” by introducing the concept of indirect discrimination which occurs without “any bias or discriminatory intention on part of the agent” (Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 55; see also Khaitan, Reference Khaitan and Lippert-Rasmussen2018, 31). In reaction to these legal developments, philosophers have devised their own understanding of the phenomenon. A paradigmatic example is Altman’s conception, which stipulates that an action or policy indirectly discriminates against some socially salient group G if and only if it disproportionately disadvantages members of G relative to members of other socially salient groups even though it neither explicitly targets nor is intended to disadvantage members of G qua being members of G (see Altman, Reference Altman and Zalta2020; for a slight variation, see Lippert-Rasmussen, Reference Lippert-Rasmussen2014, 72).

My claim is that definitions like this one fall short of doing justice to the challenge posed by indirect discrimination because, in effect, cases like “Ambivalent” elicit not one, but two intuitive reactions. Yes, they strike us as discriminatory, but they do not do so in a straightforward way. Instead, much like the famous rabbit–duck illusion and other reversible figures, our intuitive assessment of such cases oscillates depending on whether we focus on the operating intention behind these allegedly discriminatory actions (and, specifically, on the agents’ lack of bigoted beliefs and motives) or on the predictably uneven distribution of burdens and disadvantages they result in. To illustrate the point, imagine a person trying to make up their mind about “Ambivalent.” If they concentrate on the fact that the employer in question never hires a woman, they will in all likelihood consider the case to be one of gender-based discrimination. If, in contrast, they conceptualize it as a case in which an employer is just trying to keep their business afloat, they might even consider the employer’s reluctance to hire potentially unreliable people justified. In any case, their deliberation will be just that: a complex, multilayered deliberation and not a snap judgement. To incorporate this second intuition, we need an understanding of discrimination that not only allows us to identify cases like “Ambivalent” as (some kind of) discrimination but one that also takes into account that these are hard cases, cases in which we have to squint a little to see their discriminatory nature, so to speak.

As they are tailor-made to identify cases like “Ambivalent” as instances of indirect discrimination, conceptions like Altman’s evidently do justice to the first intuition, but precisely because they are tailor-made for such cases, they necessarily fall short of doing justice to the second. On conceptions like this, evaluating cases like “Ambivalent” will never be hard: they are by definition examples of indirect discrimination and there is no squinting required to see this.

On my comprehensive conception, matters are less straightforward: “Ambivalent” qualifies as an instance of discrimination if the employer wrongs their target and if they do so in a way that expresses disrespect toward the social group their target is perceived to belong to. Since the employer’s hiring practices dismiss a great number of people without even considering their qualifications, they certainly wrong at least those women who would have been qualified to fill the vacant positions. The first condition is thus met. When it comes to the second condition, though, things are less obvious. To establish whether the employer’s actions express disrespect, we need to ask ourselves how they would reasonably be interpreted, that is, which intentions people would reasonably ascribe to the employer. If the employer seems truthful and reasonably self-aware, one should simply believe what the employer tells them about their intentions as they are obviously best positioned to know about their own mental states. By this standard, the employer’s refusal to hire women would not express anything disrespectful about women: they would not mean anything else by it but just try to avoid additional costs. Yet circumstances conspire to undermine the employer’s credibility because, first, many people still harbor sexist prejudices either explicitly or implicitly and, second and more importantly, the employer’s willingness to bring about a situation that starkly disadvantages women fits right into this pattern. In light of this additional evidence, one would be justified in distrusting the employer’s avowed intentions and in believing that what they really intend to do is to keep women in their “proper place” (i.e., the home), which renders their actions disrespectful and, consequently, discriminatory.Footnote ¹⁸ But even though “Ambivalent” on my account ultimately turns out to be an instance of discrimination, we need to weigh conflicting evidence to reach this conclusion (i.e., the employer’s testimony and its presumptive first-person authority versus statistical evidence about prevailing sexist beliefs and the fact that the employer’s actions fit well with a sexist mindset). This accounts for the duck–rabbit quality typical of cases like “Ambivalent.”

I contend that these three features are present in all cases discussed in the literature under the label of indirect discrimination: that an agent wrongs their target by unfairly disadvantaging them; that there is an innocuous way to interpret their action (based on their own description of the matter) as well as one that renders it disrespectful toward the social group the target is perceived to belong to (based on the pervasiveness of stereotypes about said social group and on the fact that the agent is willing to unfairly disadvantage members of said group); and that the agent’s willingness to bring about this disadvantage ultimately renders more reasonable the interpretation of their action that implies disrespect. If this is so, then my account can satisfy both intuitions: it can identify such cases as instances of discrimination albeit not in a straightforward way.

5. Objections

I want to close by discussing two objections that would try to show that my comprehensive conception is in fact not comprehensive enough as it excludes cases that intuitively should count as discrimination. Each objection takes aim at a different condition of my conception. The first centers on condition (3) and thus on my claim that, to be discriminatory, an action must express that the members of the targeted social group have lower status. What about actions that wrong their targets and are performed by their agents out of group-centered animus or contempt but fail to express this negative stance in any way a critic might ask. To see what the critic is driving at, consider “Committee₂,” in which A, again, dismisses B’s application and does so because he passionately hates women. In fact, A turned into a misogynist after B had ended their relationship. While nobody knows about A’s hateful stance toward women, the fact that B has left him and that their separation left him shattered is well established. Under these circumstances, people will reasonably interpret A’s dismissal of B as an instance of personal (rather than group related) spite that expresses contempt of B as a person, but not of women as a social group. This would make his action an instance of unjust differential treatment but not, on my account, of discrimination. Yet since B’s dismissal is in fact driven by misogyny against all women, this assessment is false, or so a critic might contend.

I admit that, intuitively, cases like “Committee₂” seem to be prime examples of discrimination and that intention-based accounts like Lippert-Rasmussen’s and Hellman’s capture this intuition nicely. But as I stated in the introduction, I consider my conception to be ameliorative and thus am willing to accept a certain number of counterintuitive implications as long as these are offset by other theoretical benefits. As I stressed in section 3, the biggest theoretical benefit of a comprehensive conception like mine, which is tailored toward the all-of-us intuition rather than to the one elicited by “Committee₂,” is that it identifies actions that are morally problematic in a particularly pernicious way in that they not only wrong their direct target but also try to legitimize this kind of wrongful behavior by expressing falsehoods about how a whole social group can legitimately be treated. They thus include the risk of an untold number of future wrongs. In contrast, definitions that seek to capture cases like “Committee₂” focus more on the perpetrator. Specifically, they aim to identify those instances of wrongful group-based mistreatment that are perpetrated by “true believers,” that is, by people motivated by bigoted group-centered beliefs or motives—something my account is unable to do. Put bluntly, while my account strives to detect particularly bad actions, intention-based conceptions want to detect particularly bad actors. In my opinion, there is value to both approaches. From the point of view of improving matters for those suffering from group-based mistreatment, however, mine seems more tailored to the target because to prevent further group-based mistreatment, it is more important to identify (and consequently prevent) those instances that try to justify this kind of mistreatment rather than those that are perpetrated out of a particular motivation, however, despicable. So even though my comprehensive conception of discrimination cannot identify cases like “Committee₂” as instances of discrimination and therefore falls short of accommodating a plausible intuition, it does so in order to focus on actions that, from a practical point of view, matter more.

A critic might take these comments to heart and on their basis try to craft a second objection that centers on my condition (2). If I am so concerned about actions that try to legitimize the mistreatment of certain social groups, then what about actions that do just that by being disrespectful toward some social group, but do not in addition wrong any of its members? To illustrate this idea, consider “Committee₃,” in which A, again, makes a number of sexist remarks about women and their (alleged) lack of qualifications. In the end, though, A votes for B because he wants the department to hire a (supposedly) weak candidate so that his own authority will not be threatened. In this case, B and all other women have certainly been disrespected, but since B has not been passed over unfairly, she has not been wronged in any additional way.Footnote ¹⁹ On my account, A’s comments will not count as discriminatory, as they fall short of condition (2). But what is a hiring committee where people openly make sexist remarks if not discriminatory, or so a critic might ask.

My reply to this second objection somewhat resembles my reply to the first. Again, I concede that it seems intuitive to describe A’s comments in “Committee₃” as discriminatory and that my account cannot accommodate this intuition. But again, this failure does not strike me as problematic but instead as the result of focusing on a distinction that morally matters. Because even though group-based disrespectful actions are, as noted, particularly pernicious, group-based disrespectful actions that also wrong one of the group’s members in some additional way are worse. To see this, compare the situation of B in “Committee” and in “Committee₃”: obviously, in both cases, B and all women are the target of disrespect—but in “Committee,” B is also unfairly cheated out of a job she is qualified for. An account of discrimination that did away with condition (2), as the critic’s comment suggests, would not be able to pick up on this difference. This difference, however, matters a great deal, particularly to B, and a useful conception of discrimination should reflect this. Rather than accommodating the intuition elicited by “Committee₃,” I would therefore discount it by pointing out that while the critic rightly detects a similarity between “Committee” and “Committee₃” (both depict instances of group-related disrespect), they fail to pay sufficient attention to an important difference (in one of them the direct target is wronged twice).