wave of the US opioid epidemic, from the mid-1990s to 2010, involved counterposed racial disparities.1 Opioid mortality increased 167% among white Americans while declining 5.5% among Black Americans (Furr-Holden et al. 2021, 680).
The first wave of the US opioid epidemic, from the mid-1990s to 2010, involved counterposed racial disparities.2 Opioid mortality increased 167% among white Americans while declining 5.5% among Black Americans (Furr-Holden et al. 2021, 680).
What caused these disparate outcomes? The first cause involved differential prescription:
African-Americans received fewer opioid prescriptions […] because doctors believed, contrary to fact, that black people 1) were more likely to become addicted […] 2) would be more likely to sell the drugs and 3) had a higher pain threshold than white people because they were biologically different.3
Alexander et al. (2018) estimate that if doctors had prescribed opioids to Black Americans at the same rates as they were prescribed to white Americans, an additional 14,124 Black Americans would have died of overdoses between 1999 and 2017.
The second cause involved differential trafficking. Black tar heroin from Xalisco, Mexico, spread across US markets in the 1990s, but among the new addicts “almost everyone was white” (Quinones 2015, 8). This was because “the Xalisco cells never deal with African Americans. They don’t sell to black people; nor do they buy from blacks, who they fear will rob them. They sell almost exclusively to whites” (2015, 45). For example, in the late 1990s, a major Xalisco cell set up in Dayton, Ohio, then left that market for Columbus because it “had more white people.” Why? A leader of the cell said “the thing I didn’t like in Dayton was we were dealing with a lot of blacks. I don’t deal with blacks if I don’t have to. […] They’ll pistol-whip you” (quoted in Quinones 2015, 144).
The disparate outcomes involved a dramatic increase in white opioid mortality, with no corresponding increase in Black mortality. But the differential treatment involved doctors and dealers acting on the basis of anti-Black stereotypes, not anti-white stereotypes. One epidemiologist, Matthew Kiang, commented that differential prescription is “an extremely rare case where racial biases actually protected the population being discriminated against.”4 Intuitively, that’s true in both cases. If it is true in either case, we have a counterexample to a widely held view that is meant to play significant explanatory roles in contemporary theories of discrimination, namely:
Disadvantage: Discrimination against y necessarily disadvantages y.
My goal is to explore this challenge. Is Disadvantage true? And can it be explanatory?
I. The Target
Disadvantage is the received view. This does not mean that it is universally endorsed. For example, Niko Kolodny writes: “what is ordinarily meant by ‘discriminating against you’, I suggest, is giving you lesser regard as part of a system of differential regard that tracks a basing trait” such as race or disability (2023, 188). There’s no appeal to Disadvantage here. But that is exceptional. Disadvantage is very widely endorsed.
Disadvantage to y is said to partly define discrimination against y. Here’s Kasper Lippert-Rasmussen: “discrimination against someone simply is disadvantageous differential treatment” (2013, 15). Here’s John Gardner (2018, 62): “Discriminating against is discriminating to someone’s disadvantage”, in that “the discriminator (D) disadvantages the person discriminated against (E)”, and “the fact of E’s disadvantage that is built into the ‘against’.” Many even use “discrimination” and “disadvantageous treatment” interchangeably: see Ben Eidelson (2015, 16) and Michael Foran (2022, 9).
Perhaps some only mean to endorse the weaker view that disadvantage is a typical feature of discrimination against someone. But when disadvantage to y is said to partly define discrimination against y, many are explicit that a biconditional holds between the definiendum and the definiens: see, e.g., Sangiuliano (2024, 65). Hence, it is unsurprising that many explicitly say disadvantage is a necessary or universal feature of discrimination against someone. Here’s Lippert-Rasmussen: “Discrimination necessarily involves disadvantageous treatment” (2013, 190, fn. 3). Here’s Gardner: “in discriminating against someone, one is necessarily acting to the disadvantage of that someone” (2018, 67). Here’s Eidelson: “to discriminate against someone, one must treat him differently in some dimension than others, and this differential treatment must be comparatively disadvantageous to him in some respect” (2015, 15). Here’s Shu Ishida: “X discriminates against Y if and only if X disadvantageously treats Y” (2021, 485). Here’s Sophia Moreau: it is “a feature of all cases of discrimination qua discrimination” that “one or more people are disadvantaged, in relation to others, on the basis of certain traits” (2020, 157).
There is plenty more evidence that Disadvantage is widely endorsed. Consider: “When P is not disadvantaged […] one cannot speak of discrimination against her” (Statman 2021, 320); “discrimination consists of acts, practices, or policies that impose a relative disadvantage on persons based on their membership in a salient social group” (Altman 2020); “a person is said to discriminate if she disadvantages others on the basis of their race, ethnicity, or other group membership” (Wasserman 1998); and others.5
Disadvantage also plays important dialectical roles in discussions of discrimination. For one example, consider Shlomi Segall’s defense of an equality of opportunity account of the badness of discrimination. Segall writes: “when discrimination is bad it is surely so because someone else … has been discriminated against”, and if so, “then surely the badness of the overall act must be somehow associated with that other (disadvantaged) person” (2012, 86). The assumption that the person discriminated against is the disadvantaged person is central to Segall’s equality of opportunity view: “To suffer unequal opportunity (that is, to be disadvantaged relative to others) is the one and only bad accounting for the badness of discrimination” (2012, 90). For other examples, I could point to recent work on intersectional, algorithmic, and epistemic discrimination.6
In a literature rife with disagreement, Disadvantage is a rare point of convergence. And a surprising one, too, in three respects. First, as Berndt Rasmussen (2019, 874) notes—citing Deborah Hellman (Hellman 2008), among others—most define discrimination against someone in terms of disadvantage while explicitly denying that disadvantage is a core wrong-making feature of discrimination. Second, Disadvantage is assumed, not argued for. Thomsen (2015, 304), for example, makes this explicit, asserting that the “only significant difference” between “discrimination against” and “discrimination between” is that the former “disadvantages the discriminatee”, then declining to “argue this point.” Third, many accept Disadvantage while saying “disadvantage has been left undefined, in order not to preempt debates about the kind of disadvantage that counts” (Khaitan 2015, 34). As Gardner put it, we can appeal to Disadvantage “without being wedded to any sectarian account of what counts as a relevant disadvantage, let alone of how to go about counting it” (2018, 79). To my mind, it is unclear why Disadvantage should be a point of theoretical convergence when it does not follow from common commitments about why discrimination is wrong, is not argued for, and relies upon one central notion that has been left undefined. Nevertheless, Disadvantage remains the received view. So, we should ask: Is it true?
This is rarely challenged. Indeed, the only direct, extensive challenge I know of comes from Oscar Horta in the aptly titled “Does Discrimination Require Disadvantage?”. Horta writes that it is “commonly assumed that discrimination entails the infliction of some kind of disadvantage upon discriminatees” (2015, 277). Horta then appeals to counterexamples such as a “racist viewer” of a cartoon who thinks that the animated fictional European character is “cool” but the animated fictional Native American character “is a fool and ugly” (2015, 281). But a proponent of Disadvantage can deny that this is a case of discrimination; arguably, it is a thought crime, whose “victim” is not even a moral patient.7 Similar problems arise for Horta’s other examples.
Several cases that are already discussed in the literature provide better fodder for potential challenges to Disadvantage. But this potential remains unrealized. Consider Søren Midtgaard and Viki Pedersen’s ‘Paternalistic Discrimination’. They begin: “Discrimination is typically characterized by disadvantageous differential treatment” (Midtgaard and Pedersen 2024, 1). Then they ask whether, say, prohibiting women from bullfighting can constitute “paternalistic discrimination against” women (2024, 5). They answer that it can: even though it is advantageous “globally”, “paternalistic discrimination [is] disadvantageous to the paternalizee(s)”, because it is disadvantageous “with respect to” some important interests (2024, 6). This exemplifies a trend: when theorists consider examples that seem hard to square with Disadvantage, they preserve the received view by adopting a more sectarian account of what counts as a relevant disadvantage. (Whether that trend has been ill-advised will be a major theme going forward.) Few seriously consider whether such cases should instead push us to reject Disadvantage.
II. Who was Discriminated Against?
Since it began in the 1990s, the opioid epidemic has been largely ignored in philosophy, outside of a small body of work in bioethics.8 This is surprising.9 In 2022, The Lancet still described this epidemic as “one of the worst public health disasters affecting the USA and Canada,” comparable to Covid-19 (which is widely discussed in philosophy).
This is also unfortunate. Like Moreau (2020, 29), I think “it is very important to use … real cases when theorizing about discrimination,” to “test proposed theories against our moral intuitions.” True, it can help to consider simple hypothetical cases—e.g., where A fires B for being Black. But we miss something if we only test theories against, in Moreau’s words, “hypothetical cases in which there is no extended social context to analyze.” Furthermore, I think we miss something if we only test theories against cases where that extended social context is familiar. We must also attend to surprising cases like the opioid epidemic, where various racial disparities do not move in lockstep. Such cases may be atypical, but they are not peripheral. Indeed, because they are atypical, they promise to illuminate familiar phenomena in an unfamiliar light, and thereby help us see the nature of discrimination more clearly. So, intuitions about such cases contain important philosophical data. Indeed, I think they are more trustworthy than any intuitions about the truth of an abstract theoretical generalization like Disadvantage.
The present challenge to Disadvantage starts with the intuition that both differential prescription and differential trafficking were cases of discrimination against Black Americans, but did not disadvantage them. We can see the force of the challenge by considering, in turn, the two possible responses. In the next section, I’ll consider the second response: affirming that differential prescription and trafficking involved discrimination against Black Americans, by saying both practices disadvantaged Black Americans. In the rest of this section, I’ll consider the first response: biting the bullet by denying that either case constitutes discrimination against Black Americans.
That is a rather large bullet to bite. Differential prescription and trafficking are similar to paradigm cases of anti-Black discrimination. In providing goods and services, doctors and dealers were motivated by insufficient regard for Black people, and acted on prejudicial racial stereotypes of Black people as untrustworthy, criminal, dangerous, and biologically abnormal.10 Black Americans have a legitimate complaint about being treated differently on this basis, and it seems to be a complaint about discrimination.
This response faces a further, subtler problem. It suggests that these cases involved discrimination against white Americans. Differential prescription and trafficking are cases of differential treatment that satisfy common definitions of direct discrimination (see, e.g., Lippert-Rasmussen 2013, 15, 26–27; Eidelson 2015, 17; Gardner 2018, 56).11 Who is disadvantaged is meant to settle who was discriminated against (§IV). If these cases did not disadvantage Black Americans, they certainly disadvantaged white Americans. So, suppose Disadvantage is true and differential prescription and trafficking are cases of discrimination between white and Black Americans that disadvantaged white but not Black Americans. It follows that doctors and dealers directly discriminated against white Americans. That’s a bigger bullet to bite. (Other responses to the challenge also lead to this problem, so I’ll say more about it in §IV.)
It would, of course, be far too quick to hold that any philosophical theory must be false if it has counterintuitive implications. Perhaps consequentialism is true, and we should bite the bullet about turning the trolley. But if you are tempted to bite the bullet here, two comparisons to consequentialism are worth noting. First, Disadvantage is not treated as one option among others; it is the received view. Second, no one has offered an argument for Disadvantage. Why should an undefended theoretical assumption remain an orthodoxy once we see its contentious implications for real-world cases?
III. The Kind of Disadvantage that Counts
The second response is to affirm that differential prescription and trafficking involved discrimination against Black Americans, by positing that both practices disadvantaged Black Americans. This is more promising. But it is also trickier to evaluate. Its plausibility depends on what kind of disadvantage counts, and as we saw, proponents of Disadvantage leave this undefined. Disadvantage is, in that way, a moving target.
Worse yet, when we try to pin it down as a target, several cross-cutting distinctions between kinds of disadvantage can be made, making the theoretical options proliferate. In this section I focus on the distinction that is also most frequently applied by proponents of Disadvantage to other challenging cases. We saw an illustration of it above: paternalistic discrimination disadvantages the paternalized not “globally” but “with respect to” some important interests (Midtgaard and Pedersen 2024, 6).
What is the distinction at issue? Lippert-Rasmussen glosses the difference between “local” and “global disadvantage” as being “whether it suffices that the relevant act or policy is disadvantageous in some respect, or whether it needs to be disadvantageous all things considered” (2013, 61). Many say local disadvantage is sufficient. Here’s Thomsen (2015, 305): “the disadvantage at stake may be restricted to a particular domain rather than all things considered.” Here’s Gardner (2018, 62 fn. 9): “[d]iscriminating against is discriminating to someone’s disadvantage”, but “need not be to anyone’s overall disadvantage.” See also Eidelson (2015, 32) and Ishida (2021, 490).
Holding that local disadvantage is sufficient seems like a promising response to the differential prescription and trafficking. The dramatic increase in white opioid mortality swamps other important considerations, making it implausible that differential prescription and trafficking disadvantaged Black Americans relative to white Americans globally or all things considered. Despite this, differential prescription and trafficking may have locally disadvantaged Black Americans relative to white Americans. Let’s consider some possible candidates for why this would be.
Analgesia. “Because African-Americans were less likely to receive [opioid] prescriptions, they were less likely to become addicted—though they were more likely to endure unnecessary and excruciating pain for illnesses like cancer” (Frakt and Monkovic 2019). This may seem like the most promising candidate: it is hard to deny that being subject to excruciating pain is a local disadvantage, or that in some cases of differential prescription doctors subjected Black Americans to this harm. The problem, though, is that Disadvantage is not merely a claim about some cases. It is a claim about all cases. Yet this local disadvantage is not present in differential trafficking. Nor is it present in many, if not most, cases of differential prescription. The opioid epidemic arose partly due to the explosion of opioid prescriptions for common conditions such as chronic back pain, for which opioids are no more effective than other analgesics.12 So this proposal may explain why oncologists disadvantaged the Black patients that they discriminated against, but it does not explain why most other doctors did so. If you go to a doctor with back pain and they prescribe you nothing, it seems odd to say that their failure to prescribe you OxyContin is what caused you unnecessary pain. The problem was their failure to prescribe effective, non-addictive analgesics. So, when doctors prescribed opioids to white but not Black patients with chronic back pain, that act of discrimination against Black patients did not cause them unnecessary pain.
Goods. A simpler proposal does have the requisite generality. Differential prescription and trafficking may have locally disadvantaged Black Americans relative to white Americans by denying the former a good—opioids—regardless of whether that good is good for the recipient. It is beyond dispute that differential prescription and trafficking involved this kind of local disadvantage: black tar heroin for recreation and OxyContin for back pain are a good, in this sense. But this kind of local disadvantage is trivial. Proponents of Disadvantage do not think disadvantage is trivial. Like Ishida (2021, 487, fn. 13), many “use ‘harm’ and ‘disadvantage’ interchangeably.”13 Sangiuliano similarly says that by definition “a discriminatory act against a victim disadvantages her; it sets back an aspect of the victim’s wellbeing”, as a “local harm” (2024, 66). After saying that disadvantage to y is understood in terms of setting back y’s “specific interests”, Eidelson asks whether a serial killer who only kills white women “discriminates against men and non-white women, then, by not targeting them for killing”, and answers “I think we should not. According to my theory, that is explained by the fact that people do not have a specific interest in being targeted for killing” (2015, 33). We should similarly deny that someone who posts anthrax only to women thereby disadvantages or discriminates against men. Best not to save Disadvantage by trivializing disadvantage. Aside from seeming unwelcome on its face, we’ll see that this move exacerbates the problem that Disadvantage ceases to be explanatory (§IV and §V).
Preferred goods. A similar, but better, proposal is that opioids were good for the recipients simply in virtue of being desired. Eidelson writes that “paradigm cases of discrimination involve differential satisfaction of such an interest in being granted an opportunity one has sought” (2015, 32). Black tar heroin for recreation and OxyContin for back pain can be good in this non-trivial sense. If white and Black Americans both sought opioids but only the former received them, their specific interest in receiving opioids was differentially satisfied even if not getting heroin, or getting a different analgesic, would have been better for them. As such, this sets a low bar for disadvantage. But it still does not have the requisite generality. For one, not all patients sought opioids specifically, as opposed to pain relief generally. For another, consider the modus operandi of Xalisco cells from the 1990s. They “discovered that methadone clinics were, in effect, game preserves. […] Every new cell learned to find the methadone clinic and give away free samples to the addicts” (Quinones 2015, 64–65). When Xalisco cells avoided operating in predominantly Black cities, they subjected fewer recovering Black addicts to this predatory behavior. Those recovering addicts had not sought and did not want this opportunity to take heroin recreationally—hence being at methadone clinics.
Stigma. In general, the differential provision of goods and services can be stigmatizing even when the goods in question are harmful—see Arneson (2006, 790), Hellman (2008, 23, fn. 13), and Eidelson (2015, 32). Take casinos. The expected return from gambling is negative, but barring Black Americans entry to casinos is locally disadvantageous in this non-trivial sense. It blocks a marginalized group from participating in public life and thereby sends a message that they are unfit for such participation. Stigma is a significant setback to interests. But it does not make differential prescription and trafficking locally disadvantageous. This is partly because the practices were not sufficiently public. What doctors write on prescription pads is confidential. When dealers covertly set up shop in majority white cities but not majority Black cities, they don’t exactly announce this. Perhaps these practices risked causing stigma, but this move raises tricky issues.14 Setting them aside, any such risk is massively outweighed. Increasing opioid prescriptions and trafficking would have increased addiction among Black Americans, and thereby risked far more stigma. There’s plenty of evidence to support this (Murakawa 2014). The crack epidemic in the 1980s is Exhibit A. The opioid epidemic in the 1990s would have been Exhibit B, if racial biases had not protected the population being discriminated against from this important dimension of disadvantage.
Discrimination is a disadvantage. A final proposal is that people have an interest in not being discriminated against. Several proponents of Disadvantage make this move. Ishida (2021, 494), for example, writes: “Someone may reasonably endorse a life in which he is not subjected to discrimination by virtue of his skin colour, even when it is not disadvantageous” in other respects. Eidelson (2015, 32–33) makes a similar move. (For critical discussion and further references, see Sangiuliano (2024, esp. 71–72).) If the fact that an act discriminates against y explains why it disadvantages y, it is easy to see how Disadvantage can be true in the face of literally any putative counterexample.
What’s hard to see is how Disadvantage can remain explanatory once we make this move. That someone’s differential treatment of you and me disadvantaged you is meant to explain why it was discrimination against you, not me (§IV). If the act disadvantaged you because it discriminated against you, this explanation becomes viciously circular.
Is all of this conclusive? Of course not. I have not yet touched on two other important distinctions between kinds of disadvantage (see §IV and §V). Nor will I address every distinction between kinds of disadvantage. But my goal is to advance a challenge, not offer an impossibility proof. The onus is on proponents of Disadvantage to take up the gauntlet and show why it is true, in the face of these cases, that discrimination against you necessarily disadvantages you.
As we just saw above, any such response is also subject to a significant constraint: it had better not salvage the truth of Disadvantage by sacrificing its theoretical significance. Disadvantage is meant to explain both the directionality and unity of discrimination. If Disadvantage cannot do so, it becomes an idle wheel in theories of discrimination. Let’s turn, then, to exploring what the received view says about these explanatory roles.
IV. The Directionality of Discrimination
By explaining the directionality of discrimination, I mean the following. Take any case of differential treatment: x treats y differently than z. What makes x’s act a case of discrimination against y rather than against z? Most answer: disadvantage to y!
This answer was, to my knowledge, first offered by Radcliffe Richards (1985, 64). It is also offered by Thomsen (2015, 304): “discrimination against” differs from “discrimination for or discrimination between”, and the “difference between these” is that “the first disadvantages, the second advantages and the third neither advantages nor disadvantages.” Eidelson (2015, 17, 30–36) and (Lippert-Rasmussen 2006, 170; also Lippert-Rasmussen 2013, 15) also provide accounts of discrimination against y in terms of disadvantage to y. As do many others. For this reason, many emphasize that disadvantage is strictly speaking not a necessary feature of discrimination; it is a necessary feature of discrimination against someone. “Disadvantaging is not an essential element of discriminating, but it is an essential element of discriminating against” (Gardner 2018, 62); “‘discrimination against’ (as opposed to ‘discrimination between’ as in giving people of different sizes clothes of different sizes) always involves … differential, disadvantageous treatment” (Räsänen and Lippert-Rasmussen 2024, 683). Disadvantage explains who is discriminated against.
This explanatory role is significant. Radcliffe Richards (1985, 64) wrote that “the kind of discrimination needed to give a group grounds for complaint” is “discrimination against [that group].” As such, Lippert-Rasmussen (2013, 14)’s “book focuses on discrimination against individuals.” One way to put this is that we have a directed duty to not discriminate. This duty is owed to someone, who is wronged when it is violated. Who is wronged? Whoever was discriminated against! As Andrew Altman explains the idea:
Differential treatment is symmetrical: if blacks are treated differently from whites, then whites must be treated differently from blacks. But it is implausible to hold that Jim Crow and South Africa’s apartheid system discriminated against whites. The system arguably held back economic progress for everyone in the South, but that point is quite different from the implausible claim that everyone was a victim of discrimination. Accordingly, it is better to think of discrimination in terms of disadvantageous treatment rather than simply differential treatment. Discrimination imposes a disadvantage on certain persons relative to others, and those who are treated more favorably are not to be seen as victims of discrimination (2020, §1.1).
Why were Jim Crow and Apartheid cases of discrimination between white and Black people but only discrimination against Black people? Because they disadvantaged Black but not white people. Why does this matter? Because it is false that these practices “discriminated against whites”, such that “everyone was a victim of discrimination.” The victim, who bore the legitimate complaint, was the group discriminated against.
Suppose Disadvantage is true. Does it explain the directionality of discrimination? Put differently: could it be true that when x treats y differently than z and thereby discriminates against y rather than z this necessarily disadvantages y, but false that this disadvantage explains why x discriminated against y rather than z? It could. Indeed, we already have a helpful illustration of how Disadvantage could be true without being explanatory in this sense. Some propose that an act that differentially treats you and me disadvantages you because it discriminated against you. If so, the act cannot constitute discrimination against you because it disadvantaged you. That’d be circular.
The opioid epidemic suggests two further, less obvious reasons to think that even if Disadvantage is true, it does not plausibly explain the directionality of discrimination.
Symmetry
When x treats y differently than z, this is symmetric: x also thereby treats z differently than y. What makes x’s act a case of discrimination against y rather than against z? If Disadvantage provides the answer, then the kind of disadvantage that counts should be asymmetric. That is, when x disadvantages y, x should not also disadvantage z!
This is, I think, why everyone from Radcliffe Richards to Altman—see Eidelson, Moreau, Segall, and others quoted in §I—says that the kind of disadvantage that counts is relative (a.k.a., “comparative”, “interpersonal”), not absolute (a.k.a., “non-comparative”, “intrapersonal”). This is close to a consensus in the literature.15 Here’s Lippert-Rasmussen (2013, 16): “The relevant disadvantage is interpersonal, not intrapersonal”, as “an act of discrimination against Y may improve Y’s situation relative to the way it was prior to the discriminatory act and relative to how it would have been had the discriminatory act not taken place.” Here’s Mary Kate McGowan (2019, 167 n. 24): “To say that non-whites are discriminated against in the United States today … is to say that they are disadvantaged relative to whites in the United States today.”
It is surprising, then, that many proponents of Disadvantage say that the kind of disadvantage that counts must be relative rather than absolute but can be local rather than global.16 Absolute and local disadvantage pose a problem for the same reason. Both can be symmetric. When x differentially treats y and z, this can disadvantage both y and z in absolute terms. And it can also disadvantage y relative to z in one respect, while doing the inverse in another respect.
This initial observation is the basis for a dilemma for proponents of Disadvantage. Differential prescription and trafficking constitute discrimination against Black Americans. So, suppose both cases disadvantaged Black Americans. In what sense? Consider appeals to absolute global disadvantage: in these cases, Black Americans were made worse off than they were or would otherwise have been (all-things-considered). As we saw, the trouble here is that the same is plausibly true for white Americans, given the catastrophic increase in white opioid addiction and mortality. So differential prescription and trafficking disadvantaged both groups, in this sense. If disadvantage is what explains who was discriminated against, we’re now led to the verdict that these cases constitute direct discrimination against both Black and white Americans.
Now consider appeals to relative local disadvantage. There are several ways of defending this, some of which are more attractive than others (§III). But we can here set such nuances aside, as they all face a single objection. If differential prescription and trafficking disadvantaged Black Americans relative to white Americans in any respect (denying goods, increasing pain, risking stigma), they also disadvantaged white Americans relative to Black Americans in other respects (increasing addiction and overdoses). In different respects, each group was disadvantaged relative to the other and treated more/less favorably than the other. So, appealing to absolute or local disadvantage makes these cases of symmetric discrimination against Black and white Americans, such that both groups were the victims of discrimination.
This problem generalizes. For example, barring Black citizens from jury duty disadvantages them on many important dimensions, but it disadvantages white citizens along some important dimensions too, as being more regularly summoned for jury duty reduces one’s liberty. Affirmative action is a particularly tricky case, since it tends to have relative advantages and disadvantages for underrepresented groups (and correlative relative disadvantages and advantages for other groups). In myriad cases, it is hard to show that the group discriminated against is disadvantaged without lowering the bar for the kind of local disadvantage that counts. In any case of differential treatment between Black and white Americans that discriminates against the former, such accounts make it easier to locate local disadvantage to the former. But for the same reason, they also make it easier to locate local disadvantage to the latter. (If ‘disadvantage’ is trivialized, we will always locate disadvantage to the latter!) As such, appeals to local disadvantage make it easier to see how Disadvantage can be true in the face of putative counterexamples, but harder to see how it can be explanatory.
What if we instead appealed to relative global disadvantage? This kind of disadvantage is asymmetric: if x’s act disadvantaged y relative to z all-things-considered, then x’s act cannot also have disadvantaged z relative to y all-things-considered. (More carefully: it is asymmetric, subject to a caveat that I discuss in §V.) So, this move makes it easier to see how Disadvantage can be explanatory. But it makes it harder to see how it can be true in the face of putative counterexamples: it is implausible that differential prescription and trafficking involved relative global disadvantage to Black Americans.
To be clear, my case for this dilemma does not turn on my adopting a sectarian view about the kind of disadvantage that should count. Instead, my conjecture is that on any account of the kind of disadvantage that counts, proponents of Disadvantage face one of two problems. One problem is failing to explain why the relevant cases involved disadvantage to Black Americans, and hence discrimination against Black Americans (if relative global disadvantage is what counts, Disadvantage faces counterexamples). The other is failing to explain why the relevant cases did not also involve disadvantage to white Americans, and hence discrimination against both Black and white Americans (if absolute disadvantage or relative local disadvantage are the kind that counts, then Disadvantage becomes ill-suited to explaining the directionality of discrimination).
Consider two responses. First, perhaps in the relevant cases discrimination is asymmetric, even though disadvantage is symmetric. If so, something other than disadvantage must explain why these cases constitute discrimination against Black but not white Americans, but that’s consistent with holding that disadvantage explains the directionality of discrimination in other cases—which is still an explanatory role. But notice that proponents of Disadvantage, from Radcliffe Richards to Altman, endorse something quite strong and ambitious. They define discrimination against someone in terms of disadvantage to them. Thomsen (2015, 304) does not say that disadvantage occasionally distinguishes discrimination between from discrimination against; it is the “only” difference between the two. If we do not claim that disadvantage necessarily explains the directionality of discrimination, we do not need to posit the strong commitment that discrimination against you necessarily disadvantages you. All we need to posit is the modest thesis that discrimination against you often disadvantages you.
Second, perhaps in the relevant cases discrimination and disadvantage are both symmetric after all. Perhaps differential prescription and trafficking did discriminate against both Black and white Americans, and barring Black Americans from jury duty did discriminate against both white and Black Americans. So-called “anti-classificationists” may embrace this verdict, but most discrimination theorists offer trenchant objections to that view (e.g., Hellman 2016, 914–917). Eidelson also indicates a willingness to accept such verdicts, while notably still treating them as bullets to bite.17
But this is not a bullet we should be willing to bite. To see why, let’s switch gears and turn to why directionality of discrimination matters: its relevance to who is wronged.
Wronging
In §I, I noted that Disadvantage is a surprising point of convergence in the literature partly because none of the prominent views of what makes discrimination wrong appeal to disadvantage. Consider two examples. For Hellman, what makes discrimination wrong is that it demeans, where “to demean” is an expressive act—i.e., “to put down, to diminish and denigrate”, or “to treat another as lesser”, relative to someone else (2008, 29). For Eidelson, what makes discrimination wrong is that it disrespects, which is a “certain kind of deliberative failing” (2015, 80). To demean someone or disrespect someone, they hold, is not in itself to disadvantage them.
This has two significant implications. First, it explains why we should not respond to the present challenge by holding that doctors and dealers disadvantaged white Americans, so this group were (also) discriminated against. Of course, white Americans have a legitimate complaint about the outcomes in epidemic, given the material harms they suffered. But it is not a complaint about direct discrimination. Put simply: nothing in the causes of these disparate outcomes seems to be anti-white.18 This makes it hard to see why white Americans were thereby subject to the wrong of discrimination, on any theory. Doctors who overprescribed opioids to white patients did not expressively demean them or treat them with lesser regard than Black patients. Doctors and dealers were guilty of a deliberative failing, but not of the right kind. For example, doctors’ medical negligence was disrespectful to patients, but was not comparatively disrespectful to white patients—and discrimination “distinctively involves comparative disrespect” (Eidelson 2015, 91). You may demur. But if you say suffering harmful consequences for any reason generates a complaint about direct discrimination, what does that complaint distinctively involve? Why think discrimination is a distinctive moral wrong?
The second and more substantial implication is that we come to a further reason to think that even if Disadvantage is true, it does not plausibly explain the directionality of discrimination. The directionality of discrimination captures who is wronged. We saw this claim above many times. Discrimination against a group is what’s needed for them to be “victims” of discrimination (Altman 2020) who have “grounds for [a] complaint” about discrimination (Radcliffe Richards 1985). Now suppose that an act of discrimination is wrong because it demeans or disrespects but constitutes discrimination against someone because it disadvantages them, and wrongs someone because it discriminates against them. This is an odd theoretical package: what makes an act of discrimination wrong is untethered from what makes an act of discrimination wrong someone.
We can frame this problem as a further dilemma. Suppose we say that disadvantage explains why discrimination is against someone, but does not thereby explain why it wrongs them. This gives up the common account of why the directionality of discrimination matters: that it tells us who is the victim, who has the legitimate complaint. If we’re pushed to give up on that account, we must ask: Why does it matter whether discrimination between you and me is against you rather than against me? If the disadvantage to you explains why you are discriminated against, but this does not make you a victim or give you a legitimate complaint, what exactly have we explained?
Alternatively, suppose we say that disadvantage explains why discrimination is against someone and thereby explains why it wrongs them. But aside from generally being in tension with common views about why discrimination is wrong (e.g., because it demeans or disrespects) this will further constrain our options with respect to the kind of disadvantage that counts in understanding Disadvantage itself. Recall some proposals from §III: differential prescription and trafficking disadvantaged Black Americans by denying them goods, or preferred goods. These set a low bar—sometimes a trivial bar—for disadvantage. This makes it easier to show how Disadvantage can be true in the face of putative counterexamples. But they now make it harder to see how Disadvantage can be explanatory, leaving aside concerns about symmetry. These kinds of disadvantage hardly look promising as explanations of who is wronged. Conversely, to make Disadvantage more plausibly explanatory, we would need to set a higher bar for the kind of disadvantage that counts. But that will make it harder to show how Disadvantage can be true, as it is more vulnerable to counterexamples.
More can be said here. But I hope to have illustrated why the opioid epidemic can call into question whether Disadvantage is true and whether it is explanatory. This should make us apprehensive about defenses of Disadvantage that offer pyrrhic victories at best: they risk upholding the truth of the claim that discrimination against someone necessarily disadvantages them, by robbing it of its explanatory significance.
V. The Unity of Discrimination
Even if Disadvantage does not explain the directionality of discrimination, it may still explain the unity of discrimination. So far, we have been concerned with direct discrimination (differential treatment), not indirect discrimination (disparate impact). What, if anything, makes direct and indirect discrimination form a unified kind?
Notably, some proponents of Disadvantage deny the unity of direct and indirect discrimination. Gardner argues that the duty not to discriminate directly is a duty of corrective justice, but the “duty not to discriminate indirectly focuses centrally on the relative disadvantage of some applicants as such”, so it is a “duty of distributive, not corrective, justice” (1996, 360). Eidelson’s stance is similar (2015, 39).
But theirs is the minority view. Most say direct and indirect discrimination are “forms of the same thing, the same kind of injustice” (Moreau 2010, 176), and that theories of discrimination should be rejected if they cannot cast direct and indirect discrimination as species of a genus (Lippert-Rasmussen 2013, 36, ch. 2). How can we explain the unity of direct and indirect discrimination? If both definitionally involve disadvantage, they are plausibly species of a genus (Lippert-Rasmussen 2013, 74–76), and the duty not to discriminate directly against y also focuses centrally on the disadvantage to y and hence involves the same kind of injustice (Moreau 2020, 6). Differential treatment concerns being directly disadvantaged because we possess a protected trait, while “disparate impact” concerns being “disadvantaged even indirectly because we possess these traits” (Moreau 2010, 155).
Disadvantage thus seems to provide a powerful explanation of the unity of discrimination. Of course, it only succeeds if Disadvantage is true. If it is false, the argument is turned on its head. It is a given that indirect discrimination necessarily involves disadvantage. If direct discrimination doesn’t, we have further evidence for disunity.
More interestingly, though, is the challenge that Disadvantage might once again turn out to be true without being explanatory. That is, it could be true that direct and indirect discrimination necessarily involve disadvantage, without this explaining why they are species of a genus or forms of the same kind of injustice. Once again, this challenge turns on the kind of disadvantage that counts. Direct and indirect discrimination could both necessarily involve disadvantage, but not involve the same kind of disadvantage; if so, Disadvantage only gets us to a nominal form of unity.
Here’s one illustration. Suppose we trivialize the kind of disadvantage that counts for direct discrimination. What about indirect discrimination? Either it definitionally involves non-trivial disadvantages (so we lose the explanation why it is the same kind of injustice), or it is similarly trivialized (so every disparity in outcomes suffices for indirect discrimination, which ceases to constitute any kind of injustice at all).
The best illustration of the problem, however, comes from the third distinction between kinds of disadvantage: objective and subjective disadvantage. So far, we have considered objective disadvantage, which is determined by actual effects. Subjective disadvantage, by contrast, is determined by the expected effects. A similar distinction has long been made in various ways by proponents of Disadvantage. Here’s Lippert-Rasmussen: “it is only when someone is treated in what is, or is believed by the agent to be, a disadvantageous way that we speak of discrimination against” (2006, 170).19
I don’t think differential prescription or trafficking disadvantaged Black Americans in either sense. The racial disparity in opioid prescriptions arose because, recall, “doctors believed, contrary to fact, that black people […] were more likely to become addicted […] had a higher pain threshold,” and so on. Prescribing opioids to white but not Black patients did not make the latter locally or globally worse off relative to the former according to the agents’ (false, prejudicial) beliefs about the groups. The discriminators in differential trafficking who pushed black tar heroin on recovering white addicts while avoiding Black addicts also surely did not believe this made the latter group worse off than the former. This still leaves some ways of holding that differential prescription or trafficking subjectively disadvantaged Black Americans.20 But my main concern here is not whether that claim is true, but whether it can be explanatory.
Here’s the problem. Indirect discrimination definitionally involves objective disadvantage. Whether an action constitutes disparate impact discrimination is determined only by its actual effects. The agent need not intend “to disadvantage the members of the group” or have any “other objectionable mental state” (Altman 2020, §2.2). The appeal to Disadvantage is meant to show that the central concern with direct discrimination, as with indirect discrimination, is not that agents had objectionable mental states—that they engaged in “inappropriate differentiation”—but that victims were “disadvantaged in ways that fail to treat them as the equals of others” (Moreau 2020, 6). Now suppose indirect discrimination definitionally involves objective disadvantage while direct discrimination definitionally involves subjective disadvantage. Subjective disadvantage is not about the effects of actions on the world; it is about what happens in the discriminator’s head. It is, in other words, about objectionable mental states. So, what, exactly, is the theoretical advantage of positing that indirect discrimination concerns disadvantage understood in terms of its actual effects and direct discrimination concerns disadvantage understood in terms of what the agent believes the likely effects to be? They’re still different injustices.
Proponents of Disadvantage could respond that the kind of disadvantage that counts for both direct and indirect discrimination must be subjective, not objective. That generates a more than nominal form of unity. But it is wildly inconsistent with the common understanding of the nature of disparate impact discrimination. Moreover, it converts all complaints about discrimination into complaints about objectionable mental states. It is hard to see why that should be attractive by these theorists’ lights.
Alternatively, proponents of Disadvantage could respond that the kind of disadvantage that counts for direct and indirect discrimination must be objective, not subjective. That makes it easier to defend the view that Disadvantage is explanatory. Indeed, this move may be essential for explaining the unity of discrimination and its directionality. In §IV, I said that disadvantage is asymmetric if it is relative and global, subject to a caveat. Here’s the caveat. Suppose the kind of disadvantage that counts is relative, global, and either objective or subjective. That form of disadvantage can easily be symmetric! The actual effects of x’s act could disadvantage y relative to z all-things-considered while x believed or expected the same act would disadvantage z relative to y all-things-considered. If the kind of disadvantage that counts is relative, global, and subjective, this can still be symmetric: x can believe that an act disadvantages y relative to z all-things-considered and inconsistently believe that it does the opposite. (Would it be surprising if some of the licensed physicians who believe that Black patients have higher pain thresholds also turn out to have inconsistent beliefs?) So, if Disadvantage is meant to explain the unity and directionality of disadvantage, there is significant pressure to hold that the kind of disadvantage that counts must be objective.
But as before, responding that disadvantage is objective makes it easy to see how Disadvantage can be explanatory while making it harder to see how it can be true. Young (1990, 196) was concerned about appeals to indirect discrimination—as a “broadened meaning of discrimination”—because they involve a “results test”: the legitimacy of the complaint depends on what actual effects materialize. On this response, a “results test” also applies to direct discrimination! Take a case where direct discrimination against someone expectedly disadvantages the victim, but the disadvantage happens not to materialize. (There are many such cases: see Arneson (2017), as well as Garcia (2018, 697) on the connections to ‘moral luck’.) When expected harms to victims of direct discrimination fail to materialize, surely victims do not lose their legitimate complaint! Perhaps this response is still defensible; but it at best tries to salvage the received view by saddling it with highly contentious commitments. Which should at least make us wonder why Disadvantage remains an orthodoxy.
VI. Conclusion
The opioid epidemic provides extremely rare cases where racial biases protected the population being discriminated against. Focusing on these cases brings to light a significant challenge to the received view that discrimination against someone necessarily disadvantages them. But the received view does not only say that this claim is true; it says it is explanatory. It is meant to explain the directionality and unity of discrimination. Hence, my challenge poses a dilemma. On any view you like about the kind of disadvantage that counts, the received view will turn out to be either false (as it is subject to counterexamples) or a theoretical idle wheel (as even if it is true it cannot explain the directionality and unity of discrimination).
Is the challenge insurmountable? Maybe not. Few philosophical challenges are. But my goal is much more modest than that. I want to push proponents of the received view to say more in its defense. Since the received view is widely endorsed but has never been argued for, my hope is that opening up a debate about the connection between discrimination and disadvantage can help make progress in theorizing about what discrimination is and why it matters. This can include defending the received view by taking a careful stance about the kind of disadvantage that counts. But it can also include comparing the received view to a wide range of possible alternatives.
In closing, I’ll briefly sketch how we might progress towards developing such alternatives. Recall the connection between the directionality of discrimination and directed duties: we should expect that what makes an act of discrimination against you is tied to what makes it wrong you, and what makes it wrong you is tied to what makes it wrong simpliciter. We have many plausible theories of what makes discrimination wrong simpliciter. Maybe it is because it is demeaning, or because it is disrespectful. I suggest that we can use these theories to generate promising alternatives to Disadvantage: if an act of discrimination between you and me is an act of discrimination against you rather than me, this is plausibly because you are the one who is demeaned, or disrespected (or …). Each proposal would need to be developed and defended,21 and it is premature to assess their comparable virtues. But they strike me as promising ideas, including in their prospects for explaining why differential prescription and differential trafficking constituted discrimination against Black but not white Americans.
Notes
- This paper spent a while in the cooker. A great many helpful conversations over the years have significantly shaped its content. I’d particularly like to thank Shalom Chalson, Garrett Cullity, Chris Howard, Alex Sarch, audiences at the 8th Australasian Workshop in Moral Philosophy at ANU, at the University of Adelaide, and at the Center for Eksperimentel-Filosofiske Studier af Diskrimination Conference, at Aarhus Universitet, and the anonymous reviewers and editors at Philosophers’ Imprint. ⮭
- This paper spent a while in the cooker. A great many helpful conversations over the years have significantly shaped its content. I’d particularly like to thank Shalom Chalson, Garrett Cullity, Chris Howard, Alex Sarch, audiences at the 8th Australasian Workshop in Moral Philosophy at ANU, at the University of Adelaide, and at the Center for Eksperimentel-Filosofiske Studier af Diskrimination Conference, at Aarhus Universitet, and the anonymous reviewers and editors at Philosophers’ Imprint. ⮭
- Frakt and Monkovic (2019). See also Hoffman et al. (2016), Wyatt (2013). ⮭
- Quoted in Frakt and Monkovic (2019). ⮭
- See also Scanlon (2008, 73), Simons (2016, 89), and Akhtar (2024, 37). See also Altaian (2015, 157)—which is by Altman, but was published with a typo in the name. ⮭
- The centrality of Disadvantage is reflected in the title of Loets (2024): ‘Intersectional Disadvantage’. See also Hu (2024, 24) on algorithmic racial discrimination and Byskov (2021, 116, 118) on epistemic injustice. ⮭
- More can be said here. Beeghly (2021) defends the minority view that thoughts can be discriminatory; cf. Sangiuliano (2024, 68). Lippert-Rasmussen (2013, 51) says discrimination requires a victim. ⮭
- For a representative, and especially relevant, example of this work, see James and Jordan (2018). ⮭
- And, I expect, due to the enormous education disparities in the epidemic. See Powell (2025). ⮭
- By comparison, it is at least contestable that ‘paternalistic discrimination’ involves insufficient regard for the paternalized. For that reason, Statman (2021) denies that it discriminates against the paternalized. ⮭
- It is true that in one case, the activity was illegal. But this is a red herring. One can discriminate in the course of illegal activities (Eidelson 2015, 32–33). It would be discriminatory for a casino to bar Black Americans from entry; an illegal poker game does not have carte blanche to engage in such bigotry. ⮭
- See Van Zee (2009) and Jenkin et al. (2021). ⮭
- See also, e.g., McGowan (2019, 167 n. 24). Some say Lippert-Rasmussen denies that disadvantage necessarily involves harm, but that is a misinterpretation. Berndt Rasmussen (2019, 887 fn. 32) writes: “[Lippert-Rasmussen] leaves open the possibility of a non-welfarist sense of disadvantageous treatment, although he doesn’t develop this idea further.” But the cited passage—from Lippert-Rasmussen (2013, 18)—starts with “Disadvantageous treatment should be distinguished from treatment that causes harm” and ends with “If actual effects do not (completely) determine whether a certain treatment is disadvantageous, what then does?” It does not distinguish disadvantage from harm, but from causing harm as an actual effect. The point is that the kind of disadvantage that counts may be subjective, not objective (see §V below). Thanks to Lippert-Rasmussen for confirming this interpretation (p.c.). ⮭
- An anonymous reviewer helpfully suggests a response, drawing on Hellman’s view that “it is capacity that matters”, in the sense that an “action, if known, would … express[] denigration” (2017, 105, emphasis added). But here the source of that capacity would not be the agents’ actions. If we knew a doctor did not prescribe a patient with back pain opioids, that would not itself be demeaning; the same holds if we knew that drug dealers left Dayton. In other words, these acts, on their own, do not have conventionally derogatory social meanings towards any racial group. The risk of stigma is, I think, only generated by the possibility of knowing the agents’ reasons for acting. What if we instead said that an act risks causing stigma when it is the case that the agents’ actions and reasons, if known, would express denigration? This is a more promising proposal. But I think it is in tension with Hellman’s view that when actions are “independently morally permissible” but for the agents’ private “bad reasons” for acting, this does not “render the [actions] impermissible” (2008, 166). See further, Cullity and Wodak (2025). ⮭
- Sangiuliano (2024, 65, 70) is the exception, holding that disadvantage can be relative or absolute. ⮭
- E.g.: “being discriminated against must be comparatively disadvantageous to a person in some specific respect, but being comparatively disadvantaged in that respect need not be globally comparatively disadvantageous to him (and may even be globally comparatively advantageous …)” (2015, 32). ⮭
- Eidelson (2015, 35) holds that it is true that some cases of discrimination are symmetric: “the exclusion of African Americans from jury service in the United States” was discrimination against white Americans and against Black Americans; however, Eidelson tries to soften the blow by adding that “to say” the latter is “pragmatically more appropriate”. This seems to posit that saying the former is true but infelicitous. But why is it infelicitous or pragmatically inappropriate? Eidelson does not identify the source of the infelicity, or appeal to any standard test for infelicity, to support this position. One common test is cancellability. For example, if the source of the infelicity were a violation of the maxim of quantity, we could cancel it with: “The exclusion of Black Americans from jury duty was discrimination against both white and Black Americans, but it was worse discrimination against Black Americans.” That sentence still seems false. Eidelson’s response also does not address the point about wronging, below. ⮭
- Here’s a potential cause for doubt. Much of the literature on the demographics of the opioid epidemic succumbs to a Black-white racial binary. (This is one of several qualms I have about the common framing, which I have set aside: see Wodak (2022), and Singh and Wodak (2023).) But while opioid mortality increased dramatically among white Americans, it did not increase among Asian-Americans and Pacific Islanders (“AAPIs”). What explains this disparity? Perhaps, an anonymous reviewer suggests, it was due to some “prejudice” about “white Americans.” If this were so, such anti-white causes plausibly give white Americans a legitimate complaint about being victims of discrimination. However, I can find no evidence that prejudices about white Americans played such a causal role. The white–AAPI racial disparity is understudied, but the most comprehensive overview of all racial disparities in the epidemic only explicitly attributes it to differences in educational attainment and regulatory environment (rather than prejudice): see Powell (2025, §4.1.1). ⮭
- Cf. Lippert-Rasmussen (2013, 18, 157–8). Others gloss subjective disadvantage differently. Statman (2021, 315) appeals to “intended effects” but that’s too narrow. Dealers may expect but not intend overdoses. ⮭
- One obvious response is that instead of appealing to the discriminators’ beliefs, we should appeal to their evidence. This move filters out doctors’ false, unjustified beliefs about how their Black patients differed from their white patients. But it also filters out their unjustified, negligent beliefs about the relative efficacy of opioids for treating conditions such as chronic back pain. Such beliefs were clearly contrary to the evidence at the time (see Leung et al. (2017) and Pappin et al. (2022)), including for the high-volume physicians that Van Zee describes as “the least discriminate prescribers” (2009, 222). When such physicians prescribed opioids to white patients with back pain, this was not supported by good evidence; so, when they discriminated against similar Black patients by not prescribing them opioids, this was not expectedly worse for their Black patients given their evidence. ⮭
- It is worth noting that this approach can be neutral on the contested issue of whether we should prefer a moralized account of discrimination. It only requires a moralized account of discrimination against. ⮭
References
Akhtar, Sahar (2024). Immigration and Discrimination: (Un)welcoming Others. Oxford University Press.
Alexander, Monica J., Mathew V. Kiang, and Magali Barbieri (2018). Trends in Black and White opioid mortality in the United States, 1979–2015. Epidemiology, 29(5), 707–715. doi: http://doi.org/10.1097/EDE.0000000000000858.
Altaian, Andrew (2015). Discrimination debated. Jurisprudence, 6(1), 156–168. doi: http://doi.org/10.1080/20403313.2015.1018029.
Altman, Andrew (2020). Discrimination. In The Stanford encyclopedia of philosophy, edited by Edward N. Zalta. Metaphysics Research Lab, Stanford University. https://plato.stanford.edu/archives/win2020/entries/discrimination/.
Arneson, Richard (2006). What is wrongful discrimination?. San Diego Law Review, 43, 775–808.
Arneson, Richard (2017). Discrimination and harm. In The Routledge Handbook of the Ethics of Discrimination, edited by Kasper Lippert-Rasmussen (151–163). Routledge.
Beeghly, Erin (2021). Stereotyping as discrimination: Why thoughts can be discriminatory. Social Epistemology, 35(6), 547–563. doi: http://doi.org/10.1080/02691728.2021.1930274.
Berndt Rasmussen, Katharina (2019). Harm and discrimination. Ethical Theory and Moral Practice, 22(4), 873–891. doi: http://doi.org/10.1007/s10677-018-9908-4.
Byskov, Morten Fibieger (2021). What makes epistemic injustice an ’injustice’?. Journal of Social Philosophy, 52(1), 116–133. doi: http://doi.org/10.1111/josp.12348.
Cullity, Garrett, and Daniel Wodak (2025). But thinking makes it so: How bad attitudes can make discriminatory actions wrong. Philosophy and Phenomenological Research, 111(3), 1081–1104. doi: http://doi.org/10.1111/phpr.70058.
Eidelson, Benjamin (2015). Discrimination and Disrespect. Oxford University Press.
Foran, Michael (2022). Grounding unlawful discrimination. Legal Theory, 28(1), 3–34. doi: http://doi.org/10.1017/S1352325221000264.
Frakt, Austin, and Toni Monkovic (2019). A ’rare case where racial biases’ protected African-Americans. The New York Times, November 25.
Furr-Holden, Debra, Adam Milam, Ling Wang, and Richard Sadler (2021). African Americans now outpace whites in opioid-involved overdose deaths: A comparison of temporal trends from 1999 to 2018. Addiction, 116(3), 677–683. doi: http://doi.org/10.1111/add.15233.
Garcia, J. L. A. (2018). Wrongful racial discrimination in moral analysis: Some recent accounts, an alternative conception, and attempts to extend theoretical models. Georgetown Journal of Law & Public Policy, 16, 697–730.
Gardner, John (1996). Discrimination as injustice. Oxford Journal of Legal Studies, 16(3), 353–367. doi: http://doi.org/10.1093/ojls/16.3.353.
Gardner, John (2018). Discrimination: The good, the bad, and the wrongful. Proceedings of the Aristotelian Society, 118, 55–81. doi: http://doi.org/10.1093/arisoc/aoy002.
Hellman, Deborah (2008). When Is Discrimination Wrong?. Harvard University Press.
Hellman, Deborah (2016). Two concepts of discrimination. Virginia Law Review, 102, 895–952.
Hellman, Deborah (2017). Discrimination and social meaning. In The Routledge Handbook of the Ethics of Discrimination, edited by Kasper Lippert-Rasmussen (97–107). Routledge.
Hoffman, Kelly M., Sophie Trawalter, Jordan R. Axt, and M. Norman Oliver (2016). Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between blacks and whites. Proceedings of the National Academy of Sciences, 113(16), 4296–4301. doi: http://doi.org/10.1073/pnas.1516047113.
Horta, Oscar (2015). Does discrimination require disadvantage?. Moral Philosophy and Politics, 2(2), 277–297. doi: http://doi.org/10.1515/mopp-2014-0032.
Hu, Lily (2024). What is 'race' in algorithmic discrimination on the basis of race?. Journal of Moral Philosophy, 21(1-2), 1–26. doi: http://doi.org/10.1163/17455243-20234369.
Ishida, Shu (2021). What makes discrimination morally wrong? A harm-based view reconsidered. Theoria, 87(2), 483–499. doi: http://doi.org/10.1111/theo.12285.
James, Keturah, and Ayana Jordan (2018). The opioid crisis in Black communities. Journal of Law, Medicine and Ethics, 46(2), 404–421. doi: http://doi.org/10.1177/1073110518782949.
Jenkin, Deanne E., Justine M. Naylor, Joseph Descallar, and Ian A. Harris (2021). Effectiveness of oxycodone hydrochloride (strong opioid) vs combination acetaminophen and codeine (mild opioid) for subacute pain after fractures managed surgically. JAMA Network Open, 4(11), e2134988. doi: http://doi.org/10.1001/jamanetworkopen.2021.34988.
Khaitan, Tarunabh (2015). A Theory of Discrimination Law. Oxford University Press.
Kolodny, Niko (2023). The Pecking Order. Harvard University Press.
Leung, Pamela T. M., Erin M. Macdonald, Matthew B. Stanbrook, Irfan A. Dhalla, and David N. Juurlink (2017). A 1980 letter on the risk of opioid addiction. New England Journal of Medicine, 376(22), 2194–2195. doi: http://doi.org/10.1056/NEJMc1700150.
Lippert-Rasmussen, Kasper (2006). The badness of discrimination. Ethical Theory and Moral Practice, 9(2), 167–185. doi: http://doi.org/10.1007/s10677-006-9014-x.
Lippert-Rasmussen, Kasper (2013). Born Free and Equal?. Oxford University Press.
Loets, Annina (2024). Intersectional disadvantage. Australasian Journal of Philosophy, 102(4), 857–878. doi: http://doi.org/10.1080/00048402.2024.2329972.
McGowan, Mary Kate (2019). Just Words: On Speech and Hidden Harm. Oxford University Press.
Midtgaard, Søren Flinch, and Viki Møller Lyngby Pedersen (2024). Paternalistic discrimination. Law and Philosophy, 43, 417–446. doi: http://doi.org/10.1007/s10982-024-09521-4.
Moreau, Sophia (2010). What is discrimination?. Philosophy & Public Affairs, 38(2), 143–177. doi: http://doi.org/10.1111/j.1088-4963.2010.01181.x.
Moreau, Sophia (2020). Faces of Inequality: A Theory of Wrongful Discrimination. Oxford University Press.
Murakawa, Naomi (2014). The first civil right: How liberals built prison America. Oxford University Press.
Pappin, Jessie, Itai Bavli, and Matthew Herder (2022). On what basis did Health Canada approve OxyContin in 1996? A retrospective analysis of regulatory data. Clinical Trials, 19(5), 584–590. doi: http://doi.org/10.1177/17407745221108436.
Powell, David (2025). Understanding the demographics of the opioid overdose death crisis. Journal of Population Economics, 38, 54. doi: http://doi.org/10.1007/s00148-025-01108-0.
Quinones, Sam (2015). Dreamland: The true tale of America’s opiate epidemic. Bloomsbury.
Radcliffe Richards, Janet (1985). Discrimination. Aristotelian Society Supplementary Volume, 59(1), 53–82. doi: http://doi.org/10.1093/aristoteliansupp/59.1.53.
Räsänen, Joona, and Kasper Lippert-Rasmussen (2024). Personality discrimination and the wrongness of hiring based on extraversion. Journal of Business Ethics, 195(3), 681–694. doi: http://doi.org/10.1007/s10551-024-05643-w.
Sangiuliano, Anthony (2024). Harmless discrimination, wrongs, and rules. Law and Philosophy, 43(1), 61–88. doi: http://doi.org/10.1007/s10982-023-09479-y.
Scanlon, Thomas (2008). Moral Dimensions: Permissibility, Meaning, Blame. Harvard University Press.
Segall, Shlomi (2012). What’s so bad about discrimination?. Utilitas, 24(1), 82–100. doi: http://doi.org/10.1017/S0953820811000380.
Simons, Kenneth (2016). Discrimination is a comparative injustice: A reply to hellman. Virginia Law Review Online, 102, 85–100.
Singh, Keshav, and Daniel Wodak (2023). Does race best explain racial discrimination?. Philosophers’ Imprint, 23(24), 1–22. doi: http://doi.org/10.3998/phimp.2463.
Statman, Daniel (2021). Why benefitting a person cannot constitute a form of discrimination. Ratio Juris, 34(4), 315–330. doi: http://doi.org/10.1111/raju.12321.
Thomsen, Frej Klem (2015). Stealing bread and sleeping beneath bridges—indirect discrimination as disadvantageous equal treatment. Moral Philosophy and Politics, 2(2), 299–327. doi: http://doi.org/10.1515/mopp-2014-0027.
Van Zee, Art (2009). The promotion and marketing of OxyContin: Commercial triumph, public health tragedy. American Journal of Public Health, 99(2), 221–227. doi: http://doi.org/10.2105/AJPH.2007.131714.
Wasserman, David (1998). Discrimination, concept of. In Encyclopedia of Applied Ethics, edited by Ruth Chadwick (805–814). Academic Press.
Wodak, Daniel (2022). Of witches and white folks. Philosophy and Phenomenological Research, 104(3), 587–605. doi: http://doi.org/10.1111/phpr.12799.
Wyatt, Ronald (2013). Pain and ethnicity. AMA Journal of Ethics, 15(5), 449–454. doi: http://doi.org/10.1001/virtualmentor.2013.15.5.pfor1-1305.
Young, Iris Marion (1990). Justice and the Politics of Difference. Princeton University Press.