1. A brief survey of epistemic injustice and epistemic harm

Miranda Fricker’s 2007 book Epistemic Injustice has become the locus classicus in the epistemic injustice literature. There, she posits two categories of epistemic injustice: testimonial injustice and hermeneutical injustice.Footnote ¹ The former occurs when “prejudice causes a hearer to give a deflated level of credibility to a speaker’s word” (Fricker Reference Fricker2007, 1). Fricker offers as a paradigm case the trial of Tom Robinson in Harper Lee’s To Kill a Mockingbird. The all-white jury predictably disbelieves Tom Robinson’s testimony, and convicts him; their racist prejudice leads them to unjustly deflate his credibility. He is “…unable to convey to them the knowledge he has…” (28). Critical to Fricker’s account is that it is an identity prejudice that causes the jury to disbelieve him, namely, their racist prejudice against black people. Fricker distinguishes between incidental credibility deficits that are the results of, for example, mere bad epistemic luck and those that are grounded in larger systems of social injustice.Footnote ² These are those features by which prejudices “Track the subject through different dimensions of social activity” (27). Since Fricker’s original publication, numerous modifications to the concept have been proposed, however the central case of testimonial injustice remains a wrongful, lowered assessment of another’s credibility as a function of prejudice: one wrongfully treats them as less credible than one should.

Hermeneutical injustice is “the injustice of having some significant area of one’s social experience obscured from collective understanding owing to a structural identity prejudice in the collective hermeneutical resource” (155). As an example, Fricker cites Carmita Wood’s account of the origin of the term sexual harassment. Prior to this coinage, despite sexual harassment being an overwhelmingly common experience, there was no name for that experience, diminishing women’s ability to understand a distinctive social experience. This cognitive disadvantage constituted a hermeneutical injustice. Note that while harassees and harassers alike had the same cognitive impairment due to this lacuna, it constitutes a hermeneutical injustice only to the harassee because for the harasser, this disablement is not a significant disadvantage (151). Not all instances of absent conceptual resources constitute a hermeneutical injustice; the lack of those resources must be wrongful. Prior to the development of germ theory, many people lacked the cognitive resources to make sense of their lived experience vis-à-vis disease, but this was not wrongful, because it was due to technological and conceptual limitations, rather than structural identity prejudice.

In both kinds of injustice, the subject is stated to be the victim of epistemic harm. However, there is ambiguity between epistemic harming and epistemic wronging in Fricker’s text.Footnote ³ She states,

Here, the harm and the wrong are clearly conflated—the harm is the wrong—the distinction simply isn’t made. However, elsewhere in Fricker’s work, as well as in the broader literature, it’s clear that what’s at play is something more like a grounding relationship. The reason that epistemic injustices are epistemic injustices is because of the harms involved. The existence of epistemic harm is not a sufficient condition for epistemic wronging (Fricker and others give cases where epistemic harm exists without epistemic wrong), but it is both necessary and explanatory.

Fricker believes however that we must also sort out from the “primary” epistemic harms from the “secondary” harms: harms extrinsic to the primary injustice “in that they are caused by it rather than being a proper part of it” (44). They are harms that result from instances of epistemic injustice. She argues that we can divide these downstream harms into two classes, practical and epistemic. The former are the deleterious effects of being disbelieved: one is falsely convicted of a crime, one loses out on a job opportunity, etc. The latter consist of those effects that epistemic injustice may have on one’s own epistemic cognition. One may

Contra Fricker, some accounts of epistemic injustice ground the wrong thereof in these secondary epistemic harms—epistemic injustice is wrong because of the injurious downstream effects that it has on peoples’ lives. In “What’s Wrong with Epistemic Injustice” (Reference Congdon, Kidd, Medina and Pohlhaus2017), Matthew Congdon observes that this Utilitarian sort of analysis “leaves something out of the moral picture,” following similar observations from, among others, Fricker herself (Reference Fricker2007, 43-46); Haslanger (Reference Haslanger2014); and Wanderer (Reference Wanderer2012). Epistemic injustice is wrong even when these kinds of downstream harms don’t obtain. Wanderer provides a useful case wherein it is the perpetrator, rather than the victim, who bears the practical consequences of unjust disbelief. Suppose someone chooses not to heed a lifeguard’s warning concerning a shark sighting, because of a negative identity prejudice, and is then killed by a shark (149-153). The lifeguard is the victim of epistemic injustice; the perpetrator is the one that suffers the downstream consequences.

I’ll not dispute the existence or significance of secondary epistemic harms. That being disbelieved often carries with it these secondary harms is inarguable, as is that these secondary harms can be disastrous, such as the murder of innocents by the state when wrongfully disbelieved.Footnote ⁴

Let us move on to the primary harm of epistemic injustice. Fricker characterizes the primary harm as

She defends this on the grounds that “The capacity to give knowledge to others is one side of that many-sided capacity so significant in human beings: namely, the capacity for reason.”

She claims this harm has a further symbolic power that “adds a layer of harm of its own: the epistemic wrong bears a social meaning to the effect that the subject is less than fully human” (Fricker Reference Fricker2007, 44, emphasis original). She adds, “When one is wrongfully mistrusted… one is dishonored—this would not be an inappropriate term in connection with the primary harm of testimonial injustice” (46, emphasis original).

This primary harm, according to Fricker, is that “testimonial injustice demotes the speaker from informant to source of information, from subject to object. This reveals the intrinsic harm of testimonial injustice as epistemic objectification” (133). While not affirming, to use her language, the whole of Kant’s “considerable philosophical apparatus,” she characterizes this as a species of treating the person as a mere means (134). Gaile Pohlhaus Jr. usefully diagnoses the badness of this harm:

Here we can see the grounding relationship between the wrongness of epistemic injustice and epistemic harm. Epistemic injustice is wrong in virtue of the epistemic harm involved. This isn’t to say that epistemic harm is sufficient for epistemic injustice. Fricker writes,

Thus, it is only when epistemic harming is the result of a negative identity prejudice that it also results in epistemic wronging. This establishes this structure as a species of the larger genus, wherein harming is permissible only with adequate justification. To use Fricker’s example, we are justified in harming Matilda when we epistemically objectify her, because she has given us good reason to do so, but not Tom Robinson, where only prejudice is working as the justifier (42).Footnote ⁸

Objections have been raised against Fricker’s account of epistemic harm. Congdon (Reference Congdon, Kidd, Medina and Pohlhaus2017) offers the case of Mr. B, a Muslim Bosnian citizen being held without charge in Guantanamo for seven years – his captors refuse to believe, on the basis of his being Muslim, that he does not have any knowledge concerning a bomb plot. However, they do not objectify him in the relevant way. Rather, they rely on his being a reliable informant, respecting his subjectivity. Gaile Pohlhaus Jr. Reference Pohlhaus2014) observes that this recognition is a critical part of epistemic injustice. She writes,

With this in mind, Pohlhaus Jr. introduces her own account of epistemic harm. She contends that while epistemic harm treats the victim as less than a full epistemic subject, it doesn’t wholly treat them as an object. She suggests that instead we draw from de Beauvoir, and ground it in a subject/other relationship: the harm of epistemic injustice is that it diminishes the subjectivity of the victim such that they are treated as a full subject only when doing so accords with the subjectivity of the dominant knower. When it doesn’t, their experience is disregarded, and then they become, to use Fricker’s language, a mere source of information. She labels this, following Cahill (Reference Cahill2012), as being derivatized, stating, “Under this model, the primary harm of testimonial injustice is defined as: being relegated to the role of epistemic other, being treated as though the range of one’s subject capacities is merely derivative of another’s.” (Pohlhaus Jr. Reference Pohlhaus2014, 107).

Other accounts of epistemic harm exist. Shannon Sullivan (Reference Sullivan, Kidd, Medina and Pohlhaus2017) offers an account using Deweyan pragmatist epistemology. According to Dewey, rather than knowledge being a matter of objectively representing the world, beliefs are true or false as a function of their enabling us to pragmatically transact in the world. Borrowing an example from Sullivan, knowing the width of a table is a function of our ability to use that knowledge to do certain actions. If one wants to know if certain items will fit on it, it’s true that the desk is thirty inches wide. However, if a carpenter wishes to construct an alcove in which to fit the desk, instead one might measure it to be 29.9 inches. For the function of carpentry, it’s true that the desk is 29.9 inches wide (208-209).

This framework, Sullivan thinks, makes better sense of epistemic harm than Fricker’s representationalist epistemology, stating that, “working with a transactional epistemology is more likely to improve human lives by eliminating the harms of epistemic injustice.” She characterizes epistemic harm in several ways, stating that, “the speaker isn’t allowed to epistemologically transact with the world in ways that enable her own, as well as others’ flourishing” and that “the harm is that being dismissed as crazy impacts the perceived reasonableness of the entire culture or community in question” For Sullivan, it is a multifaceted harm. She adds, “epistemic injustice conceived as a failure of transactional suffering can happen as a deprivation… but even more importantly it can occur as a production, in which a person is produced as a kind of knower required to epistemologically engage the world in ways that undercut her” (210).

2. Underinclusivity

Fricker’s objectification account allows epistemic harm to be decoupled from epistemic wronging—there are cases of epistemic harming which are justified. But can epistemic wronging exist without epistemic harm? Dunne and Kotsonis (Reference Dunne and Kotsonis2024) aim to demonstrate the necessity of distinguishing epistemic wronging from epistemic harming, attempting to demonstrate that epistemic wronging can exist without epistemic harming, and vice-versa. They conclude that while wronging can exist without harming, harming cannot exist without wronging; epistemic harm is a sufficient condition for epistemic wrong.

We are asked to consider the following case:

The authors answer in the affirmative: “On Fricker’s conception of epistemic injustice, the answer is yes. He is wronged in his capacity as a knower; he is wronged as a giver of knowledge.” They argue that he is not harmed per Fricker’s (or Pohlhaus’s) conception of epistemic harm, saying, “There is no widely held identity prejudice against white, heterosexual, middle-aged male doctors.Footnote ¹⁰ The doctor is not a marginalized speaker who “is likely to be discredited and/or silenced when the information she contributes moves beyond the scope of the world experienced from dominant subject positions.”

But Fricker’s understanding of epistemic harm doesn’t require prejudice. As discussed in the foregoing, Fricker’s account of epistemic harm is epistemic objectification, “…this reveals the intrinsic harm of testimonial injustice as epistemic objectification: when a hearer undermines a speaker in her capacity as a giver of knowledge, the speaker is epistemically objectified.” While this objectification may be the product of prejudice, it need not be. Fricker gives as example the case of Matilda, who is disbelieved to due a long history of dreadful lies. (Fricker Reference Fricker2007, 42) She is epistemically objectified, but justifiably so. Fricker tells us, “If a hearer cannot be blamed for the grounds of her flawed credibility judgement… then she cannot be blamed for the harm that may result.” Her framework makes sense of this case by stating that the doctor in the case may be harmed (because epistemically objectified) but not wronged (because not a victim of identity prejudice). Nor does this threaten Pohlhaus Jr.’s framework; her overall structure of the relationship of harming and wronging is consistent with Fricker’s, and nothing turns on which of the two accounts we accept. Thus, Dunne and Kotsonis fail to give a case of epistemic wronging without epistemic harming.

From here, they move to the question of epistemic harming without epistemic wronging. After considering some possible cases they state “Perhaps there are cases of epistemic injustice that we have not considered where an agent is harmed qua knower without been wronged [sic]. We do not think that such a case exists. This is because Fricker (Reference Fricker2007) frames the concept of epistemic injustice in a manner that requires primary epistemic wrongs to follow from epistemic harms” (11). However, recall that Fricker explicitly offers a number of cases where harming does exist without wronging. Matilda’s history of lying justifies her hearers, both ethically and epistemically, in reducing their credibility of her, even to the extent of treating her as an epistemic object. Dunne and Kotsonis recognize that Fricker allows for certain exceptions (12), but don’t adequately recognize these counterexamples, stating instead that “to be objectified, more often than not requires that one is wronged… More often than not, this view of objectification holds true even if, for some non-epistemic reason, we believe that we are justified in treating others in an epistemically unjust way.” This however, is a petitio principii. The question at hand is whether or not there are epistemically just ways of epistemically harming, not whether we are justified in treating others in an epistemically unjust way. According to Fricker, there are cases where we may justifiably objectify people epistemically, and it is here that we see instances of epistemic harming without epistemic wronging.

Of course, it certainly seems like the doctor has been wronged; he is distrusted for morally suspect reasons, and that, perhaps, he hasn’t been harmed, precisely because “the doctor has not experienced any adverse impact on his epistemic agency as a knower or giver of knowledge via the wrong he suffered from the patient” (Dunne and Kotsonis Reference Dunne and Kotsonis2024, 8). To take this strategy though is to simply deny Fricker’s account: Fricker defines epistemic harm as epistemic objectification, and this would be to say that not all instances of epistemic objectification are epistemic harms. This may be true, but once we abandon Fricker’s view, we cannot use it alone to demonstrate underinclusivity. To show that epistemic wrong can exist without epistemic harm outside of Fricker’s framework, one must show that no account of epistemic harm can ground all cases of epistemic wronging.

Dunne and Kotsonis fail to adequately decouple epistemic wrong from epistemic harm. Can there exist epistemic wrong without epistemic harm? In the remainder of this section, I will argue that it can. To develop their case, Dunne and Kotsonis draw on the literature on credibility excess, especially the work of Emmalon Davis. While I believe that their efforts fail, I believe that they correctly identified wrongful credibility excess as a plausible locus for decoupling harming and wronging.

Early in her work, Fricker dismisses the possibility of epistemic injustice in cases of credibility excess. While it is the case that sometimes people are afforded more credibility than they deserve, she denies that this is an instance of epistemic injustice:

This has been challenged by a number of authors, including Medina (Reference Medina2011, Reference Medina2012) and Coady (Reference Coady, Kidd, Medina and Pohlhaus2017), who illuminate the reciprocal relationship between credibility deficits and excess: Sometimes, someone’s being the victim of a credibility deficit implies that someone else is undeservedly benefitting from a credibility excess. That Tom Robinson is accorded less credibility implies that his accuser Bob Ewell is afforded unjust epistemic esteem. Coady and Medina both argue that this constitutes its own form of epistemic injustice; Coady illustrates it in usefully distributive terms—when we must choose who to believe between two conflicting claimants, a credibility excess translates into a credibility deficit insofar as credibility is a scarce resource.

This kind of relative, distributive excess is more intuitive than the possibility of an absolute excess. Typically, we find that simply giving people more of a good isn’t the sort of thing that can wrong them, unless it involves an unjust deprivation on the part of another. Furthermore, in these kinds of cases, while it’s clear that injustice has occurred, it’s not clear that the person who is benefited is themselves wronged. We don’t necessarily think that Bob Ewell is wronged when the jury affords him a credibility excess. Rather, it is unjust insofar as it wrongs Tom Robinson.

But there are cases where absolute, nonrelative benefits can wrong a person. Suppose that a computer science professor affords all the women in his class additional time on an exam, assuming women are more likely to struggle with the material. Suppose too that one of the women really were struggling, and succeeded because of the extension. She is nevertheless wronged by the professor. She is benefitted for the wrong reason – because of a negative identity prejudice held by the professor.

Taking wrongful credibility excess as a legitimate form of epistemic injustice, we should not just consider the cases where dominant knowers are given unwarranted epistemic esteem because of the negative identity prejudices held about nondominant knowers, but also the case where a nondominant knower is afforded a credibility excess because of a negative identity prejudice. This phenomenon has been discussed in depth by (Davis Reference Davis2016). I will offer two such examples, the latter adapted from Davis.Footnote ¹¹

Suppose that Aubrey is seeking advice concerning polyamorous lifestyles. They ask their coworker Shannon for advice, and, without critical reflection, accept what Shannon has to say. However, suppose that the only reason that they afford this credibility to Shannon is not because they have evidence that Shannon is knowledgable about polyamory, but rather because Shannon is queer, and Aubrey holds a negative stereotype of queers as promiscuous.Footnote ¹² This is intuitively wrong for the same reasons that affording a credibility deficit to Shannon would be if they were to opine about sports, believing queer folk wouldn’t be knowledgeable about sports.

Alternatively, we might consider the perception of Asian-Americans as “model minorities.” This is clearly a wrongful, racist stereotype and if it should afford an Asian person a credibility excess, this is plausibly an instance of epistemic injustice. Suppose that Jordan is struggling academically, and so turns to their friend Rin, a Japanese-American student, for advice on more effective studying. They do so not on evidence that Rin is a more successful student than they, but because Rin is Asian, and they hold the racist stereotype that, as such, Rin is a good student. They wrong Rin before even listening for their answer. Their asking the question itself constitutes affording Rin an unjust credibility excess.Footnote ¹³

It’s clear Aubrey has wronged Shannon and Jordan has wronged Rin. But have they been harmed? According to the three theories discussed, I don’t think so. For Fricker, this would mean Aubrey has downgraded Shannon from an informant to a mere source of information, not treating them as an epistemic subject.Footnote ¹⁴ But clearly, this is not so. We can apply a similar criticism here as was applied by Congdon and Pohlhaus Jr.– Aubrey treats Shannon as a subject, but in a wrongful manner.

Given this inadequacy in Fricker’s account, Davis (Reference Davis2016) instead endorses Pohlhaus Jr.’s formulation,Footnote ¹⁵ arguing that

However, Pohlhaus Jrs.’ account doesn’t fare any better here. While it’s possible Aubrey might derivatize Shannon in this instance, it’s not clear that this must be the case. Shannon is certainly being othered, but not in a way that necessarily derivatizes them. Rather, it privileges their subjective experience. Their experience is wrongfully and unjustly considered superior to the hearer’s. Now, it is plausible that should the victim challenge them on this, that they will then derivatize them. If Shannon challenges Aubrey, Aubrey might well then disbelieve that Shannon is not familiar with polyamory, because, to use Pohlhaus Jr. Reference Pohlhaus2014) language, “…a dominantly situated subject, is not disposed to seeing the world from the eyes of nondominantly situated subjects in ways that displace the centrality of his own subjective experiencing of the world” (109).

Recognizing this problem, Davis modifies Pohlhaus Jr.’s account, noting that in these sorts of cases, “it is only because a marginalized speaker possesses what dominant others perceive to be socially and epistemically distinct experiences that she is acknowledged at all.” She offers a modified account of the primary harm of epistemic injustice, at least in cases of credibility excess:

This makes more sense of the phenomenon than does Pohlhaus Jr.’s original account, but I think it is still inadequate. To show why, I’ll first show that de dicto inclusion is not necessarily problematic. As an ethicist, I’m approached for my views on ethical conundra in various discursive contexts; it’s assumed that I can provide insight into the problem at hand in a way that a non-ethicist might not be able to. In such circumstances, I am granted de dicto, rather than de re, participation in the conversation. I am not asked to give my view on the conundrum in my capacity as me, but rather in my capacity as a relevant expert. Echoing Davis, any ethicist would do.

This is saliently different than the case of compulsory representation that Davis discusses; it’s reasonable to expect that an ethicist provide insight into ethical conundra, just as it is to ask a chemist to provide information about chemistry, or a composer about music. Given norms of expert deference, it’s critical that relevant experts be solicited in this de dicto fashion, especially in contentious and evaulative fields like ethics. Calling upon experts to perform acts of epistemic labor isn’t problematic, because that is the expert’s social epistemic role. That some invitations to participate de dicto are pernicious does not mean that all are.

So, we must determine if the kind of de dicto participation that Davis proposes is pernicious, not because they are de dicto, but for some further reason. Key to Davis’s account is the idea that speakers are invited to participate epistemically only insofar as they speak with a voice of distinction: it is not just that they are invited to speak as a representative of that group, but they are invited to speak only as a representative of that group, and excluded from the conversation otherwise. The marginalization occurs because the speaker is not invited to speak as a full subject.

But we should recognize that in certain epistemic circumstances, even this truncating of subjectivity is warranted. Consider an ethicist being asked to give an assessment of a situation, not in a casual environment, but in their capacity as a member of an ethics board for a hospital. Here, it is both epistemically and morally critical that they give the best answer that they can given their training, not merely their own view. This limiting of one’s subjectivity is a necessary function of their social epistemic role.

The case of the medical ethicist is useful because it helps to clarify that limited subjectivity in one discursive environment does not translate, necessarily, into a broader marginalization. One can be a full participant in an epistemic system generally, and yet be denied full subjectivity within some contexts. The ethicist asked to speak as an ethicist may be asked to speak in a de dicto fashion, even though in the larger epistemic context they have been accorded de re status.

Ergo, one must go further than simply showing that this limited subjectivity exists in cases of wrongful credibility excess, but that it is the bad sort of limited subjectivity. Beyond the requirements Davis sets out, it must be the case that it is not reasonable to ask the person to speak in a de dicto capacity, and that they are not more broadly granted de re status.

In most cases, I think that both conditions will be met, given the general structure of oppression. Nevertheless, Davis’s account is underinclusive too. The case of Aubrey and Shannon usefully demonstrates why. Aubrey does ask that Shannon speak in a de dicto capacity. However, they do not ask them to speak as a compulsory representative of the queer community. Rather, by dint of their assumptions of queer promiscuity, they ask that they speak in their assumed capacity as an expert about non-monogamy.

Suppose that Shannon rebukes Aubrey, informing them that they are monogamous, and don’t know anything about polyamory. Further suppose that Aubrey is unusually receptive to evidence, and upon being rebuked by Shannon, actually change their view, appropriately lowering their assessment of Shannon’s credibility vis-a-vis polyamory. Aubrey might then ask what Shannon thinks about polyamorous lifestyles, inviting them to contribute to the discussion in a de re manner; they are not asked to speak from a position of distinction, because they are no longer presumed an expert.

Aubrey’s asking Shannon to speak from a position of distinction came from this presumption of expertise, a presumption which came from a problematic othering of Shannon prior to this discursive act. Critically, Aubrey’s solicitation, while only inviting Shannon to speak de dicto, was not in their capacity as a queer person, but in their capacity as someone familiar with polyamorous lifestyles. It is not just the case that any queer would do, but that anyone with that presumed expertise would do. The moral and epistemic failing was in the presumption that Shannon had the relevant expertise.

Admittedly, Aubrey, upon being rebuked, might not be receptive; they might insist that Shannon must know something and continue to press them, denying them de re status in the epistemic community, and demanding that they speak in the presumed authoritative capacity. This would be a wrong and would meet the criteria for harm posited by Pohlhaus Jr. and Davis. However, this would be a separate epistemic injustice. Aubrey’s assessment of Shannon’s credibility has flipped from an excess to a deficit. This isn’t surprising, it’s precisely this sort of inconsistency that Pohlhaus Jr.’s account is better set up to explain than Fricker’s. But this second epistemic injustice does not adequately explain the first. Aubrey still wrongs Shannon when they first overinflate Shannon’s credibility.

What about Sullivan’s view? It’s at least plausible that there is no harm here. It’s clear Shannon is not being required to epistemologically transact in the world in ways that stymie their own or others flourishing. Since this is not a dismissal, it doesn’t seem that this wrong negatively impacts the reasonableness of the epistemic community as a whole. Perhaps, though, Shannon is produced as a kind of knower required to epistemologically engage in the world in ways that undercut them. However, let us add a single stipulation to the case: suppose Shannon actually is polyamorous and does have a great deal of knowledge about polyamory – Aubrey just doesn’t have that information: the epistemic wrong remains in Aubrey’s problematic assumption. Though Shannon is called upon to act as an expert, it doesn’t seem like they are undercut by so doing.

But let us suppose that we accept that Shannon’s treatment here does cause them to be undercut. Sullivan’s account will be underinclusive still. Sullivan’s account, unlike, those of Fricker and Pohlhaus Jr., is subject to an experience requirement.Footnote ¹⁷ It requires that a victim of epistemic harm be phenomenally affected by that harm. Davis’s view is similarly constrained. Suppose Aubrey never actually asks Shannon to speak. Rather, in casual conversation, Shannon opines on polyamory, and Aubrey assesses them as more credible on the subject than they otherwise would have, because of Shannon’s queer identity.

Here, Shannon was already participating in a de re capacity, but because of queerphobic assumptions regarding promiscuity, Aubrey accords them a greater degree of credibility that they should; Shannon may never be aware of this. Let us further suppose Aubrey never actually asks Shannon, nor hears Shannon speak to the point; they merely think that if they did ask Shannon, they would get valuable information, but never broach the subject. The mere assumption is sufficient for epistemic injustice, even though it has no experiential effect on Shannon.

Neither of the other accounts have such a requirement. According to either, someone can be epistemically harmed without being aware of it. Someone may objectify or derivatize me without affecting me phenomenally in any way. Consider social etiquette. There are many circumstances where politeness demands we not make our disbelief apparent. If I, a perennially tardy individual, assure my friend that I will arrive for a meeting at three, they might (justifiably) disbelieve me, but it may still be incumbent upon them to suppress any outward signs of that disbelief to avoid rudeness.

This wouldn’t be an epistemic injustice, merely an epistemic harm; they have good reason to disbelieve my claim. If however they disbelieve me not because of my perennial tardiness, but rather because they find trans persons to be inherently untrustworthy, holding that our gender presentation constitutes a constant form of dishonesty, this would by all accounts be an epistemic injustice. However, if they are a stoic enough individual who feels bound by politeness, I may be still wholly unaware of this.

Let’s return to Sullivan’s view. In these sorts of cases of wrongful esteem, the plausible remaining harm was that the victim of the harm was priced into transacting epistemologically in a way that undercut them. In these no-experience cases, they aren’t. They aren’t transacting at all. But, plainly, they are still wronged. Thus, all four accounts on the table are underinclusive.

Before proceeding, it is worth addressing a possible objection to what I’ve argued so far. Even if I have successfully identified cases of wronging without harm, perhaps I haven’t identified cases of epistemic wronging, without epistemic harming.Footnote ¹⁸ That is, it’s obvious that the victims in the foregoing have been wronged, but conceivably they have been wronged in a way that doesn’t constitute epistemic injustice.

This problem arises as a function of Fricker’s base definition of epistemic injustice; in the original framework she tells us “The primary harm is a form of the essential harm that is definitive of epistemic injustice in the broad” (44). According to Fricker’s framework, without a harm, there definitionally can’t be an instance of testimonial injustice.

There are two ways one can interpret this objection. First, one could reject that these are cases of epistemic injustice because they are cases of credibility excess, which Fricker rejects as candidates for epistemic injustice. But we must remember that while authors working in epistemic injustice studies frequently uncritically accept Fricker’s original formulation,Footnote ¹⁹ significant progress has been made since her monograph. Intuitively, these are cases of epistemic injustice, just as were the counterexamples offered by, among others, Pohlhaus Jr. and Davis. Rejecting them qua credibility excess would be an uncritical rejection not only of my arguments but those of many other authors working in the field.

Alternatively, one could reject these counterexamples insofar as they do not contain an epistemic harm, which has been taken to be a necessary condition for epistemic injustice in the field so far. However, I have not at this point shown that these cases do not involve an epistemic harm at all, merely that they cannot be explained by any extant account of epistemic harm. The arguments in the foregoing are compatible with the possibility that some other account of epistemic harm exists – it is only in concert with the arguments in the following that I reach the claim that these cases lack an epistemic harm tout court. To deny that these cases are instances of epistemic injustice isn’t the claim that epistemic injustice must involve epistemic harm, but is rather the claim that some extant account of epistemic harm is true. Rejecting the cases on this ground would be a petitio principii; we could have rejected Pohlhaus Jr.’s, Davis’s, or Sullivan’s on the same grounds.

Of course, this reply leaves open the question of what makes these cases examples of epistemic injustice if not epistemic harm. We will return to this question in §4, but first let us discuss the general relationship between wronging and harming.

3. Wronging without harming

None of these four accounts of epistemic harm adequately explain all cases of epistemic injustice—they cannot explain cases like Shannon and Aubrey’s. Intuitively, we might think that our next step should be to find a better account of epistemic harm, which will act as a better ground for epistemic injustice.

However, looking at harm more broadly, we find that this may not be the case. The motivation to provide a better theory of epistemic harm operates under the assumption that wrongs are grounded in harms. But we have good reason to doubt that this is the case. As David Boonin (Reference Boonin2014) states, “There are many cases in which an act does not harm a person but is at least widely believed to wrong that person. If I point a gun at you and pull the trigger in an attempt to kill you… it seems clear that I have wronged you even if the gun misfires and you are not harmed” (108). There are a number of examples from the analytic literature on harming that demonstrate that wronging doesn’t require harming. I’ll briefly discuss three.

The most obviously relevant case comes from Woodward (Reference Woodward1986).

4. Are epistemic harms in principle possible?

Let us begin by analogizing a bit further between posthumous harms and epistemic harms. Gardner (Reference Gardner, McMahan, Campbell, Goodrich and Ramakrishnan2021) notes that at least part of the question of whether or not the dead can be harmed hinges on what Tadros (Reference Tadros and Oberdiek2014) calls the currency of harm. Tadros distinguishes between this currency of harm, and what he calls the measure of harm (Gardner refers to these as the substantive component and the formal component of harm, respectively.) As she puts it, “the substantive component of the theory tells us what it is about you or your life that I interfere with when I harm you… [the formal component] tells us how a particular harm might be related to an action or event” (Gardner Reference Gardner, McMahan, Campbell, Goodrich and Ramakrishnan2021, 381).

Recall the counterfactual comparative theory of harm. According to this theory, the measure of the harm (the substantive component) is the victim’s being worse off as the result of an action than they otherwise would have been. The currency of harm (the formal component) would be the respect in which the victim is worse off. The question of posthumous harm is partly determined by whether or not the dead are appropriate subjects for the currency of harm. For example, Gardner (Reference Gardner, McMahan, Campbell, Goodrich and Ramakrishnan2021) offers an argument that the currency of harm must be well-being.Footnote ²⁷ If this be correct, and assuming the problem of backward causation cannot be solved, the dead cannot be harmed—the dead are not well-being subjects.

When we extend this notion to epistemic harm, the relevant question is whether or not epistemic harm trades in the currency of harm more generally; to be worthy of the name, it seems that an epistemic harm must be a harm, and as such impinge upon us vis-à-vis the currency of harm.Footnote ²⁸ It must make us worse off in the relevant respect. Some possible currencies of harm in the literature include well-being (Gardner Reference Gardner, McMahan, Campbell, Goodrich and Ramakrishnan2021), rights (Woodward Reference Woodward1986), interests (Kleinig Reference Kleinig1978; Perry Reference Perry2003),Footnote ²⁹ or overall quality of life (Kagan Reference Kagan1994). Some of these are more narrow; Kleinig specifies that only welfare-interests are candidates, while Perry argues that only those morally relevant interests are.

Whether or not these currencies could be impinged upon by epistemic harm will typically depend on the particularities of the theory. For example, whether or not epistemic harm can impinge on our welfare will depend on one’s substantive theory of well-being. Likewise, whether or not it can impinge upon one’s rights will depend on one’s theory of rights. This presents an initial worry for the notion of epistemic harm—it is not, to adopt Bradley’s (Reference Bradley2012) language, axiologically neutral, being incompatible with many theories of well-being, of rights, of interests. If well-being is indeed the currency of harm, as Gardner (Reference Gardner, McMahan, Campbell, Goodrich and Ramakrishnan2021) convincingly argues, then epistemic harm is plausible only under a fairly narrow subsets of theories of welfare. Notably, epistemic harm will be incompatible with any theory that has the so-called experience requirement: as discussed in §2 we can be the victim of epistemic injustice without being aware that we are such a victim. Likewise, desire satisfactionism is out because, plainly, we don’t always want to be believed, even when disbelieving us wrongs us (see the example in the following). Only certain objective-list theories and perhaps perfectionism would be compatible. If Gardner is correct that the currency of harm is well-being (as I believe she is), then epistemic harm is on shaky ground.

Other theories of the currency of harm don’t fare better. It’s certainly not clear that being wrongfully disbelieved infringes on our rights, interests, or overall quality of life. This becomes more apparent when we consider that sometimes epistemic harms, per the above accounts, actually work in favor of the victim, in at least some respect. There are cases where we do not want to believed. Suppose that I consider myself honor-bound to tell the truth to someone, but I am hoping they will disbelieve me. If they do, this both satisfies my desire and works in favor of my interests. I might even strategically use the fact that I know they are likely to wrongfully disbelieve me. If they disbelieve me because of a negative identity prejudice, they’ve surely wronged me, but according to an interests or desires account of harm, they won’t have harmed me. Conversely, they may harm me according to such an account if they do the right thing and believe me.

The notion of epistemic harm then is far from axiologically neutral; for the concept to make sense, quite a lot has to go “right.” It’s not compatible with most plausible accounts of the currency of harm. If welfare is the currency, then hedonism, desire-satisfactionism, and most objective list theories are incompatible. If rights, for example, are the currency, it must be that epistemic harm always constitutes an infringement upon our rights; this is dubious, and involves appeal to a right to credibility.

Why does it matter if our theory of harm is axiologically neutral? Simply put, it is because epistemic injustice is manifestly important. If epistemic harm is a necessary condition for epistemic injustice, then the plausibility of epistemic injustice hinges on the plausibility of epistemic harm, and ipso facto, what the correct theory of well-being, or interests, or rights is. But plainly, epistemic injustice is real and significant regardless of what the correct theory of well-being, interests, or rights is. If we take it as true that epistemic injustice exists (as we should), one of two things is also true. Either, epistemic harm is not required for epistemic injustice, or hedonism, desire-satisfactionism, and most objective list theories are all false; if we can affirm the existence of epistemic injustice, and we hold that epistemic harm is necessary for epistemic injustice, we will have solved much of the debate on harm and well-being more generally.Footnote ³⁰ While not definitive, I find the alternative, that epistemic injustice can exist absent epistemic harm, to be more plausible. After all, there are already theories of epistemic wronging that don’t appeal to epistemic harm. While I don’t want to endorse his view as correct, Congdon (Reference Congdon, Kidd, Medina and Pohlhaus2017) does, for example, argue instead for a basis of Hegelian misrecognition.

In §2, I showed that there are cases of putative epistemic injustice that cannot be explained by current theories of epistemic harm. In the interim, I’ve argued that epistemic harm likely does not exist, and if so we cannot explain those casesFootnote ³¹ with any theory of epistemic harm. What remains then is the problem of what makes epistemic injustices genuinely epistemic, as opposed to some other kind of injustices, if not epistemic harm. How can we explain one’s “being wronged in their capacity as a knower?” While there is no space in the present work to fully explore this question, I offer three general possibilities. First, a sufficiently circumscribed account of wronging may be able to do that job. If Congdon’s misrecognition account is true, perhaps that misrecognition is the unifying feature. Perhaps it’s even possible that a small adjustment to an extant theory of epistemic harm may do the job. Suppose that McGlynn (Reference McGlynn2019; Reference McGlynn2020) adequately rescues Fricker’s objectification account. It may be that a kind of epistemic objectification is indeed the ground for the wrong of epistemic injustice, just without constituting a harm. This might seem to come down to semantics—whether or not we use the term “harm.” However, by not appealing to harm, even just as a matter of verbiage, we also preempt objections to cases of epistemic injustice where the victim does not seem worse off, just as we insist that Quinn is wronged but not worse off in Airline.

The second possibility is that the ground of epistemic injustice isn’t what unifies the concept. Perhaps sometimes the ground is objectification, in other cases derivatization, etc. If epistemic injustice encompasses all cases where someone is wronged in their capacity as a knower, it may not be the wrong-making feature that matters, but just that the victim is wronged in this capacity.

Finally, it’s possible that there just isn’t a unifying property that makes epistemic injustice epistemic. It could instead be a Wittgensteinian cluster concept. These are simply preliminary possibilities, and this question should be explored in greater depth in further work in epistemic injustice studies. Regardless though, it is plausible that if we abandon the concept of epistemic harm, we do not threaten the existence of epistemic injustice more generally.

Some may think that we are giving up too much if we abandon epistemic harm. What about epistemic violence? Kristie Dotson (Reference Dotson2011) defines epistemic violence as “a failure of an audience to communicatively reciprocate, either intentionally or unintentionally, in linguistic exchanges owing to pernicious ignorance” and pernicious ignorance as “a reliable ignorance or a counterfactual incompetence that, in a given context, is harmful” (242). At first blush, abandoning epistemic harm might seem like abandoning epistemic violence. But this needn’t be the case.

Dotson notes that the harms involved in pernicious ignorance and epistemic violence are quite varied. She states,

Citing Fricker (Reference Fricker2007), Townley (Reference Townley2003), and Collins (Reference Collins2000), Dotson discusses specific harms to intellectual courage, epistemic agency, and the intellectual traditions of entire groups, respectively, as relevant harms that might be involved in cases of epistemic violence. Elsewhere, discussing a case from Cassandra Byers Harvin (Reference Harvin1996), she states, “the harm that results from the situated ignorance in the Harvin example can come from many fronts. Some argue that part of the “stress” of living as a black person in the United States is the persistence of racial microaggressions” (Dotson Reference Dotson2011, 250)

My theory doesn’t given us reason to doubt that these are harms. Recall Fricker’s distinction between primary and secondary harms. What I’ve argued for casts doubt only on the primary harm: the epistemic harm. Under this primary-secondary framework, these harms would be secondary. They are the downstream, damaging effects of epistemic injustices, rather than the objectification, derivatization, or undermining testimonial practice, as described by Fricker, Pohlhaus Jr, or Sullivan.

Wouldn’t abandoning epistemic harm undermine some cases of epistemic violence? Surely some cases of pernicious ignorance are harmful only in virtue of putative epistemic harms. Suppose this is the case. This may give us reason, rather than doubting cases of epistemic violence, to expand our understanding thereof. Perhaps reliable ignorance needn’t be harmful to be pernicious. This would give us reason to believe that it’s possible that harm is not a necessary condition for epistemic violence, just as it is not for epistemic injustice more broadly.