1. The primary harm of testimonial injustice

Miranda Fricker characterises an epistemic injustice as ‘a wrong done to someone specifically in their capacity as a knower’ (Reference Fricker2007: 1). Her focus is primarily on what she calls testimonial injustice, where a speaker's word isn't given due weight because her audience unjustly regards her as less credible on the subject matter at hand due to their biases. The main topic of this paper concerns the kind of harm suffered by someone who experiences testimonial injustice. In particular, I will criticise Fricker's proposal that the primary harm involved is a kind of epistemic objectification, where this is understood on the model provided by Martha Nussbaum's influential analysis of sexual objectification, and where this is taken to involve the denial of someone's epistemic agency. I am not the first to criticise Fricker's account of the primary harm, and I examine the existing objections, arguing that some put pressure on Fricker's account while others do not. I then try to formulate what I take to be the most difficult problem for Fricker's account, arguing that one of Fricker's own central examples of testimonial injustice in fact offers the basis of a particularly telling objection. Finally, I turn to an assessment of what our reaction should be to these problems for Fricker's account. Fricker's other critics have mostly concluded that they motivate a move to an alternative account of the primary harm of testimonial injustice, one drawing on the resources of recognition theory. My aim in the final section of the paper is to show that this conclusion is premature. Both Fricker and her critics have underestimated the resources that are provided by Nussbaum's analysis of objectification when offering an account of the primary harm, and something very much in the spirit of Fricker's account survives the objections.Footnote ¹

The remainder of section 1 more fully introduces the notion of testimonial injustice and Fricker's central examples, discusses the worry that the examples don't seem to illustrate a unified phenomenon, and sets out Fricker's account of the primary harm of such injustices as in part a response to that worry. Section 2 critically assesses Fricker's account of the primary harm of testimonial injustice, examining recent objections due to José Medina, Gaile Pohlhaus Jr., and Emmalon Davis, as well as developing my own objection. Finally, section 3 outlines a Nussbaum-inspired account of the primary harm as a kind of epistemic objectification that avoids the problems that the objections raise to Fricker's own version, and compares it to the currently popular rival account offered by recognition theory.

What Fricker calls the ‘central case’ of testimonial injustice involves a systematic identity-prejudicial credibility deficit (Reference Fricker2007: 28). We can unpack this characterisation with Fricker's own examples. In To Kill a Mockingbird, Tom Robinson faces a credibility deficit, since he is not believed by the all-white jury when truthfully testifying to his innocence at his trial for the rape of Mayella Ewell, a young white woman. This injustice is identity-prejudicial since it's due to the prejudices about black men held by the jurors.Footnote ² Moreover, it's systematic because these racial prejudices ‘track’ him ‘through different dimensions of social activity—economic, educational, professional, religious, and so on’ (Fricker Reference Fricker2007: 27). Non-systematic – and so non-central – cases of testimonial injustice involve identity-prejudicial credibility deficits that are grounded in prejudices about someone's identity that do not track in this manner; for example, Fricker describes philosophers of science being subject to identity-prejudicial credibility deficits at a conference dominated by working scientists (Fricker Reference Fricker2007: 28–9). Fricker's other central example is that of Marge Sherwood in (the screenplay of) The Talented Mr Ripley, who has her (initially well-founded) suspicions about the role of Tom Ripley in her fiancé Dickie Greenwood's death dismissed by Dickie's father Herbert on the grounds that they're merely based on ‘female intuition’ (Fricker Reference Fricker2007: 9). Here again we have a case of a systematic identity-prejudicial credibility deficit, but this time it's prejudices associated with Sherwood's gender rather than her race that are at work, and which track her across her social interactions.

That's one difference between Fricker's two cases, but it's not the most important difference for our purposes here. Greenleaf very clearly doubts (and wants to cast doubt on) Sherwood's capacity to know what has happened to his son. That marks a contrast with the case involving Tom Robinson; the jurors think that Robinson knows fine well what happened between himself and Mayella Ewell, but they think that he is being insincere. While it's easy to see how the former kind of case is one in which Sherwood is treated unjustly in her capacity as a knower, it seems like much more of a stretch to insist that the latter kind of case also falls under this characterisation (Pohlhaus Reference Pohlhaus2014: 101).

Fricker is well aware that there is a potential worry here, and she can be read as attempting to address it in two complementary ways. First, she endorses a view about the function of knowledge ascriptions due to Edward Craig (Reference Craig1990), according to which (roughly) to attribute knowledge that P to someone is to tag them as a good informant on the question of whether P, and this gives the ‘core’ of our concept of a knower. To be excluded from the community of epistemic trust is, given this thesis, ‘a matter of exclusion from the very practice that constitutes the practical core of what it is to know’ (Fricker Reference Fricker2007: 6). Adopting this thesis yields a clear sense in which one who is unjustly regarded as insincere when testifying is subject to harm in their capacity as a knower. However, neither Craig's methodology for epistemology nor his specific claim about the purpose of knowledge attributions has been widely accepted, and both have been subject to substantial criticisms (see e.g. Kappel Reference Kappel, Haddock, Millar and Pritchard2010; Kelp Reference Kelp2011; Lackey Reference Lackey, Brown and Gerken2012). For example, Kelp (Reference Kelp2011) argues that a priest who takes the seal of confession appropriately strictly can be a knower without being a remotely good informant, and that a conspiracy theorist who masks her real beliefs about global warming by asserting what the best science tells us may be an excellent informant while falling far short of knowing. Kelp is careful not to present these as counterexamples to Craig's thesis, since Craig isn't engaged in the business of trying to offer strict necessary and sufficient conditions for being a knower, but Kelp does think they should give us pause and that they motivate seeking an alternative to Craig's account of the function of knowledge attributions.Footnote ³ So Fricker is appealing to a controversial and implausible claim as a load-bearing element of her account of testimonial injustice, and she does so without providing any defence of this claim; the burden of proof would seem to be on her to persuade us that Craig's account is much more plausible than it appears to be.

Fricker's second line of reply brings us directly into contact with my focus in this paper, since she suggests that another, related sense in which there is unity between cases of unjust imputation of incompetence and those of unjust imputation of insincerity emerges when we look at what she calls the primary harm of testimonial injustice:

Fricker describes several secondary harms of testimonial injustice, including both practical harms, such as being found guilty and sentenced to prison in Tom Robinson's case, and epistemic harms, such as having one's confidence unjustly undermined to the extent that one actually ceases to know (Fricker Reference Fricker2007: 46–8). However, these harms are contingent, in the sense that one might be subject to a testimonial injustice and not be harmed in these ways; for example, one might be subjected to a testimonial injustice, and instead of this weakening one's conviction in the truth of the claims one is making, it might make one more convinced than before. The primary harm is not contingent in this manner – it would be ‘a form of the essential harm that is definitive of epistemic injustice in the broad’ (Fricker Reference Fricker2007: 44) – and it would (partly) explain why these secondary harms arise when they do.

Fricker first offers an initial characterisation of the primary harm of testimonial injustice, designed to explain why it is ‘a form of the essential harm that is definitive of epistemic injustice in the broad’, and she then develops this account in more depth later in the book. Fricker's initial characterisation of the primary harm is given in the passage quoted above, but here's a more explicit statement:

This characterisation of the primary harm does make it clear how cases of testimonial injustice, including cases of presumed insincerity such as Fricker's example from To Kill a Mockingbird, are instances of epistemic injustice: that is, how they are instances of harm to one specifically in one's capacity as a knower. However, that's not to say that all is well with Fricker's account. First, there's still room to wonder whether this characterisation of the primary harm cleanly applies to all cases of testimonial injustice. As we have seen, Fricker's principal characterisation of testimonial injustice is in terms of credibility deficits, and to treat someone as less credible than one ought to on some issue isn't the same as unjustly treating their testimony on that issue as failing to express knowledge (Luzzi Reference Luzzi2016). There are several kinds of cases in which these come apart. Perhaps, due to identity prejudices, I unjustly lower my confidence in what a speaker is telling me while still (perhaps grudgingly) regarding them as expressing knowledge. Or a speaker might issue a prediction that P on grounds which are reasonably good but manifestly insufficient for knowledge that P, and due to prejudice I give their prediction no weight whatsoever.Footnote ⁴ In both these kinds of examples, I do the speaker an injustice by reducing my credence in what they say on the basis of my identity prejudices, but I don't unjustly fail to regard their assertions as expressing knowledge; in the first kind of example I still regard the speaker as expressing knowledge, and in the second it's mutually known by speaker and audience that the speaker is not expressing knowledge.Footnote ⁵ Given this point, we might be sceptical of the proposal that the primary harm of identity-prejudicial credibility deficits involves wronging someone in their capacity as a giver of knowledge, as Fricker's characterisation suggests.

However, my main goal in this paper is not to target Fricker's initial characterisation of the primary harm of testimonial injustice, but rather to critically examine her development of an account of that harm as a kind of epistemic objectification. Before turning to that task, it is worth saying a little about the significance of this aspect of Fricker's account. Why can't we just rest content with Fricker's initial characterisation of the primary harm of testimonial injustice (assuming that the misgivings aired in the previous paragraph can be silenced)?

I think that there are at least two points to be made in reply to this question, and they're closely related. First, Fricker's characterisation of the primary harm as involving a subject being harmed in their capacity as a giver of knowledge spells out the (primary) epistemic harm involved in testimonial injustice, but it doesn't say anything about the ethical harm involved. The further claim that testimonial injustice involves epistemically objectifying someone makes a step towards a unified account of the epistemic and ethical harms that are intrinsic to testimonial injustice. Second, classifying the harm of testimonial injustice as a kind of objectification puts it in a category together with other unjust behaviours and practices, perhaps including, but certainly not limited to, pornography, prostitution, and slavery. Of course, by itself neither labelling certain behaviours and practices as objectifying nor likening it to practices such as the making and use of pornography suffices to pinpoint the ethical harm involved, since there's no consensus on what makes objectification ethically problematic (or even whether objectification is always ethically problematic; see Nussbaum (Reference Nussbaum1995) for an influential defence of the claim that it is not). But it does put it in a familiar category, and looking at what it has in common with other behaviours and practices in that category may help us in theorizing about the nature of the ethical wrong involved in objectification. So while Fricker's account of the primary harm as a kind of objectification doesn't amount to a full theory of the ethical harm involved in testimonial injustice, it would still mark progress on that score if it were correct.

With this reminder of what's at stake here in mind, let us turn to Fricker's development of the idea that the primary harm of testimonial injustice is that it involves a kind of epistemic objectification, to be understood along the lines suggested by Nussbaum's influential analysis of sexual objectification (Nussbaum Reference Nussbaum1995).Footnote ⁶ Fricker again looks to Craig's discussion of the function of knowledge attributions, this time to a distinction he draws between informants and mere sources of information:

Fricker's proposal, then, is that testimonial injustice involves treating an epistemic subject as a mere source of information, and she argues that this in turn involves a kind of objectification. There's objectification here, according to Fricker, since there's a denial of epistemic agency; it's to treat a person as being epistemically on a par with a tree trunk from which one can infer information about the tree's age if one has the right background information (Fricker Reference Fricker2007: 132–3), and tree trunks exercise no epistemic agency.

Fricker isn't suggesting that treating someone as a source of information is always disrespectful or problematic. Her example of the bedraggled guest in the passage just quoted makes this clear; there's nothing in the slightest untoward about that interaction, as Fricker describes it. Whether treating someone as a source of information is problematic is going to depend on the broader context – as Nussbaum says, ‘In the matter of objectification, context is everything’ (Reference Nussbaum1995: 271). One of Nussbaum's examples to illustrate this point involves her using her lover's stomach as a pillow as he sleeps; with his consent and in the context of a relationship in which he is usually recognised as more than a pillow this seems fine, and perhaps even a sign of trust and intimacy, but if Nussbaum were to consistently treat a person as if they are no more than a pillow that would be different (Reference Nussbaum1995: 265). Likewise, Fricker wants to suggest, treating someone as a source of information is alright as long as one doesn't treat them merely as a source of information; it's crucial that in one's interactions with them that one also sometimes treats them as an informant (or at least that one would be disposed to so treat them were one to have further interactions with them).

2. Epistemic objectification and the denial of epistemic agency

Now that we have Fricker's account of the primary harm of testimonial injustice on the table, I want to turn to criticisms of it.

A first objection to Fricker's account comes from José Medina (Reference Medina2012: 203–4). He argues that the account is incomplete since there are possible cases of testimonial injustice that involve someone being treated as merely an informant, rather than as a mere state of information. Medina accepts Fricker's idea that the harm in testimonial justice involves a kind of denial of epistemic agency, but he argues that the agency awarded a mere informant is still limited when we compare it with that of an inquirer:

Medina concedes more to Fricker here than I intend to in what follows. As noted above he takes his point to show that Fricker's account requires supplementation, and below I'll aim to show that there are cases that suggest there's something more fundamentally wrong with her account – indeed, that Fricker has described such cases herself. Still, I do think that a satisfactory account of the primary harm should have something to say about the kinds of case Medina has in mind here, and I'll come back to it in the final section.

The second argument against Fricker's account I want to consider comes from Gaile Pohlhaus Jr. (Reference Pohlhaus2014: 102–4), who argues that it's ‘odd’ to characterize the harm done to a victim of epistemic injustice as a kind of objectification. Her worry is that mere sources of information make epistemic claims on us that are absent when we subject a person to testimonial injustice. Of Fricker's own examples of sources of information, she writes:

A number of Pohlhaus's claims here might seem overstated. I can fail to believe that the number of rings is as I have just counted if I have certain kinds of collateral information: I have taken a drug that impairs counting; someone I regard as an epistemic peer (or superior) came up with a different result; and so on. It takes a slightly more sceptical turn of mind to come up with such possibilities for the umbrella example, but we can do it. However, Pohlhaus makes it clear that we're meant to set aside these kinds of considerations (Reference Pohlhaus2014: 103, 113 n4). Even still, her argument seems suspect. My worry starts from the observation that Pohlhaus subtly but significantly changes Fricker's examples. Fricker discusses gleaning the age of a tree from counting the number of rings on its trunk and that it has been raining from the wetness of one's guest and her umbrella (Reference Fricker2007: 132–3). In contrast, in the passage just quoted Pohlhaus asks us to consider whether a subject who has had the relevant experiences can deny the propositions that the tree has n growth rings (when one has just counted n rings) and that one's guest's umbrella is wet; these are propositions that form part of the bases of the beliefs that interest Fricker when she offers the examples.

This observation is sufficient, I think, to cast doubt on Pohlhaus's objection to Fricker. The worry is that when Pohlhaus contrasts clear cases of treating something as a source of information to cases of testimonial injustice, the differences she finds are artificial, generated by the changes she has quietly made to Fricker's own examples. When we keep the examples more parallel, the contrast Pohlhaus claims to find is missing. In cases of testimonial injustice I don't believe what I should, not because I doubt that the person has testified that P to me, but because prejudices I carry cause me to make an inappropriately low credibility assessment of that person, and so I don't give their word the weight I should. In Fricker's cases of treating something or someone as a source of information I fail to believe what I ought, not because I doubt that the tree has the number of rings that I just tallied or that my guest's umbrella is wet, but because I have inappropriate doubts about whether on that basis I can glean information about the age of the tree or the weather outside. There no longer seems to be any contrast, and so Pohlhaus fails to establish that there's an ‘oddity’ to Fricker's account of the primary harm of testimonial injustice here.

Despite this conclusion, I think that Fricker's account is in trouble. In order to bring out the problem with the account, it is useful to have it more fully laid out in Fricker's own words:

Here Fricker draws a link between treating someone as a source of information and denying them epistemic agency. The proposal, as I understand it, is that treating someone as a source of information is to epistemically objectify them, and the reason that it's a form of objectification is that sources of information are epistemically passive: that is, they are lacking in epistemic agency or subjectivity, unable to exercise a capacity for knowledge. And the main problem with Fricker's account is that treating someone as a source of information, even when one is perpetrating a testimonial injustice against them, doesn't seem to involve treating them as lacking in epistemic agency in anything like this sense.

Importantly, we can illustrate the point with one of Fricker's own examples.Footnote ⁷ Recall that earlier we distinguished cases in which a speaker's competence to know is slighted and cases in which it's only their sincerity that is questioned. The point I want to make is clearest with respect to the latter, so let's consider again the example of Tom Robinson. Thinking that someone is lying doesn't seem to involve thinking of them as an epistemically passive object, or thinking of them as unable to exercise a capacity for knowledge. It plausibly does involve treating them as a source of information rather than an informant; I cannot simply accept their word, but can at best glean information, either by inferring the opposite of what they say (if I take them to be systematically and predictably trying to mislead me) or from other clues such as their appearance, body language, and behaviour (both non-verbal and verbal). Fricker's claim that this is a kind of harm to one in one's capacity as a giver of knowledge is very plausible. However, her specific account of that harm, as involving a kind of objectification through being treated as epistemically passive rather than as a locus of epistemic agency, seems much less plausible as an account of the harm done to Tom Robinson.

Fricker explicitly claims otherwise, writing:

Let's grant to Fricker that Atticus Finch has indeed got the measure of the jurors here, with regards to the prejudices at work in their assessment of Tom Robinson.Footnote ⁸ However, the prejudices we find summarized in Finch's summing-up speech are that all black men lie, are inherently immoral beings, and are not to be trusted around women (Lee Reference Lee1960: 225, quoted in Fricker Reference Fricker2007: 25). These are (as Finch says, ‘evil’) assumptions about the morality and behaviour of black men, but there's no ‘obvious’ slight to Robinson's status as an epistemic subject, where that would imply a thoroughgoing epistemic passivity.

This suggests that treating someone as a source of information rather than an informant in the context of subjecting them to a testimonial injustice seems to be compatible with robust recognition of a considerable degree of epistemic agency, and at least one of Fricker's own central examples of testimonial injustice illustrates this compatibility. Though Tom Robinson is harmed by the jury's treatment of him in his capacity as a giver of knowledge, Fricker's account of this harm seems implausible since he isn't, contrary to Fricker's suggestion, treated as epistemic passive or as akin to a felled tree. Given this, there seems to be little prospect of diagnosing the primary harm of testimonial injustice as a kind of epistemic objectification, when the objectification in question is taken to essentially involve this kind of denial of epistemic agency.

Given the slipperiness of the notion of epistemic agency, as well as uncertainty about what's involved in failing to recognize someone's epistemic agency in a problematic way, it's difficult to say anything decisive in favour of this conclusion (though the point cuts both ways – it will be hard for Fricker to establish that the cases in question do involve a substantial slight to a subject's epistemic agency for the same reason). Perhaps it will be helpful, though, to briefly again compare and contrast the case of Tom Robinson with that of Marge Sherwood. This will serve to further bring out the differences suggested in the previous section, but I also hope that the example of Marge Sherwood will further illustrate what's involved in denying someone's epistemic agency, in a way that will prove useful for assessing the claims I have made about Tom Robinson. Sherwood has exercised epistemic agency, as I understand this notion, in all kinds of ways. She has acquired evidence relevant to the issue at hand (what has happened to her fiancé Dickie Greenleaf), and moreover she has appreciated the significance of that evidence. Taking that evidence as a basis she has drawn a reasonable – and indeed, correct – conclusion. Dickie's father's put-down to Sherwood explicitly signals that he doesn't take Sherwood to have exercised agency in any of these ways, and he does this because of prejudices about her gender. She's just reporting a ‘female intuition’ that bears no relation to the facts of the matter (Fricker Reference Fricker2007: 14). In his eyes, she doesn't even count as really having engaged in inquiry on the matter. Fricker's analogy to a tree-stump is still somewhat hyperbolic, but there's at least a sense in which Herbert Greenleaf treats Sherwood as not having done – and of being incapable of doing, on account of her gender – any better than the stump with respect to the issues he is investigating. The prejudices involved in the trial of Tom Robinson, real and dangerous as they are, don't have anything like this character, and that's the basis on which I claim that the jury are not failing to recognize his epistemic agency. Robinson's status as an informant is, of course, called into question by the racist presumption that he can't help but lie, but I argued in section 1 that, contrary to Fricker's Craig-inspired proposals, to slight someone in their capacity as an informant isn't always to slight them in their capacity as a knower.

This is, I think, the strongest kind of objection to Fricker's account of the primary harm. However, in a recent paper Emmalon Davis (Reference Davis2016) offers a related and interesting objection to Fricker's account, on the basis of considering cases of identity-prejudicial credibility excesses rather than deficits. Fricker herself had argued that while being given an excess of credibility might have negative downstream epistemic consequences for one, which might cumulatively amount to an injustice, these are not cases in which one suffers the primary harm of testimonial injustice (Fricker Reference Fricker2007: 20–1). However, Davis argues that cases of prejudicial credibility excess (PCEs) can equally be cases in which a speaker is immediately and intrinsically epistemically harmed. In one of Davis's examples, an Asian-American student is asked for help with a difficult algebra problem, due to the so-called ‘positive stereotype’ that Asian-Americans are good at mathematics (Davis Reference Davis2016: 487).Footnote ⁹ Such a case, Davis argues, shares much in common with Fricker's examples of prejudicial credibility deficits – both kinds of unjust treatment are epistemically flawed in relying on prejudicial stereotypes and ethically flawed in failing to recognise the individuality of different members of the same social group (Reference Davis2016: 487–8) – and so we should look for an account of the primary harm of testimonial injustice that covers both excesses and deficits. However, she plausibly contends that Fricker's account doesn't extend to cases of PCE, and for much the same reasons that I've argued it struggles with the case of Tom Robinson:

This objection leaves a little more room for Fricker to manoeuvre than the one I developed above, since it remains open to Fricker to continue to refuse to regard cases of PCE as genuine examples of epistemic injustice. Since I employed one of Fricker's own central examples above, there's no such move available there. That said, as with Medina's case, Davis's cases are persuasive, and in what follows I'll accept that they are problematic for Fricker's account as it stands and need to be accommodated by any replacement or modification.

3. The nature and significance of epistemic objectification

In the previous section, I have criticised Fricker's account of the primary harm of testimonial injustice as a kind of objectification which involves a denial of a person's epistemic agency, and which is to be understood on the model provided by Nussbaum's analysis of sexual objectification. If I've succeeded in putting that proposal in doubt, it's natural to wonder what positive proposal (if any) might be put in its stead, and how far an alternative that offers a satisfactory treatment of the problem cases needs to depart from Fricker's own account.

A relatively extreme response would be to suggest that Fricker's whole project is misconceived – that there's no such thing as the primary harm of testimonial injustice since we're not dealing with a unified phenomenon here, as suggested by the kinds of differences between the different cases that we discussed in section 1. I suggested in that section that perhaps none of Fricker's attempts to unify the different kinds of example is entirely successful, and so I think that this remains a possibility we should take seriously. That said, I hope we can do better.

Another possibility we should consider is whether there's an alternative account of the primary harm of testimonial injustice that unifies the two kinds of cases, showing the differences to be relatively superficial, as well as accommodating the problem cases reviewed in the previous section. This is the approach taken by virtually all of Fricker's critics discussed in this paper. They suggest that the harm of testimonial injustice is instead that it involves a kind of othering of another person, in a sense inspired by Simone de Beauvoir's and Franz Fanon's developments of recognition theory (Pohlhaus Reference Pohlhaus2014; Davis Reference Davis2016; Congdon Reference Congdon, Kidd, Medina and Pohlhaus2017); as Pohlhaus puts it, ‘the intrinsic epistemic harm of testimonial injustice is more aptly described in terms of a subject/other relation rather than the subject/object relation proposed by Fricker’ (Reference Pohlhaus2014: 100). Elaborating on how recognition theory might be fruitfully applied here, Pohlhaus writes:

This kind of account seems well placed to accommodate examples in which we might want to say that someone is being treated as less than a full epistemic subject in their own right, without being regarded as a mere epistemic object, more or less devoid of epistemic agency. However, although Davis also favours this kind of account of the primary harm of testimonial injustice, she notes that as developed by Pohlhaus it fails to accommodate Davis's examples of prejudicial credibility excesses. In Davis's examples of PCEs, as she understands them, the ‘othering’ of a speaker doesn't involve only giving them credibility when their testimony coheres with and helps to make sense of ‘the experienced world of those deemed fully as subjects’ (as Pohlhaus suggests in the quote above and elsewhere in her paper); on the contrary, they are given credibility excesses just so long as their testimony goes beyond the experiences of the dominant perspective:

In both kinds of cases, one's epistemic agency, while recognised, is only recognised in relation to the dominant, deemed full epistemic subjects, and so Davis offers a relatively unified account of the epistemic harm across credibility deficits and excesses.

However, there's also a third option that departs much less dramatically from Fricker's own approach to these issues. We might wonder whether we should really give up so easily on Fricker's idea that there's a kind of epistemic objectification involved in testimonial injustice. If the arguments of this paper are correct, we shouldn't conceive of this kind of objectification in the way that Fricker suggests – as primarily involving a kind of denial of epistemic agency – but that doesn't put the more general thought to rest.

There is, in any case, something rather disappointing about Fricker's conception of epistemic objectification. On Nussbaum's analysis, objectification is a very rich, multifaceted cluster concept – she identifies no fewer than seven ways that one can treat someone as an object, and Rae Langton (Reference Langton and Langton2005) adds three more – and we would naturally expect epistemic objectification to be analogously interesting and complex. But we don't find any notion with comparative richness in Fricker's discussion. That suggests that we haven't come close yet to fully mapping out the different ways that our behaviour and attitudes can be epistemically objectifying. Though I won't have space to develop such an account in anything like full detail here, in this final section I want to suggest that a Nussbaum-inspired account of epistemic objectification may in fact have the resources to diagnose the harm in the cases of testimonial injustice that evade Fricker's own account.

As mentioned in the previous paragraph, Nussbaum distinguishes seven ways of treating someone as something. These are:

1. 1. Instrumentality: the objectifier treats the object as a tool of his or her purposes.

2. 2. Denial of Autonomy: the objectifier treats the object as lacking in autonomy and self-determination.

3. 3. Inertness: the objectifier treats the object as lacking in agency, and perhaps also in activity.

4. 4. Fungibility: the objectifier treats the object as interchangeable with other objects of the same type, or with objects of other types.

5. 5. Violability: the objectifier treats the object as lacking in boundary-integrity, as something it is permissible to break up, smash, break into.

6. 6. Ownership: the objectifier treats the object as something that is owned by another, which can be bought or sold, and so on.

7. 7. Denial of Subjectivity: the objectifier treats the object as something whose experiences and feelings (if any) need not be taken into account.

There's much to be said about each of these ways of objectifying, the relations between them, Nussbaum's conception of objectification as a cluster concept involving all seven, her claim, discussed briefly above, that objectifying in some of these senses can be morally neutral and perhaps even a good thing, and Langton's three additions to the cluster. I will leave these issues aside on this occasion, however. Instead, I'll consider how enriching the notion of epistemic objectification in ways suggested by Nussbaum's full analysis might enable us to retain Fricker's suggestion that the primary harm of testimonial injustice is a kind of epistemic objectification.

With Nussbaum's list on the table, we can see that Fricker's notion of epistemic objectification is largely an epistemic spin on inertness and instrumentality; her diagnosis of the harm of testimonial injustice is that it's to treat someone as epistemically passive and as merely a source of information for one's own epistemic purposes. I suggest that we take another look at Nussbaum's list for ways to accommodate the cases that Fricker's account struggles with, namely Medina's case of a testifier who is treated as nevertheless lacking full epistemic agency, Davis's cases of credibility excess, and cases of presumed insincerity such as Tom Robinson.

Let's start with Davis's cases, since these are the most straightforward. In fact, although Davis expresses a preference for an alternative account, she actually offers the seed of an account of her cases in terms of epistemic objectification. Identity prejudicial credibility excesses (PCEs), in her view, involve a kind of epistemic fungibility:

Sexual fungibility involves treating members of a certain social group as interchangeable for sexual purposes. Epistemic fungibility involves treating members of a certain social group as interchangeable for epistemic purposes; testimony from one member of the group (on the subject at hand, at least) is as good as any other. It's also natural to think that there is a kind of instrumentality here (and perhaps this is true in general of fungibility); one is treating the person as fungible relative to one's own epistemic goals and projects. There's no reason to look outside of Nussbaum's analysis of objectification to accommodate such cases, once we avail ourselves of its full resources.

The other two kinds of examples are similar, but trickier. It's not that Nussbaum's analysis leaves us without things to say about these cases, but what it does let us say may seem thin and unsatisfactory – we'll return to this worry below. Medina's examples, recall, involve someone who is treated as an informant without being allowed to count as an inquirer; in the simplest kind of case, one treats them as a source of answers to one's own questions, rather than as the author of their own questions. Tom Robinson illustrates the possibility that someone's testimony might be unjustly deemed worthless without this involving a slight to their epistemic agency, due to their being regarded as mendacious. Above I noted that Fricker's account treats the kind of epistemic objectification involved in testimonial injustice as involving at least a combination of instrumentality and inertness, and the reason the two kinds of cases just mentioned seem to fall outside of her account is that they don't seem to involve inertness in any significant respect.

Both kinds of cases do seem to involve an epistemic form of instrumentality, however. In Medina's case, this is explicitly built into the setup; one treats someone's epistemic agency as merely a means for carrying out one's own epistemic projects, and fails to see them as capable of pursuing their own epistemic agenda, from which one might stand to learn. With Tom Robinson, the idea is one expressed by Fricker herself and introduced already above; rather than listening to what Robinson has to say and granting him the credibility he deserves, one uses him as a source of fodder for one's own information-gathering activities. Fricker goes further, of course, adding that he is thereby treated as a passive source of information, but we should stop short of that claim. Given the presumption that Robinson knows what happened but is dishonest, his testimony will furnish the premises of all kinds of inferences that he himself does not intend, and his attempts to communicate what he in fact knows will be doomed to fail – that's at least part of the dynamic we see played out in the courtroom scenes in the book. The suggestion is that we can think of this too as a kind of epistemic instrumentality; Robinson's attempted testimony is treated as material in the service of the epistemic projects of others, and his own intention to communicate his knowledge of what happened is systematically thwarted. Consider, for example, an exchange that Fricker (Reference Fricker2007: 24) and others have picked out.Footnote ¹⁰ Under pressure from the prosecutor, Mr Gilmer, Robinson says that he regularly helped Mayella Ewell because he felt sorry for her. Mr Gilmer responds:

Here, Robinson is trying to simply and straightforwardly answer Mr Gilmer's question. However, due to Mr Gilmer's prejudices and assumptions, he takes Robinson's answer as confirmation of those prejudices (that Robinson is a liar, is untrustworthy, has delusions of superiority (Fricker Reference Fricker2007: 24–5), and so on). Moreover, Mr Gilmer clearly frames Robinson's testimony in such a way so as to maximise the chances it will have the same effect on others in the courtroom (including the jurors).

Of course, this is a fictional, and so artificial example, and so even if one agrees that Robinson and his testimony are subject to a kind of harmful instrumentalization in this example, one may resist the idea that this is a plausible general diagnosis of the epistemic-ethical harm involved in such cases of testimonial injustice. But I do think that this example, artificial though it may be, is pointing us in the right direction. Generalised, the thought is that to systematically and prejudicially dismiss what someone is trying to convey on a particular topic as lies is to be disposed to treat what they say as of epistemic relevance – as potentially yielding some worthwhile information – only insofar as one treats it as fodder for one's own information-gathering projects and insinuations, as Mr Gilmer does with Tom Robinson's testimony. As I've already noted, this is close to Fricker's own account, as detailed in the long quote above, but stripped of her insistence that this is to treat someone as epistemically ‘passive’. As in Medina's case, this is instrumentality without inertness, and this is why Fricker's account, with its focus on inertness and passivity, can't accommodate it.

Interestingly, there's epistemic fungibility present in the case of Robinson too. Normally, the fact that a person was present during and involved in some event would make them a particularly good person to listen to if one hoped to find out what happened. But if all black men lie, as the deeply racist presumption Finch imputes to the jury would have it, then Robinson is no longer a particularly good person to listen to if one wants to find out what happened between him and Mayella Ewell, despite his first-hand experience (indeed, his involvement); the jury might as well listen to any other black man on this matter.

Viewed through the lens of Nussbaum's cluster analysis of objectification, the disputed cases are ones in which there is epistemic fungibility and/or epistemic instrumentability, but where inertness is absent. Is that enough to rescue Fricker's thesis that the primary harm of testimonial injustice is a kind of epistemic objectification? As noted above, this is likely to be a matter of dispute. It's certainly less clear that treating someone this way merits being called objectification; one may be treating someone badly if one treats them as an instrument for one's own purposes and as fungible, but it's less clear that one is thereby treating someone as a mere something. For my own part, whether or not one is willing to apply the label of ‘objectification’, I think that Nussbaum's analysis does plausibly offer us resources to formulate a unified account of the primary epistemic harm of testimonial injustice.

To try to bring out the kind of harm involved in testimonial injustice, on this account, let's consider the analogy to sexual objectification again. Suppose that one has a sexual partner that one invariably treats as merely a means to satisfy one's own sexual wants, without any consideration of their sexual pleasure, and as fungible with other people who would be able to satisfy one's sexual wants. Such a relationship seems unpleasantly one-sided, and we might even think that there's something objectifying about it, but it's not clear that there's anything inherently morally problematic. As Nussbaum stresses repeatedly, as already noted above, ‘In the matter of objectification, context is everything’ (Reference Nussbaum1995: 271), and so long as the wider relationship is one of rough equality and mutual respect, and the one-sidedness of the sexual relationship is something both parties have willingly signed up to, we might feel that no moral line has been crossed.Footnote ¹¹ But if the one-way nature of the sexual relationship isn't consensual, or if the wider relationship involves a lack of respect or a significant power differential, then there may well be something morally problematic here. The suggestion is that a Nussbaum-inspired account of epistemic objectification as the primary harm of testimonial injustice should see the cases that were problematic for Fricker's account as analogous. The shape of the overall account of the primary harm this yields is that of Nussbaum's cluster analysis of objectification; the primary harm of testimonial injustice is that it involves epistemically objectifying someone, where this in turn involves consistently treating them in a way that's an epistemic analogue of one of the ways of treating someone as something outlined by Nussbaum. In both the sexual and the epistemic cases, this might involve treating someone as if they were inert in the relevant respects; one might treat someone as a mere sexual object or as a mere object from which one can glean information via inference. But the cases suggest that it might not: that one can recognise someone as a sexual or epistemic agent, and still treat them like an object in a problematic way.Footnote ¹²

This suggestion invites the following worry.Footnote ¹³ Sexual objectification of the kind described in the previous paragraph will likely seem clearly morally problematic to most of us. It's much less obvious, however, that the epistemic analogues of such behaviour are morally problematic. I think there's something right about this objection, and it suggests that it would be a mistake to take the sexual and the epistemic to be too closely analogous here. In particular, the significance of consent seems different; treating someone as a mere means to satisfy my sexual ends without their consent seems morally problematic in a way that treating their utterances as mere fodder for my own inquiries without their consent often does not. However, there are plausibly still important parallels between the sexual and the epistemic. Let's return once more to our central examples of testimonial injustice, Tom Robinson and Marge Sherwood. Both are cases in which the other two contextual features that Nussbaum identifies as wrong-making are present. Both involve power-differentials; as Fricker writes of Sherwood, ‘[i]t is easy to see that Greenleaf's silencing of Marge here involves an exercise of power, and of gender power in particular’ (Reference Fricker2007: 9).Footnote ¹⁴ And, grounded as they are in systematic identity prejudices, both credibility deficits can plausibly be taken as manifestations of a general disposition to treat Sherwood and Robinson – and anyone who shares their social identities – as epistemically instrumental, with respect to certain matters at least. Neither Sherwood nor Robinson are treated as incapable of having their own epistemic ends with respect to the relevant issues but, due to their prejudices, their audiences are disposed to systematically treat them as mere instruments to their own ends when it comes to these issues, and that does seem to involve a problematic kind of dismissiveness.

That's the direction that I think we should take Fricker's Nussbaum-inspired account of the primary harm in. So developed, this kind of account has important aims and elements in common with the kind of account offered by Fricker's critics in terms of recognition theory. Both accounts are attempts to articulate ways in which someone can be treated as less than a full epistemic subject in their own right without failing to recognise them as an epistemic subject, in possession of epistemic agency and capacities. Which account is superior? I won't try to settle this question here, since I take it to be part of a much larger question about the theoretical resources that best enable us to explain the wrong involved in various kinds of oppression and injustice, and this larger question is not one I can engage with in any kind of depth here. However, the discussion offered in this paper does, I hope, justify two more modest conclusions. First, the case against Fricker's account of the primary harm in terms of epistemic objectification, where the latter is understood along the lines suggested by Nussbaum's analysis, is stronger than the recent literature suggests, since the cases that Fricker's account mishandles include her own central examples of testimonial injustice. However, Fricker's critics have been much too quick to conclude that a proposal along the lines Fricker suggests is untenable, and that we need to invoke the resources of recognition theory. As I have stressed, Nussbaum's analysis of objectification is richer and more flexible than either Fricker or her critics have acknowledged thus far, and epistemologists should fully explore and take advantage of its resources when thinking about ways in which people can be epistemically harmed.Footnote ¹⁵