Insult and Injustice in Epistemic Partiality

1 Introduction

Proponents of epistemic partiality in friendship argue that friendship makes demands of our epistemic lives that are at least inconsistent with the demands of epistemic propriety, and perhaps downright irrational. In this paper, I focus on the possibility that our commitments to our friends distort how we respond to testimony about them, their character, and their conduct. Sometimes friendship might require us to ignore (or substantially underweight) what others tell us about our friends. However, while this practice might help promote your friendship and protect your friends, refusing someone’s testimony can constitute a special kind of insult towards them, according to Allen Hazlett (2017) and Finlay Malcolm (2018). It seems to follow that friendship can require us to insult people by refusing their testimony. This should count against treating your friends with epistemic partiality, one might think. It’s plausible that we have moral reasons not to insult other people, whether they’re our friends or not. The risk of insulting non-friends, it turns out, is just the beginning of the problems for the epistemic partiality in friendship. Recent discussions in social epistemology, led by Miranda Fricker (2007), Kristie Dotson (2011) and José Medina (2012), among others, have brought renewed attention the distinctly epistemic nature of some kinds of injustice. Drawing on this work, I’ll suggest a principled way to determine when epistemic partiality in friendship is morally impermissible because of the genuine risk of harmful epistemic injustice.

The paper proceeds as follows. First, I’ll argue that treating our friends with epistemic partiality can involve insulting other people by refusing their testimony. Then I’ll consider whether we ever ought to commit more morally serious testimonial injustices in the name of friendship. After considering some putative cases, I’ll introduce the concept of testimonial injustice, referring primarily to the work of Dotson (2011). I’ll explain how and why the epistemic bias which I’ve prescribed in friendship could be thought to cause such injustice. Drawing on the work of Sheila Lintott (2015), I’ll consider some ways in which the extent of the testimonial injustice caused by bias in friendship might be worse yet, namely, because of the role that homophily plays in determining who we become (and remain) friends with. In particular, I’ll consider whether epistemic partiality perpetuates racist or sexist forms of testimonial injustice. Finally, I’ll attempt to alleviate some of these concerns and determine when we’re morally required to treat friends with epistemic partiality. I’ll argue that whilst protecting a friend’s well-being might require you to cause testimonial injustice, such injustice would be morally unjustified. Accordingly, if you commit such injustice in the name of friendship, you’re morally blameworthy for doing so. This, I suggest, imposes an additional limit on epistemic partiality in friendship.

2 Epistemic partialism: an overview

In this section, I’ll give a brief overview of the main positions regarding epistemic partiality in friendship. Now, this view has proven to be controversial, to say the least.¹ Since the subject matter of this essay is located downstream from the question of whether we in fact ought to treat our friends with epistemic partiality, I’ll assume for the sake of argument that some version of this position is true. In any case, we usually take it for granted that good friends have a rather more optimistic view of each other’s character and conduct, and give each other the benefit of the doubt when it comes to potential flaws. This seems like an unobjectionable part of the common sense of friendship. Epistemic partialism is the view that friendship makes genuine demands of our epistemic lives that are at least inconsistent with the demands of epistemic propriety, and perhaps downright irrational. Simon Keller (2004) and Sarah Stroud (2006) are credited with igniting philosophical interest in the epistemic demands of friendship. In the following paragraphs, I’ll offer a very brief overview of their respective positions. Keller’s view of epistemic partiality is relatively mild, and paints a picture of a moderate bias in favour of our friends:

[W]hen good friends form beliefs about each other, they sometimes respond to considerations that have to do with the needs and interests of their friends, not with aiming at the truth … The fact that a person is someone’s friend can sometimes explain why that person is inclined to believe certain sorts of falsehoods. (Keller 2004, 330)

For Keller, it seems, friendships sometimes involve a weaker epistemic bias, which can go some way to explain certain of our ‘sub-optimal’ epistemic performances, but which hardly debases us as epistemic agents. Rather, Keller suggests that good friends show a kind of leniency to their friends, in that they demand less evidence to believe something positive about them than they would of someone with whom they didn’t share a close, personal relationship. When looking on the bright side, so to speak, epistemic partiality may involve over-weighing evidence which supports a beneficial judgement about one’s friends. When giving the benefit of the doubt, epistemic partiality can involve ignoring or under-weighing evidence which would cast doubt on a positive judgement about a friend. At times, Stroud’s view seems rather more radical. She writes:

Friendship positively demands epistemic bias, understood as an epistemically unjustified departure from epistemic objectivity. … Or, to put the point as succinctly—and brutally—as possible, friendship requires epistemic irrationality. (Stroud 2006, 518)

On her view, there is a tension between the norms of friendship and the norms of epistemic propriety, such that we cannot always live up to the demands of friendship and the demands of epistemic propriety. To put it in a slogan, the view seems to be that being a good friend requires being a bad knower. Elsewhere, Stroud’s position seems less radical: she suggests that epistemic partiality is so much a matter of denying the facts, but rather exercising ‘interpretative charity’ when we think about the facts about our friends’ characters and conduct (Stroud 2006, 507). On this view, epistemic partiality isn’t always a matter of shutting out the ‘base-level facts’ about a friend’s character and conduct. It’s also possible to protect them by suspending judgement about the moral status of their actions, or by granting ad hoc exemptions from our normal moral standards.

Crucially, for this paper at least, it seems the fact that we’re friends with someone can make a substantial difference to how we respond to testimony about them. If epistemic partialism is true, then the fact that we’re friends with someone makes a difference to how we should respond to testimony about them too. Let’s consider an example proposed by Stroud, in which someone tells you that your friend, Sam, ‘recently slept with someone and then cruelly never returned any of that person’s calls, knowingly breaking that person’s heart’ (Stroud 2006, 504). In her discussion of this case, which I’ll return to throughout this paper, Stroud suggests that we show partiality in the way that we process the evidence we gain from such reports. Sam’s friends might require more evidence before they accept the allegation against him than they would otherwise, for instance, if Sam were not their friend. They might even withhold judgement even despite the evidence they possess. Of course, there may other options open to Sam’s friends. According to the accusation, Sam knowingly broke someone’s heart by failing to return their calls after a one-night stand. Even if his friends can’t deny that Sam ultimately caused this person’s heartbreak, you might suspend judgement about his blameworthiness for it. Sometimes we attempt to rationalise our epistemically partial judgements about our friends. Stroud continues:

Your reaction to this story might be, for instance, “There’s never any artifice with Sam. You know where you stand with him: if he doesn’t want to see you, he makes that clear. There’s no false politeness, no pussyfooting, no hypocrisy, no stringing you along—Sam’s too genuine for any of that.” In a similar vein, what other people might classify as compulsive womanizing on Sam’s part, you might see as irrepressible but fickle enthusiasm and appetite for female charm in all its many varieties. (Stroud 2006, 508.)

What’s so interesting about this element of epistemic partiality, that is, the way friendship can distort the way we interpret other people’s testimony, is that in these testimonial cases, unlike non-testimonial cases, there seems to a straightforward conflict between what we owe our what we owe out friends and what we owe speakers qua sources of knowledge. Moreover, the upshot of this conflict is not just that people might form unjustified beliefs about their friends, but also that they might insult the people whose testimony they seem to ignore when they do so.

2.1 Does epistemic partiality require us to insult non-friends?

Treating your friends with epistemic partiality can cause you to insult non-friends, in particular, when you refuse to accept what they tell you about your friends, or so I’ll argue you in this section. I’ll offer some considerations in support of the claim that Sam’s friends insult the speaker by refusing her testimony, particularly drawing on the work of Hazlett (2017) and Malcolm (2018). Discussions about Stroud’s Sam example (2006) tend to focus on what Sam’s friends owe him, epistemically speaking, and why. Now I’ll consider the case from the perspective of the third-party speaker who reported Sam’s misbehaviour to his friends, but whose testimony was rejected.

2.2 Insulting someone by refusing their testimony

Refusing someone’s testimony can insult them. To illustrate this point, Allen Hazlett (2017) relates the events of David Foster Wallace’s ‘Oblivion’ (2004). In ‘Oblivion’, Randall Napier reports ‘the strange and absurdly frustrating marital conflict between Hope and myself over the issue of my so-called “snoring”’ (Wallace 2004, 122). Hope tells Randall that he snores; Randall refuses to accept this and instead insists that she must be dreaming it. His refusal to accept her testimony insults her, and she’s insulted, because (from her perspective) his refusal expresses his judgement that cannot distinguish her dreams from reality. This drives Randall Napier into a kind of sceptical fury: it’s not just about snoring for him. It’s about his ability to perceive the external world. It’s unbearable for Randall that Hope casts doubt on his credibility with respect to his own consciousness. Randall understands that Hope is trying to tell him something, but he rejects it because, by his lights, he has stronger evidence against her testimony than she has in support of it. It damages Hope and Randall’s marriage when he refuses her testimony. An irony of this case is that, by the end, Hope and Randall are in very similar situations. They have both told each other what they believe about the ‘snoring situation’, and they both seem to be sincere. And they reject each other’s testimony. This example is amusing but it points to a deeper truth: it damages our relationships with people when we don’t accept their testimony.

Before we continue, we must clarify whether you really can insult someone by refusing their testimony. You insult someone by refusing their testimony when your refusal expresses you doubts about their credibility, at least according to Hazlett (2017, 42) and Malcolm (2018). Credibility, in this context, is understood as comprising the qualities of trustworthiness and competence.² In the epistemology of testimony, trustworthiness is typically understood to be the quality of consistently telling people what you take to be true, or justified, or knowledge. Competence is the quality of reliably coming to form true beliefs, or justified beliefs, or knowledge.³ This puts us in promising position to establish sufficient conditions for the special insult of refusing someone’s testimony: You insult a speaker by refusing her if and only if your refusal expresses your unreasonable doubts about the speaker’s credibility.

Let’s flesh this out. First, you insult someone by refusing their testimony if your refusal of their testimony expresses the fact that you doubt that they actually know what they take themselves to know. Refusing someone’s testimony in these conditions is an attack against their competence. Second, you insult a speaker by refusing their testimony if your refusal of their testimony expresses the fact that you doubt that they’re trustworthy. Refusing someone’s testimony in this way attacks their character. Of course, the insult of refusing someone’s testimony isn’t so simple. It’s not the case that you insult someone every time you refuse their testimony—after all, sometimes you have very good reasons to do so.

To understand Hazlett and Malcolm’s accounts of the special insult of refusing testimony, it’s important to have a clear understanding of the relevant notion of credibility. Malcolm turns to Fricker for this, in particular her claim that ‘knowledge […] is enshrined in the figure of the good informant’ (1998, 163). The good informant, on Fricker’s view, is someone who possesses both credibility and rational authority (Fricker 1998, 167). Credibility, as described above, comprises competence and trustworthiness. Competence is the property of being a reliable source of truth. Trustworthiness is the property of reliably sharing the truth, once you’ve got it. Good informants must be competent and trustworthy, of course, but that’s not enough. Their intended audiences also need to know that they’re competent and trustworthy. It’s no good being a credible source of valuable information if nobody recognises you as such. It’s in these terms that we can start to unravel the special insult of refusing someone’s testimony. First, we might refuse someone’s testimony because we judge that they’re not likely to know what they take themselves to know. When we refuse someone’s testimony for these reasons, this expresses a negative judgement about their competence. Malcolm writes:

The hearer implies that the speaker lacks competence in a fundamental area of human intellectual life. She implies that he has failed to attain knowledge with respect to his belief, and since the speaker takes himself to have knowledge—to have acquired his belief in a justified way—it is an insult to treat him as though he has not. To do so is to imply that he is incompetent as a knower with respect to this testimony. (Malcolm 2018, 54)

This is quite straightforward: in many aspects of life, it’s insulting to be wrongly deemed incompetent, and our epistemic lives are no exception. Imagine that a tourist visiting your hometown refuses your testimony when you give them directions to their hotel, because they doubt your sense of direction. If their refusal of your testimony expresses an unreasonable judgement that you’re incompetent, then their refusal of your testimony is insulting and their refusal insults you. Of course, it’s not obvious that we always insult people when we refuse their testimony, even if we reject it because we judge them to be incompetent. Possibly, it’s not an insult to refuse someone’s testimony for this reason if they really are incompetent, or your judgement that they are is reasonable. Indeed, this consideration is central to Hazlett’s view of the special insult of refusing someone’s testimony (2017).

Sometimes, though, you refuse someone’s testimony because you think that they’re untrustworthy. In situations like this, you judge that you can’t rely on the speaker to tell you the truth, even though you suspect it’s in their possession. Correspondingly, your refusal of someone’s testimony insults them if it expresses an unreasonably negative judgement about their trustworthiness. Malcolm writes:

If a hearer treats a speaker as though he is attempting to intentionally deceive her—as though he is lying with his testimony—she may insult him by implying that he is performing a morally objectionable action. As with other immoral acts, it is offensive to accuse someone of performing one. So, it would be offensive to suggest that someone is a thief, or abusive, or racist, or a liar, even if it turns out that she is none of these things. As such, when we reject a speaker’s testimony, and do so because we take the speaker to be lying, we can insult the speaker by implying that she has acted immorally. (2018, 56)

Here we have two, compatible explanations of how you can insult someone by refusing their testimony. Of course, as Malcolm points out, the reason you refuse their testimony are relevant. It can’t always be the case that refusing someone’s testimony is an insult.

There are many ways we can reasonably come to doubt someone’s credibility. Malcolm (2018) identifies three primary sources of reasons for refusing someone’s testimony, namely, defeaters, inductive reasons, and prejudgement reasons. I’ll begin by explaining Malcolm’s view that defeaters can give you a reason to doubt someone’s testimony and what this means for our understanding of the insult of refusing someone’s testimony (2018, 57-58). Sometimes you’re just in a better evidential position than the speaker. Their evidence for believing that p is outweighed by your evidence to the contrary. It’s epistemically rational to refuse their testimony in these conditions. If you have strong evidence for believing that not-p, then, when someone with weak evidence for believing that p tells you that p, you should refuse their testimony. In this situation, you possess rebutting defeaters for their testimony. Undermining defeaters are different. Rather than weighing against your interlocutor’s evidence, undermining defeaters weaken the evidential-support relation between their evidence and their belief. Whilst the speaker believes according to their evidence, you’re in possession of additional information, according to which their evidence offers diminished support their belief. When you possess defeaters of either kind for a speaker’s testimony that p which they don’t possess, it’s not so obvious that you insult them by refusing their testimony. Certainly, it would seem unreasonable on the part of the speaker if they felt offended by your refusal of their testimony, at least once they’d been informed that you possess overwhelming defeaters for the justification of their belief.⁴

Returning on Stroud’s example, we can ask what kinds of defeaters Sam’s friends might have for the third party’s testimony. One possibility is that they have rebutting defeaters for her testimony. Indeed, it’s tempting to think that, as his friends, they might be better positioned to ascertain the truth about their friend Sam’s character than the third-party speaker. If it’s reasonable for them to believe that Sam is a caring person, then this will go some way to rebut the allegations of the third party. We can imagine Sam’s friends’ reasoning:

• (1) If the third party’s testimony is correct, then Sam’s not a caring person.

• (2) But Sam is a caring person, and we should know; we’re his friends.

• (3) Therefore, the third party’s testimony is not correct.

For Sam’s friends, there are two possible explanations of the speaker’s utterance. Either the speaker is incompetent, or she is untrustworthy. With sufficiently strong prior evidence that Sam is a caring person, his friends are in a strong evidential position to refuse the allegations of the third party.

This way of rebutting criticism of our friends won’t always work. We can’t speak for every case, or even most cases, simply by pointing out that it’s possible that Sam’s friends could have legitimate rebutting defeaters for the speaker’s testimony. Indeed, granted a pessimistic view of human nature, we might think that it’s quite improbable that Sam’s friends have legitimate rebutting defeaters for the testimony of the third party. As Stroud herself writes (2006, 516), some of the force of the example of Sam’s misdemeanour come from the fact that people are quite prone to selfish or cruel behaviour in their sex lives. We cannot be sure that Sam’s friends will have strong enough evidence of Sam’s good character to the defeat the third party’s testimony, at least, not in most cases. Now, if that’s the case, and they refuse the speaker’s testimony on that basis, then then their refusal of the third party’s testimony is an insult to her.

Of course, it’s possible that you refuse someone’s testimony and thereby insult them, not because you’re unreasonably confident that you possess defeaters for their testimony, but because we have doubts about their credibility, perhaps because they’ve deceived or misled us in the past. Malcolm (2018, 58-60) argues that inductive reasons can give you grounds to doubt someone’s testimony. If a speaker has demonstrated incompetence in the past, then it’s epistemically rational to doubt their credibility. Likewise, if a speaker has demonstrated their untrustworthiness in the past, then it’s epistemically rational to doubt their credibility. Plausibly, then, what you believe about someone’s track record can give you good epistemic reasons to doubt their credibility. In reality, of course, things are a little more complicated. For one thing, it’s worth noting that you can at a given time have inductive reasons for doubting someone’s credibility without their having a bad track record. A speaker’s credibility is temporarily compromised if, for instance, they’re drunk, or they’re being blackmailed. The thought is that if you possess inductive reasons to doubt someone’s credibility, then your refusal of their testimony doesn’t insult them so strongly (on Malcolm’s view) or at all (on Hazlett’s view). For example, if you have good evidence that a speaker is an incompetent arithmetician, then you shouldn’t accept their testimony when they tell you how much you owe after you decide to split the bill for dinner. Likewise, Malcolm points out (2018, 59), this is why it’s not such a strong insult to refuse someone’s testimony in boy-who-cried-wolf cases. If you’ve got a good reason to suspect that someone’s lying to you, then it’s not a strong insult to be wary what they tell you, but rather, what we might call a slight (Malcolm 2018, 51). Relatedly, when you receive inconsistent testimony, for instance, when one person tells you that p and the other that not-p, you may find yourself in the unfortunate position of having to decide whom to insult. However, the extent of the insult may be mitigated somewhat by the fact that, other things being equal, the very fact that you have inconsistent testimony from testifiers that you judge to be equally reliable is evidence that you should reduce your confidence in both speakers’ credibility.

Now we should consider how this fits into the wider project of this paper. Is it plausible that Sam’s friends have inductive reasons to refuse the speaker’s testimony? And if so, when, if ever, does their response to these reasons insult the third-party speaker? Keeping with Stroud’s Sam example (2006), suppose that Sam’s friends unreasonably believe that the speaker lacks credibility. They don’t have sufficient inductive reasons to support their judgement. Under these conditions, if they refuse the speaker’s testimony, then they insult them. Given the theory of testimonial insult presented here, they insult them because their refusal of their testimony expresses an unreasonably critical view of their competence or trustworthiness.

Unfortunately, this isn’t the most promising way of understanding the insult of refusing someone’s testimony because of friendship. We can imagine that before hearing the third party’s judgement about Sam’s behaviour, his friends have a reasonable, positive judgement about the third party’s credibility. For them to insult the speaker by refusing their testimony about Sam’s behaviour for inductive reasons, they would have to have reached an unreasonable, negative judgement about their testimony on the basis of inductive reasons before she tells them what Sam did. Otherwise, it’s hard to see how inductive reasons play a part in their refusal of their testimony at all.

A better explanation can be found in the insight that prejudgement reasons lead us to doubt a speaker’s testimony. This account of the insult of refusing someone’s testimony best fits the cases we’ve discussed so far in this paper, and best explains the nature of the insult caused when someone’s testimony is refused for reasons of friendship. We often seem to rely on prejudgement reasons in our reasoning about other people’s testimony. In fact, it’s common enough that we only have prejudgement reasons to inform our reasoning. In support of this claim, Malcolm (2018, 60) quotes Fricker:

Barring a wealth of personal knowledge of the speaker as an individual [a hearer’s] judgement of [a speaker’s] credibility must reflect some kind of social generalization about the epistemic trustworthiness—the competence and sincerity—of people of the speaker’s social type, so that it is inevitable (and desirable) that the hearer should spontaneously avail himself of the relevant generalizations in the shorthand form of (reliable) stereotypes. (Fricker 2007, 32)

To refuse someone’s testimony for prejudgement reasons is to refuse someone’s testimony because of judgements you have about their credibility which are based on your understanding of their social identity. Malcolm carefully uses the term ‘prejudgement’ rather than prejudicial to emphasise the fact that not all cases in which a hearer has reasons to doubt someone’s testimony based on their social identity are cases of deep injustice (2018, 60). The speaker’s social identity might comprise an intersection of factors, including their gender, race, social class, and religion. I would suggest that we consider friendships as factors which contribute to a person’s social identity too. At very least, friendships reflect our social identity—they can and are tracked by other people as a means of making judgements about you.

2.3 Is epistemic partiality insulting to non-friends?

What does this tells us about the cases in which a speaker’s testimony is refused by the listener for reasons of friendship? Well, when Sam’s friends refuse the testimony of the third party, it’s plausible that they do so for prejudgement reasons. The speaker’s testimony is prejudged and, ultimately, refused because of her social position. The relevant aspect the speaker’s social position is that she a non-friend. When Sam’s friends identify that the speaker isn’t one of Sam’s friends, they judge that they are therefore not a credible source of some information about Sam. Indeed, the deficit in the credibility they attribute to the speaker is likely to be subject-specific: they pre-judge that non-friends are not credible sources of moral criticism of friends.

When the third party tells Sam’s friends that he slept with someone and knowingly broke their heart, his friends refuse their testimony. More specifically, it’s because they’re not Sam’s (or their) friend that they don’t accept criticism of Sam from them. ‘Since they’re not Sam’s friend, they simply aren’t in a position to know’, they might say. What’s more, they might reason that it’s neither her business to judge Sam’s behaviour (‘They’re not even his friend!’) nor to share their judgements with them (‘And they’re not our friend either!’). Indeed, armed with these assumptions about who does and who doesn’t have the right to form an opinion about Sam’s character and conduct, his friends might construe the very fact that they, a non-friend, have shared their information about Sam’s conduct as a reason not to believe them.

How, then, do Sam’s friends insult the third-party speaker when they refuse their testimony? Suppose that the strategies that Sam’s friends use to rationalise their prejudgement of her criticism are epistemically defective. Accordingly, the judgements they end up with are epistemically unreasonable. Suppose, as before, that the third party reasonably believes that Sam slept with someone and then knowingly broke her heart. What’s more, suppose that the third party is quite sincere in their effort to tell Sam’s friends about his behaviour. The third party, then, is not treated as a rational authority, even though they deserve to be. This is an insult.

It seems, then, that if epistemic partialism is correct, friendship can be a source of pro tanto moral reasons to insult non-friends by refusing their testimony, in particular, when the alternative, that is, accepting their testimony, would damage our friend’s well-being. In the section that follows, I’ll consider whether treating friends with epistemic partiality risks causing testimonial injustices. I’ll argue that it does, drawing on the work of Fricker and Kristie Dotson (2011). Against this theoretical background, I’ll consider and respond to an objection owed to Sheila Lintott (2015), who claims that epistemic partiality in friendship could be morally problematic for reasons because it risks perpetuating injustice.

2.4 Can showing our friends epistemic partiality perpetuate injustice?

Writing in The Guardian, Ijeoma Oluo (2015) explains how she empathises with those who find themselves unable to process the possibility that a close relation could be guilty of treating another with terrible cruelty. Oluo’s close relation, whom she refers to as Steve, was twice convicted of sexual assault. It took Oluo and some of her family members twelve years to accept the truth about Steve. She explains how they were able to maintain their belief in his innocence for so long:

When the victim showed up to court covered in cuts and bruises and her neck in a brace, Steve’s parents argued to us that it was a ruse: the bruises were makeup; the neck brace was for show. My mom, who hadn’t even been particularly close to Steve, believed all of their stories and justifications. She never said anything to vilify the victim, but she would say that something about it just “didn’t seem right.” (Ijeoma Oluo 2015)

It seems that by explaining away evidence through a kind of post hoc rationalisation, Steve’s close relations were able to cultivate the belief that he was innocent despite the public availability of overwhelming evidence.

When we’re presented with accusations about the wrongdoing of our friends and loved ones, it seems we demand a great deal of evidence before we believe the accusations against them. Steve’s close relations suspend judgement about the credibility of the victim’s testimony and deny the veracity of her injuries.⁵ It’s plausible that they wouldn’t have treated the victim’s testimony in this way, were Steve not a close, personal relation. Therefore, this case provokes serious questions about the moral permissibility of epistemic partiality. Other cases of morally objectionable epistemic partiality in friendship are conceivable. Thus, refusing to accept someone’s testimony can constitute a harm more serious than insult, particularly when the content of that testimony relates to serious moral misconduct or when there are vulnerable people involved.⁶

Let’s start with an example. Imagine that, in a private conversation, a non-friend (hereafter the speaker) accuses your friend of serious moral misconduct. Suppose that your friend is accused of repeatedly using a racial slur towards a third party. This is, of course, a morally grave version of Stroud’s Sam case. Let’s suppose, as before, that the speaker’s belief about Sam’s conduct is reasonable and that she has no reason to make false accusations. For the sake of argument, let’s suppose that it would unreasonable, at least for a neutral hearer, not to accept the speaker’s testimony in this case: The speaker’s testimony provides the neutral hearer with normative epistemic reasons to believe, and they don’t possess sufficient non-testimonial evidence to the effect that the speaker is unreliable or trustworthy.

But what should a friend do when presented with such a serious accusation of moral misconduct? In the case discussed above, borrowed from Stroud (2006), the harm that would be caused by incorrectly believing the accusation against your friend outweighs the harm the that would be caused by incorrectly rejecting it. Now it’s plausible that if you came to believe the false accusation that your friend repeatedly used a racial slur to abuse someone, you would harm your friend. It’s also plausible that if you believed that your friend had racially abused a third party, then your commitment to friendship with them would be undermined. That’s because, plausibly, learning that your friend is racist will cause you to feel what Ward Jones (2011) calls lover’s shame. Coming to feel ashamed of someone can cause you to withdraw your commitment to the relationship which you share with them, thereby frustrating your friendship with them. To put it simply, what this suggests is that, to the extent that your friendship with the accused is your priority, you should reject the accusation.⁷

It hardly needs stating that this conclusion is morally objectionable. It would be an unacceptable implication of epistemic partialism, that we ought to ignore accusations of serious moral misconduct against our friends. But it’s difficult to make sense of this, if we grant that beliefs (and omissions of belief) can harm, and we have some kind of special moral duty towards our friends.

Why is it so dangerous to refuse this kind of testimony? First, let’s work within the framework established in the first half of this paper, according to which you insult someone by refusing their testimony if and only if your refusal expresses unreasonable doubts about their credibility. When the speaker shares her testimony with someone in this case, it’s highly unlikely that the hearer will possess defeater-reasons, inductive reasons, or prejudgement reasons, to refuse her testimony. Since it’s unlikely that the hearer possesses good reasons to refuse the testimony of the speaker, it’s likely that the hearer insults the speaker by refusing her testimony. It’s worth considering the severity of the insult paid by the hearer in these circumstances. Arguably, it’s extremely insulting to refuse someone’s testimony in a case like this because of the gravity of the implication made by that refusal: It implies that the hearer judges that the speaker is either not competent enough to know what happened to her, or otherwise is insincere. It’s a very serious insult to judge unreasonably that a victim of racist abuse is lying about what happened to them. The insult is so much more serious because the judgement about their credibility is so much more critical. To doubt that someone is telling the truth about an incident of racist abuse is a serious insult because of how bad it would be to lie about such an event. To deny that someone is competent enough to know whether someone else has been the victim racist abuse is a serious insult, at least because of the high level of incompetence implied.

We can also consider the negative social and political consequences of refusing the third-person reports of racially or sexually aggravated violence or discrimination. It’s highly plausible that rejecting reports of racism can be extremely harmful, not only to the victim herself, but also to other victims. By refusing to accept a report of racism, you immediately deny the victim a vital opportunity for support. It’s also reported that many victims of racism are reluctant to report the abuse they have suffered, because they believe that their reports will not be accepted and that, if they are accepted, they will be misunderstood. (We’ll return to this topic in the next section.) The point is that even before we’ve delved very deeply into the ethics of bias, we should be reluctant to accept a theory of epistemic bias in friendship which would promote rejecting such testimony.

What I’ve tried to show in the previous section is that refusing morally important testimony silences voices that ought to be heard. It would be extremely problematic if friendship required this. The thing is, to the extent that your care for your friend is frustrated by your forming negative moral judgements about them, friendship does appear to require you to silence these voices. In the following section, I’ll explain in more detail how this claim fits into the theoretical framework of testimonial injustice.

2.5 Epistemic injustice and testimonial silencing

In the previous section I considered a case in which morally serious testimony was, in a sense to be explained now, silenced. The silencing of this testimony appears to be a result of the demand for epistemic partiality in friendship. We considered the possibility that the speaker would withhold her testimony as a result of her believing that it wouldn’t be accepted anyway. Both cases seem like cases of serious injustice. In the section that follows, I’ll explain why it’s correct to understand these as cases of a distinctly epistemic kind of injustice, namely, testimonial injustice.

In archetypal cases of testimonial injustice, a speaker who knows that p, testifies that p in a context where it is entirely reasonable for them to do so, but their testimony is rejected because on prejudices about the speaker’s social identity that are held by their audience. Testimonial injustice is thus caused by negative epistemic stereotyping. Fricker proposes a model of everyday conversational interaction to explain the nature of testimonial injustice, according to which, in ordinary, successful conversational interactions, both speakers and hearers rely on stereotypes and heuristics to make judgements about the trustworthiness and competence of their interlocuter. To be recognised as an accurate and trustworthy source of testimony, a speaker must be recognised as such by their audience; but these properties are opaque. Rather, we must rely on indicator properties, such as gender, race, social class, to determine, as best we can, whether a speaker is a competent and trustworthy source. The problem is that these stereotypes and heuristics are susceptible to distortion by pernicious (and sometimes violent) force of social power.

In testimonial injustice, we have seen, the speaker is granted the opportunity to speak, but their testimony is rejected (Fricker 2007, 26). Sometimes, however, a speaker doesn’t get as far as sharing their knowledge at all, but rather, they withhold it. We’ll follow Dotson in calling the phenomenon of the coerced withholding of testimony ‘smothering’. And what is smothering? A speaker’s testimony is smothered when she is illegitimately coerced into withholding her testimony. ‘Testimonial smothering’, Dotson explains, ‘is the truncating of one’s own testimony in order to insure that the testimony contains only content for which one’s audience demonstrates testimonial competence’ (2011, 244). In an archetypal case, someone who is the victim of smothering possesses the relevant intellectual virtues to justify (in some sense) their telling their audience that p (i.e., their belief that p is true and they are trustworthy), but they withhold their testimony because they are confident that their testimony will be refused or misunderstood anyway. For an instance of withheld testimony to count as testimonial smothering, according to Dotson’s account, three conditions must be met.

• (1) the content of the testimony must be unsafe and risky;

• (2) the audience must demonstrate testimonial incompetence with respect to the content of the testimony to the speaker; and

• (3) testimonial incompetence must follow from, or appear to follow from, pernicious ignorance. (Dotson 2011, 244)

It’s worth taking a moment to consider each of these conditions. On Dotson’s view, unsafe testimony ‘is testimony that an audience can easily fail to find fully intelligible’ (2011, 244). Risky testimony, for Dotson, ‘runs the risk of leading to the formation of false beliefs that can cause social, political, and/or material harm’ (2011, 244). As an example of unsafe and risky testimony, Dotson (2011, 244-5) gives the example of testimony about domestic violence in non-white communities. Referring to the work of Kimberlé Crenshaw (1991, 1256-1257), Dotson explains how, when women of colour consider speaking about domestic violence in African American communities, they often withhold their testimony because of the risk that what they say will be taken to justify harmful stereotypes about African Americans (2011, 245). The content of the testimony is unsafe because audiences are likely to misunderstand it, incorrectly taking reports of discrete instances of wrongdoing as evidence of more general behavioural trends. The content of the testimony is risky because, if the hearers of the testimony misunderstand it, they’re likely to form damaging false beliefs about African American men.

The second condition for testimonial smothering, on Dotson’s account (2011, 245), is that the hearer fails to show the speaker that she is a competent recipient of testimony. Testimonial competence involves some degree of proficiency in at least the following two skills: the ability to understand what you’re told, and the ability to recognise when you don’t (or you’re not likely to) understand what you’re told. Dotson (2011, 245) reflects on what it’s like to listen to a lecture on theoretical physics as a non-expert. The competent, non-expert can not only understand some of what she’s told in the lecture, but also recognise when she doesn’t understand. Testimonial smothering requires that the speaker withholds or truncates her testimony because of the hearer’s failure to demonstrate that she’s got what it takes to interpret the speaker’s testimony and to appreciate her own limitations.

A third condition for testimonial smothering is that the testimonial incompetence of the hearer must result from (or appear to follow from) pernicious, situated ignorance. What, then, is situated ignorance? Someone in a state of situated ignorance lacks knowledge as a result of their social positioning.

Situated ignorance, which follows from one’s social positioning, is a result of epistemic limitation that fosters a kind of epistemic distance between those not in possession of that limitation and those who do possess the limitation. (Dotson 2011, 248)

Epistemic distance between two or more people exists when there is a gap between their respective worldviews, such that they have different ways of seeing and understanding the world. Differences in race, gender, and social and economic status can all contribute to the growth of epistemic distance between persons or peoples, at least according to Dotson (2011, 248). According to Dotson’s account, testimonial smothering occurs only if the failure of the hearer to demonstrate testimonial competence results from situated ignorance.

To conclude this section, let’s consider how this relates to the cases we considered earlier. When the speaker tells Sam’s friend that Sam used a racial slur against them, the friend refused their testimony. Unless Sam’s friend has good epistemic reasons to think that the speaker is lying or mistaken, their refusal of this testimony has the appearance of an instance of testimonial injustice. Similarly, in cases where the speaker withholds their testimony from Sam’s friends because they’re confident that, as his friends, they won’t accept their testimony, we find the appearance of testimonial smothering. A piece of the puzzle is missing, however, if the epistemic bias showed by Sam’s friends is to count as testimonial injustice. Testimonial injustice and testimonial smothering both require pervasive ignorance or identity prejudice, but it isn’t immediately obvious that that’s what’s happening when we refuse someone’s testimony for reasons of friendship. To be clear, it isn’t obvious that it’s Sam’s friend’s racist prejudices that cause them to refuse the testimony of Sam’s accuser when they say that Sam used a racial slur. Nor is it clear that it’s Sam’s friend’s misogynistic worldview that causes them to refuse the testimony of the third party who accuses Sam of sexual assault. Although racism and misogyny undoubtedly lead to the silencing of victims, in the cases we’ve considered, another explanation for the refusal or withholding of testimony is available, namely, that Sam is the hearer’s friend. In the section that follows, I’ll make the case that, in fact, pervasive ignorance can account for the fact that we refuse testimony for such reasons. Then, I’ll conclude, it’s reasonable to think that there’s a serious moral problem with the conclusion that we should treat our friends with epistemic partiality.

2.6 Homophily, bias, and friendship

For friendship to present a serious risk of testimonial injustice, it must be the case that something about the nature of friendship increases the risk of pernicious ignorance or identity prejudice. This doesn’t mean that epistemic partiality logically implies epistemic injustice. The point is rather that, given the society we live in and the systemic injustices that pervade social and political life, epistemic partialism has a contingent (but not trivial) causal relationship with epistemic injustice. In fact, as Sheila Lintott explains, it’s surprisingly likely that friendship could perpetuate pernicious ignorance. To begin, Lintott (2015) draws attention to the fact of homophily in social relations. She writes:

The basic idea is that if I see another as a member of “my group,” whatever group that might be, I am likely to be positively biased toward and attracted to that person. Positive emotions are more strongly felt and commitment and loyalty are more active for “in-group” people. As a corollary, I am more or less indifferent to and sometimes negatively disposed toward those I perceive as not members of my group, as, in other words, members of an “out-group.” (2015, 321)

What Lintott describes here fits well with some intuitive claims about nature of commitment in friendship. Feeling a shared bond is essential for friendships to flourish (Telfer 1970). To be friends with someone requires more than just liking them; you must feel connected by some common appreciation of what’s good and bad, what’s right and wrong, and so on. The thing is, the conditions which create these bonds are liable to perpetuate prejudice based on social identity. Lintott continues:

Shared past experiences and interests definitely bond us: they can be shortcuts to friendship. But there is a downside to relying too heavily on innocent and obvious seeming connections, for they can be barely veiled shorthand for racial, ethnic, class, gender, or other similarities that may trigger various biases and prejudices. (2015, 324)

It’s plausible that a person’s social identity influences the friendships they form. If that’s correct, then, at least in many cases, it seems totally innocuous. Some groups of friends bond over a shared hobby. Their engagement in a shared activity provides that ‘sense of a bond’ which was important in Telfer’s account. This is what brings together friends who play sport together or attend the same book club. Lintott doesn’t pull her punches:

[W]e need to pause to ask what determines to whom we “already feel a special concern.” “Because he is my friend” may provide some moral justification for differential treatment while attempts to justify differential treatment “because he’s white,” “because she’s Catholic,” or “because he’s heterosexual,” patently fail. However, if a major reason a friendship has formed concerns social identity, then justifying partiality “because he is my friend” is not entirely distinct from justifying partiality “because he’s white.” (2015, 325)

What Lintott says about homophily and bias in friendship suggests that it’s not unreasonable after all to think that the preference we have for our friends’ interests over the interests of non-friends could lead to morally serious cases of testimonial injustice. Now I’ll suggest an explanation why.

Why, we might ask, would Sam’s friends refuse testimony which alleges that he racially abused a third party? It’s tempting to answer that they would refuse the testimony because Sam is their friend and leave the matter there. However, to echo Lintott’s words, we should ask ourselves to what extent the fact that Sam is their friend is explained by relevant facts about social identity, in this case, race. If the fact that Sam is the hearer’s friend is caused by the fact that Sam is white, then the quieting of the third party’s testimony might well be explained by social identity.

In the previous section, I sketched two examples of testimonial injustice, which I take to be both plausible and morally problematic. More to the point, these examples suggest that the requirement for epistemic partiality in friendship is morally problematic because it seems to demand that we engage in practices of silencing that risk propagating testimonial injustice, particularly when a given friendship is formed and maintained in ways (e.g., via homophily) that reinforce situated ignorance. Drawing on the work of Fricker (2007), Dotson (2011) and Lintott (2015), I’ve tried to provide a framework for understanding how the practices of epistemic partiality can be seen as causal factors for testimonial injustice. To the extent that your friendship with someone, and specifically the fact that you’re close to them, is a result of your own prejudices and/or morally pernicious ignorance, epistemic partiality in friendship risks perpetuating testimonial injustice. This is morally unacceptable, so we should carefully consider the extent to which epistemic partiality in friendship is morally permissible.

2.7 The problem: It’s morally impermissible to treat your friends with epistemic partiality

We find ourselves in a bind. On the one hand, it seems like the harm caused by insulting someone by refusing their testimony is morally excusable in some cases, in particular, when the harm done to the speaker is minimal. On the other hand, when your refusal to accept a speaker’s testimony for reasons of friendship causes epistemic injustice through testimonial silencing, it seems the harm is by no means morally excusable. Given this tension between our intuitions, it’s necessary to develop a principled way of determining why it’s morally permissible to pay minor insults for reasons of friendship, but not to cause testimonial injustice.

3 A response: It’s a matter of moral priorities

In this section, I’ll propose a way of determining when it’s morally permissible to treat your friends with epistemic partiality. I’ll argue that it’s a matter of moral priorities. Treating your friends with epistemic partiality is morally wrong when they stand accused of serious moral misconduct because to do so is to get your moral priorities wrong. This moral wrongness, I’ll argue, is a source of pro tanto moral reasons against treating your friends with epistemic partiality when they stand accused of serious misconduct. However, treating your friends with epistemic partiality isn’t morally wrong when they stand accused of minor moral misdemeanours.

There seems to be something morally objectionable about treating close friends with epistemic partiality when they’re accused of serious moral misconduct. For instance, it’s plausible that if your friend is accused of racist violence but you treat them with epistemic partiality, and thereby excuse them from blame, at least in your view, you’ve done something morally objectionable. I’ve already suggested that one reason why this is wrong is that it would perpetuate the injustice of testimonial silencing. But it doesn’t always seem morally objectionable to treat close friends with epistemic partiality. Sometimes, it’s true that treating our friends with epistemic partiality risks insulting, but to the extent that this insult harms them, it nevertheless seems excusable.

This is puzzling. How do we explain the wrongness of treating friends with epistemic partiality? It’s a matter of moral priorities, or so I’ll argue in the following section. First, let’s consider a different but closely related question: Why is it wrong to be friends with morally bad people? Jessica Isserow (2018) defends a moral-priorities account of the wrongness of friendship with morally bad people, which can be fruitfully applied to the problem under discussion in this paper.⁸ The central claim of Isserow’s view is the following: While it’s not always immoral to overlook a friend’s flaws, it is when your overlooking their flaws causes you to lose sight of your moral priorities or renege on your moral commitments. Isserow writes:

One could understand an individual who was willing to forgive a friend’s failure to recycle; for this is a fault in spite of which we could plausibly accept someone. But an individual who discounted a friend’s rampant racism would suggest to us that she could not care less about the values which tell against racism, or for the potential victims of racist attitudes. At the very least, she would suggest to us that she does not stand for (or is not standing up for) such values in the fullest sense. Her willingness to discount vices of this extreme sort would suggest that there are certain values to which she is not properly responsive. (2018, 3112)

On Isserow’s view, someone who is knowingly maintains their friendship with a morally bad person is improperly responsive to important moral values. It’s here, Isserow claims, that they exhibit the vice of moral complacency. Moral complacency, to be clear, is the moral vice of having your moral priorities out of order. Morally complacent friends excuse what shouldn’t be excused (Isserow 2018, 3112). Someone who remains friends with a known racist has their moral priorities out of order because, Isserow claims (2018, 3112), they judge the value of their friend’s redeeming features to outweigh the disvalue of their racism, found in the harmfulness of their very attitudes and in harmful actions that result from their attitudes.

This puts us in a better position to understand when it’s immoral to treat a friend with epistemic partiality. You exhibit moral complacency when you treat a friend with epistemic partiality with respect to an accusation of serious misconduct. What’s wrong with epistemic partiality in these cases is that you’re incorrectly allowing the value of your friendship to outweigh the disvalue of their racism. In other words, you’re giving your friendship too high a place in your moral priorities. The thought is that if you prioritise clearing their name rather than finding out the truth, you express a vicious lack of concern for the wrongness of racism. When it the question whether your friend is racist becomes salient, your priority should focus on establishing whether they really are racist, and not protecting their well-being.

Treating your friends with epistemic partiality when presented with accusations of serious moral misconduct is morally complacent. When the people harmed by your moral complacency are already vulnerable to identity prejudice and its epistemic consequences, there is a heightened risk of perpetuating testimonial injustice via silencing. This must be taken into consideration when considering whether we should treat friends with epistemic partiality. Now, if we take moral reasons to be a species of pragmatic reasons, then, when it would be morally complacent to treat a friend with epistemic partiality, you have moral, pragmatic reasons not to treat your friend with epistemic partiality when they’re accused of serious moral misconduct. In practice, of course, this will likely turn out to be more easily said than done. You might harm your friend by refusing to give them the benefit of the doubt when they face such morally serious accusations. For instance, they may be wounded by your lack of loyalty. However, when the question whether they’re racist becomes salient, it’s reasonable to conclude that there’s more to be lost than there is to be gained from this friendship. In summary, it’s possible that the moral considerations against treating a friend with epistemic partiality outweigh the considerations in favour of it. Under these conditions, you should not treat your friend with epistemic partiality.

We might question the gravity of moral complacency. Isserow claims we don’t typically rank moral complacency among the most serious vices (2018, 3113). I think that the seriousness of moral complacency depends on the seriousness of the wrongdoing that is excused as a part of that complacency. The more morally abhorrent the accusations against your friend, the worse it is, morally speaking, to treat them with epistemic partiality. Let’s reflect on Carol Reed’s The Third Man (1949). At first, Martins thinks that Lime is just another petty criminal selling bootlegged cigarettes. In response, he’s disappointed in his friend, but he gets over it. And although it’s morally complacent of Martins to excuse Lime’s behaviour, I don’t think we’d expect him to break off his friendship. When Martins only suspects Lime of bootlegging, the cost of moral complacency is outweighed by the value of their friendship. It’s noteworthy that Martins’ friendship with Lime is damaged irreparably when he finally grasps the severity of the accusation against Lime. Lime is suspected of stealing penicillin, diluting it, and selling it back to hospitals. Many people die because of his crimes. When Martins is given a tour of a children’s hospital where many of Lime’s victims are apparently bound to lie in agony until their deaths, he cannot be mistaken any longer—the veil is lifted. At this point, he slackens his defensive epistemic standards and comes to accept the allegations against Lime.

Let’s consider Stroud’s original Sam example (2006), in which he’s accused of cruelly breaking someone’s heart. When Sam’s friends treat Sam with epistemic partiality, it’s true that they risk insulting the speaker. But plausibly, the harm caused by this insult is mild. It shouldn’t occupy a very high place in his friends’ moral priorities. The thought is that Sam’s well-being should occupy a higher place in their moral priorities than the feelings of non-friends.⁹ It’s plausible that our friends are more vulnerable to our harmful beliefs and judgements than non-friends, other things being equal. At least, then, when we must adjudicate between the testimony of a friend and the testimony of a non-friend on a non-serious matter, it’s plausible that our friend’s well-being is at greater risk, that is, greater risk of being harmed and at risk of greater harm. Crucially, the harm caused by their treating Sam with epistemic partiality is negligible in comparison with the harm of treating someone accused of racial violence with epistemic partiality. This should occupy a much higher place in Sam’s friends’ moral priorities. In summary, you shouldn’t treat friends with epistemic partiality when they’re accused of serious moral misconduct. To do so would be to risk causing serious epistemic injustice in the form of testimonial silencing. Preventing—or at least, not personally causing—epistemic injustice of this kind should be high among your moral priorities.

4 Summary

In this paper, I’ve discussed a moral objection to treating your friends with epistemic partiality. Treating your friends with epistemic partiality can involve insulting non-friends. In particular, you risk insulting non-friends when you refuse their testimony out of partiality to a friend’s well-being. But sometimes, treating your friends with epistemic partiality can perpetuate serious injustice. If a friend is accused of racially or sexually aggravated violence or discrimination, but you refuse the accusation out of loyalty to your friend, then, I argued, it’s plausible that you contribute to the testimonial silencing of the victims of racism or sexism. Our intuitions differ about each of these cases. On the one hand, insulting non-friends is bad, but it doesn’t seem right that the risk of insulting a non-friend outweighs the risk of harming a friend. We should treat your friends with epistemic partiality in these circumstances. On the other hand, the risk of perpetuating epistemic injustice seems to outweigh the risk of harming a friend. We shouldn’t treat our friends with epistemic partiality in these cases. How do we explain this difference in our intuitions? It’s a matter of moral priorities, or so I argued. It’s morally permissible that your friends’ well-being should matter more to you than the well-being of a non-friend, especially when your friend is at greater risk of harm. But when protecting your friend’s well-being requires you to risk perpetuating serious epistemic injustice, it’s not morally permissible for you to prioritise your friends’ well-being. This suggests that proponents of epistemic partialism ought to accept an additional limit on epistemic partiality in friendship, namely, one that’s imposed by morality.¹⁰