Jump to ContentJump to Main Navigation
Epistemic InjusticePower and the Ethics of Knowing$

Miranda Fricker

Print publication date: 2007

Print ISBN-13: 9780198237907

Published to Oxford Scholarship Online: September 2007

DOI: 10.1093/acprof:oso/9780198237907.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy).date: 24 May 2019

Testimonial Injustice

Testimonial Injustice

(p.9) 1 Testimonial Injustice
Epistemic Injustice

Miranda Fricker (Contributor Webpage)

Oxford University Press

Abstract and Keywords

This chapter formulates a working definition of social power, and identifies and defines a sub-type — identity power. The first kind of epistemic injustice is explored: testimonial injustice, wherein a speaker receives an unfair deficit of credibility from a hearer owing to prejudice on the hearer's part. A distinction between systematic and incidental testimonial injustice is explained.

Keywords:   social power, identity power, credibility, testimony

In Anthony Minghella's screenplay of The Talented Mr Ripley, Herbert Greenleaf uses a familiar put‐down to silence Marge Sherwood, the young woman who, but for the sinister disappearance of his son, Dickie, was soon to have become his daughter‐in‐law: ‘Marge, there's female intuition, and then there are facts.’1 Greenleaf is responding to Marge's expressed suspicion that Tom Ripley—a supposed friend of Dickie and Marge, who has curried much favour with Greenleaf senior—is in fact Dickie's murderer. It is easy to see that Greenleaf's silencing of Marge here involves an exercise of power, and of gender power in particular. But what do we mean by power? And how does gender power relate to the general notion of social power? In order to paint a portrait of testimonial injustice and to home in on its distinctive central case, we need to answer these questions about the nature of social power in general and the particular kind of social power (of which gender power is one instance) that I shall call identity power.

1.1 Power

Let us begin from what I take to be the strongly intuitive idea that social power is a capacity we have as social agents to influence how things go in the social world. A first point to make is that power can operate actively or passively. Consider, for example, the power that a traffic warden has over drivers, which consists in the fact that she can fine them for a parking offence. Sometimes this power operates actively, as it does when she actually imposes a fine. But it is crucial that it also operates passively, as it does whenever her ability to impose such a fine influences a person's parking behaviour. There is a relation of dependence between active and (p.10) passive modes of power, for its passive operation will tend to dwindle with the dwindling of its active operation: unless a certain number of parking fines are actively doled out, the power of traffic wardens passively to influence our parking behaviour will also fade. A second point is that, since power is a capacity, and a capacity persists through periods when it is not being realized, power exists even while it is not being realized in action. Consider our traffic warden again. If a driver, in a crazy state of urban denial, pays no heed one afternoon to what traffic wardens can do, parking wantonly on red lines and double yellow lines entirely without constraint, then we have a situation in which the traffic warden's power is (pro tem) quite inoperative—it is idling. But it still exists. This should be an unproblematic metaphysical point, but it is admittedly not without dissenters, for Foucault famously claims that ‘Power exists only when it is put into action’.2 We should, however, reject the claim, because it is incompatible with power's being a capacity, and because even in the context of Foucault's interests, the idea that power is not a capacity but rather pops in and out of existence as and when it is actually operative lacks motivation. The nearby Foucauldian commitment to a metaphysically light conception of power, and the idea that power operates in a socially disseminated, ‘net‐like’ manner do not depend on it, as we shall see.

So far, we have been considering power as a capacity on the part of social agents (individuals, groups, or institutions) exercised in respect of other social agents. This sort of power is often called ‘dyadic’, because it relates one party who is exercising power to another party whose actions are duly influenced. But since it might equally be pictured as influencing many parties (the traffic warden's power as constraining all drivers in the area), I shall focus on what is essential: namely, that this sort of power is exercised by an agent. So let us call it agential power. By contrast, power can also operate purely structurally, so that there is no particular agent exercising it. Consider, for instance, the case where a given social group is informally disenfranchised in the sense that, for whatever complex social reasons, they tend not to vote. No social agent or agency in particular is excluding them from the democratic process, yet they are excluded, and their exclusion marks an operation of social power. It seems in such a case that the power influencing their behaviour is so (p.11) thoroughly dispersed through the social system that we should think of it as lacking a subject. Foucault's work presents historical examples of power operating in purely structural mode. When he describes the kind of power at work in historical shifts of institutionalized discursive and imaginative habits—as when a practice of categorizing certain criminals as ‘delinquents’ emerges as part of a professionalized medical‐legal discourse3—he illustrates some of the ways in which power can operate purely structurally. These sorts of changes come about as the result of a system of power relations operating holistically, and are not helpfully explained in terms of particular agents' (persons' or institutions') possession or non‐possession of power. Further, in purely structural operations of power, it is entirely appropriate to conceive of people as functioning more as the ‘vehicles’4 of power than as its paired subjects and objects, for in such cases the capacity that is social power operates without a subject—the capacity is disseminated throughout the social system. Let us say, then, that there are agential operations of social power exercised (actively or passively) by one or more social agents on one or more other social agents; and there are operations of power that are purely structural and, so to speak, subjectless.

Even in agential operations of power, however, power is already a structural phenomenon, for power is always dependent on practical co‐ordination with other social agents. As Thomas Wartenberg has argued, (what he calls) dyadic power relationships are dependent upon co‐ordination with ‘social others’, and are in that sense ‘socially situated’.5 The point that power is socially situated might be made in a quite general way as a matter of the importance of social context taken as a whole: any operation of power is dependent upon the context of a functioning social world—shared institutions, shared meanings, shared (p.12) expectations, and so on. But Wartenberg's point is more specific than that, since he argues that any given power relationship will also have a more significant, direct dependence on co‐ordination with the actions of some social others in particular. He presents the example of the power that a university teacher has over her students in grading their work. This power is of course broadly dependent upon the whole social context of university institutions and systems of grading, and so on. But it is also more directly dependent upon co‐ordination with the actions of a narrow class of social others: for instance, the potential employers who take notice of grades. Without this co‐ordination with the actions of a specific group of other social agents, the actions of the teacher would have no influence upon the behaviour of the students, for her gradings would have no bearing on their prospects. Co‐ordination of that more specific kind constitutes the requisite social ‘alignment’ on which any given power relation directly depends. Or rather, the social alignment is partly constitutive of the power relation.

Wartenberg's point is clearly right. It also helps one see what is right about the Foucauldian idea that power is to be understood as a socially disseminated ‘net‐like organisation’—even while it may equally lead one to reject as a piece of exaggeration his claim that power is ‘never in anybody's hands’.6 The individual teacher indeed possesses the power to grade the student; but her power is directly dependent upon practical co‐ordination with a range of social others. She possesses her power, if you like, in virtue of her place in the broader network of power relations. Now, the mere idea of such practical co‐ordination is thoroughly generic, applying to the power required to get anything at all done in the social world—my power to cash a cheque is dependent on practical co‐ordination with the cashier at the bank and a range of other social agents. But we are trying to establish a conception of something called ‘social power’, which is on anybody's reckoning more specific than the mere notion of ‘social ability’ (such as is involved in my cashing a cheque). What, then, is distinctive of social power? The classical response to this question is to say that power involves the thwarting of someone's objective interests.7 But this seems an unduly narrow and (p.13) negative conception of power, for there are many operations of power that do not go against anyone's interests—in grading their work the university teacher need not thwart her students' interests. Wartenberg's response to the question is to say that what makes the teacher's ability to grade her students' work a matter of social power is that the student encounters it ‘as having control over certain things that she might either need or desire’.8

This way of putting it is appropriate for many agential relations of power; but the present aim is to establish a working conception of social power that is sufficiently broad to cover not only agential but also purely structural operations of power, and Wartenberg's idea of social alignment is not designed to do this. However, I believe that there is such a conception available, and that the notion of control, in slightly more generic guise, remains essential. The fundamental feature of social power that Wartenberg's notion of social alignment reflects is that the point of any operation of social power is to effect social control, whether it is a matter of particular agents controlling what other agents do or of people's actions being controlled purely structurally. In agential relations of power, one party controls the actions of another party or parties. In purely structural operations of power, though the power has no subject, it always has an object whose actions are being controlled—the disenfranchised group in our example of informal disenfranchisement, the ‘delinquents’, of Foucault's Discipline and Punish. In such cases there is always a social group that is properly described as being controlled, even while that control has no particular agent behind it, for purely structural operations of power are always such as to create or preserve a given social order. With the birth of the ‘delinquent’, a certain subject position is created as the subject matter for a certain professionalized theoretical discourse; with the disenfranchisement of a given social group, the interests of that group become politically expendable.

Putting all this together, I propose the following working conception of social power:

a practically socially situated capacity to control others' actions, where this capacity may be exercised (actively or passively) by particular social agents, or alternatively, it may operate purely structurally.

Although we tend to use the notion of social power as a protest concept—on the whole, we cry power only when we want to object—the (p.14) proposed conception reflects the fact that the very idea of social power is in itself more neutral than this, though it is never so neutral as the mere idea of social ability. It is right, then, to allow that an exercise of power need not be bad for anyone. On the other hand, placing the notion of control at its centre lends the appropriate critical inflection: wherever power is at work, we should be ready to ask who or what is controlling whom, and why.

1.2 Identity Power

So far the kind of social co‐ordination considered has been a matter of purely practical co‐ordination, for it is simply a matter of co‐ordination with others' actions. But there is at least one form of social power which requires not only practical social co‐ordination but also an imaginative social co‐ordination. There can be operations of power which are dependent upon agents having shared conceptions of social identity—conceptions alive in the collective social imagination that govern, for instance, what it is or means to be a woman or a man, or what it is or means to be gay or straight, young or old, and so on. Whenever there is an operation of power that depends in some significant degree upon such shared imaginative conceptions of social identity, then identity power is at work. Gender is one arena of identity power, and, like social power more generally, identity power can be exercised actively or passively. An exercise of gender identity power is active when, for instance, a man makes (possibly unintended) use of his identity as a man to influence a woman's actions—for example, to make her defer to his word. He might, for instance, patronize her and get away with it in virtue of the fact that he is a man and she is a woman: ‘Marge, there's female intuition, and then there are facts’—as Greenleaf says to Marge in The Talented Mr Ripley.9 He silences her suspicions of the murderous Ripley by exercising identity power, the identity power he inevitably has as a man over her as a woman. Even a flagrant active use of identity power such as this can be unwitting—the story is set in the Fifties, and Greenleaf is ingenuously trying to persuade Marge to take what he regards as a more objective view of the situation, a situation which he correctly sees as highly stressful and emotionally charged for her. He may not be aware that he is using gender to silence Marge, and (p.15) what he does is perhaps well‐intentioned and benevolently paternal. But it is no less an exercise of identity power.

Greenleaf's exercise of identity power here is active, in that he performs an action which achieves the thing he has the power to do: silence Marge. He pulls it off by effectively invoking a collective conception of femininity as insufficiently rational because excessively intuitive.10 But in another social setting a man might not need to do anything to silence her. She might already be silenced by the mere fact that he is a man and she a woman. Imagine a social context in which it is part of the construction of gender not merely that women are more intuitive than rational, but, further, that they should never pitch their word against that of a man. In that sort of social situation, a Herbert Greenleaf would have exercised the same power over a Marge—his power as a man to silence her as a woman—but passively. He would have done it, so to speak, just by being a man. Whether an operation of identity power is active or passive, it depends very directly on imaginative social co‐ordination: both parties must share in the relevant collective conceptions of what it is to be a man and what it is to be a woman, where such conceptions amount to stereotypes (which may or may not be distorting ones) about men's and women's respective authority on this or that sort of subject matter. Note that the operation of identity power does not require that either party consciously accept the stereotype as truthful. If we were to interpret Marge as thoroughly aware of the distorting nature of the stereotype used to silence her, it would still be no surprise that she should be silenced by it. The conceptions of different social identities that are activated in operations of identity power need not be held at the level of belief in either subject or object, for the primary modus operandi of identity power is at the level of the collective social imagination. Consequently, it can control our actions even despite our beliefs.

Identity power typically operates in conjunction with other forms of social power. Consider a social order in which a rigid class system imposes an asymmetrical code of practical and discursive conduct on members of different classes, so that, for instance, once upon a time (not so long ago) an English ‘gentleman’ might have accused a ‘member (p.16) of the working classes’ of ‘impudence’, or ‘insolence’, or ‘cheek’, if he spoke to him in a familiar a manner. In such a society the gentleman might exercise a plain material power over the man by, say, having him sacked (maybe he was a tradesman from a company that needed the gentleman's patronage); but this might be backed up and imaginatively justified by the operation of identity power (the social conception of him as a gentleman and the other as a common tradesman is part of what explains his capacity to avenge the other's ‘impudence’). The gentleman's identity carries with it a set of assumptions about how gentlemen are to be treated by different social types, and in virtue of these normative trappings the mere identity category ‘gentleman’ can reinforce the exercise of more material forms of social power. The identity power itself, however, is something non‐material—something wholly discursive or imaginative, for it operates at the level of shared conceptions of what it is to be a gentleman and what it is to be a commoner, the level of imagined social identity. Thus identity power is only one facet of social identity categories pertaining to, say, class or gender, since such categories will have material implications as well as imaginative aspects.

Could there be a purely structural operation of identity power? There could; indeed, identity power often takes purely structural form. To take up our disenfranchisement example again, we can imagine an informally disenfranchised group, whose tendency not to vote arises from the fact that their collectively imagined social identity is such that they are not the sort of people who go in for political thinking and discussion. ‘People like us aren't political’; and so they do not vote. Conversely, we can imagine that among those groups that do vote, identity power plays its part here too. Part of what encourages many of us to vote is a social self‐conception in the collective imagination such that ‘People like us are politically engaged’. Identity power, like social power in general, may be agential or purely structural; it may work positively to produce action or negatively to constrain it; and it may work in the interests of the agent whose actions are so controlled, or again it may work against them.

The reason for our particular interest in identity power is that we shall be concerned with how it is involved in the sort of discursive exchange in which knowledge can be imparted from speaker to hearer—in the broadest sense, testimonial exchange. I shall argue that identity power is an integral part of the mechanism of testimonial exchange, because of the need for hearers to use social stereotypes as heuristics in their (p.17) spontaneous assessments of their interlocutor's credibility. This use of stereotypes may be entirely proper, or it may be misleading, depending on the stereotype. Notably, if the stereotype embodies a prejudice that works against the speaker, then two things follow: there is an epistemic dysfunction in the exchange—the hearer makes an unduly deflated judgement of the speaker's credibility, perhaps missing out on knowledge as a result; and the hearer does something ethically bad—the speaker is wrongfully undermined in her capacity as a knower. I now turn to the exploration of this dual epistemic and ethical dysfunction. The task is to home in on what is perhaps the most ethically and socially significant moment of identity power's impact on our discursive and epistemic relations, and to paint a portrait of the distinctive injustice that it entails: testimonial injustice.

1.3 The Central Case of Testimonial Injustice

Broadly speaking, prejudicial dysfunction in testimonial practice can be of two kinds. Either the prejudice results in the speaker's receiving more credibility than she otherwise would have—a credibility excess—or it results in her receiving less credibility than she otherwise would have—a credibility deficit. Consider the immediate discursive impact of a speaker's accent, for instance. Not only does accent carry a social charge that affects how a hearer perceives a speaker (it may indicate a certain educational/class/regional background), but very often it also carries an epistemic charge. Accent can have a significant impact on how much credibility a hearer affords a speaker, especially in a one‐off exchange. I do not mean that someone's accent is especially likely to lead a hearer, even an intensely prejudiced one, automatically to reject outright some manifestly believable assertion or, conversely, to firmly believe some otherwise incredible assertion. No doubt these things are possible, but given that for the most part it is generally in the interests of hearers to believe what is true and not believe what is false, it would be a strong prejudice in an unusual context that would be single‐handedly powerful enough to have that sort of effect. The idea is rather that prejudice will tend surreptitiously to inflate or deflate the credibility afforded the speaker, and sometimes this will be sufficient to cross the threshold for belief or acceptance so that the hearer's prejudice causes him to miss out on a piece of knowledge.

(p.18) In face‐to‐face testimonial exchanges the hearer must make some attribution of credibility regarding the speaker.11 Such attributions are surely governed by no precise science, but clearly there can be error in the direction of excess or deficit.12 On the whole, excess will tend to be advantageous, and deficit disadvantageous. As a qualification, however, we should note that in localized contexts excess could bring disadvantage in its wake, and deficit could conceivably bring advantage. With regard to the former, consider an overburdened GP whose patients ask him medical questions that call for a more specialist training. He is not in a position to answer them fully responsibly; yet he must do his best to answer them, since the patients need an answer, and he is the only source they have access to. His patients assume that he is in a position to provide the information they need, and thus they attribute to him an excess of credibility on the matters in question. Let us add that any attempts to disabuse them of their inflated view of his expertise would damage the doctor–patient relationship by unduly undermining their confidence in him. All this is an ethical burden for our GP, because he is aware that his best advice might yet mislead them about an important health issue. For this GP, the credibility excess he receives from his patients brings an unwanted ethical burden, and so we see that credibility excess can be disadvantageous.13 Alternatively, consider the example of a professor who gives a more junior colleague some work for comments and who is relying on that colleague's critical feedback to get the thing straight before a conference presentation. If the junior colleague is an admirer and gives too much benefit of the doubt, then his comments will be less critical than they might otherwise be, and the professor is effectively let down. Again, the credibility excess she receives on this occasion is only (p.19) a disadvantage to her. In such circumstances as these, then, credibility excess can be disadvantageous, though on the whole it is surely more usually an advantage.

What of the possibility that credibility deficit can in unusual circumstances be an advantage? Consider the stuttering Claudius, destined one day to be emperor of Rome, but who repeatedly escapes political murder on the way up owing to the fact that he is generally taken to be a fool. Or alternatively, recall that inimitable character from Seventies TV crime detection, Lieutenant Columbo, whose bumbling and shambolic style lures those he is investigating into a false sense of security and enables him to quiz them off‐guard. Credibility deficit, then, in such specific and localized contexts, can be advantageous. In general, however, we shall see that credibility is a good that one needs to get enough of for all manner of well‐functioning, and accordingly we should think of its deficit as generally disadvantageous.

On the face of it, one might think that both credibility deficit and credibility excess are cases of testimonial injustice. Certainly there is a sense of ‘injustice’ that might naturally and quite properly be applied to cases of credibility excess, as when one might complain at the injustice of someone receiving unduly high credibility in what he said just because he spoke with a certain accent.14 At a stretch, this could be cast as a case of injustice as distributive unfairness—someone has got more than his fair share of a good—but that would be straining the idiom, for credibility is not a good that belongs with the distributive model of justice. Unlike those goods that are fruitfully dealt with along distributive lines (such as wealth or health care), there is no puzzle about the fair distribution of credibility, for credibility is a concept that wears its proper distribution on its sleeve. Epistemological nuance aside, the hearer's obligation is obvious: she must match the level of credibility she attributes to her interlocutor to the evidence that he is offering the truth. Further, those goods best suited to the distributive model are so suited principally because they are finite and at least potentially in short supply. (Recall Hume on the genealogy of justice: a situation of plenty is not one in (p.20) which the distributive concept will naturally arise.15) Such goods are those for which there is, or may soon be, a certain competition, and that is what gives rise to the ethical puzzle about the justice of this or that particular distribution. By contrast, credibility is not generally finite in this way, and so there is no analogous competitive demand to invite the distributive treatment.

Accordingly, in cases of credibility deficit, the injustice we are aiming to track down is not to be characterized as non‐receipt of one's fair share of a good (credibility), as this would fail to capture the distinctive respect in which the speaker is wronged. The idea is to explore testimonial injustice as a distinctively epistemic injustice, as a kind of injustice in which someone is wronged specifically in her capacity as a knower. Clearly credibility deficit can constitute such a wrong, but while credibility excess may (unusually) be disadvantageous in various ways, it does not undermine, insult, or otherwise withhold a proper respect for the speaker qua subject of knowledge; so in itself it does her no epistemic injustice, and a fortiori no testimonial injustice. On the contrary, our imagined professor and GP are overly esteemed in their capacity as knowers.

Yet, could it be (we should press the question) that there are some circumstances in which being overly esteemed in one's capacity as a knower would do one harm of a sort that merits the label ‘testimonial injustice’? Suppose we imagine someone growing up who, because of various social prejudices overwhelmingly in his favour, is constantly epistemically puffed up by the people around him. Let's say that he is a member of a ruling elite, and that his education and entire upbringing are subtly geared to installing this message firmly in his psychology. Perhaps the pupils who attend his school even wind up with a distinctive accent and certainly a confident air that helps mark them out as epistemically authoritative. No doubt the credibility excess he tends to receive from most interlocutors in his class‐ridden society will be advantageous: it is very likely to bring him lucrative employment and a certain automatic high status in many of his discursive exchanges, and so on. But what if all this also causes him to develop such an epistemic arrogance that a range of epistemic virtues are put out of his reach, rendering him closed‐minded, dogmatic, blithely impervious to criticism, and so on? Is it not the case that such a person has in some degree quite literally been made a fool of? And if so, is there not something to the idea that (p.21) the catalogue of credibility excesses that have malformed his epistemic character amounts to some sort of testimonial injustice? Is he not, after all, precisely wronged in his capacity as a knower? I think the answer is probably Yes, and we are perhaps confronted with an interesting special case of testimonial injustice. Note, however, that it is cumulative, whereas our focus has been on token cases of the injustice. I do not think it would be right to characterize any of the individual moments of credibility excess that such a person receives as in itself an instance of testimonial injustice, since none of them wrongs him sufficiently in itself. It is only if enough of them come together in the semi‐fanciful manner described that each moment of credibility excess takes on the aspect of something that contributes to the subject's being epistemically wronged over the long term. Consequently, I would suggest that while the example does indicate that some people in a consistently privileged position of social power might be subject to a variant strain of testimonial injustice: namely, testimonial injustice in its strictly cumulative form; none the less it does not show that any token cases of credibility excess constitute a testimonial injustice. The primary characterization of testimonial injustice, then, remains such that it is a matter of credibility deficit and not credibility excess.

Let us begin to home in on the concept of testimonial injustice, now duly conceived as a form of credibility deficit. A first point to notice is that prejudice is not the only thing that can cause credibility deficit, and so not all sorts of credibility deficit are cases of testimonial injustice. A credibility deficit might simply result from innocent error: error that is both ethically and epistemically non‐culpable. One reason why there will always be cases of innocent error is that human judgement is fallible, and so it is inevitable that even the most skilled and perceptive hearers will on occasion come up with a mistaken judgement of a speaker's credibility. More specifically, a hearer may simply have a false belief about the speaker's level of expertise and/or motives, so that she gives him less credibility than she might otherwise have done. So long as her false belief is itself ethically and epistemically non‐culpable (it does not, for example, result from an immoral hatefulness or from epistemic carelessness), there will be nothing culpable in her misjudgement of his credibility. It is simply an unlucky epistemic mistake of one or another familiar kind.

Consider an example in which the hearer—let us say that she is a philosopher, an ethicist—knows that her interlocutor is an academic at a certain institution, and having looked him up on the web she (p.22) believes him to be a medic, since his name was listed under medical sciences. When the conversation turns to a certain current debate in the literature pertaining to her own specialism, moral fictionalism, and to her surprise he expresses a forthright critical view on the fictionalist approach, she affords his word a lower credibility than she would if she took him for a fellow ethicist. In fact, however, unbeknownst to her, he is an ethicist, with a specialism in medical ethics, employed in a medical department, and so her false belief about his professional identity has put him in credibility deficit for the duration. Yet I would suggest that her misjudgement does him no real testimonial injustice. It is simply an innocent error. An unlucky mistake of this sort, then, can cause a credibility deficit that does not constitute a case of testimonial injustice. At least, I suggest that we circumscribe the concept in this manner. Of course it would not be linguistically outrageous for our imagined hearer, embarrassed on learning the true professional identity of her interlocutor, to say she felt bad for doing him such an ‘injustice’. But this would be a very weak sense of injustice; so much so that it is a mere shadow of our ordinary ethical and political sense of the word and lacks the usual implication of moral badness. This is largely a terminological point, so if others disagree, then they can regard cases of innocent error as producing a weak form of testimonial injustice. For my part, however, I shall reserve the term for cases in which there is something ethically bad about the hearer's misjudgement.

What about credibility deficit caused by ethically innocent but epistemically culpable error? If we revisit our example and alter it so that we picture our philosopher making her mistake as the result of a hopelessly careless web search, I suggest that we find that the credibility deficit she assigns her interlocutor still does not amount to a testimonial injustice. Her unduly deflated credibility judgement of him does not insult or undermine him as a knower, for she has simply made a stupid mistake. While her error is epistemically culpable, its ethical non‐culpability still seems to prevent the resultant credibility deficit from constituting a testimonial injustice: an ethically non‐culpable mistake cannot undermine or otherwise wrong the speaker. It seems that the ethical poison of testimonial injustice must derive from some ethical poison in the judgement of the hearer, and there is none such wherever the hearer's error is ethically non‐culpable. The proposal I am heading for is that the ethical poison in question is that of prejudice. From different points in history one might draw on many depressing examples (p.23) of prejudices obviously relevant to the context of credibility judgement, such as the idea that women are irrational, blacks are intellectually inferior to whites, the working classes are the moral inferiors of the upper classes, Jews are wily, Orientals are sly . . . and so on in a grim catalogue of clichés more or less likely to insinuate themselves into judgements of credibility at different moments in history. But in order to furnish the philosophical imagination less crudely, let us turn to an example from literature that provides us with a historically truthful fiction.

The example is from Harper Lee's To Kill a Mockingbird. The year is 1935, and the scene a courtroom in Maycomb County, Alabama. The defendant is a young black man named Tom Robinson. He is charged with raping a white girl, Mayella Ewell, whose family's run‐down house he passes every day on his way to work, situated as it is on the outskirts of town in the borderlands that divide where whites and blacks live. It is obvious to the reader, and to any relatively unprejudiced person in the courtroom, that Tom Robinson is entirely innocent. For Atticus Finch, our politely spoken counsel for the defence, has proved beyond doubt that Robinson could not have beaten the Ewell girl so as to cause the sort of cuts and bruises she sustained that day, since whoever gave her the beating led with his left fist, whereas Tom Robinson's left arm is disabled, having been injured in a machinery accident when he was a boy. The trial proceedings enact what is in one sense a straightforward struggle between the power of evidence and the power of racial prejudice, with the all‐white jury's judgement ultimately succumbing to the latter. But the psychology is subtle, and there is a great complexity of social meanings at work in determining the jury's perception of Tom Robinson as a speaker. In a showdown between the word of a black man and that of a poor white girl, the courtroom air is thick with the ‘do's and ‘don't’s of racial politics. Telling the truth here is a minefield for Tom Robinson, since if he casts aspersions on the white girl, he will be perceived as a presumptuous, lying Negro; yet, if he does not publicize Mayella Ewell's attempt to kiss him (which is what really happened), then a guilty verdict is even more nearly assured. This discursive predicament mirrors his practical predicament at the Ewell's house on that fateful day when Mayella grabbed him. If he pushes her away, then he will be found to have assaulted her; yet if he is passive, he will equally be found to have assaulted her. So he does the most neutral thing he can, which is to run, though knowing all the while that this action too will be taken as a sign of guilt. Mr Gilmer's (p.24) interrogation of Tom is suffused with the idea that his running away implies culpability:

  • ‘ . . . why did you run so fast?’
  • ‘I says I was scared, suh.’
  • ‘If you had a clear conscience, why were you scared?’16
Running away, it seems, is something a black man in Maycomb County cannot do without incriminating himself. Similarly, there are many things he cannot say in court and stand a chance of being heard as truthful. At a pivotal moment during the prosecution's interrogation, for instance, Tom Robinson makes the mistake of being honest about his kindly motivations for stopping off at Mayella Ewell's house as regularly as he did to help her out with odd jobs. The scene, like the whole story, is reported from the point of view of Scout, Atticus Finch's young daughter, who is secretly surveying the proceedings with her brother, Jem, from the ‘Negro gallery’. Mr Gilmer, the prosecutor, sets him up:

‘Why were you so anxious to do that woman's chores?’

Tom Robinson hesitated, searching for an answer. ‘Looked like she didn't have nobody to help her, like I says—’

. . . Mr Gilmer smiled grimly at the jury. ‘You're a mighty good fellow, it seems—did all this for not one penny?’

‘Yes suh. I felt right sorry for her, she seemed to try more'n the rest of ‘em—’

You felt sorry for her, you felt sorry for her?’ Mr Gilmer seemed ready to rise to the ceiling.

The witness realized his mistake and shifted uncomfortably in the chair. But the damage was done. Below us, nobody liked Tom Robinson's answer. Mr Gilmer paused a long time to let it sink in.17

Here the ‘damage’ in question is done to any epistemic trust that the white jury has so far been human enough to feel towards the black testifier. For feeling sorry for someone is a taboo sentiment if you are black and the object of your sympathy is a white person. In the context of a racist ideology structured around dogmas of white superiority, the fundamental ethical sentiment of plain human sympathy becomes disfigured in the eyes of whites so that it appears as little more than an indicator of self‐perceived advantage on the part of the black subject. A black man is not allowed to have feelings that imply a position of any (p.25) sort of advantage relative to any white person, no matter how difficult and lonely her life might be. The fact that Tom Robinson makes the sentiment public raises the stakes in a way that is disastrous for legal justice and for the epistemic justice on which it depends. The trial is a zero‐sum contest between the word of a black man against that of a white girl (or perhaps that of her father who has brought the case to court), and there are those on the jury for whom the idea that the black man is to be epistemically trusted and the white girl distrusted is virtually a psychological impossibility—Robinson's expressed sympathy in feeling sorry for a white girl only reinforces that impossibility.

As it turns out, the members of the jury stick with their prejudiced perception of the defendant, formed principally by the racial stereotypes of the day. Atticus Finch challenges them to dispense with these prejudicial stereotypes; to dispense, as he puts it, with the ‘assumption—the evil assumption—that all Negroes lie, that all Negroes are basically immoral beings, that all Negro men are not to be trusted around our women’.18 But when it comes to the verdict, the jurors go along with the automatic distrust delivered by the prejudices that structure their perception of the speaker. They find him guilty. And it is important that we are to interpret the novel so that the jurors really do find him guilty. That is to say, they do not privately find him innocent yet cynically convict him anyway. Even allowing that the psychology here may be to some degree indeterminate, it is crucial that they genuinely fail to do what Atticus Finch in his summing‐up describes as their ‘duty’:

‘ . . . In the name of God, do your duty.’

Atticus's voice had dropped, and as he turned away from the jury he said something I did not catch. He said it more to himself than to the court. I punched Jem.

‘What'd he say?’

‘ “In the name of God, believe him,” I think that's what he said.’19

Finch is trying to impress upon the jury that they have a duty to believe Tom Robinson, and this supports my interpretation of the jurors' psychology. Finch evidently takes it that what the jury need to be urged to do is to make the right judgement, to do the right epistemic thing. He does not urge them to focus on their moral and legal duty to convict only if they truly judge the defendant guilty, for he is aware that their prejudice goes psychologically deeper than that, all the way to the jurors' (p.26) very powers of judgement. When they do deliver the guilty verdict, this attests to their failure in their duty to make the proper testimonial judgement, in the light of the evidence. They fail, as Atticus Finch feared, precisely in their duty to believe Tom Robinson. Given the evidence put before them, their immovably prejudiced social perception of Robinson as a speaker leads at once to a gross epistemic failure and an appalling ethical failure of grave practical consequence. As it turns out, Tom Robinson does not survive long enough to go ahead with any appeal, for he is shot in the back as he tries, we hear it said, to escape over the prison fence right in front of the guards.

It is perhaps worth remarking that even the most hateful prejudicial ideologies may be sustained not only by explicitly hateful thought and talk but also by more domestic stereotypical ideas that are almost cosy in comparison. There is a relatively light‐hearted theme of epistemic untrustworthiness that runs through the book as a leitmotif, softly echoing the deadly serious racist exclusion from epistemic trust of the sort that leads ultimately to the killing of Tom Robinson. We see this, for instance, when Scout is talking to her family's friend and neighbour, Miss Maudie, about the reclusive and mysterious young Boo Radley (aka Mr Arthur), about whom spooky stories abound and who is an object of unfailing fascination for the children. Scout quizzes Miss Maudie about him:

‘Do you think they're true, all those things they say about B—Mr Arthur?’

‘What things?’

I told her.

‘That is three‐fourths coloured folks and one‐fourth Stephanie Crawford,’ said Miss Maudie grimly.20

Given a culture where it is so utterly natural for white people to associate ‘coloured folks’ in general with irresponsible gossip (even in a spirit of independent‐mindedness, as is the case with Miss Maudie's response to Scout), it is not hard to imagine a relation of support between this comparatively cosy side of the ideology and the far harsher, more squarely unjust associations that work to undermine the epistemic trustworthiness of black people. While there may be nothing hateful in the more light‐hearted side of these attitudes, still it may be a significant nutrient to the hateful ideology overall.

Tom Robinson's case represents an extreme example of the sort of testimonial injustice I am aiming to portray philosophically. An initial (p.27) sketch might lead us to capture it as prejudicial credibility deficit. But while this may serve as a general definition of testimonial injustice, it misses a crucial feature of the sort of testimonial injustice that Tom suffers. There are all sorts of prejudices that can cause credibility deficit, yet where the resultant testimonial injustice is highly localized and therefore lacking any of the structural social significance that a case such as Tom Robinson's clearly has. Imagine, for instance (I adapt an example proposed to me by a scientist), a panel of referees on a science journal who have a dogmatic prejudice against a certain research method. It might reasonably be complained by a would‐be contributor that authors who present hypotheses on the basis of the disfavoured method receive a prejudicially reduced level of credibility from the panel. Thus the prejudice is such as to generate a genuine testimonial injustice (writing being one medium of testimony). Although such a testimonial injustice may be grievous for the careers of the would‐be contributors, and perhaps even for the progress of science, none the less its impact on the subject's life is, let us assume, highly localized. That is to say, the prejudice in question (against a certain scientific method) does not render the subject vulnerable to any other kinds of injustice (legal, economic, political). Let us say that the testimonial injustice produced here is incidental.

By contrast, testimonial injustices that are connected, via a common prejudice, with other types of injustice, might appropriately be termed systematic. Systematic testimonial injustices, then, are produced not by prejudice simpliciter, but specifically by those prejudices that ‘track’ the subject through different dimensions of social activity—economic, educational, professional, sexual, legal, political, religious, and so on. Being subject to a tracker prejudice renders one susceptible not only to testimonial injustice but to a gamut of different injustices, and so when such a prejudice generates a testimonial injustice, that injustice is systematically connected with other kinds of actual or potential injustice. Clearly the testimonial injustice suffered by Tom Robinson is systematic, for racial prejudice renders him susceptible to a panoply of injustices besides the testimonial kind. Systematic testimonial injustice constitutes our central case—it is central from the point of view of a guiding interest in how epistemic injustice fits into the broader pattern of social justice.

The main type (the only type?) of prejudice that tracks people in this way is prejudice relating to social identity. Let us call this sort of prejudice identity prejudice. It can come in positive or negative (p.28) form—prejudice for or against people owing to some feature of their social identity—but since our interest is in cases of credibility deficit rather than excess, we shall be concerned only with negative identity prejudice. (Indeed, I shall tend to use ‘identity prejudice’ as short for ‘negative identity prejudice’.) The influence of identity prejudice in a hearer's credibility judgement is an operation of identity power. For in such a case the influence of identity prejudice is a matter of one party or parties effectively controlling what another party does—preventing them, for instance, from conveying knowledge—in a way that depends upon collective conceptions of the social identities in play. In our Mockingbird example, racial identity power is exercised in this way by members of the jury as they make their deflated credibility judgements of Tom Robinson, with the result that he is unable to convey to them the knowledge he has of what happened at the Ewells' place. This is the essential exercise of identity power in the courtroom that seals Tom's fate, though of course it is not the whole story, for this operation of identity power is crucially supported by Mr Gilmer's simple but highly effective prosecution strategy, which is to invoke the usual collective negative imaginings of the Negro. Gilmer deliberately controls the jurors, and sure enough the jurors go on to control what Tom Robinson does, preventing him from conveying his knowledge to them.

With the concepts of identity prejudice and systematicity in place, we are now in a position to propose a refined characterization of the central case of testimonial injustice—the systematic case. The speaker sustains such a testimonial injustice if and only if she receives a credibility deficit owing to identity prejudice in the hearer; so the central case of testimonial injustice is identity‐prejudicial credibility deficit. We should note, however, that there could be exceptions; that is, one can imagine cases of identity‐prejudicial credibility deficit that are not cases of systematic testimonial injustice, and so not examples of our central case. Consider the following case (an anecdote recounted to me by a philosopher of science). There is a large international conference dominated by research scientists and some historians of science, with only a smattering of philosophers of science. It becomes clear that the philosophers of science are regarded by the majority of the other delegates as out of touch with the realities of scientific practice, so much so that they are, frankly, held in some intellectual disdain. In this context, it would seem, simply falling into the identity category ‘philosopher of science’ renders one's word likely to be dismissed as the vain speculations of an out‐of‐touch academic. Thus there are genuine (p.29) cases of identity‐prejudicial credibility deficit going on here. These testimonial injustices, however, do not instantiate our central case, for they are not systematic. Despite the prejudice's being an identity prejudice, it does not concern the kind of broad identity category that makes for a tracker prejudice; on the contrary, its social significance is highly localized to the specific conference context described. It therefore produces only an incidental testimonial injustice.

To categorize a testimonial injustice as incidental is not to belittle it ethically. Localized prejudices and the injustices they produce may be utterly disastrous for the subject, especially if they are repeated frequently so that the injustice is persistent. If, for instance, the practical context in which the injustices occur is that of a project, professional or otherwise, which is crucial to the person's life being worth living, then the accumulation of incidental injustices may ruin their life. The importance of systematicity is simply that if a testimonial injustice is not systematic, then it is not central from the point of view of an interest in the broad pattern of social justice. ‘Persistent’ labels the diachronic dimension of testimonial injustice's severity and significance, whereas ‘systematic’ labels the synchronic dimension. The most severe forms of testimonial injustice are both persistent and systematic. Such is the case for Tom Robinson, who lives in a society in which the prejudice that devalues his word also blocks his everyday pursuits repeatedly and in every social direction. By contrast, cases of testimonial injustice that are neither persistent nor systematic are on the whole unlikely to be very disadvantageous. Generally speaking, systematic injustice tends towards persistence, because the imaginative conceptions of social identity that feature in the relevant tracker prejudices are likely to be enduring features of the social imagination.

Now that I have identified our central case as systematic testimonial injustice, let us now inquire further into how identity prejudice enters in to make its impact on the discursive exchange. We must explore the role of stereotypes in hearers' judgements of speakers' credibility.


(1) Anthony Minghella, The Talented Mr Ripley—Based on Patricia Highsmith's Novel (London: Methuen, 2000), 130.

(2) Michel Foucault, ‘How Is Power Exercised?’, trans. Leslie Sawyer from Afterword in H. L. Dreyfus and P. Rabinow, Michel Foucault: Beyond Structuralism and Hermeneutics (Hemel Hempstead: Harvester Press, 1982), 219.

(3) ‘Now the “delinquent” makes it possible to join [two figures constructed by the penal system: the moral or political “monster” and the rehabilitated juridical subject] and to constitute under the authority of medicine, psychology or criminology, an individual in whom the offender of the law and the object of scientific technique are superimposed’ (Foucault, Discipline and Punish: The Birth of the Prison, trans. Alan Sheridan (London: Penguin Books, 1977), 256; originally published in French as Naissance de la prison by Editions Gallimard, 1975).

(4) ‘[Individuals] are always in the position of simultaneously undergoing and exercising this power. . . . In other words, individuals are the vehicles of power, not its points of application’ (Michel Foucault, Power/Knowledge: Selected Interviews and Other Writings 1972–1977, ed. C. Gordon, trans. C. Gordon, L. Marshall, J. Mepham, and K. Soper (Hemel Hempstead: Harvester Wheatsheaf, 1980), 198).

(5) Thomas E. Wartenberg, ‘Situated Social Power’, in T. Wartenberg (ed.), Rethinking Power (Albany, NY: State University of New York Press, 1992), 79–101.

(6) ‘[Power] is never localised here or there, never in anybody's hands, never appropriated as a commodity or piece of wealth. Power is employed and exercised through a net‐like organisation. And not only do individuals circulate between its threads; they are always in the position of simultaneously undergoing and exercising this power’ (Foucault, Power/Knowledge, 98).

(7) See Steven Lukes, Power: A Radical View (London: Macmillan, 1974).

(8) Wartenberg, ‘Situated Social Power’, 89.

(9) Minghella, The Talented Mr Ripley, 130.

(10) For an argument to the effect that intuition is not in general a source of cognitive failing but rather an essential cognitive resource, see my ‘Why Female Intuition?’, Women: A Cultural Review, 6, no. 2 (Autumn 1995), 234–48; a shorter version of which appears as ‘Intuition and Reason’, Philosophical Quarterly, 45, no. 179 (Apr. 1995), 181–9, without the discussion of female intuition in particular.

(11) Pace two well‐known views in the epistemology of testimony. First is Reid's view according to which we naturally operate counterpart principles of veracity and credulity in our testimonial exchanges (see Thomas Reid, Inquiry into the Human Mind, ch. 6, sect. xxiv: ‘Of the Analogy between Perception and the Credit We Give to Human Testimony’ (first published 1764)). Second is Tyler Burge's view, according to which we have an a priori entitlement for believing what others tell us, other things equal (see his ‘Content Preservation’, Philosophical Review, 102, no. 4 (Oct. 1992), 457– 88). I shall discuss these views in Ch. 3, as I situate the phenomenon of testimonial injustice in the epistemology of testimony more generally.

(12) I sympathize with Coady's scepticism about there being any precise science here, any precise ‘credibility ratio’ to determine what degree of belief the hearer is entitled to (see C. A. J. Coady, Testimony: A Philosophical Study (Oxford: Clarendon Press, 1992), 210).

(13) I thank Hugh Mellor for this example, which I have elaborated somewhat.

(14) In ‘Rational Authority and Social Power: Towards a Truly Social Epistemology’, Proceedings of the Aristotelian Society, 98, no. 2 (1998), 159–77, I wrote as if both deficit and excess were cases of epistemic injustice (the only type of which I considered was what I am here more specifically calling testimonial injustice), but the considerations I present here have changed my mind. I am also using the notion of ‘credibility’ rather more generically than I did in that paper.

(15) David Hume, A Treatise of Human Nature, III. ii. 2, ed. L. A. Selby‐Bigge, 3rd edn. (Oxford: Clarendon Press, 1975).

(16) Harper Lee, To Kill a Mockingbird (London: William Heinemann, 1960), 202.

(17) Ibid. 201.

(18) Ibid. 208.

(19) Ibid. 210.

(20) Harper Lee, To Kill a Mockingbird (London: William Heinemann, 1960), 51.