2.1. Central cases

For Fricker, central cases of testimonial injustice are marked by a ‘structural social significance’ (Fricker Reference Fricker2007, 27), embodied in their systematic and persistent nature. Testimonial injustices are systematic if, ‘they are connected, via a common prejudice, with other types of injustice’ (27). These prejudices will track a speaker through various social realms, e.g. the political, economic, sexual, legal, etc. Testimonial injustices are persistent in so far as the prejudices responsible are, ‘enduring features of the social imagination’ (29). The systematic and persistent features of these central cases are the synchronic and diachronic aspects, respectively, of this especially pernicious form of epistemic injustice.

By way of elucidation, Fricker provides examples to give the reader a clear picture of her focus. One such example is the trial of Tom Robinson in Harper Lee’s To Kill a Mockingbird. Though the evidence is clearly in favor of acquitting Robinson of any charges, he is nonetheless found guilty due to the jury’s (mis)perception of his testimony. Robinson’s words on his own behalf are not given the credibility they deserve in virtue of prejudicial stereotypes that unjustly characterize black males as untrustworthy, sexually aggressive, etc. The prejudices working against Robinson will track him through various social spheres, meaning the biased and unjust stereotypes of black males will negatively affect Robinson in the myriad contexts in which he interacts with whites. Also, the prejudices represent deeply ingrained views of the social imagination (especially in the American South where the book takes place). So, what we categorize as a central case will depend on our theory of the types of prejudices that (1) can track a speaker through various social contexts, (2) are deep-seated in the social imagination, and (3) are connected with other types of injustice outside of the epistemic.

2.2. Prejudices

We can think of prejudices broadly as, ‘judgments, which may have a positive or a negative valence, and which display some (typically, epistemically culpable) resistance to counter-evidence owing to an affective investment on the part of the subject’ (Fricker Reference Fricker2007, 35). But Fricker believes the prejudices relevant to the central cases of testimonial injustice must be of a negative valence. Also, importantly, these prejudices must arise from an ethically bad form of motivated cognition, meaning the cause of one’s resistance to counter-evidence must be ethically noxious. As Fricker states, ‘we have a special interest in negative identity prejudices, and these are, I take it, always generated by some ethically bad affective investment’ (35).

Fricker does not spend much time developing what it is for the affective investment that generates the prejudice, and thereby the resistance to counter-evidence, to be ethically noxious, but examples she cites are suggestive. Fricker borrows a thought experiment from Nomy Arpaly involving a young boy, Solomon, from a poor country living in an isolated farming community. Solomon is taught that women are inferior with respect to abstract thinking (Arpaly Reference Arpaly2003, 103). Arpaly suggests that Solomon is not guilty of any irrationality if he believes the dominant view of his community—Solomon is merely epistemically unlucky to occupy a region in which such a falsehood is disseminated as a well-confirmed truth. We are asked to imagine further that Solomon attends university where he studies alongside women who are adept abstract thinkers. If Solomon continues in his false belief that women are inferior abstract thinkers, this would expose his belief as a prejudice, as his judgment persists despite adequate reason to think otherwise. Fricker comments that the relevant ethical flaw in Solomon’s judgment,

It seems Fricker takes the situation to be similar with respect to the white jury and Tom Robinson. The jury’s prejudice—the reason they persist in their negative judgments about black males—is grounded in an ethically noxious contempt for certain racial groups. In sum, the type of prejudice relevant for testimonial injustice must (1) have a negative valence with respect to the object of the judgment, and (2) the judgment must be resistant to counter-evidence in virtue of an ethically virulent motivation. We can summarize Fricker’s notion, which she dubs negative identity prejudice, as follows:

2.3. Stereotypes

Fricker draws on evidence from the work of Daniel Kahneman and Amos Tversky to argue that in everyday exchanges in which one determines how far to trust a speaker, one will rely on stereotypes used as heuristics.³ Stereotypes are roughly analyzable as associations of a social group with a set of descriptive attributes. The class of stereotypes relevant to the central cases of testimonial injustice must be, according to Fricker’s account, unreliable, and the attributes ascribed to members of the social group must be ‘disparaging.’⁴ Combining this notion of stereotype with NP, we get negative identity-prejudicial stereotypes:

On Fricker’s account, central cases of testimonial injustice will involve a hearer perceiving the testimony of a speaker as less credible than deserved in virtue of the effects of an NPS. As I will argue, there are clear cases of testimonial injustice that exhibit persistence and systematicity and are certainly central from the point of view of an interest in the broad pattern of social justice, but do not arise from an NPS.

4.1. Social influence and prescriptive credibility deficits

Though there is a robust body of literature on social influence, which I take to be directly relevant to Fricker’s account, there is little attempt to provide an analysis of the central concept of social influence. There are several research paradigms that are traditionally taken to fall under the heading of ‘social influence’ and, presumably, each address an aspect of the concept as it is employed in social psychology. For our purposes, two of these paradigms are of special interest, namely, persuasion research and research involving the use of information in a group setting. In a persuasion paradigm, a speaker gives an argument for some position, p, and study subjects are asked to indicate their level of agreement with p in light of the speaker’s argument. The influence of the speaker is operationalized as the change in the subject’s stance with regards to p after hearing the argument. Various trials can involve speakers of different social groups who nonetheless keep the wording and presentation style of the argument identical to test the effects of a speaker’s perceived group membership on the ability of that speaker to persuade. In a group paradigm, several subjects are tasked with collectively generating a solution to a problem. Subjects are each given a different set of information relevant to solving the problem prior to the group discussion stage. Researchers then observe whose proposed information and suggestions are subsequently utilized in group deliberation. In a group paradigm, a speaker’s level of social influence will be related to the perceived relevance or force of the information she offers the group.

Though gender differences in social influence are mediated by various context sensitive variables, it is safe to claim that, in general, arguments or attempts to persuade are more likely to be ignored if the arguer is a women, and contributions by men in a group setting are more likely to influence group decision and factor into group deliberation (Carli Reference Carli2001). For example, in a study performed by Propp (Reference Propp1995), mixed-sexed groups were more likely to utilize certain information in group deliberation if the individual who presented the information was male. As Propp was able to demonstrate, gender, ‘biases the use of information; a male’s relatively high source credibility increases the perceived validity of the information he introduces’ (453). In her review of the social influence literature, Linda Carli claims,

Research involving persuasion and group paradigms provides a rather straightforward empirical demonstration of the phenomena Fricker is trying to capture. The gender disparity in persuasion and information use is partially mediated by assessments of competence. In virtue of a prejudicial stereotype that characterizes women as less competent than men with respect to a range of domains, individuals assign women less credibility, which in turn leads study participants to view arguments offered by women as less persuasive and information offered as less relevant. This decrease in the ‘perceived validity’ of information or arguments presented can be straightforwardly analyzed in terms of unjustly low credibility being assigned to female speakers.

Though I have presented the social influence literature as straightforwardly applicable to Fricker’s work, the relevance of the literature may be challenged on several grounds. (1) It could be objected that neither the persuasion nor group research paradigms, strictly speaking, address gender differences in a speaker’s ability to transmit information through testimony. The paradigm case of testimony involves a speaker telling or asserting some proposition, p, and a hearer coming to believe p on the basis of perceiving and understanding the speaker’s assertion (E. Fricker Reference Fricker, Sintonen, Wolenski and Niiniluoto2004). Neither arguing for p nor mentioning p as relevant in a group problem-solving context qualifies as a bald assertion or an instance of a telling. To empirically verify the existence of testimonial injustice involving gender stereotypes we would need to examine the variable willingness of subjects to believe some proposition, p, on the bases of various speakers baldly asserting that p. (2) Alternatively, one might argue that it is not clear from the experimental design whether female speakers are given a credibility deficit or whether males are assigned a credibility excess. Given that, under Fricker’s account, testimonial injustice must involve the administration of a credibility deficit, we would need to establish that the social influence literature clearly involves credibility deficits before taking the literature to vindicate Fricker’s account.

I take both of these objections to largely be a cavil for several reasons. It’s not clear that Fricker’s central cases of testimonial injustice, like the exchange between Marge Sherwood and Herbert Greenleaf in The Talented Mr. Ripley, deal solely in testimony, excluding other speech acts such as providing evidence for previous claims. Nor is it clear that driving a hard conceptual boundary between testimonial injustice and injustice with respect to other speech acts would prove fecund, as it is doubtful that significantly different ethical issues would arise surrounding testimony, arguments, suggestions, musings, etc., or that a relevantly different psychological process would govern the interaction of credibility assignments versus various other speech acts.¹⁶ With respect to the second objection, the issue of determining whether a case involves a credibility excess assigned to one party or a credibility deficit assigned another will arise for any potential case of testimonial injustice, so I don’t take this issue to be a serious objection.¹⁷

Though the work on gender disparities in social influence provides solid empirical evidence for Fricker’s claims, assessments of competence are not the only mediating factor that affects the perceived strength of an argument or the perceived relevance of information. The likeability of a speaker also has an effect on social influence and isn’t reducible to assessments of speaker competence or sincerity. In turn, the presentation style and nonverbal behavior of a speaker can affect her perceived likability. Women (but not men) risk backlash in the form of decreased credibility assignments when overtly displaying assertiveness or competence; ‘women who attempt to assert authority outside of traditionally female contexts, particularly over men, encounter a resistive, “backlash” reaction that reduces their ability to achieve influence over others’ (Ridgeway Reference Ridgeway2001, 649).

A bevy of studies demonstrate that women can prove more influential and persuasive to men (and thereby be assigned a higher level of credibility) when utilizing a more tentative style of communication, even if such will hurt evaluations of competence.¹⁸ For example, in a persuasion paradigm study performed by Carli, LaFleur, and Loeber (Reference Carli, LaFleur and Loeber1995), women who demonstrated a more task-oriented manner of nonverbal communication (e.g. rapid speech, taunt posture, firm voice, etc.) as opposed to a social style (e.g. leaning toward the listener, friendly facial expression, etc.) were found to be less influential by men than male speakers who were similarly task oriented. Female and male speakers who exhibited a task-oriented style were not necessarily judged to be differing in competence, but women who displayed this more ‘traditionally masculine’ manner of communication, thereby flouting prescriptive stereotypes, were found to be more threatening and, in turn, proved less influential. The prescriptive aspect of stereotypes that dictate women should exhibit communal characteristics can affect how one perceives the credibility of a female speaker. As Carli, LaFleur, and Loeber found, ‘with male audiences, likableness is a more important determinant of influence for female speakers than male speakers. For women, being merely competent is not enough’ (Reference Carli, LaFleur and Loeber1995, 1038).

Further studies have shown the effects of nonverbal communication to be sensitive to the perceived salience of gender in a given context (Reid et al. Reference Reid, Palomares, Anderson and Bondad-Brown2009). When, through some means, gender is made salient, female speakers will tend to prove more influential with male listeners if they utilize a tentative style of presentation. On the other hand, if certain other qualities are made salient, such as one’s educational history, female speakers will tend to prove more influential relative to male listeners if they utilize a more assertive presentation style.

Returning to the example of Sally, her suggestions for company policy may be unduly devalued in virtue of her more assertive, ‘masculine’ presentation style in a situation in which her gender as opposed to business credentials is made salient. Sally, like Hopkins, doesn’t have to be stereotyped as having qualities thought to be incompatible or unfit for the male dominated business world. In fact, Sally may be judged as equivalent (or superior) in competence, assertiveness, independence, etc., to her male peers, but her words can nonetheless be devalued because her counter-stereotypic behavior is perceived as threatening or a ploy for power by her male counterparts. Sally is assigned a credibility deficit for acting in a counter-stereotypic manner and thus punished and censured for her behavior as opposed to having her competence (or competence related attributes) devalued.

Sally’s case aligns with the spirit of what Fricker is trying to capture, as Sally is being harmed qua knower in virtue of flouting a (prescriptive) stereotype that tracks her through various social realms and represents deeply ingrained views about how women should behave. Sally’s case is rooted in negative affective states in response to women who act in counter-stereotypic ways, but her peers misperceive her level of credibility in virtue of stereotypes about how women should be as opposed to how women are. The case of Sally doesn’t fit the picture of NPS Fricker takes to characterize central cases of testimonial injustice. The prescriptive component of stereotypes deserves separate mention in our account of testimonial injustice—otherwise, it will prove too narrow to capture cases like Sally’s.

I propose splitting testimonial injustice into two further categories, namely, descriptive credibility deficits and prescriptive credibility deficits. Descriptive credibility deficits will operate on the basis of NPS as Fricker describes. As in the first case involving Sally, her male business partners attribute communal traits to her in virtue of her perceived gender, which conflict with the more agentic qualities thought to be important for effective business strategizing. This attribution causes Sally’s business partners to misperceive her level of credibility in the context of the business meeting. On the other hand, prescriptive credibility deficits arise from a backlash effect in which the speaker is punished by being assigned a lower level of credibility than deserved in virtue of flouting a relevant prescriptive stereotype. As in the second case involving Sally, her male business partners react negatively to Sally’s more task-oriented or assertive communication style, which is incommensurate with how women should act according to the relevant prescriptive stereotype. This negative affective response can manifest as an assignment of a lower level of credibility than deserved—in effect, a punishment for acting in a counter-stereotypic manner. Both credibly assignments involve a misperception of the credibility of a speaker, but these misperceptions arise by different means and therefore are worthy of conceptual distinction.

5.1. Credibility perceptual capacity

In an average conversational exchange, our initial credibility judgment of a speaker will be, according to Fricker, formed by a credibility perceptual capacity (CPC). A CPC is a nonconscious, non-deliberative mental process that presents a speaker in ‘epistemic colour’—to use Fricker’s turn of phrase. In normal conversation, we do not consciously reflect on body language, vocal intonation, social group membership, or other cues in service of deliberating about the credibility of a given speaker—these cues are processed subpersonally with the result that a speaker will appear to have a certain level of credibility.¹⁹ This epistemic, perceptual capacity, according to Fricker, is informed by a background ‘theory’ consisting of a body of generalizations including (1) stereotypes associated with certain social groups, (2) assessments of potential motivations of the interlocutor, and (3) estimates of the likelihood that such an individual will tell the truth in a given context. Of course we can proceed to alter our initial credibility judgments through deliberation and conscious reflection on relevant evidence, but these initial credibility assignments will be made automatically and subpersonally.

Fricker treats CPC as nigh exclusively truth directed. Under Fricker’s characterization, CPC is purely a function of seemingly relevant information at one’s disposal that bears directly on the credibility that should be assigned a speaker (assuming the information is accurate). But we do not use assignments of credibility solely for the end of assessing the likelihood that a speaker is telling the truth. Of course counter-veridical motives can affect our treatment of evidence, regardless of whether that evidence is testimonial in nature or derived from another source. There is a large body of empirical work that demonstrates our proclivity to formulate self-serving beliefs.²⁰ But prescriptive credibility deficits are not merely further instances of motivated cognition. In a case involving a prescriptive credibility deficit, the relevant evidence is not simply reinterpreted to support a favored hypothesis or ignored to prevent an undesired conclusion. The evidence is manipulated as a means of punishment for the source, namely, the counter-stereotypic individual whose actions or behavior presents some perceived psychological threat. Prescriptive credibility assignments can serve an ego defensive role by silencing counter-stereotypic individuals. As Schimel et al. (Reference Schimel, Simon, Greenberg, Pyszczynski, Solomon, Waxmonsky and Arndt1999, 907) claim, ‘by derogating those who are different, the individual deflects any threat posed by this deviance.’

Extending the point, there is evidence that physical attraction can boost one’s perceived credibility and can enhance an interlocutor’s chance of persuading a listener (Horai, Naccari, and Fatoullah Reference Horai, Naccari and Fatoullah1974). In a study performed by Matschiner and Murnen (Reference Matschiner and Murnen1999), the researchers found that male participants tended to judge one and the same female speaker to be more physically attractive when she exhibited an exaggerated, ‘feminine’ style of communication, which in turn boosted the likelihood of persuading her male audience. Physical attraction runs orthogonal to competence and expertise—it’s not the case that the increased social influence afforded the attractive is due to people judging comely individuals as more intellectually capable or sincere. The effect of something like physical attraction on credibility proves hard for Fricker’s framework to integrate. Physical attraction alters credibility assignments independently of any generalizations about levels of competence, expertise, likelihood of telling the truth, or other epistemically relevant associations. This boost in credibility is arguably a form of praise, just as prescriptive credibility deficits are a form of punishment. More forcefully put, prescriptive credibility assignments can be used to punish those we find threatening and, in turn, bolster those we favor. Punishment and reward are non-epistemic functions of credibility assignments that will go overlooked if we characterize CPC as determined exclusively by background information concerning the likelihood that an interlocutor will assert a truth in a given context.

5.2. The responsibilities of a virtuous listener

The responsible listener doesn’t passively accept whatever testimony comes her way. This doesn’t mean the responsible listener consciously moves through a chain of inference, given her evidence in the testimonial context, to the conclusion that she should either believe or withhold assent in what has been testified to.²¹ Instead, as Fricker claims, ‘without actively assessing or reflecting on how trustworthy our interlocutor is, the responsible hearer none the less remains unreflectively alert to the plethora of signs, prompts, and cues that bear on how far she should trust’ (Reference Fricker2007, 66). The responsible hearer is unreflectively yet critically engaged when receiving testimony—she remains vigilant for any potential indicators of the speaker’s level of competence and sincerity and to any distorting effects of an NPS.

Competence and sincerity do seem, normatively, to be the correct dimensions along which to assess a speaker, but they do not exhaust the dimensions that factor into assessments of speaker credibility. As I’ve argued, the perceived likeability of a speaker, as assessed by a listener, is an independent dimension that affects credibility judgments. In turn, if the responsible hearer must remain unreflectively aware of how an NPS may affect her assessments of competence and sincerity, she must also remain vigilant for how the perceived likeability of a speaker may lead to backlash in the form of a credibility deficit. With a more nuanced, descriptive understanding of the ways our credibility judgments can go systematically awry we gain a subtler, normative understanding of what the virtuous listener must do to avoid perpetrating testimonial injustice.

This expanded notion of what a responsible listener needs to monitor leads to a potential critique of Fricker’s account that mirrors recent criticisms of what Henderson (Reference Henderson2008) dubs an acceptance with monitoring (AM) analysis of justified, testimonial belief.²² AM views like that of Elizabeth Fricker’s (e.g. Reference Fricker1995, Reference Fricker, Sintonen, Wolenski and Niiniluoto2004) hold that one is justified in accepting the testimony of a speaker only if one’s acceptance is grounded (at least partially) in monitoring (though not necessarily reflectively or consciously) for cues diagnostic of the trustworthiness of the speaker. Drawing on social psychological research in deception detection, Shieber (Reference Shieber2012) and Michaelian (Reference Michaelian2010) argue that, though we do typically monitor for deception, we aren’t particularly reliable in determining speaker sincerity.²³ If we aren’t reliable at detecting mendacity, then monitoring for deception shouldn’t be required for the justified acceptance of testimony, pace E. Fricker.

An analogous objection may be raised against Fricker’s account of how the virtuous listener should compensate for the potential deleterious effects of NPS and backlash. In the case of a prescriptive credibility deficit in which one’s credibility assignment acts as a way to suppress the threat of a counter-stereotypic speaker, it is unlikely that there will be some internal cue or phenomenological marker one could reliably detect, even with training, that would indicate whether one’s credibility assignments are being negatively affected by counter-veridical motives. The motivated cognition involved in something like backlash operates at a subpersonal level outside of one’s conscious awareness. Fricker’s suggestion that one might be able to be, ‘reflectively aware at the time that prejudice is shaping one’s perception of an interlocutor, ’ (Reference Fricker2007, 97) and proceed to alter the judgment accordingly, isn’t likely. Just as cues for deception remain hard to monitor and recalcitrant to detection training, cues that one is using one’s credibility assignment as punishment for a counter-stereotypic interlocutor will be equally hard to expose.

Alcoff (Reference Alcoff2010) makes a somewhat similar argument (though not with respect to backlash effects), questioning the effectiveness of attempting to bring prejudice and stereotype that operate at a nonconscious level under volitional control. In response, Fricker (Reference Fricker2010) cites two papers, including a series of four studies performed by Monteith et al. (Reference Monteith, Ashburn-Nardo, Voils and Czopp2002) that attempts to elucidate how ‘automatic processes can be controlled and changed in the context of stereotyping’ (1029). Though there is some empirical support for Fricker’s claims, there is as yet no univocal account of how to combat prejudice, and there has been little if any work that directly deals with combating backlash as a defensive psychological response (Rudman and Fairchild Reference Rudman and Fairchild2004; Rudman et al. Reference Rudman, Moss-Racusin, Glick, Phelan, Devine and Plant2012). Also, many of the suggestions for counteracting prejudice, such as fostering egalitarian attitudes or promoting opportunities for emotional bonding with those of other social groups, aren’t the type of things that one can employ in the context of a testimonial exchange—it’s not as if the virtuous listener can spontaneously grow an emotional attachment to her interlocutor, for example. But, importantly, in Fricker’s response to Alcoff she claims,

As this passage shows, Fricker is not restricting her analysis of the virtue of testimonial justice to certain moves one should make in the testimonial context. The virtuous listener is prepared to engage in whatever psychological exercises could lead to a reduction of prejudice, within the bounds of practicality. Fricker’s virtue of testimonial justice differs from a more strict AM position where the focus is on what one does or monitors for in the testimonial exchange, and, therefore, Fricker’s account may escape worries about the effectiveness of a requirement that one remain alert for signs of backlash or counter-veridical motives.

The virtuous listener should certainly be aware of contemporary and future social psychological work on potential ways to combat prejudice and backlash, but this still leaves us with the question of what we should actually do in the context of receiving testimony. Again, there is little if any literature that deals with methods of effectively combating the possibility of assigning a prescriptive credibility deficit as a form of backlash, and prior to the discovery of some definitive, behavioral protocol we can employ to (at least partially) neutralize the possibility of backlash, it is still an open question as to what the virtuous listener should do in a context where testimonial injustice may occur. Neither requiring that we monitor for cues of prejudice and backlash nor requiring that we follow whatever procedure is best supported by our empirical evidence, currently gives us a clear and useful policy to utilize when receiving testimony.

Of course it’s not as if we have, or could even hope to possess, some definitive set of information (e.g. precise estimates of the base rate of false testimony, precise estimates of the likelihood that we will incorrectly perceive a speaker’s credibility in a given context, etc.) that could be used to calculate how to adjust our credibility assignments. Instead, we have a vast body of psychological literature that supports the claim that we may mishandle the words of our conspecifics in virtue of the irrational effects of deep-seated prejudice and defensive psychological responses. A rigorous analysis of the literature will fail to produce any type of precise suggestion in terms of how to alter our credibility assignments in a context in which an NPS or backlash may infect our epistemic perception. Fricker’s best proposal for how the virtuous listener should respond to the threat of prejudice might very well be that, ‘all we are able to do is to render our judgment more vague and more tentative’ (Reference Fricker2007, 92). There is certainly more philosophical and psychological work to be done, but it isn’t clear just how definite of a set of suggestions for combating something like backlash we can hope to generate or what shape these suggestions will take with further research. In turn, we need to be careful to make sure that our theory about how the virtuous listener operates and what we require of the virtuous listener is both psychologically feasible and compatible with the relevant empirical literature.