Kristie Dotson uses the term epistemic violence to denote the harm that audiences do when their ignorance prevents them from understanding the testimony of a speaker (Reference DotsonDotson 2011). A man who has never experienced gender discrimination firsthand may struggle to take a woman's reports of sexism seriously, and his failure to understand her may discourage her from reporting sexism again in the future. Dotson's work highlights the importance of being a generous and well‐informed listener, able to understand speakers as they want to be understood.

Like most philosophers who work on silencing, Dotson focuses on linguistic communication: on how speakers’ words can be ignored, dismissed, or twisted and used against them. Yet in many communicative settings, we communicate with more than just our words. Our facial expressions, gestures, and tones of voice convey a great deal of information about ourselves, much of it about what we are currently experiencing. These acts of nonlinguistic communication are no less susceptible to being silenced, distorted, and exploited. Just as audiences can do harm when they fail to understand a speaker's words, so too can they do harm when they fail to understand a speaker's nonlinguistic behavior. Thus, in addition to being generous and well‐informed listeners, we ought to invest due care in the reading of other people's facial expressions, gestures, and tones of voice, lest we harm them through our ignorance.

My aim in this article is to expand upon Dotson's analysis by identifying another type of epistemic violence that arises in the context of reading nonlinguistic expressions of emotion. Emotional misperception, as I call it, occurs when the following conditions are met:

1. 1 A misreads B's nonlinguistic expression of emotion,

2. 2 owing to reliable ignorance,

3. 3 harming B.Footnote ¹

Imagine that a white woman speaks politely but forcefully during a job interview. The interviewer expects her to be more demure, however, owing to a sexist ideology, and this expectation causes him to hear her tone as subtly contemptuous. As a result, he considers her a bad fit for the position and offers the job to someone else. In this case, the interviewer has misread the woman's emotion (condition 1), done so as a result of reliable ignorance (condition 2), and then used this misreading as a justification for denying her the job (condition 3). Allow me to clarify each of these conditions.

Condition 1 states that emotional misperception requires the misreading of emotion. I distinguish among three basic ways of misreading emotion (adapted from Reference Ekman and FriesenEkman and Friesen 2003, 141):

1. 1 False negative: B expresses an emotion, yet A perceives B's behavior as emotionally neutral.

2. 2 False positive: B does not express an emotion, yet A perceives B's behavior as emotionally expressive.

3. 3 False categorization: B expresses one emotion, yet A perceives B's behavior as expressing another emotion.

The interviewer's misreading of the woman's tone of voice is an example of a false positive: the woman did not speak in an emotional tone of voice, yet the interviewer heard her tone as subtly contemptuous.

What counts as a false categorization depends on how we individuate emotions, which is no simple matter. If we say that “fear” and “terror” name distinct emotions, for instance, then reading fear as terror would count as a false categorization. But if we say that “fear” and “terror” name the same emotion, then reading fear as terror would not count. There are many different taxonomies of emotion on offer, but as a general rule, the richer our taxonomy, the more opportunities there will be for false categorizations. For the purposes of this article, I will follow Basic Emotions Theory in assuming that there are, at the very least, eight distinct emotions that can be distinguished on the basis of facial expressions: joy, sadness, fear, surprise, anger, disgust, contempt, and shame (Reference EkmanEkman 1992). Thus, I shall assume that misreading fear as anger counts as a false categorization, but I shall remain agnostic as to whether misreading fear as terror also counts. I encourage readers to supplement this taxonomy as they see fit. Regardless of how one individuates emotions, a false categorization occurs when an audience perceives an emotion of one type as an emotion of another type.

Condition 2 states that the misreading of emotion must result from reliable ignorance. In this article I identify three sources of reliable ignorance that are responsible for the misreading of emotion. In cases of emotion stereotyping, a false generalization creates expectations about which emotions a person will or will not express in a given context, and these expectations can skew the observer's perception (Reference BlumBlum 2004, 261–62). A famous study, discussed later, found that observers perceived an infant's emotional expressions differently depending on whether they believed the infant was male or female, and that these differences reflected cultural stereotypes concerning the emotional lives of men and women (Reference Plant, Hyde, Keltner and DevinePlant et al. 2000). In the interview example, the interviewer hears the woman's tone as contemptuous because, given various cultural stereotypes, he expects her voice to be much softer and gentler than it is.

In cases of emotion apathy, an observer could invest due care in the reading of someone's emotions, but for lack of trying does not. Research suggests that people who occupy dominant social positions make more mistakes when perceiving the emotions of people who occupy subordinate social positions than vice versa (Reference Elfenbein and AmbadyElfenbein and Ambady 2002b). A plausible explanation for this asymmetry is that those with greater social standing have less to lose from making mistakes, and so take less care in reading emotional behavior. With power comes the privilege of being unconcerned with the emotions of those who lack power, and with this apathy comes an increase in emotional misperception.

Finally, in cases of emotion parochialism, an observer is ignorant of a community‐specific style of emotional expression, and so fails to read an expression correctly. Research suggests that although there are substantial cross‐cultural similarities in nonlinguistic emotional expression, there are also subtle cultural differences, amounting to distinct “dialects” of expression (Reference Elfenbein and AmbadyElfenbein and Ambady 2002a; Reference Elfenbein, Beaupré, Lévesque and HessElfenbein et al. 2007). Failures to appreciate these cultural differences can lead to misreadings of emotion.

To be clear, not all misreadings of emotion result from reliable ignorance. Sometimes we misread a person's emotions because we are distracted or because the observational conditions are poor. Sometimes we misread a person's emotions because this person has expressed her emotion in an unusual way or because the context suggests that she would be feeling a different emotion. These misreadings of emotion may even be as harmful as those that result from reliable ignorance. However, when the misreading of emotion does not result from reliable ignorance, I do not classify it as an instance of emotional misperception (condition 1 is met, but condition 2 is not).

Condition 3, finally, states that the misreading of emotion must harm the person being misread. A person can be harmed in many ways, and I use the term harm in its broadest sense. However, following Dotson and her focus on harms related to silencing, I focus on the ways in which emotional misperception can negate, undermine, and distort an individual's experience and expression of emotion. Some of these harms are analogous to those analyzed by Dotson in the realm of linguistic communication, but others are unique to nonlinguistic communication. One of the unique harms concerns “emotional labor,” or the work that individuals do to conform their emotions and expressions to social norms, often in the context of employment (Reference HochschildHochschild 1983). I suggest that emotional misperception can deny a person the fruits of her emotional labor.

In what follows, I review Dotson's definition of epistemic violence (section ), which contains three technical terms: communicative reciprocity, reliable ignorance, and pernicious ignorance. I then work through each of these terms, modifying them so that they may be applied to the misreading of emotion (II–IV). Finally, I conclude by distinguishing emotional misperception from closely related phenomena (section V).

I. Dotson on Epistemic Violence

Dotson defines epistemic violence as “a refusal, intentional or unintentional, of an audience to communicatively reciprocate a linguistic exchange owing to pernicious ignorance” (Reference DotsonDotson 2011, 238). Audiences exhibit communicative reciprocity when they “recognize one another's speech as it is meant to be taken” (Reference HornsbyHornsby 1995, 134; quoted in Reference DotsonDotson 2011, 237). In other words, reciprocity occurs when audiences successfully grasp what speakers mean to communicate. Reciprocity fails when audiences take speakers to mean something other than what they intended to mean. Suppose that I email a student and write “I have not received your paper yet; please send it to me by midnight.” Although I have used the word “please,” I intend the email to be taken as a command rather than a request. If the student responds with “I'll try to get it to you this weekend,” then evidently he has interpreted my email as a mere request, and thus he has failed to reciprocate, regardless of whether he is blameworthy for doing so.

A failure of reciprocity enacts epistemic violence only if it is the result of pernicious ignorance. Dotson defines pernicious ignorance as “any reliable ignorance that, in a given context, harms another person” (Reference DotsonDotson 2011, 238). Reliable ignorance, in turn, is “ignorance that is consistent or follows from a predictable epistemic gap in cognitive resources” (238). Does the student's failure to reciprocate my command result from pernicious ignorance? Probably not. First, the student's failure to grasp my command as a command does not seem to harm me, or at least not much. I may feel frustrated, and I may spend a few minutes trying to figure out how best to tell him that he cannot wait until the weekend to send me the paper, but I would not characterize these nuisances as harms. Second, the student's failure to grasp my command as a command is probably not the result of a reliable ignorance. It would be the result of a reliable ignorance if I were a member of a marginalized group and the student held the false belief that people like me are not in a position to issue commands to him, but because I am a straight white cis man, it is most likely that the student has simply read too much into my use of the word “please.” Thus, although the student has failed to reciprocate in our exchange, his conduct does not constitute an act of epistemic violence against me.

Dotson identifies two types of epistemic violence that arise in the context of giving and receiving testimony. Testimonial quieting occurs when audiences fail to acknowledge the credibility of speakers (242). Testimonial smothering occurs when speakers refrain from testifying to something because audiences have indicated that they will not understand or believe this testimony (244). Testimonial quieting is something that audiences do to speakers, whereas testimonial smothering is something that speakers do to themselves, given the evident hostility or incompetence of their audiences.

To these categories of epistemic violence I add another: emotional misperception. Reliable ignorance can lead audiences to misread a person's emotions, which, in turn, can harm this person. To develop this concept further, I will show how the concepts of communicative reciprocity, reliable ignorance, and pernicious ignorance may be modified to apply to the misreading of emotion.

II. Reciprocity and Emotion

Dotson's concept of communicative reciprocity, drawn from Jennifer Hornsby, is specific to illocutionary speech acts. (Illocutionary speech acts are the basic units of linguistic communication, and may take the form of assertions, commands, requests, promises, and so on.) Here is how Hornsby defines reciprocity:

In order for a speaker to issue a command, for instance, audiences must recognize the speaker's intention to have her words interpreted as a command (Reference MaitraMaitra 2009). If the audience fails to recognize this intention—perhaps the audience thinks that the speaker intends to perform a request (Reference KuklaKukla 2014)—then the speaker has failed to issue a command. Reciprocity is required not only for the successful uptake of a speech act, then, but also for the successful performance of that speech act in the first place.

I am interested in a broader notion of reciprocity, which corresponds to a broader notion of communication. In everyday face‐to‐face conversations, we communicate with more than just our words. Our facial expressions, tones of voice, and body language convey a great deal of information about what we are thinking and feeling. The smirk that flashes on a man's face after he has gotten his way betrays his smugness. A woman's closed stance reveals her distrust of others. And the raised pitch of a child's voice manifests her excitement. None of these behaviors is an illocutionary speech act, but each communicates information about a person's thoughts, motivations, and emotions. I use the term comportment to refer to any and all nonlinguistic behaviors that convey information about a person, and the term emotional comportment to refer to any and all nonlinguistic behaviors that convey information about a person's emotions, in particular. Facial expressions and tones of voice are the best examples to keep in mind.

Historically, philosophers have conceived communication in ways that exclude emotional comportment. According to the inferential model of communication, a person communicates by saying or doing something with the intention of having audiences recognize the intention to communicate (Reference GriceGrice 1957). In saying “Pass the salt,” I intend for you to recognize my intention to request the salt from you. According to the commitment model of communication, by contrast, a person communicates by making a “move” in a “language game,” thereby altering the normative statuses of participants (Reference Kukla and LanceKukla and Lance 2009). In asserting “The nearest bus stop is two blocks north,” I accept justificatory responsibility for the truth of this statement and I license you to use this claim in your practical deliberations. These models are well‐suited to account for how illocutionary speech acts communicate, but poorly suited to account for how nonlinguistic emotional expressions communicate. None of the examples of emotional comportment that I mentioned above (the smug smirk, the distrustful posture, the excited tone of voice) is performed with communicative intentions, and none of them is a move in a language game.

Another theory of communication, first developed in the biological sciences, can accommodate emotional comportment. According to the signaling model of communication, A signals B just in case A conveys information about B and was designed to do so (Reference GreenGreen 2007, 212). Some signals are designed by natural selection; others are designed by human intentions and conventions. In every case, a person communicates by saying or doing something that conveys information and that was designed to do so. This model is very well suited to account for nonlinguistic expressions of emotion, regardless of whether we think that they were designed by natural or cultural forces.

Corresponding to this broader model of communication is a broader conception of communicative reciprocity. Here is how I define reciprocity as it pertains specifically to emotional comportment (using Hornsby's definition as a template):

In order for a person to signal sadness with a pout, for instance, the pout must convey information about sadness and have been designed to do so. If the audience misreads the pout, then the pout has not done what it was designed to do. The audience's failure to reciprocate thus prevents the pout from having successfully signaled sadness in the first place.

III. Reliable Ignorance and Emotion

A remarkable—and indeed, often remarked—feature of emotional comportment is that it transcends cultural and linguistic boundaries. Charles Darwin wondered whether “the same expressions and gestures prevail… with all the races of mankind” (Reference DarwinDarwin 2009, 22), and contemporary research suggests there are substantial cross‐cultural similarities in both the production and perception of emotional comportment.Footnote ² One striking study found that blind paralympic athletes representing twenty‐three countries from six continents produced the same spontaneous facial expression of joy upon winning their final match and the same spontaneous facial expressions of sadness and disgust upon losing their final match—and that these spontaneous facial expressions were the same as those produced by sighted athletes, regardless of nationality (Reference Matsumoto and WillinghamMatsumoto and Willingham 2009). Another study, a meta‐analysis of around one hundred published experiments, found that almost 90% of subjects across multiple cultures agreed in their reading of photographs depicting facial expressions of joy, that 60–70% agreed in their reading of photographs depicting facial expressions of anger, disgust, sadness, and surprise, and that just under 60% agreed in the reading of photographs depicting facial expressions of fear (Reference Elfenbein and AmbadyElfenbein and Ambady 2002a). In line with these findings, neurobiologists hypothesize that there are species‐typical neural networks dedicated to the production and perception of emotional comportment (Reference de Gelderde Gelder 2016).

Despite these cross‐cultural similarities, there is also evidence of an in‐group advantage. In other words, individuals are slightly—but only slightly—better at reading the emotional comportment of people like them—that is, people who share the same race, culture, and nationality (Reference Elfenbein and AmbadyElfenbein and Ambady 2002b). Why is that? Three hypotheses have been proposed and tested. The Stereotype Bias Hypothesis holds that individuals are more likely to stereotype out‐group members, and so more likely to misread their emotional comportment (Reference Kirouac, Hess, Philippot, Feldman and CoatsKirouac and Hess 1999). The Ethnic Bias Hypothesis holds that individuals are less motivated to understand out‐group members’ emotions, and so more likely to misread them (Reference Kilbride and YarczowerKilbride and Yarczower 1983). And the Emotion Dialect Hypothesis holds that there are subtle differences in emotional comportment across groups that make it harder—but only a little harder—to read out‐group members’ emotions correctly (Reference Elfenbein, Beaupré, Lévesque and HessElfenbein et al. 2007). The available evidence suggests that all three hypotheses are partly right—that in some situations errors result from stereotyping, that in other situations they result from a lack of motivation, and that in still other situations they result from out‐group members expressing themselves in different ways from in‐group members.

Anytime there is an error in reading a person's emotions, there has been a failure of emotional reciprocity. In order for a failure of emotional reciprocity to enact epistemic violence, however, it must be the result of pernicious ignorance. Pernicious ignorance has two core features: it is reliable and it is harmful. I'll discuss reliable ignorance in this section, and harmful ignorance in the next.

Dotson defines reliable ignorance as “ignorance that is consistent or follows from a predictable epistemic gap in cognitive resources” (Reference DotsonDotson 2011, 238). Often, reliable ignorance is situated ignorance, or ignorance that “follows from one's social positioning” (248). In what follows, I identify three sources of reliable, situated ignorance responsible for the misreading of emotion, corresponding to the three hypotheses I mentioned earlier, which I call emotion stereotyping, emotion apathy, and emotion parochialism. (The first two involve “consistent” ignorance; the third involves “a predictable epistemic gap in cognitive resources.”) These three sources often conspire with one another, and many instances of emotional misperception can be traced to a combination of two or three.

A stereotype is a “fixed, often simplistic generalization about a particular group or class of people” (Reference CardwellCardwell 2014, 227). Examples include “Black men are dangerous” and “women are bad at math.” One way that stereotypes influence behavior is by distorting perception (Reference BlumBlum 2004, 261–62). People who hold the stereotype that Black men are dangerous perceive Black men to be larger than they in fact are (Reference Wilson, Hugenberg and RuleWilson, Hugenberg, and Rule 2017). Parents who hold the stereotype that women are bad at math are more likely to perceive their daughters as struggling with math homework, even when the daughters’ math skills are equal to or greater than those of their peers (Reference Eccles, Jacobs and HaroldEccles, Jacobs, and Harold 1990).

An emotion stereotype, in turn, is a fixed, often simplistic generalization about the emotive tendencies of a particular group of people. In Black Feminist Thought, Patricia Hill Collins discusses two common stereotypes of Black women: the “Mammy” and the “Matriarch.” The Mammy is “the faithful, obedient domestic servant,” who is “loving, nurturing, and caring [toward] her White children and ‘family’” (Reference CollinsCollins 2008, 73). The Matriarch, by contrast, is “overly aggressive” (74). Later we shall meet a Black professor who feels caught between these two emotion stereotypes: her white students expect her to be either loving and nurturing or aggressive toward them, and when she isn't loving or nurturing, the students misread her comportment as aggressive.

Emotion stereotypes, like other types of stereotypes, can distort perception. Women are believed to be more emotional than men, for instance, and although there is little evidence to support this stereotype, research confirms that both men and women perceive women to be more emotional than men (Reference Barrett and Bliss‐MoreauBarrett and Bliss‐Moreau 2009). At the same time, women are thought to experience sadness, embarrassment, and guilt more frequently than men, whereas men are thought to experience anger and pride more frequently than women. In a famous experiment, subjects were shown a video recording of an infant and asked to identify the emotion that the infant expressed (Reference Plant, Hyde, Keltner and DevinePlant et al. 2000). When the subjects were told that the infant was female, they were more likely to see sadness. When told that the same infant was male, they were more likely to see anger. In another experiment, subjects were shown a photograph of a person performing an ambiguous anger/sadness expression (Reference Plant, Kling and SmithPlant, Kling, and Smith 2004). When the photograph was altered to make the person appear female, observers were more likely to see sadness. When it was altered to make the person appear male, observers were more likely to see anger.

Emotional misperception will often be driven by emotion stereotypes like these. When the interviewer mishears the woman's tone as subtly contemptuous, it is because he holds a stereotype that causes him to expect her tone to be softer and so to interpret her assertiveness as contemptuousness. The stereotype creates a reliable ignorance that can, in turn, result in acts of emotional misperception.

A second, related source of reliable ignorance is emotion apathy. Emotion apathy occurs when an individual does not invest due care in the reading of someone's emotions, owing to a lack of trying. In order to invest due care in the reading of someone's emotions, one ought typically to pay attention to the person's emotional comportment, take into account one's knowledge of this person's emotional style, and evaluate the context in which this comportment occurs. We often do these things without thinking about them, but they do require cognitive effort. Research suggests that although people can reliably identify emotional comportment correctly when they take the care to do so, accuracy levels sink as people become distracted or occupied with other tasks (Reference Stein, Peelen, Funk and SeidlStein et al. 2010). In cases of emotion apathy, a person has the cognitive resources to invest due care in the reading of someone's emotions, but doesn't for lack of trying.

Earlier I mentioned the in‐group advantage. Individuals are, on average, slightly better at reading the emotions of people like them than the emotions of people unlike them. That being said, research also suggests that some groups of people are worse than others at reading the emotions of people unlike them (Reference Elfenbein and AmbadyElfenbein and Ambady 2002b). Men are more likely to misread the emotions of women than vice versa (Reference HenleyHenley 1977). White people are more likely to misread the emotions of Black people than vice versa (Reference Gitter, Black and MostofskyGitter, Black, and Mostofsky 1972). And natural‐born citizens are more likely to misread the emotions of immigrants than vice versa (Reference MachidaMachida 1986). One plausible explanation for why men, white people, and the native‐born are more likely to misread the emotions of women, people of color, and immigrants, respectively, is that given the unequal distribution of social power and privilege, the latter typically have more to lose from misreading the emotions of the former, and thus will typically dedicate greater attention and care to reading the former's emotions. Inversely, those who enjoy social power and privilege typically have less to lose, and thus may be more likely to exhibit emotion apathy. This asymmetry is important because it explains why emotional misperception will disproportionately affect those who are already disadvantaged. Since men are more likely to misread the emotions of women, for instance, women are more likely to be disadvantaged from having their emotions misread.

Emotion apathy, as a source of reliable ignorance, often drives emotional misperception, either independently of or in conjunction with emotion stereotyping. When a person does not invest due care in the reading of someone's emotional comportment, it is more likely that this person will fall back on low‐effort cognitive processing, like stereotyping, to make sense of this comportment. So emotion apathy and emotion stereotyping will often be co‐conspirators in the pernicious misreading of emotion.

A third source of reliable ignorance is emotion parochialism. Evidence suggests that there are subtle differences in emotional comportment between groups—distinct “dialects” rather than distinct “languages,” as it were—and that because of this, errors are more frequent in inter‐ than in intra‐group readings of emotion (Reference Elfenbein, Beaupré, Lévesque and HessElfenbein et al. 2007). One recent study reported that Western Caucasian subjects produce—and look for—greater movements around the mouth in episodes of anger, fear, and disgust, whereas East Asian subjects produce—and look for—greater movements around the eyes in episodes of these emotions (Reference Jack, Garrod, Yu, Caldara and SchynsJack et al. 2012). Were a Western Caucasian individual to misread an East Asian individual's expression of fear, given that the former was focusing narrowly on the latter's mouth, then the misreading would owe to emotion parochialism. Applying Chomsky's competence/performance distinction, we could say that emotion stereotyping and emotion apathy lead competent readers of emotion to commit performance errors, whereas emotion parochialism leads to a lack of competence in reading certain emotions.

The writer Collier Meyerson observes that Black women are especially susceptible to having their emotions misread:

Meyerson argues that white teachers and police officers do not know how to read Black women's expressions—that this is a case of emotion parochialism. I suspect that all three sources of reliable ignorance may be involved. The white teachers and police officers may be unfamiliar with the distinctive ways in which Black women express their emotions (emotion parochialism). But they may also be investing less effort in reading these emotions than they would invest in reading white women's emotions (emotion apathy). And they may be drawing on emotion stereotypes to make sense of the unfamiliar behavior that they are not taking enough care to interpret (emotion stereotyping). Multiple sources of reliable ignorance may be present, and may fuel and feed off one another. For this reason (among others), emotional misperception is no easy fix. Training white teachers and police officers to read different dialects of emotional comportment may not reduce the frequency of misreadings, given that pernicious stereotypes and social dynamics may also drive these misreadings.

To recap: there are at least three sources of reliable ignorance responsible for the misreading of emotion (emotion stereotyping, emotion apathy, and emotion parochialism), where reliable ignorance is a necessary, but not sufficient, condition for the enactment of epistemic violence.

IV. Pernicious Ignorance and Emotion

Reliable ignorance becomes “pernicious ignorance,” and thus enacts epistemic violence, only if it also harms the person whose emotions have been misread. But how can the misreading of a person's emotions be harmful? In her discussion of how epistemic violence can harm speakers in the context of giving testimony, Dotson focuses on the harms of silencing. She argues that repeated instances of testimonial quieting (the audience's refusal to regard a speaker's testimony as credible) can lead to testimonial smothering (the speaker's refusal to present testimony, given the evident incompetence or hostility of audiences) (Reference DotsonDotson 2011, 250).

Emotional misperception can also silence. Earlier I cited research suggesting that people often misread women's concentration on math problems as frustration (Reference Eccles, Jacobs and HaroldEccles, Jacobs, and Harold 1990). This is an example of emotion stereotyping, since the false generalization that women are bad at math generates the expectation that women will express frustration when doing math problems, and this expectation may cause audiences to see frustration where there is none (false positive). This misreading of emotion silences if it prevents audiences from believing women when they claim to understand the math problems. More generally, emotional misperception silences when an audience refuses to believe a person's claims about what she is experiencing, given that the audience reads her comportment as evincing something else. If a man thinks that a woman sounds upset, then he may disregard her repeated claims not to be upset. Famously, research shows that audiences reliably take tone and body language to trump speech when they send mixed messages, so the misreading of tone and body language can reliably undermine a person's testimony (Reference Mehrabian and WienerMehrabian and Wiener 1967).

Another way in which emotional misperception can silence is related to what Sue Campbell calls emotional dismissal, or the refusal to take women's emotions seriously based on the sexist assumption that women are oversensitive. Campbell observes that when a man labels a woman's anger as “bitterness,” he has already dismissed it as irrational, seeking to explain it in terms of the woman's deviant emotional dispositions rather than in terms of the injustice that elicited it (Reference CampbellCampbell 1994; see also Reference Spelman, Garry and PearsallSpelman 1989). This is not a case of emotional misperception, however, since the woman's emotion is not misread; rather, this emotion is used as a ground for dismissing the woman as irrational. Related examples that do count as emotional misperception can be found in the writings of Audre Lorde and bell hooks, who observe that white feminists are quick to exclude Black feminists on account of their supposedly destructive anger (Reference hookshooks 1984, 12–13; Reference LordeLorde 1984, 124–33; see also Reference AhmedAhmed 2010, 50–87). In these cases, the Black feminists are not angry, but white feminists misread them as angry (false positive, owing to emotion stereotyping), and then use this imagined anger as a ground for dismissing them. Emotional dismissal overlaps with emotional misperception, then, when one misreads a woman as having an emotion that one devalues, and then dismisses her on the basis of that purported emotion.

A third way in which emotional misperception can harm the person whose emotion is misread is related to what the sociologist Arlie Hochschild calls “emotional labor” (Reference HochschildHochschild 1983). Emotional labor is the effort that individuals go to in order to emote in accordance with display rules (typically in the context of employment). Display rules are spoken or unspoken norms governing the emotional comportment that a person ought (or ought not) to produce in specific circumstances. Examples of display rules include “women ought to smile while in public,” “one ought not to express (unwarranted) contempt in a job interview,” and “men ought never to express sadness or fear.”Footnote ³ In her classic study of Delta flight attendants, Hochschild observes that these flight attendants are required to attend seminars that teach proper emotional comportment, and that they are repeatedly told to “Really work on your smiles” and to “Relax and smile” (Reference HochschildHochschild 1983, 104–105). Today, Walmart hires “greeters” to meet customers at the door with a warm smile. For many service‐industry positions, emotional labor is written into the job description: beyond assisting customers, one is expected to appear cheerful and to suppress negative emotions. If one fails to emote in accordance with these display rules, then one may be disciplined.

The misreading of emotion can harm someone if she is expected to emote in accordance with display rules, does so, but is misperceived as violating these rules, and thereby becomes subject to reprimand or other disciplinary measure. Recall the example with which this article began. A woman emotes in accordance with display rules at a job interview, but the interviewer misreads her expression owing to an emotion stereotype, judges her to have violated these rules, and removes her from consideration for the job as a result.

The connection between emotional misperception and emotional labor is significant, given that Black women are overrepresented in the service industry. The individuals who are most likely to make a living by managing their emotional comportment are also the most likely to have their emotional comportment misread, and thus also the most likely to be unjustly disciplined on the basis of this misreading. In this way, emotional misperception creates a double bind that poses a threat to Black women's labor‐power. If you do not emote in accordance with workplace norms, then you may be punished. Even if you do emote in accordance with these norms, then, depending on the attentiveness and good will of your manager, you may be punished anyway. The effect of emotional misperception here is broadly analogous to silencing: expressions of job competence go unrecognized as such, and, preposterously, are potentially punished as indicating job incompetence.

I have discussed a number of ways in which emotional misperception can harm the person whose emotion is misread. In some cases, the person is silenced, such that her testimony is dismissed or disbelieved on account of how an audience perceives her emotional comportment. In other cases, the person is punished for failing to emote in accordance with display rules, despite the fact that she has emoted in accordance with them. I suspect that there are many other ways in which emotional misperception can do harm. My view is that emotional misperception harms people in multiply different, often interlocking and mutually reinforcing ways, which cannot be easily disentangled or distinguished.

V. Epistemic Violence and Emotional Misperception

Let's review. Dotson defines epistemic violence as “a refusal, intentional or unintentional, of an audience to communicatively reciprocate a linguistic exchange owing to pernicious [i.e., reliable and harmful] ignorance” (Reference DotsonDotson 2011, 238). I have defined emotional misperception as a failure to communicatively reciprocate someone's emotional comportment owing to pernicious ignorance. This ignorance is reliable if it is grounded in emotion stereotyping, emotion apathy, or emotional parochialism. It is harmful if it silences, delegitimizes, or otherwise undermines the person whose emotion has been misread. In sum, emotional misperception is a type of epistemic violence that occurs when an audience's reliable ignorance causes the misreading of a person's nonlinguistic expression of emotion, which, in turn, harms that person.

Consider another example of emotional misperception. In a fascinating study of how race affects professors’ experiences in the classroom, Roxanna Harlow reports one professor's thoughts on the stereotype of the “angry Black woman”:

This professor, who teaches at a predominantly white university, finds that many white students expect her to be maternal. When she doesn't express positive emotion toward them, however, they mistake her lack of expression for an expression of negative emotion—of “attitude.” And as a result, they cease to respect her authority in the classroom, which can harm her in many different ways. The professor reasonably speculates that they misread her behavior because they expect her to fit the stereotype of either the “Mammy” or the “Matriarch” (Reference CollinsCollins 2008, 71–75; see also Reference HarrisHarris 1982; Reference LordeLorde 1984, 126). This expectation skews their perception, leading them to see “attitude” where there is none. So we have a false positive, resulting from emotion stereotyping, and this false positive harms the teacher by undermining her authority in the classroom.

I have taken care to distinguish emotional misperception from a number of closely related phenomena. First, emotional misperception is distinct from cases in which a person is harmed as a result of having her emotions accurately read. In her discussion of emotional dismissal, Campbell describes how women who express their emotions are routinely dismissed as being irrational (Reference CampbellCampbell 1994). (Women who do not express their emotions are often no better off, as they may be dismissed as being cold and uncaring.) To qualify as emotional misperception, a person's emotions must be misread.

Second, emotional misperception is distinct from cases in which the misreading of emotion that harms someone is caused not by reliable ignorance, but by nonreliable ignorance. I might take care to read someone's emotions, without the influence of stereotypes, and with full knowledge of this this person's expressive style, yet still make a mistake. This mistake may even harm the person in the ways I have explored. But to the extent that the reliable ignorance condition is not met, I would characterize this case not as emotional misperception, but as a closely related phenomenon.

Finally, emotional misperception is distinct from cases in which the misreading of emotion results from reliable ignorance yet does no harm. I might reliably mistake a person's surprise expression for amusement, yet not respond in a way that silences her, delegitimizes her, or otherwise disadvantages her. The category of emotional misperception includes only misreadings that do harm.

In distinguishing emotional misperception from these related phenomena I do not mean to imply that the latter are any less important or worthy of study. Rather, my aim is simply to highlight a phenomenon that is analogous to Dotson's categories of epistemic violence, yet that appears in the context of nonlinguistic communication.

A question that I have not explored in this article is whether emotional misperception is always bad. Although emotional misperception is, by definition, harmful, it may occur in circumstances in which it is, all things considered, morally justifiable. For instance, emotional misperception could be an act of resistance or solidarity within systems of oppression. Given the precedent of men dismissing women who complain about sexism as oversensitive “snowflakes,” many women have begun dismissing men who complain about “reverse sexism” as oversensitive “broflakes.” Both dismissals may at times result from emotional misperception (when the complaints that are dismissed are not actually impassioned), but I approve of the dismissal of “broflakes” as a clever strategy for resisting sexism. There is much more to say on the virtues and limitations of practicing epistemic nonviolence, but here I leave it as a tantalizing prospect for future research.

My goal in writing this article has been to awaken a sense of responsibility for how we read and respond to the emotions of those around us. Before dismissing someone for overreacting, before complaining to a manager about an employee's aggressive tone, and before judging someone on her ability to manage her emotions, one should pause to think: how might my prejudices be coloring my perception of this person's comportment, and how might my judgment be harmful? The misreading of emotion may not seem like a grave injustice, all things considered, but the ease with which it feeds on and fuels other injustices makes it a proper object of moral consideration and, more important, moral reform.