Journal of Controversial Ideas

It ought to go without saying that perception is not necessarily the same as reality. Perception is subject to error. In The Brothers Karamazov the volatile Dmitri storms into his father’s house, assaults a servant who stands in his way, and seems prepared to murder his father on the spot, all because the woman who obsesses him has paid the senior Karamazov a visit – or so he believes. “She’s here! I saw her turn toward the house just now,” he exclaims. In point of fact, Grushenka is not there. Sometimes our eyes deceive us.

As uncontroversial as the non-identity of perception and reality ought to be, many appear to believe that this principle should be waived in at least one instance, that of racial discrimination. In the medical literature there seems to be an unwritten rule that racial discrimination is in the eye of the beholder, and that the distinction (if any) between actual discrimination and the perception of discrimination by minority patients is nominal. Here I argue that the distinction between real and perceived discrimination should be handled with care, not least because it matters. If perceived discrimination “has been associated with delays or failures to seek treatment” for many conditions,1 and yet perception is fallible, then the difference between actual and perceived discrimination becomes consequential indeed. Significantly, and characteristically, the article just quoted uses “discrimination” and “perceived discrimination” as interchangeable terms.

A general practice in the literature concerned with racial disparities in medicine, the use of “discrimination” and “perceived discrimination” as equivalents implies that any distinction between them is purely verbal. Consider this passage in the literature’s founding document, the Institute of Medicine’s report Unequal Treatment (2003):

It is reasonable to assume that experiences of real or perceived discrimination in healthcare settings, as evidenced by providers’ overt behavior ... or more subtle, subjective mistreatment (e.g., healthcare providers’ low expectations for compliance or expressions of low empathy for minority patients) can affect patients’ feelings about their clinical relationships and thereby dampen their interest in vigorous diagnostic and therapeutic measures.2

The posited distinction between real and perceived discrimination dissolves immediately into one between “overt” and “subtle” discrimination – that is, between discrimination which is self-evident and discrimination whose existence must be inferred, but which is no less real. Given the authors’ stated belief that unconscious bias expresses itself in “subtle and indirect ways,”3 the latter of the two categories of discrimination is by no means illusory. (The subtlety of implicit bias soon became the axiom of an entire literature.) The distinction between “overt” and “subtle” discrimination thus boils down to the difference between rare outbursts of undiluted racism and common manifestations of disguised racism. What happened to the distinction between reality and perception? Before our eyes, the possibility of perceiving discrimination where it does not exist has vanished.

The equation of perceived and real discrimination runs through the medical literature. A typical study, one concerning diabetes, concludes both that (a) “racial/ethnic discrimination is an important barrier to diabetes management” and (b) “patients’ perceptions of discrimination may be a significant barrier to the effective management of disease, particularly for such diseases as diabetes that require collaboration between the patient and provider.”4 As if to confirm the identity of perceived and real discrimination, this study has been cited in turn as evidence of the “systemic” racism of American medicine,5 even though only 4% of surveyed respondents, 26% of whom were black or Hispanic, reported feeling they had been treated unfairly or disrespectfully because of race or ethnicity by “the doctor or medical staff” over the preceding two years.

Perceived and actual discrimination come to the same thing in the eyes of the medical literature not because people are entitled to their errors of perception (if any) but because the experience of oppression has sharpened the perceptions of minority patients, enabling them to discern the signs of discrimination where other observers would not. If “survey research … indicates that minority patients perceive higher levels of discrimination in healthcare than non-minorities,”6 this means not that minority patients overperceive discrimination but that they perceive discrimination invisible to those less familiar with it. Ordinarily, one would suppose that something as ambiguous as a subliminal bias would be easy to misinterpret, but in the literature on racial bias in medicine the objects of this bias possess an unerring ability to see through subtly distorted behavior. But assuming that racial bias in the doctor’s office is now likely to manifest itself not in expressions of outright contempt but in the nonspecific form of discomfort or anxiety on the part of the doctor (as some maintain),7 how is the patient to distinguish between a bigot and someone who simply appears uneasy? In fact, if “It’s hard to identify discrimination because they don’t show it” (as a black participant in a focus group remarks in Unequal Treatment),8 then how can anyone, much less an entire subset of patients, discern discrimination unerringly at all?

Such questions do not arise in the literature, which tirelessly emphasizes (a) the subtlety of a bias elusive enough to evade the awareness of the biased person and (b) the accuracy of the minority patient’s perception of this bias, as if there were not so much as the possibility of a conflict between (a) and (b). If minority patients did not possess an exceptional acuity of perception – if their perception were fallible – they could lose interest in “vigorous diagnostic and therapeutic measures” on the basis of mistaken beliefs about their doctors and the institution of medicine. Like the literature as a whole, the passage from Unequal Treatment noting the risk of forgone treatment does not appear to recognize this as a possibility; that is, if a minority patient should be offended by a doctor’s behavioral signals, his or her interpretation of the signals is presumed accurate. It is presumed, for example, that the patient is correct that the doctor has targeted him or her out for certain subtle “expressions of low empathy,” even though the patient does not and cannot know how the doctor behaves in the privacy of the examining room with other patients. Have the authors of Unequal Treatment never met a cold doctor?

The literature does not concern itself with the possibility of unfounded perceptions of bias because, in its judgment, the perceptions of minority patients are attuned to reality. They are acute because they have to be in order to detect the biases that warp the attitudes and behavior of others in an era when almost everyone (doctors included) professes abhorrence of racism, and when racism that persists must therefore take refuge in the covert or the ambiguous.

Well adapted to a society that no longer tolerates overt racism, implicit bias works below the surface, outside the knowledge and control of those who harbor it. In theory, a doctor who suffers from implicit bias will tend to fall back on this automatic mechanism in the course of making decisions under stress,9 as if he or she were sleepwalking, morally speaking. Such a habit would seem to pose great dangers, and many argue that it is in fact responsible for profound disparities of outcome between black and white patients.

At present, the established means of certifying the existence of unconscious bias is an online exercise, the Implicit Association Test (IAT). While the IAT has its controversies, my concern here is with the interpretation of behavior, not psychological testing, and if a bias is so well concealed that it eludes the awareness of its holder, it seems to me improbable that it will at the same time be so manifest that others cannot fail to perceive it accurately. As it happens, there seems to be no behavioral characteristic too minor to serve as a signal of bias; two proposed in Unequal Treatment are “reduced eye contact” and an “increased rate of blinking” in the presence of a minority patient.10 While such tics might be unconscious, and while a patient would notice that the doctor isn’t making eye contact or blinks a lot, how does the patient know the doctor behaves differently with non-minority patients? Although we are informed in the literature that “Black patients are able to detect provider behaviors associated with implicit racial bias,”11 exactly how is one to say that a rate is “reduced” or “increased” without knowing what it normally is?

Like the rate of eye contact, many of the indices of implicit bias seem to be relative. It is said that the implicitly biased doctor who meets with black patients will (for example) speak faster, have shorter visits, show less warmth, and use more first-person plural pronouns,12 all of which imply a standard of comparison. How do I know that my doctor spends less time with me and speaks faster unless I know his or her ordinary practice? Ironically enough, some have found that the implicit bias of doctors is associated with longer visits and slower speech.13

If unconscious bias seems hard to pin down, it has to be hard to pin down, according to the theory behind it. As the argument goes, the unconsciously biased must disguise their behavior as normal because they might become aware of their bias if it were manifested too clearly. (The anxiety betrayed in behavioral tics like excessive blinking thus suggests the biased doctor’s fear of discovering his or her own guilty secret.) But if behavior tainted with implicit bias looks normal or ambiguously normal to the naked eye, how are observers to know that it really is an expression of bias? In the judgment of some, ambiguity itself is both a warning sign of bias and an aggravator of injury. “Several recent studies have noted stronger, more negative effects from subtle or ambiguous racial encounters than from blatant ones.”14 And so behavior that might not be biased at all becomes especially injurious in the eyes of prosecutorial critics.

Even though ambiguity invites misinterpretation, and even though there is nothing about being a minority patient that exempts one from the fallibility of human perception, the possibility of misreading ambiguous behavior as a sign of implicit bias does not exist for the literature. As we have seen, according to the authors of Unequal Treatment, when black patients react to “low empathy” or the like, they correctly pick up the signals sent out by the doctor. Their understanding of his or her behavior is as accurate as if they had happened upon a hostile comment written and signed by the doctor in their medical records. According to Dovidio et al., white people in general are “unaware of subtle cues of bias that blacks perceive.”15 According to Hagiwara et al., “research has clearly shown that Black patients do react negatively to subtle manifestations of physician implicit racial bias, although the specific behaviors that elicit such effects are yet to be uncovered.”16 That is, we can be sure black patients read the manifestations of bias correctly even though we do not know quite what they are.

Hagiwara et al. suggest that one telltale sign of bias is the physician’s excessive use of the first-person plural, as in “We need to make sure that you take your medication”17 – the one and only example of the offending practice given. Even as their study takes its place in a literature emphasizing the need to engage minority patients in “team-building,” the authors fail to ask themselves how one can build a team without using the word “we,” just as they fail to consider how much more dictatorial “You need to make sure that you take your medication” sounds than “We need to make sure that you take your medication,” and fail to consider that statements like “We’ll take good care of you” are also framed in the first-person plural.18 Taking for granted the accurate decoding of elusive cues in the clinical encounter, Hagiwara et al. efface the possibility that a patient might construe evidence as questionably as they themselves do. Some of the same authors who flag the physician’s use of the first-person plural co-authored an almost simultaneous publication that includes the following statement: “One method [of improving physician–patient communication] tries to change the perspective of people from different racial groups from an ‘us versus them’ to a sense of ‘we-ness.’”19

Viewed through a lens of suspicion, even exemplary motives appear stained with implicit bias. In a discussion of stereotypes in the clinical encounter, the authors of Unequal Treatment observe that “physicians may be less aggressive in seeking minority patients’ consent for certain medical procedures, out of a heightened (but nonetheless stereotyped) concern that minority patients’ wishes to avoid aggressive or new healthcare technologies should be respected, or because of a desire to foster a sense of empowerment among minority patients relative to treatment decisions.”20 It appears that not being overbearing, showing respect for an aversion to “aggressive” measures, even nurturing a “sense of empowerment” can all be construed as demeaning, stereotype-driven behavior. A patient who entered the clinical encounter armed with this sort of interpretive license could assess virtually anything the doctor said or did as evidence of bias. Where there’s a will there’s a way.

A small study has shown that people can indeed perceive racism where none exists, although this outcome was so unexpected that the authors seem not to know what to do with it. I refer to a 2006 study of “perceived racism” in which black subjects had higher cardiovascular responses (CVRs) to a “nonracist” than a “blatantly racist” audiotape, both depicting “the unfair treatment of a customer in a shopping scenario.”21 (The contents of the tape are not given.) Though the nonracist tape was specifically fashioned to be “without discriminatory cues,” 22 of the 37 subjects exposed to it discerned at least “a little bit” of racism, with 7 discerning at least “a great deal.” This is certainly not the result we would expect if being a minority made subjects better perceivers. However, the authors discuss this result as if the 22 subjects discerned actual covert racism, even stating in the abstract that “subtle racism is a psychosocial stressor that erodes health through chronically elevated CVR.”

Did the authors come to believe that they unwittingly planted biases of their own in the innocuous tape, which 22 of 37 subjects then detected? That seems unlikely. Nevertheless, they write as if the tape were actually laced with the sort of ambiguous racism that reportedly permeates a society where racism is rarely expressed openly. “The finding that high perceptions of racism in the NRC [nonracist condition] are associated with higher elevations in BP [blood pressure] is notable, given the observations that older, blatant forms of racism are increasingly being supplanted by more subtle forms of racism.”22 The authors are not the only ones to conclude that the neutral tape was not neutral after all. Williams and Mohammed cite this study as a demonstration that ambiguous racism can be more noxious than explicit racism.23 Also representing the misconstruction of a neutral encounter as the accurate decoding of a script tainted with subtle racism, a literature review in PLoS One describes the two-tape study as follows: “Importantly, a RCT of black normotensive men found elevated BP in response to racially ambiguous stimuli, suggesting that even subtle forms of racism (not just exposure to blatant discrimination) can induce these responses.”24

The diabetes study mentioned above also documents the perception of discrimination where none exists.

In the disparities literature many argue that poor communication between the patient and the biased doctor leads to poor adherence and, thereby, a poor outcome. (Recall, for example, the claim in Unequal Treatment that subtle disrespect shown by the doctor “can affect patients’ feelings about their clinical relationships and thereby dampen their interest in vigorous diagnostic and therapeutic measures.”) But if adherence measures bias, then no bias was actually shown by providers in the diabetes study, because white and non-white patients adhered at identical rates to recommended procedures for diabetes management. Specifically, “among Whites, 89.4%, 71.1%, 73.5%, and 87.1% reported HbA1c, foot exam, eye exam, and blood pressure tests, respectively. Correspondingly, among non-Whites the estimates were 88.7%, 74.9%, 74.4%, and 90.6%, respectively.”25 This presumptive evidence that providers showed no bias against minority patients did not keep a small number of patients from “perceiving” discrimination anyway.

While investigations like this one and the two-tape study show that it is indeed possible to perceive racism where none exists, such misperception is not just theoretical. In 2002, shortly before Unequal Treatment was published, a study reported that almost 80% of surveyed black persons believed “that someone like them would be used as a guinea pig without his or her consent”26 by their doctor: a finding that in all likelihood reflects the legacy of suspicion left by the notorious Tuskegee Syphilis Experiment, which ran from 1932 to 1972.27 The 80% figure represents a case of tragic misperception on a massive scale, not a case in which the perceivers were able to discern something that escaped the duller vision of others. In the twenty-first century, doctors were not conscripting black patients into dangerous experiments without their knowledge. However, the mistrust of medicine engendered by Tuskegee was strong enough to persist in the absence of such experiments, and perhaps regardless of the need for medical treatment itself. And it is possible that the influence of Tuskegee continues to this day. The mistrust of the Covid vaccine among black Americans, even in the face of disproportional Covid mortality in their community, may reflect a suspicion of vaccination as such, born of “the erroneous belief (despite evidence to the contrary) that the men in the Tuskegee Syphilis Study were actually injected with syphilis.”28

By continually emphasizing that white doctors are actuated by hidden biases that portend trouble for minority patients, the literature in the tradition of Unequal Treatment risks reinforcing all manner of misperceptions. It plays with fire.

If the perceptions of minority patients can be mistaken or unfounded like everyone else’s, and if patients can forgo necessary care on the basis of their impressions of medicine in general or their doctors and nurses in particular, then the authors of medical studies, reviews, and commentaries would do well not to redouble the problem by treating perceptions of discrimination as indubitable evidence of discrimination per se. If investigators find that diabetic patients’ perceptions of racial or ethnic discrimination can inhibit their management of their disease, they should not report that “racial/ethnic discrimination is an important barrier to diabetes management.” Perceptions or impressions can be inaccurate, and often are. This is no academic point. Especially at a time when allegations of discrimination are magnified by the traffickers of news and opinion, the distinction between real and alleged discrimination should be respected, not blurred on principle. The accusations of racism now regularly to be found in the medical literature do not, after all, simply remain there. They are picked up by the press and others, retailed, and established by repetition. Two decades ago, the press was quick to pronounce the Institute of Medicine report proof of the racism of American medicine.29

Given the difficulty of linking implicit bias to eventual outcomes,30 some look to the clinical encounter as the best source of evidence that cues of hostility or merely anxiety compromise treatment. Such studies characteristically identify a suspect pattern in the doctor’s behavior and suggest that it is enough to alienate patients and lead them to adhere less well to treatment regimens. As noted above, one study investigating signs of discrimination in the clinical encounter flags the doctor’s use of the word “we” as an expression of “social dominance” and a marker of implicit bias – and this even as the literature on the clinical encounter, including work by some of the same authors, calls on doctors to engage in more team-building. Two of the same authors who identify longer visits as a marker of racial bias in a 2012 study31 point out in a paper included in Unequal Treatment itself that doctors talk more with and provide more information to patients of higher social standing,32 in which case longer visits would presumably signify socioeconomic privilege. In a related 2004 paper (cited more than 1000 times), doctors showed a lack of patient-centeredness in their meetings with black patients by offering too much “biomedical information and counseling.”33 Does anyone doubt how critics would judge doctors who offered black patients too little medical information and counseling? But despite finding that doctors displayed a 23% higher level of “verbal dominance” with black patients, the study offers not a single example of a doctor lecturing, belittling, or ignoring a patient over the course of the 458 audiotaped clinical encounters in its data set. One factor in particular makes it unlikely that the marginally higher showing of “verbal dominance” with black patients has any nefarious significance. The fact is that 48% of the black patients’ meetings in this study were with black doctors, as compared to 36% with white and 14% with Asian doctors.

Who would want to be hectored by a doctor or otherwise subjected to a bias that degrades the quality of care? Ultimately, the trouble with the hunt for signs of bias is that it leads to a picture of medicine as an institution so polluted with racism that minority patients might well prefer to walk away from it. It’s a wonder any black person would entrust himself or herself to the institution now regularly deplored as racist by medical editorialists.

Just as the perception of discrimination where it may or may not exist can lead to adverse outcomes including nonadherence, so the mistrust of medicine carries serious risks and costs. As the authors of Unequal Treatment note, “If minority patients mistrust doctors’ advice, they may be less likely to follow it, potentially accounting for some part of healthcare disparities.”34 Characteristically, the authors hold providers ultimately responsible in such cases:

If patients convey mistrust, refuse treatment, or comply poorly with treatment, providers may become less engaged in the treatment process, and patients are less likely to be provided with more vigorous treatments and services. But these kinds of reactions from minority patients may be understandable as a response to negative racial experiences in other contexts, or to real or perceived mistreatment by providers.35

Note the verbal distinction between (a) real mistreatment and (b) mistreatment that is only “perceived” but nevertheless gives the patient a valid reason to opt out of medical care itself. Let it be said that minor behavioral infractions like those identified in the literature as signs of bias in the clinical encounter – providing too little or indeed too much information, speaking too slowly or quickly, blinking too much, overusing the first-person plural – do not reasonably warrant refusal of care or noncompliance with treatment. Only a literature in the habit of overinterpreting behavioral signals would attach such import to these peccadilloes that they justify a patient’s suspicion or even rejection of medicine.

Does the perception of bias lead to mistrust of healthcare, or does mistrust lead to the perception of bias? The arrow could run either way or both ways. A recent study based on a survey of Californians urges in closing that “structural racism and other forms of oppression be identified as the root causes of inequity-driven mistrust.” Without retracting this demand, the authors then – two sentences later – confess that their study’s cross-sectional design “prevents determination of any causal relationships.”36 The authors cling so tightly to the belief that the racism of American medicine engenders mistrust that they are willing to defy the acknowledged limitation of their own study. Standing with them is an entire literature committed at once to scientific inquiry and a doctrinal belief that an insidious bias causes mistrust and the poor outcomes conjoined to it.

If medical care is racist from the roots up as innumerable contributors to the literature state or imply with or without using slogans like “structural racism,” then a minority patient could not mistrust it enough. In this sense, the anti-racist medical literature constitutes an incitement to mistrust and an encouragement of every bad outcome associated with mistrust in its own pages. Somehow a profession sworn to the avoidance of harm has reached this level of recklessness.