We began with institutions and society. Now, we move to individual minds surrounded and shaped by these societal structures. Next, we then move to interacting minds, which further perpetuate societal and individual racial distinctions. Racial bias at each level supports bias at the other levels, creating a racist system.
To understand individual mental structures, we start with unconscious inference, identified by Helmholtz, and its heir, implicit bias, most relevantly as expressed by Whites associating Black racial cues with negative concepts. Socially motivated (mis)perception goes one stage earlier to bias information seeking and interpretation. More specific links among racial bias in perceiving physiognomy, linked to dehumanizing associations, and aggressive behavior close this first section on the individual.
Unconscious inference
Among the intellectuals who contributed to the emergence of experimental psychology as an independent discipline in the nineteenth century was the German polymath, Herman von Helmholtz, whose numerous contributions to science include the concept of “Unbewuste Schluss” or “unconscious inference.” Helmholtz’s concept was simple, but its implications are profound, even more so today with recent advances in the mind and brain sciences. Given the complexity of just the visual world, how are humans to represent it based on their individual-level, meager sensory and perceptual system, which entails the shunting of packets of data from the world outside, through the eyes and into the brain? Helmholtz offered two ideas. First, perception is not veridical, given the complexity of the world and the rudimentary nature of the minds attempting to make sense of it. Second, as implied by the word inference, what one deduces from the evidence provided by the senses is not a replica of what is out there. Rather, mental representations of the physical world are mere approximations.
Whittling the self-esteem of Homo sapiens down further, Helmholtz went on to say that perception is not controllable, but rather that it unfolds automatically. He used a commonplace example to make this point. We know that it is not the Sun that rises, but rather that the Earth revolves around it. But when we sit on our porch at sunrise, and look toward the horizon, we incontrovertibly experience ourselves as being fixed, and the Sun, however bulky, pushing itself up to meet us. To say about the Sun that “it rises” is completely inaccurate yet completely compelling. That incorrect perception is not something over which we have choice. To think otherwise is to delude ourselves.
Helmholtz’s two ideas contained in the phrase “unconscious inference,” with many additional levels of social complexity, summarizes the challenge when we confront systemic racism. On the one hand, we “know” the facts about an economy purportedly mounted on free labor for 250 years, the undelivered promise of 40 acres and a mule, the failure of Reconstruction, the resistance to desegregation, the history of redlining and gerrymandering, a history of unequal access to education, jobs, housing, finance, healthcare, and a lack of equal protection under the law. On the other hand, the limited sensory, perceptual, learning, and memory systems of humans set up a built-in blindness and automatic inferences that generate the illusions that, for instance, White people experience more discrimination than Black people (Norton & Sommers, 2011). Or, if Black Americans have any challenges, they have created their own situation in America today (Pettigrew, 1979) and therefore are responsible for getting themselves out of that situation. Not that minorities have no illusions, but the illusions of the higher-status group have more consequences because they usually also have more power.
The features of human minds that feed into the production of systemic racism come in two forms: ordinary errors of perception, attention, learning, memory, and reasoning that are the hallmarks of all thinking systems with human-like intelligence. In addition, we add another level of theorizing familiar to psychologists, that of motivated reasoning, the idea that our preferences, goals, and desires can bias our reasoning and lead to prejudicial decisions and outcomes (Fiske & Taylor, 2021; Kunda, 1990).
Another hallmark of human cognition is the phenomenon of loss aversion, the finding human beings much prefer avoiding losses to acquiring equivalent gains (Kahneman & Tversky, 1979). Even as White Americans resist and deny the reality of systemic racism, they nonetheless feel the loss of White privilege and social status quite keenly, creating powerful resentments that motivate them to reason away the potential existence of systemic racism (Craig & Richeson, 2014; Parker, 2021).
Implicit racial bias
Beginning in the 1980s, psychologists began to document a puzzling result. Individuals who claimed to have no racial animus showed evidence of negative attitudes and stereotypes toward Black Americans (Devine, 1989; Dovidio & Gaertner, 1986). Soon, the hunt for methods to better access “implicit bias” (as contrasted with standard, explicit bias measured in surveys) was underway, with specific calls for the invention of better technologies that could bypass conscious awareness or conscious control (Greenwald & Banaji, 1995).
One such measure, the Implicit Association Test (IAT), has demonstrated a wide array of group evaluative associations. Typically, people can pair own-group cues faster with positive concepts, and other-group cues faster with negative ones—compared with vice versa. For example, White and other non-Black Americans show robust race bias in their inability to associate “good” and “bad” equally rapidly with the social categories Black and White. The IAT has attracted considerable attention (see Greenwald et al., 2020, for best practices, reliable effects, and ongoing investigations). A public online location, since 1998, has provided data from millions of tests taken by volunteer participants at http://www.implicit.harvard.edu. Several signature results have replicated multiple times with large samples over time:
-
(a)
Race bias is consistently visible in the data.
-
(b)
A small positive correlation between stated and implicit race attitudes exists, but the two are largely dissociated, i.e., many of those who report being neutral (no negative explicit attitudes toward Black or White Americans), do carry implicit associations of Black + bad and White + good to a larger extent than White + bad and Black + good. This result prompts us to yet again note that the term “racism” has been used by contemporary psychologists to refer to conscious forms of race prejudice and to emphasize its semi-independence from less conscious or implicit forms of race bias. To make this distinction clear, researchers who study implicit race bias have gone to great lengths to reserve the term racism to only refer to conscious expressions of racial animus. Our usage of the term systemic racism in this article is undertaken is in the interest of including all levels of analysis (individual, institutional, societal) and all forms, from the most explicit to the most implicit. The result of a low correlation between explicit racism and implicit race bias makes the point empirically that the two are not the same. Of course, implicit race bias feeds into what may become racism, and for this reason it is best to think about implicit race bias as the roots of racism, not the above ground, visible structure. Implicit race bias also results from systemic racism.
-
(c)
Asian Americans show the same pattern as White Americans, even though as a third-party group in response to a Black–White test, they might be assumed to have neutrality. From the point of view of systemic racism, this is an example of what it means to live in a system of inequity at all levels. Even third-party groups will acquire negative and positive attitudes toward groups that are not their own.
-
(d)
Black Americans express strong positive feelings toward their own group but on the measure of implicit cognition, they show no preference for their own group, with scores of almost any sample of Black Americans showing relative neutrality, i.e., equal association of good and bad for Black and White Americans. This absence of ingroup-favoring attitudes—juxtaposed with the ingroup-favoring lack of neutrality in all other groups in the same society—is open to various interpretations, from moral balance to internalized racism to astute pragmatism; all await other data.
-
(e)
Tests of anti-gay bias revealed it to be quite high in 2007 but steadily dropping off (by 64% since 2013) to be at an all-time low today. By comparison, anti-Black bias has dropped, but to a much lesser extent, by about 25% (Charlesworth & Banaji, in press). A 25% drop-off in race bias is not insignificant, and although the genders differ in magnitude of bias, both men and women are losing bias at equal speed. Although all demographic groups are changing, young Americans are changing faster than older Americans, suggesting that the world they inhabit is signaling a less biased set of attitudes.
Together, these data point to the individual manifestation of systemic racial bias, hidden from view but robustly present. However, psychologists have also gone beyond such demonstrations of basic cognitive associations as markers of implicit mental content to show that individual and institutional change is possible if the will to create change exists.
Socially motivated (mis)perception
The idea of motivated reasoning or motivated cognition gathers several useful ideas to understand how individual humans shape and even distort perception to deal with real or perceived threats to self. Kunda (1990), for example, posited that the individual need for accuracy is thwarted by the demand to reach a conclusion prior to the evidence being satisfactorily in place and that one’s goals and motives often drive decisions. These decisions reveal many identifiable biases that emerge to weaken the orientation toward accuracy (see Fiske & Taylor, 2021).
With more direct focus on motivated reasoning as it concerns social change, Kay et al., (2009) presented empirical evidence for a motivated tendency to view things as they are and conclude that such a state of affairs exists because it is reasonable and even representative of how things ought to be. The connection to systemic racism is quite clear, as the authors further demonstrate that motivated cognition exists in the interest of justifying sociopolitical systems that maintain inequality and resist change. People justify the status quo, preferring stability especially if they are privileged, but even if not (Jost & Banaji, 1994). Groups in a secure position show the cultural equivalent of inertia, seeking stability, but groups on the move express inertia as continuing to move (e.g., acquiring mainstream standing) (Zárate et al., 2019).
Two substantive theoretical accounts undergird these ideas as they concern complex interactions of within-person and across-person phenomena such as systemic racism. First, Sidanius and Pratto’s (1999) Theory of Social Dominance offers evolutionary and cultural evidence to support the idea that hierarchies are an almost obligatory feature of human social groups. A related but independent idea may be found in Jost’s System Justification Theory (Jost, 2020), which explicitly makes the case that individuals will sacrifice self and group interest in order to maintain larger “systems” of social arrangements and work to keep them in place. The reason, Jost argues, is that such a motivation serves to meet deep psychological needs for certainty, security, and acceptance by others. The overarching social structure is important to protect because if it is stable, then all within it will be safe, including those disadvantaged by established hierarchies.
Perception of phenotypes, deadly associations, and system-maintaining behavior
With regard to perceptions of race, the mere categorization of someone as Black shifts perceptions of their phenotype. For example, a series of experiments documented that people’s knowledge about race phenotypes drives perception of lightness of the skin tone (Levin & Banaji, 2006). In other words, experiments held skin-tone constant and varied only the features, from Afrocentric to Eurocentric; this variation in features shifts perception of skin tone, such that Afrocentric faces are viewed to be darker skinned than Eurocentric ones, despite the same gray-scale tone.
Skin tone and features are critical cues to make life and death decisions, especially in ambiguous situations that are often present in so many interactions between police and Black citizens. In simulations of police-citizen encounters, people are more likely to “shoot” unarmed Black men than otherwise equally unarmed White men (Correll, Wittenbrink, Park, Judd, & Goyle, 2010). Black men with more phenotypically Black features are more likely to receive the death penalty for murdering a White person, holding constant the features of the crime (Eberhardt, 2019). The phenotypicality effect extends even to Whites with Afrocentric features (Blair, Judd, & Chapleau, 2004). Judgments of criminality can be primed by a Black face (Eberhardt, 2019).
And there’s more: the race–crime association overlaps the dehumanizing association of Black faces with great ape faces, that Staples (2018) called the “racist trope that won’t die”; Goff, Eberhardt, Williams and Jackson (2008) provide evidence from policing that links apes and Black people, from the first moments of perception to the radio dispatch and other media, with systemic implications. In more recent work, Morehouse et al., (2021) have shown that White Americans associate White with human and Black, Asian, and Latinx with animal with greater ease than the opposite pairing (White with animal), regardless of the category of animal (generic or specific). Implicit racial biases (Whites favoring Whites) are consequential, correlating with judged trustworthiness and economic investment (Stanley, Sokol-Hessner, Banaji & Phelps, 2011).
More recently, Kurdi et al., (2021) measured attitudes toward a phenotypic feature that happens to be a dominant perceptual marker of race, Afrocentric and Eurocentric types of hair. First participants took an IAT measuring their implicit attitude toward Black women with natural or straightened hair. Then, subjects read a summary of a real legal case involving a corporation that fired a Black employee for refusing to change her natural hair (Equal Employment Opportunity Commission v. Catastrophe Management Solutions, 2016). The more negative the implicit attitude toward Afrocentric hair, the greater the sympathy with the corporation’s position rather than the plaintiff’s position in the legal case.
A relatively new approach to racial associations comes with the promise of epitomizing the term “systemic” in systemic racism. These are studies of large language corpora that are now possible using machine learning approaches to natural language. With the increasing availability of trained datasets—including large samples of the language of the Internet (content archives continuously collected by the nonprofit Common Crawl) or specific trained datasets of media such as books, TV shows, etc.—allow measuring the extent to which language contains attitudes and beliefs about Black and White Americans across time. Charlesworth and Banaji (in preparation) analyzed data from Google Books from 1800 to 1990. Setting aside the data from older books to focus on whether bias is present in the language today, these are the traits most associated with Black Americans (and not with White Americans) in the late twentieth century: earthy, lonely, sensual, cruel, lifeless, deceitful, meek, rebellious, headstrong, lazy. By contrast, these are the traits associated with White Americans (and not with Black Americans): critical, decisive, hostile, friendly, polite, able, diplomatic, belligerent, understanding, confident. Other work in natural language processing (NLP) sorts adjectives into 13 stereotype-content dictionaries (Nicolas, Bai, & Fiske, 2021). The above adjectives convey ambivalent reactions to Black Americans on several dimensions, but notably neglect competence; Whites in contrast feature several competence adjectives. NLP allows efficient analysis of language in the culture or in spontaneous, open-ended descriptions (Nicolas, Bai, & Fiske, under review).Footnote 4
Words have an important role to play. People often express surprise about implicit biases in the minds of individuals who have no intent to harbor them. Considering how and why it occurs—plausible mechanisms—may prove convincing. One causal candidate is language, the predominant way humans communicate and express themselves. Words undertake much of the labor of creating racism in thoughts and feelings that are reflected in speech. Machine learning approaches to understanding racial bias in language will likely be a critical method to objectively uncover how words, spoken and written, create systemic racism. That is, linguistic patterns connect groups with valenced concepts, and the repeated pairings create associations. Without awareness, language produces the inbuilt in the architecture of social cognition (as an example, the NLP stereotype-dimensions dictionaries capture more than 80% of spontaneous stereotype content; Nicolas, Bai, & Fiske, under review).
From cognitive racial bias to aggregate racialized behavior
Individual implicit attitudes have been repeatedly shown to predict behavior; Kurdi et al. (2019) offer the largest number of studies included in a meta-analysis to date. However, as the authors note, the actual attitude–behavior relationship is marred by the poor quality of many studies, especially given the lack of psychometric control over the predicted behavior. Among the controversies that have marked this work is an intriguing idea put forth by Payne, Vuletich and Lundberg (2017), who proposed that the small correlations between individual attitude and behavior must be acknowledged as a function of what they call the “bias of crowds,” the idea that an individual’s behavior is determined by the larger social context in which that individual exists. A number of studies have appeared recently to challenge the idea that individual attitude–behavior correlations is the right place to be looking. That the actual correlation between implicit attitude and behavior is larger than it may have appeared has been revealed in a series of studies that predict behavior at the aggregate level by using aggregate IAT scores by region, such as metropolitan areas, counties, and states. Charlesworth and Banaji (2021) reviewed these studies to demonstrate more substantive relationships between IAT racial bias and consequential social outcomes.
For example, the studies reviewed reveal that the greater the implicit bias against Blacks in a region (using average IAT scores of a region) the greater is the lethal use of force by police, the greater the Black American deaths from circulatory diseases, the lower is spending on Medicaid disability programs (more likely to assist Black Americans), the greater the Black–White gap in infant low birth weight and preterm births, the greater the Black–White gap in school disciplining (suspension, law enforcement referrals, expulsions, in-school arrests), the Black–White gap in standardized testing scores (3rd–8th grade for math and English), and lower upward mobility.
To grasp the meaning of systemic racism as it exists at the individual level within larger society, not just in a single moment by across time, a study by Payne, Vuletich and Brown-Iannuzzi (2019) is illustrative. Their analysis of IAT data today yields strong correlations with the ratio of enslaved to free people in the southern US in 1860. States with a larger ratio in 1860 are the states with greater race bias today, 160 years later (r = 0.64). This correlation is much larger in magnitude than even the correlation between regional IAT race bias and Black American representation across the US (r = 0.32). As Charlesworth and Banaji (2021) note, “the result also suggests that today’s Americans who live in regions with greater historical legacies of slavery must be acquiring the particles of race bias embedded in the social atmosphere. Systemic discrimination is a useful term in this case as it helps capture the pervasiveness of race bias as it extends across both space and time.”
Summary. As explicit bias decreased, measured forms of implicit bias have persisted, potentially attributable to racial segregation. White Americans have limited direct experience with Black Americans, so cultural associations substitute for more individuated impressions. Implicit associations of “Black-bad” and “White-good” are weakening, but far from neutral. Meanwhile, socially motivated (mis)perception favors these system-justifying biases. Together, they support a syndrome linking racial phenotypes, deadly associations, and system-maintaining behavior. Further, cognitive racial biases underpin aggregate racialized behavior. These are some cognitive-motivational mechanisms of systemic racism. Other mechanisms involve everyday interactions that perpetuate bias. In particular, predictable patterns of disrespect and distrust maintain the interpersonal racial divide.
Racialized social interactions
Face-to-face behavior propagates bias. Individuals carry racial biases into their social settings largely by interacting with others. Repeated patterns of behavior that differ by race are, at a minimum, racialized (defined by race) and often experienced as racist. Individual racial biases, enacted in daily life, perpetuate bias, which then links the individual to the norms, scripts, and habits that constitute the social system. Interpersonal interaction conveys bias, intentionally or not. In scores of studies, White Americans distance themselves from Black interaction partners, express non-verbal discomfort, and avoid them (e.g., Dovidio, Kawakami & Gaertner, 2002; Richeson & Shelton, 2007; Word, Zanna & Cooper, 1974). In the aggregate, these patterns constitute the concrete manifestations of a racially biased social system.
We have already seen White people’s generically negative default associations with Black Americans, linking them to crime (untrustworthy) and to animals (incompetent). These reflect the two key stereotype dimensions in intergroup perception (Fiske, 2018): warmth and competence. These dimensions organize people’s perceptions of social systems: perceived competence reflects groups’ stereotypic status in society. The hierarchy supposedly reflects merit, so rank predicts their supposed competence and evokes respect—or supposed incompetence and disrespect. Besides groups’ status (competence), the other aspect of social structure is groups’ apparent cooperative or competitive goals, interdependencies that stereotypically predict warmth and trustworthiness. Cooperators on our side are nice; competitors are not. Stereotypes derive from social structural perceptions (status and interdependence), especially when people learn about others they might encounter (Fiske, Cuddy, Glick & Xu, 2002; Nicolas et al., 2021). Black Americans do not get a break on either dimension. And because these racialized perceptions derive from social structure, they pave the way for systemic racism. Consider the evidence for these two dimensions: competence and warmth in racialized perceptions and behavior.
Disrespect communicates Whites’ view of Blacks as low status and incompetent
The default representation of Black Americans is low status (Dupree, Torrez, Obioha & Fiske, 2021). Whites spontaneously associate Black faces with low-status jobs, compared to Whites. The structural belief that Blacks are low status appears in associating them with jobs such as janitor, dishwasher, garbage collector, taxi driver, cashier, maid, prostitute. This race–status association correlates with endorsing social dominance (believing that some groups inevitably dominate others, and it is better that way) and with meritocracy (group get what they deserve). All these judgments share a common element of disrespect and assumed incompetence.
Race–status associations emerge in behavior that maintains Black people at the bottom of the hierarchy. Respondents endorsed Black applicants for lower status jobs and withheld support for organizations and government policies aiding minorities. Thus, racialized associations, assumptions, and preferences all identify a view of Black people's structural position as low status, on average. Behavior communicates these attitudes, whether examined or not. Thus, race–status associations imply Black incompetence, covarying with feeling-thermometer (0–100) ratings of interracial bias, social dominance orientation, meritocracy beliefs, as well as hierarchy-maintaining hiring and policy preferences.
Disrespectful behavior that presumes incompetence of Blacks appears in another series of studies. Well-meaning liberals, expected to introduce themselves to a Black partner, dumbed-down their speech, as they did in vocabulary for a task assignment (Dupree & Fiske, 2019). Similarly, White Democratic presidential candidates also showed a competence downshift in speeches to minority audiences only (Dupree & Fiske, 2019).
This pattern reproduces itself when respondents imagine introducing themselves to a lower-status person (race unspecified) at work (Swencionis & Fiske, 2016). They claim their goal is to communicate their own warmth (as they downplay their competence), but this rests on the presumption of the other’s incompetence. Trying to be folksy does not communicate respect.
The presumption that structural status predicts competence is widespread (averaging r > 0.80 across US and international samples; Fiske & Durante, 2016). The implication is that for most White Americans, the association that pops into their minds will link a Black person with incompetence. People communicate such disrespect by failing to bet on or invest in the other’s performance (Walsh, Vaida, & Fiske, under review).
Structurally, this amounts to racism. Black people are widely perceived as inferior in these ways, which are baked into the social hierarchy, reflecting disrespectful patterns of interpersonal behavior. All of this perpetuates the social hierarchy and the image of Blacks as incompetent.
Worse yet, disrespect surfaces in police encountering Black drivers. From the first moment (“Hey” instead of “Sir” or “Ma’am”), police officer language shows computationally derived, measurably lower respect (Voigt et al., 2017). Given the already fraught relationships between police and Black community members, this worsens an already dangerous encounter and undermines the chances to create trust.
Distrust communicates Whites’ views of Blacks as uncooperative and not warm
Besides incompetence, the other major dimension of social cognition is warmth (trustworthy, friendly), as noted. The default stereotype of a Black person is probably also untrustworthy, but the data on this point are surprisingly indirect. Whites can be expected to distrust Blacks as part of the larger principle that, categorically, people mistrust outgroups. More specifically, as noted, Whites associate Blacks with crime, which certainly undermines trust.Footnote 5 This configuration fits survey data showing that ratings of poor (i.e., explicitly low-status) Black people allege incompetence (disrespecting them) but also lack of warmth (distrusting them).
Plotting these ratings in a warmth x competence space, poor Blacks are frequently judged as low on both. Because White Americans link race and status, the low-income Black person is the default Black person, allegedly incompetent, but also untrustworthy. Mistrust is indicated by excessive surveillance of Black Americans (driving while Black, shopping while Black, false accusations of theft or assault, police shootings…).Footnote 6
Distrust can be operationalized as behavior: In the economic Trust Game, a player must decide how much of their starting endowment to share, on the knowledge that it will be tripled, and on the hope that their partner will share back, generously. Incentivized trust-game behavior closely tracks warmth ratings; that is, societal groups rated as low warmth and untrustworthy receive less shared endowment, presumably because they are not trusted to share it back. In nationally representative samples, people of color do not fare well in the Trust Game (Walsh et al., under review). In more prosaic settings, non-verbal behavior reveals unmonitored dislike (if not specifically mistrust), as noted.
Black Americans experience repeated treatment as incompetent and untrustworthy. Because this stereotype and ensuing behavior is racially category-based and negative, as well as potentially controllable, it is racist. Because the behavior comes from societal stereotypes, which come from social structure,Footnote 7 it is systemic.
Whites’ potential control implies responsibility for reinforcing system racism
Racialized interactions could also be termed racist, in the sense that White people could potentially observe their own inequitable behavior if they chose (Fiske, 1989). People rarely examine these unwritten rules, typical behaviors, but conceivably they could, so “unexamined” bias captures the higher potential control for behavior than for implicit associations. Control implies responsibility in the minds of lay people and the law, so this interpretation of “racialized” as “racist” creates concern and is likely to be contested. But the science makes the empirical point here that racialized social behavior is demonstrably controllable, given sufficient incentive (Monteith, Lybarger & Woodcock, 2009; Sinclair, Lowery, Hardin & Colangelo, 2005). So systematically different behavior by race reflects a racist habit, script, or norm, the components of a system from the bottom up.
The challenge in controlling racist habits is that they are the cultural default. Much of this systematic behavior results from White Americans’ inexperience with Black Americans, thereby substituting societal representations for individuating information about the unique human (Fiske & Neuberg, 1990). People use especially those default representations that fit their natural human tendency to detect and prefer people they view as similar to themselves. To unpack this, consider some basic principles of affiliation that would predispose Whites to favor other Whites and exclude Black people. First is the basic tendency to categorize others and to favor those of the ingroup. For decades, principles of attraction have established its foundations in similarity (Byrne, 1971; Montoya & Horton, 2013) or homophily (McPherson, Smith-Lovin & Cook, 2001). And mere categorization suffices to produce ingroup favoritism (Tajfel & Turner, 1979). No animus is necessary, although it easily develops. As a basis for categorization, race is arbitrary (more so than gender and age; Kurzban, Tooby & Cosmides, 2001) but common (Sidanius & Pratto, 1999). Thus, race-based ingroup favoritism is a default, in the absence of other experience.Footnote 8 This makes it hard to over-ride.
Societal segregation by race makes difficulties for overcoming the racial default. Segregation limits White exposure to Blacks, undermining their direct experience, leaving Whites to rely on cognitive shortcuts to represent Blacks as a group. Indeed, the less exposure people have to outgroups, the more clearly they differentiate among them–stereotypically. That is, White Americans who know the least about other races have the clearest stereotypes about them; the less diversity, the more differentiated their cognitive representations (Bai, Ramos & Fiske, 2020).
What’s wrong with that?
As a scientific question, a skeptic might ask, what’s wrong with differentiating by stereotypes? One set of answers concerns the demeaning individual and face-to-face interaction, just addressed. The other answers pertain to sheer demographic diversity of Black Americans, covered next.
Given its racial history and ongoing systems, societal patterns and cultural stereotypes prevailing in the US tend to associate Blacks with low status and Whites with high status as noted. To the extent this race–status association has a kernel of statistical accuracy (Blacks are over-represented in low-status jobs), it fails several tests as an argument for using stereotypes as a constructive strategy of intergroup relations. First, it ignores variability, individuality, and (especially) Black diversity. Second, category-based thinking exaggerates perceived between-group variability and minimizes perceived within-group variability (Tajfel & Turner 1979; Taylor, Fiske, Etcoff & Ruderman, 1978). So “nouns that cut slices” (Allport’s, 1954 felicitous phrase for category labels) do violence to the human data. What’s more, society has civil rights laws protecting people from being judged by their group membership, so the consensus is that this is not only wrong, but illegal.
Race–status associations, in practice, ignore all the structural contributors to race–status associations, such as the neighborhood effects, already described. Whites assume meritocracy, believing that status accurately reflects individual competence (Fiske, Dupree, Nicolas & Swencionis, 2016); globally, the perceived status—perceived competence correlation hovers around 0.80. (The only countries where people are more cynical about the status-merit link are former Communist ones; Grigoryan et al., 2020.) The point here is that status has many antecedents, and not all of them are merit (or other personal, stereotypical explanations, e.g., innately good/bad at math). Systemic factors such as neighborhood, school, family resources, connections, and especially race all receive no mention in the meritocracy account.
Whites do differentiate Black Americans by subcategories, e.g., by status, specifically social class, viewing low-income Black people as incompetent and untrustworthy, but Black professionals as competent and trustworthy (Fiske, Cuddy, Glick & Xu, 2002). Black Americans themselves differentiate several subtypes of Blacks likewise along a social-class dimension (Fiske, Bergsieker, Russell & Williams, 2009).
Status-keeping shortcuts are easier to maintain without information to the contrary, such as experiencing human variability. Whites with less exposure to Blacks are more overtly prejudiced as a function of structural features such as rural residence, where they encounter less diversity (Bai et al., 2020), and lack of education, where they experience less variability of ideas. As a structural matter, segregated White rural residence also predicts lower school quality partly because of the American policy of locally funding schools; this creates an association between a weaker tax base, rural location, ethnic homogeneity, and overt bias. These systemic factors interact to produce prejudice. As an earlier section shows, the social structure permeates American arrangements since the arrival of Whites on native lands.
Nevertheless, for most Whites, their isolated lives make them inexperienced about their Black fellow citizens. Housing segregation disfavors most Whites in experience with diversity, making them often inept and naïve when speaking about issues that are facts of Black lives. This means that Whites rely on cultural shortcuts to understand the Black people whose life experience they do not know. These cognitive representations derive from perceived structural patterns such as race–status associations and race-resource unfairness (Krysan & Crowder, 2017).
We have seen that Whites’ racial beliefs are relatively automatic (implicit bias) and ambivalent (warmth/competence). The resulting associations (stereotypes) are more subtle than most people believe. They are consequently hard for anyone to detect in themselves (unexamined) or in any one person (under the radar), but the patterns appear systemically as aggregate biases. Supposing the aggregate biases are problematic, at least because they ignore variability, examine that more closely.
Aggregate bias ignores diversity
So far, this review has described the relentless systems of racism that limit opportunity and outcomes by race. Many Black Americans nevertheless succeed despite the rigged system. Black diversity thus results from those who escape the system, but also from African and Caribbean immigration, and from intermarriage. For Black students enrolled at selective colleges, especially, the diversity of their backgrounds is the main fact that underscores their success (Charles, Kramer, Massey & Torres, 2021). Any given White student’s background is far more predictable than any given Black student’s, which potentially ranges from extreme disadvantage to extreme wealth. For that minority (a third) of Black students whose segregated neighborhoods entail underfunded schools, gang violence, and concentrated police violence, their presence in college testifies to extraordinary resilience (Charles, Fischer, Mooney & Massey, 2009).
Most non-Black people do not realize that Black Americans are more diverse than most American ethnic groups. Underestimating their variety allows an oversimplified image to dominate every level, from mind to society, making it a systemic racism. This section describes diversity based on place, intermarriage, immigrant experience, parent education, and sheer escape.
A century ago, most Black Americans lived in the rural South, but after the Great Migration, most lived in cities, often in the North, usually hyper-segregated, but with family roots in both the North and South. By the turn of the current century, Black American student bodies at selective colleges were the most diverse in history, more biracial, more immigrant, more middle or upper class, and equally identifying themselves as both American and as Black (Charles et al., 2021). Black students, even as elites, show “unprecedented variation in terms of racial origins, skin tone, nativity, generation, class, and segregation” (Charles et al., 2021, Ch. 10).
Clusters of characteristics and attitudes illustrate the variety. Mixed-race students identify less with being Black, are comfortable with both Blacks and Whites, see Whites as less discriminatory, and report deep parental involvement in their schooling and cultural experiences. Mixed race students also have more White friends and fewer Black friends than their monoracial peers and are more likely to date outside the group, especially with Whites. In addition, mixed-race students are less likely to join majority-Black organizations on campus, and thus report less intense interaction with Blacks. Psychologically, the White view of biracial individuals continues to demonstrate hypodescent, i.e., the view that biracial individuals belong to the less advantaged group, or the cognitive expression of the “one drop rule.” Combining the sociological and psychological angle demonstrates the lack of consistency between how biracial Americans are viewed and the way they see themselves.
Black students with an immigrant background are most comfortable with other Black students, and report having strict parents who expect obedience, respect, hard work, and family loyalty without hands-on, hovering involvement. First-generation immigrants, especially African immigrants (versus Caribbean ones), believe in meritocracy and see Whites as not so discriminatory. After a generation, idealism gives way to pragmatism: Hard work pays off. African immigrant origins predict reliably higher grades.
As for segregation, Black students growing up with more exposure to Whites feel closer to them but also view Whites as more discriminatory, a psychologically complex mental state to manage. In contrast, living in segregated neighborhoods especially exposes Black students to higher (the top third) levels of disorder and violence, leading them to view Whites as more distant and discriminatory. But parents are more protective, relying on strict discipline but not trying to use shame or guilt as an influence strategy (more frequent in Asian families).
As with all students, high-school GPA predicts college GPA. Besides that, again as with all students, Black women do better than Black men, as do those with educated parents. Differences in academic preparation vary by segregation in two ways: the more White students in their schools, the worse Black students’ grades but the higher their SATs, suggesting more rigorous standards. Thus, the portraits of Black college students are diverse; generalizations are unreliable, except perhaps for one: resilience in the face of systemic bias and a diversity of adaptations to a variety of challenges.
We document Black diversity here for these reasons: First, to avoid making the litany of systemic Black disadvantages the sole image conveyed here. Second, because of segregation, many White people, including University faculty, see a Black person on campus and—assuming they realize this is a student—they presume the person comes from a low-income background, unprepared for college, with uneducated parents, native born, but with little experience outside the imagined ghetto, etc. This may be true for some small fraction of students, but not just the Black ones, and not true of most Black students on campus today. A third reason to remind the reader of Black diversity on campus is to highlight experiences of inter-racial contact as important one mechanism for overcoming racial bias, and—if scaled up to integrated neighborhoods, schools, and workplaces—for shifting systemic racism.
Contact: exposure to racial diversity
People with least exposure to diversity have the most differentiated images of the outgroups they have never met (Bai et al., 2020). And the prospect and first experience of diversity is not salutary; newly diverse contexts show lower well-being (Putnam, 2007; Ramos, Bennett, Massey & Hewstone, 2019). But over time, people get used to each other: well-being is higher and stereotypes melt into each, forming an undifferentiated cluster of people like us, mostly warm and competent.
Psychology has 70 years of research to explain how this works, following Allport’s (1954) contact hypothesis. In one meta-analytic perspective (Pettigrew & Tropp, 2006), intergroup contact reduces prejudice, the more it meets Allport’s conditions: shared goals, non-trivial interactions, authority sanctions, and rewarding results. Much of the process seems to be affect-driven. If the contact setting would afford the opportunity for friendship, the contact effect is stronger (Tropp & Pettigrew, 2005). This is a useful reminder that much prejudice is emotional, not cognitive. In fact, a meta-analysis of 50 years of research on racist attitudes found that they predict racist behavior the most when they are emotions (“hating them”) rather than stereotypes (“they are lazy”) or even simple evaluations (2 on a 5-point scale) (Talaska et al., 2008).
Nevertheless, the core element of successful contact, goal interdependence, does operate via information processing. In laboratory experiments, interdependence makes people attend specifically to unexpected, stereotype-inconsistent information, and they make dispositional inferences, generating an individualized coherent impression of the teammate (Ames & Fiske, 2013; Erber & Fiske, 1984). Neural signatures of mindreading prominently include the mPFC regions that reliably activate when people are inferring another’s predispositions. The mind-reading mPFC activates most for an interdependent partner’s stereotype-inconsistent attributes. Although supporting evidence includes these mechanisms, a subsequent meta-analysis (Paluck, Porat, Clark & Green, 2021) notes that few high-quality intergroup studies have focused on race per se, few look at adults, few are experiments. We have much to learn.
