Young Children Are by Nature Innocent of Race

As appealing as the notion is that 1) we are blind to race until someone draws our attention to it, and 2) once made aware of it we remain indifferent to it until someone teaches us otherwise, there is now a considerable literature establishing that neither of these claims is true. On the one hand, infants spontaneously distinguish between people of different races and show preference for those of their own race (more plausibly, preference for people of the same race as their primary caregivers). On the other hand, by early preschool age, children endorse the strident racial prejudices—believing the majority race folk are smarter, cleaner, more honest, and richer than minority race folk—prevalent in the communities in which they live.

Over the past decade, a small flood of studies has appeared—to some considerable fascination in the popular media, including Newsweek's cover story complete with a White, Gerberesque baby stoically staring out at the reader under the banner headline, “Is your baby racist?” (Bronson and Merryman, Reference Bronson and Merryman2009)—persuasively demonstrating that infants are hardly blind to race. Given the historical prevalence of studies of White children living in American college towns, this literature is refreshingly global in scope, testing, among others, American, Chinese, Israeli, Ethopian, and Korean infants. The results are consistent and clear: infants both distinguish between peoples of different (socially distinguished) races, and show a preference for faces of people of their own race.

Habituation/dishabituation studies also demonstrate that infants attend to race and discriminate between individuals of different races. Katz (Reference Katz and Leahy1983) habituated six-month-old infants to a series of faces of the same gender and same race (that is, infants were presented with a series of photographs of faces until infants began to lose interest, as measured in decreased looking time). The infants were then presented with one of three novel faces: 1) a face that was both the same race and gender as those in the original series, 2) one that differed in race but not gender, or 3) one that differed in gender but not race. Infants looked longer at the other-race and other-gender faces than at the same-race and same-gender faces, an effect that was strongest when series of photographs to which they were habituated were own-race faces.

Infants not only attend to and discriminate between people on a basis of race, they are more adept at recognizing own- rather than other-race faces. Sangrigoli and de Schonen (Reference Sangrigoli and de Schonen2004) found that French three-month-old infants were able to distinguish between a familiar and unfamiliar White face but not between a familiar and unfamiliar Asian face, although this other-race effect was readily extinguished by brief exposure to a few other-race faces prior to the experimental task. Studies with British (Kelly et al., Reference Kelly, Liu, Ge, Quinn, Slater, Lee, Liu and Pascalis2007) and Chinese (Kelly et al., Reference Kelly, Liu, Lee, Quinn, Pascalis, Slater and Ge2009; Liu et al., Reference Liu, Quinn, Wheeler, Xiao, Ge and Lee2010) infants yielded similar findings. Together these studies confirm that the perception of race involves a narrowing process: as the infant develops she doesn't become more adept at discriminating between members of her own race as opposed to other race individuals—that is, the infant does not become more “expert” in distinguishing between faces of members of a more familiar race. Rather as they get older, infants retain the ability to draw fine, intraracial distinctions primarily only among own-race faces (Kelly et al., Reference Kelly, Liu, Lee, Quinn, Pascalis, Slater and Ge2009). In addition to discriminating between racial types, three-month-old infants prefer to look at own-race faces (Bar-Haim et al., Reference Bar-Haim, Ziv, Lamy and Hodes2006).

These findings highlight an interesting quality of infant race-based judgments; viz, judgments are not based on a “simple” strategy, such as looking longer at members of one particular racial category, or that they are focusing on one particular visual property, such as lighter skin color. Recall that in the habituation studies, infants look longer at an other-race face if they were first habituated to own-race faces. On the preference task, in contrast, infants look longer at faces that racially match their own. Thus, infants do not invariably look longer at a specific type of face—own-race versus other-race. Rather looking time is task dependent. On some tasks infants look longer at an other-race face, on other tasks they look longer at a same-race face. Nor do infants prefer a particular racially relevant perceptual quality. Earlier studies with preschoolers suggested that preference for representations of Whites could be explained by an innate preference for light colors and a dispreference for dark tones (Williams et al., Reference Williams, Boswell and Best1975). Infant preferences for own-race faces, however, cannot be accounted for by a preference for a particular or idiosyncratic quality (say for lighter skin color). White infants in Israel preferred to look at White faces, whereas African infants in Ethiopia preferred African faces. In contrast, African infants living in Israel showed no preference. The latter finding suggests that a preference for own-race faces obtains only for infants living in an environment in which their own (presumably their caretaker's) race is dominant (Bar-Haim et al., Reference Bar-Haim, Ziv, Lamy and Hodes2006).

It is also worth underscoring other things that these studies do—and do not—demonstrate. First, the infant's precocious ability to recognize, categorize, and preferentially select among faces racially is consistent with the infant's capacities to recognize, categorize, and preferentially select among individuals based on a number of distinctions that later carry considerable social weight. In addition to penchants for specific individuals, particularly the newborn's preference for her mother's face (Bushnell et al., Reference Bushnell, Sai and Mullin1989), infants prefer to look at attractive faces compared to less-attractive ones (Slater et al., Reference Slater, Von der Schulenburg, Brown, Badenoch, Butterworth, Parsons and Samuels1998), and distinguish between individuals on the basis of gender (Miller Reference Miller1983), language spoken (Mehler et al., Reference Mehler, Jusczyk, Lambertz, Halsted, Bertoncini and Amiel-Tison1988), age (Brooks and Lewis, Reference Brooks and Lewis1976), as well as race. In short, predilections to attend to socially relevant information are precociously grounded in the human perceptual and conceptual system. However, these findings do not mean that the infant's attention to and processing of racially relevant information is of a piece with later-emerging beliefs, attitudes, and especially practices that favor members of one group over another. In short, these findings do not imply that infants are racist—regardless of Newsweek's recent cover article's title (Bronson and Merryman, Reference Bronson and Merryman2009).

Nor does the finding that three- but not one-month-old infants discriminate their own (i.e., most commonly encountered) race from other races mean that young infants know their own race or even have a category race. What they display is the ability to discriminate between individuals using information that older children and adults employ in diagnosing the social category of race. There is no reason to believe that the infant's capacity to distinguish between racially different individuals supports the kinds of race-based expectations that are linked to later-emerging racial prejudice among preschool children. Without that link, it doesn't make sense to call preferential looking “racist.” For example, as Katz's (Reference Katz and Leahy1983) study demonstrates, a White infant habituated to four White faces looks longer if presented with a Black face; a Black infant habituated to four Black faces looks longer if presented with a White face. Racial “preference” in this instance is dependent on an infant's race, not on a system of racial thinking that transforms racial distinctions into a portal to inequities of inclusion and exclusion. Further, as already mentioned, we know that at least some of these infant own-race preferences are readily extinguished (Sangrigoli and de Schonen, Reference Sangrigoli and de Schonen2004). Again, one of the hallmark features of racial prejudice is its tenacity. The two phenomena (infants' preferential looking by race, on the one hand, and racial prejudice, on the other) could hardly be more different in this regard.

Even if They Are “Aware” of Race, They Are Without Prejudice

Infants, as we have seen, may not be innocent of race, but they are innocent of racism. This is not surprising given limits on what an infant can have learned about the world. And, as Rodgers and Hammerstein (Reference Rodgers and Hammerstein1949) declared (at the time, controversially) in South Pacific, racism is something “you've got to be carefully taught.”Footnote ¹ Children aren't by their nature biased, prejudice is something that they acquire: good kids learn bad things from bad people setting bad examples. According to conventional wisdom “young children's attitudes toward the racially different are innately positive; … as long as adults do not actively indoctrinate racist attitudes children will grow up without prejudice” (Morris Reference Morris1981).

This notion is consistent with a more general (and often exaggerated; see below) confidence about the extent to which parents and other parochial sources of information influence development. As the twig is bent, so grows the tree; the acorn never falls far from the tree, etc. So, if children become racist, it isn't because society, as it were, made them this way, it's because they've learned it from a particular individual or local group of individuals. Similarly, learning racism isn't thought to be something that occurs among very young children. A degree of cognitive “maturity,” not possessed by the young child, is assumed to be necessary. On this view, left to their own devices, children don't invest race with much significance. A recent British publication, which stirred considerable controversy, provides a clear example of this claim: in it, the documentary filmmaker Adrian Hart argues that multicultural and antiracist school interventions do not lower race-based thinking, they actually encourage kids to think in racialist terms. Absent such well-intentioned but dangerous vigilance on the parts of teachers and administrators, Hart (Reference Hart2009) contends, children would embrace neither racialist nor racist views: “Today's anti-racist educators' … missionary zeal … exaggerate racism and profoundly misunderstands children …. [C]hildren who mixed freely with one another in an open-minded and colour-blind manner were encouraged to see themselves and their peers in racial and ethnic terms” (p. 14).

Whatever its source, there is no doubt that children are at risk for developing racial prejudice. The question is at what age? A popular blog on the Psychology Today website, entitled “Are we born racist?” declares that “an awareness of stereotypes and racism doesn't begin to happen until about age six” (citing an article by researchers who actually don't make that claim) (Mendoza-Denton Reference Mendoza-Denton2011). One reason that adults easily cling to the image of the color-blind preschooler is the young child's apparent open-mindedness. Aboud (Reference Aboud2003) writes, “Teachers and parents tend to think prejudice is low in the early school years because they observe little hostile or discriminatory behavior ….” (p. 48). Still, Aboud continues, “standardized measures reveal high levels of pro-White/anti-Black bias in White children … even as young as three years” (p. 48). Nonetheless, the color-blind preschooler remains a largely unassailed image: (White) American teachers and parents, Van Ausdale and Feagin (Reference Van Ausdale and Feagin2001) observe, “experience a deep denial when it comes to acknowledging racism, especially in young children” (p. 190).

The denial would be easier to understand if evidence to the contrary were not so robust and so long available. Studies, some dating from the 1930s (Horowitz Reference Horowitz1939), have repeatedly shown, initially that Black children and by the 1950s that White children as well (Stevenson and Stewart, Reference Stevenson and Stewart1958; Trager and Yarrow, Reference Trager and Yarrow1952), hold strident, race-based prejudice. Nor is it the case that these findings were unknown outside the academy. Studies by Kenneth and Mamie Clark (Reference Clark and Clark1939, Reference Clark and Clark1950) which established that Black preschoolers preferred and were more likely to self-identify with a White doll than a Black one were cited in the 1954 Supreme Court school desegregation ruling in Brown v. Board of Education as evidence that separate but equal education resulted in a sense of inferiority and even self-hatred among Black children (subsequent work, however, persuasively argues that the Clark's studies actually do not show this (Cross Reference Cross1991)).

Indeed, an extensive literature, relying on a range of methodologies, converges on the same conclusion: beginning around three and largely systematized by five years of age, children come to hold attitudes that reflect the prevailing racial prejudices that favor the majority and disfavor minorities (Aboud Reference Aboud1988; Katz Reference Katz and Leahy1983). (For the youngest children, evidence that they favor majority individuals is more convincing than evidence that they disfavor minorities (Aboud Reference Aboud2003).) Strikingly, there is very little variation among children in the extent to which they endorse these prejudices (Aboud and Doyle, Reference Aboud and Doyle1996), a point to which we will return below. Nor can these prejudices be attributed to a simple or “crude” mapping of blatant qualities onto race categories: young children are susceptible to a highly abstract version of racial prejudice called “stereotype threat” documented among adults in which a test-taker's performance is diminished simply by being reminded of a prevalent stereotype (e.g., when primed that Blacks struggle learning mathematics, Black test-takers perform less well than unprimed Black test-takers (Steele and Aronson, Reference Steele and Aronson1995)). In Ambady et al.'s (Reference Ambady, Shih, Kim and Pittinsky2001) study, even kindergarteners predicted that an Asian student would be more likely to do well in math than a Caucasian student. A more apt characterization than the color-blind child is the color-mute adult: (White) parents display a marked dispreference for talking about race with their young children (Katz Reference Katz and Leahy1983). No wonder parents are so easily convinced that their child hardly notices race.

Children “Discover” Race by Opening Their Eyes and Looking

The basic outline of this myth is straightforward: children become aware of race as they mature and begin to recognize patterned variation in the physical features of those around them. The cognitive processes involved are little more than the perceptual ability to distinguish racial types (which infants are clearly able to do) coupled with a general capacity to classify like things together and unlike things apart. Common sense in this regard consists of two widely held but spurious beliefs about the way children acquire knowledge of the world. On the one hand, children are thought to be tethered to perceptual information. They come to know what they can see (and touch and taste) directly and their knowledge is largely a result of efforts to bring order to perceptual experience. On the other hand, they are thought initially at least to be tethered to superficial interpretations of what they come to know. Particulars precede generalities because young children focus primarily on visual and tactile experience and bring to bear analytic skills grounded in immediate experience and the moment.

A peculiarity in the way young children acquire knowledge of race feeds these (limited) expectations about children's thinking. Young children give little evidence of noticing race and when they do their beliefs seem to be local, idiosyncratic, fragmentary, and not infrequently bizarre. Over twenty years ago, when I was working on a study I will describe in a moment, my then three-year-old daughter provided me with a telling illustration of just this point. We were living in Paris. My daughter and I were in our car, stopped at a traffic light. She looked over at the family in the car next to us and exclaimed that they all looked like her preschool friend Alexandre. Alexandre was a Eurasian child and the family in the next car appeared to be Asian. I asked her why she thought they looked like Alexandre. She mulled over the question for a moment, all the while staring intently at the people in the next car. Finally she said with great conviction and to my mind some pride, “They all have the same color hair.” Of course she was right, Alexandre and the family in the next car all had black hair—but so did much of France or for that matter so did I, at least then. Ramsey (Reference Ramsey, Phinney and Rotheram1987, p. 60) reports a similar anecdote, in this case involving teeth. These kinds of anecdotes—“Look mommy, those people have chocolate skin”—are familiar and compelling fodder for the “my-kid-doesn't-understand-race-any-more-than-nuclear-physics” sentiment that soothes the sense of a color-blind, juvenile naiveté.

Such anecdotes confirm several cherished beliefs. The first is that race is a visual judgment. We look at a face and we discern information intrinsic to the person's appearance. The idea is that racial thinking involves bottom-up processes, ones in which the stimulus information is sufficiently rich with perceptual information that the conceptual distinction follows seamlessly. In his discussion of folk beliefs about plants and animals, Brent Berlin spoke of distinctions crying out to be named. Many feel the same about race; the differences in appearance between peoples of different races cry out to be named. The fact that a three-month-old infant can distinguish her own (or her own caretaker's) race from other-race faces suggests that the patterned physical variation among racial groups jumps out at the perceiver. This would imply that visual processing is also central to learning about race categories.

Somewhat surprisingly, this is more an assumption than an empirically demonstrated finding. Although the idea that race is a visual phenomenon is almost universally held, there are few studies actually assessing whether visual information is initially relevant to learning about race (as least learning about race as a social phenomenon, presumably an infant who distinguishes between an own- and other-race face is doing so on the basis of perceptual information alone). The study I mentioned earlier was the first, and remains I believe the most comprehensive to directly examine this question. In it, we explored the relative importance of visual and verbal information in early representations of race through three- and four-year-old French children's memory for narrative (Hirschfeld Reference Hirschfeld1993). The study's underlying logic was that the more important a modality of input is in learning about race, the more likely the child is to attend to it, and hence the more likely he is to remember (encode and retrieve) it. In the study we presented children with a story in which each of the story's characters' race, gender, occupation, age, and behavior were mentioned. We then asked the children to recall the story, and then we looked at how much of this social information they had retained. The narrative was presented in one of two formats: either as a visual narrative in which the story was depicted in a picture book or as a verbal narrative that was read aloud. Half the children were presented with the visual narrative, half with the verbal one.

If the prevalent assumption were correct that race is a visual phenomenon, we would expect children to recall more racial information after viewing a visual narrative than after hearing a verbal one. If the prevalent assumption is not accurate—if children do not learn race by opening up their eyes and inducing categories from visual information—we would expect no difference. If, alternatively, children attend particularly closely to linguistic information, then verbal information would be more memorable. The results revealed that both three- and four-year-olds recalled the race of the characters twice as much following the verbal narrative than following the visual one. Memory for occupation followed essentially the same pattern as race—greater recall following the verbal rather than the visual narrative—whereas memory for gender was the inverse—substantially greater recall following the visual narrative than the verbal one. On the whole, these findings are consistent with recent work showing that young children invest social categories with greater inferential potential when they are labeled than when they are not (Birnbaum et al., Reference Birnbaum, Deeb, Segall, Ben-Eliyahu and Diesendruck2010; Diesendruck and haLevi, Reference Diesendruck and haLevi2006; Waxman Reference Waxman2010).

One interpretation of these results of the French study—the one I favor—is that children's memory for race is greater following the verbal narrative than the visual one due to the greater importance of linguistic over visual information in forming early representations of race. An alternative interpretation is that children recalled less racial information following the visual narrative not because visual information is less important but because they were unable to report visual information during verbal recall. That is, they encoded visual information but were constrained in the memory task because they could not retrieve the verbal labels for the visual phenomena. To rule out this possibility we included a second, priming condition. In the primed condition, prior to hearing or viewing the narrative, children participated in a task in which they matched labels to pictures depicting the racial kinds mentioned in the narratives. In the unprimed condition, children participated in these tasks after they heard or viewed and recalled the narrative. If the results could be attributed to difficulty retrieving labels for encoded racial information, we would expect memory for race to increase when primed for the label and its visual depiction for subjects viewing the visual narrative (we would expect no effect for subjects who were read the verbal narrative). Priming, however, had no effect for children in either the visual or verbal conditions. Children who were primed to match racial labels to pictures were no more likely to recall racial information than children who were not primed. In sum, early representations of race seem to be based on information children acquire by listening to those around them talk about social differences rather than by attending to physical differences that “cry out to be named.”

Children May Come to Notice Race on Their Own, but Have to Be Taught Prejudice

It isn't absolutely necessary that racism is something a child is “carefully taught,” as Rogers and Hammerstein, among many others, claimed. In principle, children could be born racist, as Newsweek provocatively ventured. Just as children seem to have inborn preferences for attractive over less attractive faces, inborn preferences for members of groups to which they belong over members of other groups, or inborn preferences for people who speak their own over other languages. But they don't. Children are born with innate preferences for the same reasons that other living creatures have them; namely, such preferences function to enhance survival. We prefer sweet over bitter tastes and butterflies over spiders because sweet-tasting things tend to be both ripe and nontoxic and spiders tend to be threats to our safety whereas butterflies do not. For such preferences to be inborn—to become innately-specified cognitive biases—sufficient experience is required over evolutionary time with the threatening (or soothing) events, objects, or creatures. Absent this, natural selection cannot operate and make a particular distinction into an inborn preference. Clearly our sensory systems are capable of distinguishing between racial groups; as we've seen three-month-old infants already show preference for own-race faces overs other-race faces. But human experience with other-race individuals is fairly recent, occurring within historical (rather than evolutionary) time and largely as a function of relatively large-scale migrations typically requiring significant overseas travel. (Indeed, the continent-based “races” identified by geneticists (Mountain and Risch, Reference Mountain and Risch2004) are themselves apparently quite recent phenomena (Stringer Reference Stringer2011).) Prior to such encounters, the overwhelming bulk of human experience has been with people whose bodily features vary in gentle gradations over space, so most of our neighbors had bodies which looked much like their own. The features that are diagnostic of racial groups—to the extent the latter exist as distinct entities—gradate one into the other. Instances in which sharp disjunction occur in group-relevant physical features can be maintained only by systematic social practices of exclusion and inclusion unknown until recently or by environmental obstacles to steady immigration not readily breeched prior to overseas travel. (This is not to say that humans have not long experienced neighboring peoples who looked different. Given what we know about body decoration, maquillage, masks, etc. and their cognitive relevance, it is reasonable to imagine that humans have long marked cultural differences physically (Sperber and Hirschfeld, Reference Sperber, Hirschfeld, Carruthers, Laurence and Stich2006). It is similarly likely that humans have long experienced neighbors who speak differently and for whom speech differences signal cultural, group-based differences (Hirschfeld Reference Hirschfeld2008a). Hence, it is plausible that an inborn favoritism toward members of one's own group is linked to an inborn preference for group-based differences in appearances and behavior, just not corporally inflected ones). Still, it is implausible that humans have had sufficiently long direct experience with peoples whose different corporal appearance maps onto cultural differences for natural selection to have fixed an inborn belief system supporting racial preferences.

So, if we aren't born this way, then, for those of us who have racial preferences, we've learned them. Of course, some things are so obvious we don't need to be taught them; we learn them on our own. Up from down or the fact that what goes up comes down, for instance. But much of what we have learned, according to conventional wisdom, we have been taught. Conventional wisdom also holds that much of what we're taught comes from significant local others: parents, teachers, perhaps peers. Race and racism have generally been interpreted in this light. Individuals who are racist tend to have parents who are racist. Indeed, children themselves tend to believe that their racial attitudes and those of their parents are much the same. Put in more sophisticated terms, in the words of Aboud and Levy (Reference Aboud, Levy and Oskamp2000), the conventional wisdom is that “children adopt the [racist] attitudes and behavior of significant others in their culture, through mechanisms such as imitation … and conformity” (p. 271).

Surprisingly, given how widely held this notion is, there is little evidence that it is the case (Aboud and Doyle, Reference Aboud and Doyle1996). Parental influence on children's developing racial attitudes is wildly overestimated (Bigler Reference Bigler2004). Young children, if anything, tend to acknowledge their racial prejudice more readily than adults, and tend to display a more prevalent bias than older children and adults (Baron and Banaji, Reference Baron and Banaji2006). Perhaps most surprisingly, even the children of parents who make explicit efforts to shape their children's racial attitudes tend to hold attitudes that more closely reflect those prevailing in the community in which they live than the ones their parents champion (Branch and Newcombe, Reference Branch and Newcombe1980; Spencer Reference Spencer1983). This isn't overly recent research. The absence of a close correspondence between the child's racial thinking and parochial (i.e., familial) variation has been well documented by psychologists working with the development of racial prejudice for some time. As Aboud and Doyle (Reference Aboud and Doyle1996) observe, the lack of variability in young children's racial prejudice “leaves little opportunity to find a significant correlation with parental attitudes” (p. 372). Yet these findings have had little effect, either on parental mythology of race or on efforts to change the way children think about race.

In matters of race, even quite young children are not passive participants, frank imitators, or slaves to conformity but active agents in the construction and maintenance of racial systems of belief. Racism isn't something that happens to the child, it is something he does. As it happens, this is a characteristic of many kinds of knowledge the young child acquires, not just of race. Parents generally overestimate the influence they have on their children's development, a phenomenon Judith Harris (Reference Harris1998) has called the nurture assumption. Psychologists and behavioral geneticists have explored a number of developmental outcomes and the limited role that parents—read “local” influences—play in shaping them. A useful example of this is accent: children of nonnative speakers do not develop the nonnative accents of their parents, despite the fact that their parents provide the bulk of linguistic input during the early years of language acquisition. If children model their speech on the input provided by those with whom they spend the most time, we would expect language acquisition to reflect quite local variation in speech practices. Instead what we find is that children—and we are talking about quite young children in this regard—conceptualize language, including accent, in terms of the conventions of the wider speech community. To accomplish this, children presumably are able to place higher weight to less frequent but more community-relevant input. The same seems to be the case with racial attitudes. Children learn but aren't taught—rather they teach themselves—since they bring to bear capacities to differentially weight information depending on its source with attention to the prevailing practices of the communities in which they live (a point to be expanded below). In brief, even very young children are excellent lay sociologists.

It could be argued that I am overstating the case here. Whether they can recruit algorithms that differentially give greater weight to infrequent but more informative information than to more frequent but less informative information is interesting but not necessarily an indication that the young child's path to racial thinking is laden with theoretical insight, albeit lay theoretical. True, the child might rely on cruder strategies in building an image of the racialized world, ones more akin to imitation or an urge to conform to some authority figure rather than a deep understanding of the nature and scope of the stable conventions of the community of which he is a fledgling member.

A recent study among British preschool-age children by my colleagues and myself sheds more light on the child's sense of community and the role it plays in the development of racial thinking (Hirschfeld et al., Reference Hirschfeld, Bartmess, White and Frith2007). In that study, we explored the knowledge young children have of common (i.e., community wide) race and gender stereotypes and how willing they were to use this knowledge to interpret novel situations. We compared typically developing three- and six-year-olds with a group of highly impaired autistic 6-year-olds and a group of moderately impaired autistic six-year-olds. We found that despite the global disruption to normal social interactions and learning opportunities that autism entails, there was virtually no difference between highly impaired autistics, less impaired autistics, and typically developing children of both age groups in their knowledge of racial and gender stereotypes and their willingness to use them when asked to interpret the outcome of a novel situation. Indeed, we found that under some conditions, highly impaired autistic children and typically developing three-year-olds were more likely to use prevailing community stereotypes. Specifically, in one experiment children were told about a child whose personal desire runs counter to the stereotype (e.g., a girl who prefers to play with trucks rather than dolls). We then asked children to select which of two children is more likely to play with dolls, the girl who doesn't like to play with dolls or the boy. Both the typically developing three-year-olds and the highly impaired autistic six-year-olds were more likely than older preschoolers and less impaired autistic children to choose the child on the basis of stereotypic expectations rather than personal desire (i.e., in the example, select the girl even though she doesn't like to play with dolls).

Taken together these two lines of evidence suggest that the acquisition of stereotypes and the willingness to endorse them is grounded neither in local (or parochial/familial) social interactions nor does it require sustained, systematic, and engaged social interaction. If children acquire racial knowledge through imitation, we would expect that parental attitudes would predict those of their children. If they acquire this knowledge through modeling—in which the novice actively seeks to be like the model—children whose social interactions were both highly atypical and extremely constrained would be unlikely to develop the same beliefs as children whose interactions involved sustained interpersonal engagement. Neither of these predictions is borne out by our findings. Rather young children acquire racial knowledge via a predilection to attend to input relevant to community standards and conventions, on the one hand, and via the agency of low levels of attention and engagement, on the other. In short, children seem to develop this knowledge largely on their own and do it without committing significant social (as opposed to cognitive) resources.

This tells us much about the nature of racism. Stereotypes and prejudice are inferences about an individual's qualities that derive from knowledge of groups not individuals. Hence, judgments about groups are grounded in the conventions and standards of communities that define them, regulate their relations, and establish the practices that organize inclusions and exclusions. Much of what the young child learns about the world is mediated by the peculiarities of the community of which the child and her parents are members. It is tempting to imagine that such communities could be straightforwardly recognized and the expectations and practices associated with them readily inferred. When researchers compare American, Chinese, French, and Korean infant performance on race-based discrimination tasks, they assume that these infants are drawn from communities sufficiently and easily distinguished from each other that differences in infant performance can be attributed to differences in community-based experience whereas commonalities in performance can be attributed to panhuman endowment. On this model, the community bears on the infant, who is a largely passive receptacle of both community difference and panhuman commonality.

But as we have seen, children (and infants as well (Baillargeon Reference Baillargeon, Smith and Hart2002; Csibra Reference Csibra2010; Spelke Reference Spelke1990)) are not passive receptacles but active agents in the process of acquiring knowledge of the world they inhabit. When the Clarks found that children select a white over a black doll when asked which they prefer, the children were making judgments about values peculiar to a community, or perhaps better put as Appadurai (Reference Appadurai, Lutz and Abu-Lughod1990) does a “community of sentiment” (p. 94), to which in some real sense they hardly belonged. Human social life is rich with community, not only in the sense that community conventions, standards, and practices intrinsically shape it, but also in the sense that it is shaped simultaneously by multiple communities, each with its own mix of conventions, standards, and practices. Making the task of identifying the relevant community-based conventions, standards, and practices even more challenging, communities, unlike individuals and other aggregations of individuals, are not perceptible entities. Like all generics, we experience them only through exemplars that by definition are fragments and fragmentary (Geertz Reference Geertz1973; Gelman Reference Gelman and PelletierForthcoming). Communities typically don't represent a discrete collection of folks—indeed, as Anderson's (Reference Anderson1983) influential notion of imagined communities suggests, many communities lack spatial or even temporal contiguity.

Imagine then the task facing the young child. Understanding one's own and others' actions turns not only on mental states—doing something because one wants to do it or avoiding it because one doesn't like doing it—but also on knowing the kind of person an individual is. As we saw with younger preschoolers and autistic children, this latter strategy may be actually more precocious in some respects. Humans inhabit worlds that are both socially complex and cultural; they use the ability to recognize the sorts of people there are to understand the meaning of others' behavior in negotiating these complexities. When Johnnie tries to understand an action, he can—and does—exploit expectations he has about adults, women, teachers, Blacks, or skateboard aficionados. Johnnie may expect, for example, that women are more likely than men to be secretaries without appealing to the motives—the mental states—that may underlie a choice of occupation. The kinds of people there are is largely a function of community, and much of our social life turns on acquiring the dispositions, sentiments, and predilections (to name versions of the same thing) of the communities to which we belong. Relying on a vision too narrow or too parochial—or too broad and too deracinated—one risks incorrectly inferring the community-based conventions, standards, and practices that apply. Add to this the fact that the relevant interpretation varies for different facets (and situations) of social life—as I think is unquestionable—then the child's task is even more challenging. Yet, as we have seen, children, even quite young children, are quite good at meeting this challenge. This essay is concerned with one instance of that mastery.

Children Notice Race, They May Even Acquire a Form of Racial Bias, but Ultimately They Believe that Race Is a Superficial Quality, Literally Just Skin-Deep

A corollary to the myth that children learn about race by cataloguing patterned physical variation is that children believe that race consists of little more than patterned variation in physical features. Race for young children, on this view, is literally just skin deep. Young children don't, again on this view, grasp the adult conviction that race is a deep property, governed by natural processes, and governing many facets of human thought and behavior. A fundamental feature of the American adult construal is that race has what is sometimes called great inductive potential. That is, knowing the race of a person affords us (supposedly) license to infer many other things about that person, things that have nothing obviously to do with the pattern of physical features that are typically used to identify racial category membership and that go far beyond what we otherwise know about that person from direct knowledge or experience. This is the power of stereotypes. People believe that knowing someone's race allows them to infer that he or she is likely to be honest, smart, athletic, or susceptible to sickle-cell anemia, hypertension, or prostate cancer. Children, it is generally thought, do not share this notion of race. Not knowing these things, the story goes, means that young children are essentially color-blind. Not in the sense that they don't notice difference, but in the sense that they do not invest it with the consequence that adults do.

But we now know that this portrait of young children's beliefs about race grossly underestimates what young children understand about race. Children grasp that natural processes mediate racial category membership: for instance, in a set of studies conducted a number of years ago (Hirschfeld Reference Hirschfeld1995), I found that three-year-olds expect that children inherit their racial category membership from their parents, whereas they expect that other physical properties (such as body build or hair color) are less inheritable. Similarly, they expect that racial category membership is constant across the life span, whereas membership in other social categories such as occupation is not. Similarly, in contrast to findings of earlier research, we now know that four-year-olds expect that a person's race is fixed at birth even if the child is raised by parents of a different race. In short, the weight of current research is that American preschoolers grasp that the race but not other similar physical features is governed by natural processes, is an intrinsic and bound quality of the individual, and—in virtue of work establishing their ready endorsement of racial stereotypes—that they treat race as rich in inductive potential. Moreover, these are robust cognitive “achievements”: children readily learn these things whether their parents endorse them and they learn them even if they suffer significant neurological insult.

If a Child Acts Color-Blind, He Is

Earlier it was observed that preschool age children don't act as if they are prejudiced. Experimental studies, such as those reviewed here, have shown that children endorse particularly strident racial prejudice, yet parents and teachers seldom report seeing young children discriminate on a racial basis (Aboud and Levy, Reference Aboud, Levy and Oskamp2000; Van Ausdale and Feagin, Reference Van Ausdale and Feagin2001). As already observed, parents and teachers underestimate the role of race in young children's lives, and this includes the degree to which discrimination and prejudice shape children's interactions with peers. Still, even though preschool children do display race-based discriminatory behaviors, they tend to be subtle enough to afford parental and teacher misinterpretation (Derman-Sparks and Ramsey, Reference Derman-Sparks and Ramsey2011; Fishbein and Imai, Reference Fishbein and Imai1993; Van Ausdale and Feagin, Reference Van Ausdale and Feagin2001).

More structured studies similarly have found that preschool age children seem “color-blind” when asked to nominate who they like and dislike, who they might invite to a party, with whom they are likely to play at school (Kinzler and Spelke, Reference Kinzler and Spelke2011). The explanation for the gap between attitude and action is not that young children's prejudices generally do not affect interaction: gender plays a central role in organizing children's interactions and gender stereotypes are readily voiced and frequently used by children as rationale for exclusionary and inclusionary behaviors. It isn't that preschool children don't stereotype and practice bias in interactions, rather the effects on behavior of racial stereotypes and biases seem to be fairly limited. Limited enough to convince parents and teachers that children of this age are by and large color-blind.

I suggest that two aspects of young children's racial thinking described above—preference for attending to and adopting community sentiments over local, parochial ones and a favoring of discursive over visual information during early knowledge development—help explain this otherwise paradoxical finding that young children's racial beliefs don't seem to predict behavior.

A hint of why this may happen comes from research that identifies conditions under which racial or ethnic differences do play out in children's everyday interactions. Doyle et al. (Reference Doyle, Rappard and Connolly1980) found that among preschoolers in Montreal where ethnic differences map onto language differences, everyday interactions are consistent with assessed interethnic attitudes. Thus, when children are able to identify ethnic group membership on the basis of language, ethnicity is important in predicting everyday exclusionary and inclusionary practices. Kinzler and her colleagues (Kinzler et al., Reference Kinzler, Dupoux and Spelke2007; Kinzler and Spelke, Reference Kinzler and Spelke2011; Shutts et al., Reference Shutts, Banaji and Spelke2010) also found that language-based preferences predict action. In one study, they found that younger infants looked longer at a speaker of their (caretaker's) native language and that older infants preferred to accept toys from a stranger speaking their (caretaker's) native language (Kinzler et al., Reference Kinzler, Dupoux and Spelke2007). In another study, they compared this pattern with race-based choices: in contrast to language-based preferences, they found that both ten-month old infants and two-and-a-half-year-old toddlers were as willing to accept a toy from an other-race stranger as from an own-race stranger (Kinzler and Spelke, Reference Kinzler and Spelke2011). (Yet another study revealed that by three years of age, children did prefer objects that were endorsed by someone of their own race—as well as the same gender and age—although the effect for race was weaker (Shutts et al., Reference Shutts, Banaji and Spelke2010)). In sum, of the social dimensions that infants show an early capacity to distinguish—gender, age, language, and race—race is exceptional in that it does not translate into behavioral preferences until late preschool age. Kinzler and Spelke (Reference Kinzler and Spelke2011) conclude from these findings “race-based social preferences may emerge only by the end of the preschool years” (p. 6). But we already know that this is only part of the story: children show tacit but strident race-based attitudes much younger.

Why then do such preferences emerge less precociously on behavioral measures like friendship choice or experimental ones like object-taking preference? I suggest that the answer lies in the way that children learn about the social world generally and the historical nature of race-based preferences in particular. The young child's knowledge of race cannot be innately specified in any comprehensive fashion, as noted earlier, in significant measure because racial differences are not something humans encountered prior to the advent of large-scale migrations and efficient overseas travel, a relatively recent phenomenon historically: for the bulk of human experience, people of different communities looked much alike in terms of the corporal traits associated with race (viz, face shape, skin pigmentation, hair, etc.) (Cosmides et al., Reference Cosmides, Tooby and Kurzban2003). Other social dimensions like gender and age have a much deeper evolutionary genealogy; language is more recent, but nonetheless ancient enough to be subserved by a dedicated, evolved mechanism (Hauser et al., Reference Hauser, Chomsky and Fitch2002).

Social distinctions whose acquisition can recruit innately endowed mechanisms of detection not surprisingly are more precocious in shaping behavior than social distinctions whose acquisition does not recruit such mechanisms. Hence, gender-, age-, and language-based preferences may shape behavior earlier than race-based preferences because their acquisition is facilitated by cognitive capacities already online at an earlier age. Of course, race-based preferences do recruit a dedicated cognitive mechanism, namely, a dedicated system for processing faces. But this is not a system for race-based classification per se; rather such systems of classification use a dedicated mechanism that evolved to serve another function, namely, to recognize and individuate conspecifics (Pascalis and Kelly, Reference Pascalis and Kelly2009). As observed earlier, race-based preferences among infants are a function of experience and easily extinguished with relatively minor changes in the infant's familiarity with individuals of another race; these preferences are not a function of preferences evolved specifically to racially discriminate between individuals.

Keep in mind what the child's task is as she learns about race and comes to endorse the race-based preferences of the social world she inhabits. Learning about race is part of the child's efforts at determining those aspects of group identity that are important in organizing broad swaths of human experience under the broadest range of situations. In contemporary American culture, race (along with gender) is critically important in this regard. In other cultural environments, race is almost certainly less important if present at all (consider contemporary South Asia). What the child spends his or her time on is the problem of identifying the scope and nature of the social landscape—figuring out the groups and their “natures.” Only later does the task become figuring out how to reliably distinguish between members of one versus another group.

The young child does not learn about race in isolation, but learns about race in the context of a more global task of parsing the social world. This consists of attending to information that is relevant to the scope of the social groups that exist in that world (cataloguing social entities) and attending to information that is relevant to the nature of the social groups that exist in that world (cataloguing the significance that membership in a social groups entails). Human society, like those of other primates, is marked by multiple group affiliations and a cultural system (or multiple systems) for identifying which affiliations are relevant to which situations. Human societies unlike those of other primates are marked by two distinctions: first, humans belong to many more groups than do members of other primate societies, and second, humans name the groups to which they belong.

For the young child, the task of understanding the social world is approached not by discovering what the membership criteria of all groups are, but rather mapping the social world by the groups themselves and by attention to information about their nature. In short, the child is seeking to conceptually articulate a social ontology, to identify and understand the nature and scope of the groups that constitute society. Not all groups are created equal: some human groups are relevant to more situations and are more informative of behavior than others. Often these groups tend to be seen as having a natural basis; these are groups that are thought to be part of the natural landscape. Gender and kinship are two prominent examples. Gender is something we are born with.Footnote ² A family is something that we are born into. Both statuses—one's gender and one's place in a family—are seen as determined by natural processes, grounded in the folk biology of reproduction. Race is another instance of putative natural status. Since appearance and racial group membership do not always map well onto each other, all societies in which race is an important social dimension must contrive ways to regulate disjunctions between appearance and racial group membership. In apartheid South Africa, for instance, formal courts existed to adjudicate cases in which racial ambiguities were raised. In North America, the so-called one-drop rule served both within the legal system and more generally outside it (where it continues to apply) to regulate race's porous biological borders. Under it, history (or genealogy) trumps appearance: one isn't Black because one looks Black; one is Black because one has significant Black ancestry (although one is not White in virtue of having significant White ancestry).

A child attempting to “master” racial logic in the United States would be misled if she slavishly attended solely to appearance. The logic of the system is one of natural relatedness, and that relatedness is thought to reflect an underlying logic of reproduction not appearance. All things being equal, appearance and reproduction covary, and this provides a handy but not infallible guide. In the scope of racial systems, the United States is associated with more readily apparent physical differences than are common elsewhere. Although no racialized system of thought seamlessly maps onto differences in appearance, ideological commitment to deeper differences in physical appearance is the rule even when such differences are hard for even the educated outsider to perceive. One need only read accounts of the Irish in British literature, Jews in virtually every European tradition, or discussions of Serbs by Bosnians to appreciate that actual differences in appearance count less than imagined ones.

Given all this, it makes sense that young children adopt a top-down approach to acquiring information about the prevailing social ontology, relying on a learning strategy that focuses attention on the logic of the system rather than worrying the details of individual identification. A child acquiring knowledge of race is in the business of acquiring knowledge of the kinds of things of which society is composed, not a catalogue of corporal variation. The child recognizes that society is made up both of individuals and of kinds of individuals, categorizable into groups of various sorts. There is, as already noted, strong evidence that infants attend closely to both the general (categories of people) and particular (individual persons) levels. Infants prefer representations of their own mother's face, but also female over male faces, familiar and friendly individuals over strangers and unfriendly ones, own-race over other-race faces, and children over adults (Pascalis and Kelly, Reference Pascalis and Kelly2009). During infancy knowledge growth follows a pattern that favors the acquisition of broad categories over more specific ones (Mandler Reference Mandler1992). Moreover, infants display an innate readiness to interpret efforts to focus their attention as efforts to teach about general properties rather specific ones (Csibra Reference Csibra2010). Toddlers, in turn, display a parallel, default readiness to interpret labels as referring to generic qualities of objects and creatures rather than to specific individuals and their properties (Gelman et al., Reference Gelman, Ware and Kleinberg2010). The young child's preferential attention to the broad properties of race as a kind of category of humans rather than the details of the diacritics of racial variation is of a piece with how young children develop knowledge across a number of domains.

Given greater attention to labels than physical features—i.e., talk about race rather than to limited interactions in which it may figure—means that during early stages of acquiring racial knowledge the young child comes to know, say, that he prefers Whites to Blacks but be less certain of how to identify who is Black and who is White. The force of the American racial system is a systematic favoring of the White majority and disfavoring of minorities. Learning this brute fact—particularly in the hypersegregated neighborhoods of much of the United States in which encounters by young children of other-race folk is either infrequent or buffered—is learning the fundamental nature of race in America. The cultural input is clear—Black and White refer to categories of people whose membership is grounded in natural processes; these natural processes also cause differences in inborn qualities including social capacity and potential and in outward qualities like physical appearance; and this combination of inborn and outward qualities leads to differential access to the things most valued socially. What the very young child does not know, because she approaches the question from the top-down not the bottom-up, is which physical differences are relevant to each category. Many of the robust findings we've reviewed here make sense in this light. In playmate choice or in deciding whether to accept a toy from someone, it may not be that race doesn't matter, it may be that race is still undecipherable at the level of the individual. When young White children make such “errors” they appear to be color-blind since their choices do not reveal the tacit, underlying racial prejudices they already endorse. When young Black children make a self-identification “errors”—as did Kenneth and Mamie Clark's subjects who incorrectly identified themselves as White—the same mechanism is involved: they know that White is preferred, privileged, and more attractive, and in the absence of knowledge that inhibits identification with the preferred, privileged, and more attractive option, they opt for it.

This is not to suggest that perceptual cues are irrelevant to the early development of racial concepts, they simply underdetermine them. Young children readily differentiate people on the basis of racially relevant physical features. Preschoolers also have racial categories: they recognize the existence of named, enduring groups of humans. Virtually all previous work on racial concepts has assumed that these two phenomena comprise a single system. The argument I am making here is that they represent two distinct but overlapping conceptualizations. The systems articulate in that racial concepts contain some perceptual information although it is extremely fragmentary, specifying only that certain kinds of perceptual differences are pertinent to group membership (young children probably realize that skin tone is a pertinent dimension, finger length is not). It is not that perceptual input is completely irrelevant to emerging racial categories. Rather, I propose that in building racial categories, obvious surface cues like skin color or the shape of the facial features or qualities of hair are not defining for young children. As a consequence, younger preschoolers are not terribly adept at identifying how specific categories map onto specific individuals. As a result, they are not adept at translating their race-based preferences in attitude into race-based preferences in action.

Although race-based differences are a relatively recent aspect of human social experience—and thus not subserved by an innately specified, evolved mechanism—a readiness to map differences in appearance onto groups may nonetheless be an ancient and possibly innately specified competence. Racial difference may (from an evolutionary perspective) be a recent phenomenon, but this does not mean that our ancestors inhabited a social landscape in which group-based differences were absent. Neighbors almost certainly didn't speak the same way (the peculiar human ability to learn multiple languages simultaneously as easily as to learn a single language suggests that this may have been sufficiently common to represent an evolved competence (Hirschfeld Reference Hirschfeld2008a)). Nor did they likely dress alike. It is also plausible that neighboring groups altered their physical appearance by wearing culturally distinct masks and other accouterments of cultural identity or even by altering their bodies through painting, tattoos, or scarification, to mention techniques long familiar to ethnographers. Human ancestral life was not radically isolated, as is sometimes imagined; groups engaged in long distance trade and almost certainly exchanged members through peaceful means such as marriage as well as through warfare (Gibson Reference Gibson, Mellars and Gibson1996; Tattersal and Schwartz, Reference Tattersal and Schwartz2000). Still, evidence for cultural variation is relatively recent, almost certainly limited to modern Homo sapiens (Hirschfeld Reference Hirschfeld2007). As said earlier, opportunities for encountering racially different folk is so recent as to foreclose the possibility of evolved cognitive mechanisms consecrated to race-based distinctions. It is possible, however, that noncorporal and/or uninherited corporal distinctions are something that humans were prepared by evolution to attend to and invest with social consequence (Hirschfeld Reference Hirschfeld1996).

Reducing Prejudice Is Best Achieved by Affirming that Deep Down Everyone is the Same and/or that Any Differences Between Them Should Be Celebrated, Not Scorned

Myths are cumulative. Like lies—where a first fib engenders a second, and the second a third—myths tend to bloat from isolated assertions to encumbered theories, all the more so when the object mythologized is especially significant. Myths about race and children are a good example. Start with the idea that children don't really notice race. Okay, it's not that they don't notice so much as they don't talk about it the way adults do, maybe a passing note of curiosity here or there about skin color or curly hair. Still, it's only these surface-level things that little Johnnie is concerned with. Deep down he knows that everyone is the same, even if his parents do lock the car doors when they go through certain neighborhoods, even if TV commercials for athletic shoes depict either crazed men of color performing superhuman feats of violence on inner-city netless basketball courts or determined, sculpted, disciplined White professionals working out on stainless steel exercise machines in well-heeled gyms. But yes, the best way to insure that Johnnie doesn't get the wrong message from this is to assure him that every group deserves a day, a month, a little section of the school curriculum, that we should all taste the quotidian cuisines of peoples the world over, that it's fun to dress up like the X or Y or sing their songs or dance their dances. Celebrate difference. Superficialize it. That will do the trick.

I am perhaps being unfair here. But if the child is strongly committed to a vision of race as anything but superficial and if the child has closely attended to the community conventions, standards, and practices that testify that (minority) race is not something to be celebrated, it is hard not to see multiculturalism as an attempt to plug a crumbling dike with one's thumb. Indeed, this is largely the burden of systematic research assessing this approach. As Bigler and Hughes (Reference Bigler and Hughes2010) observe:

If the goal is to change children's racial beliefs, then employing a strategy that largely makes no contact with children's prior expectations is unlikely to succeed.

The nature of the child's expectations itself suggests an alternative strategy. At risk of oversimplifying, the widely implemented, multicultural approach (sometimes unflatteringly called multicultural tourism) presumes that young children, given their cognitive and emotional immaturity, are unable to appreciate the complexities of adult representations of race. As we've seen, this is not the case. Even three-year-olds' grasp of race is markedly adultlike. This is not to suggest that preschoolers possess an adult, as opposed to adultlike, understanding. Still, as I've indicated there is considerable experimental evidence challenging the claim that young children are unable to grasp the nature of the adult image of race. Telling children that (minority) race is something to celebrate when three-year-olds already know enough of the world to reject this is not likely to succeed. By the same token, telling children that race is skin deep, that everyone is essentially the same down deep, when three-year-olds already believe that race is naturally grounded, unevenly distributed, and richly informative of other attributes, is also not likely to succeed.Footnote ³

An alternative approach would present young children with appropriate—not dumbed down—descriptions of the nature and scope of structural racial inequity, particularly its grounding in familiar quotidian acts and everyday experience. Of course this would need to be cast in contexts familiar to the young child, but given that peer culture is rife with acts of inclusion and exclusion (Hirschfeld Reference Hirschfeld, Quintana and McKown2008b; Van Ausdale and Feagin, Reference Van Ausdale and Feagin2001), it is plausible that children would be able to recruit (i.e., extrapolate from) these relations to understand those of others (Hughes et al., Reference Hughes, Bigler and Levy2007).

Understanding and empathy, however, are only a part of the process that leads to attitude, and ultimately behavioral, change. Like adults, children need to challenge their own beliefs to master the consequences of them. A crucial part of the process of behavior change turns on motivation. It is now widely acknowledged that individuals who score as having low levels of prejudice on standard measures of racial bias are in fact nonconsciously biased (Banaji and Greenwald, Reference Banaji, Greenwald, Zanna and Olson1994; Devine Reference Devine, Pratkanis, Breckler and Greenwald1989). This is not surprising given the extent to which our society is saturated with racial (and other) prejudice. What makes someone low prejudiced is not the absence of bias but the motivation to cognitively inhibit bias from shaping behavior. Frequently the motivation is a sense of fairness. Happily, children's sense of fairness—a robust sense of morality based on the notion of harm (Turiel Reference Turiel2002)—is acute, widely shared, and powerfully motivating (Derman-Sparks and Ramsey, Reference Derman-Sparks and Ramsey2011).

By the same token, despite considerable efforts to change how Americans envision race—here I don't mean efforts to reduce racism, but efforts to erode confidence in the underlying beliefs about the nature of race that support racism—surprisingly little has changed. Ann Morning's (Reference Morning2011) close and systematic examination of how the concept of race is taught in American high schools and universities is particularly telling in this regard. Her central focus is on the essentialist notion of race which other research has shown to promote stereotyping and prejudice (Bastian and Haslam, Reference Bastian and Haslam2007). An essentialist vision of race includes several linked beliefs: the idea 1) that humans can be partitioned into distinct racial types on the basis of their concrete, observable constitution; 2) that these differences are embodied, natural and enduring; and 3) that they encompass nonobvious, inner qualities (including moral and mental ones) as well as outward physical ones (Hirschfeld Reference Hirschfeld1996). Morning (Reference Morning2011) notes that after substantial educational investment in challenging this constellation of beliefs, and the “flourishing in our ostensibly postracial and multiracial era” (p. 247), racial essentialism remains largely intact:

It is perhaps less surprising when we reflect on the mixed message, and the underlying ambivalence in motivation, that antiracist discourse often delivers. Attempts to increase empathy risk fostering pity rather than respect (Aboud and Levy, Reference Aboud, Levy and Oskamp2000). Celebrations of struggle also risk confirming rather than undermining stereotypes, as Kohl (Reference Kohl1995) illustrates in his examination of treatment of Rosa Parks' life story in middle-school textbooks (note it is the Rosa Parks story, not the story of the Montgomery bus boycott). He found systematic omissions and distortions, supposedly made to render the episode in more age-appropriate terms, one consequence of which is Parks is lionized as a seamstress too tired after a day of laboring to give up her seat, not as a long-time civil rights activist. The take-away image of Parks is one of an individual, acting alone and out of fatigue, and the Montgomery Boycott as the Black community's impulsive reaction to her arrest rather than political action by design. In short, and tragically, the textbook accounts almost surely convey and reinforce more prejudice than they potentially reduce.

The troubling conclusion is that efforts to reduce racism fail because our society gets the “cure” that it's looking for. Majority society really doesn't want to lose the edge that racism affords. Earlier I discussed stereotype threat, the effect in which stigmatized minority's test performance decreases simply by virtue of the test-taker being reminded of the prevalent stereotype (Steele and Aronson, Reference Steele and Aronson1995). The effect has deservingly attracted a great deal of attention; the notion that a well-prepared individual with full confidence in her abilities is undercut merely by being reminded that others might judge her on the basis of a stereotype is indeed troubling. Less discussed but as well established is the hegemonic counterpoint to stereotype threat called “stereotype lift.” In stereotype lift, majority test-takers show enhanced performance on standardized tests when they are reminded of stereotypes that minority test-takers do less well on such exams (Chalabaev et al., Reference Chalabaev, Stone, Sarrazin and Croizet2008; Marx and Stapel, Reference Marx and Stapel2006; Walton Reference Walton2003). Of course this is hardly the only or even the most important edge that majority race individuals profit from, but it is a dramatic demonstration of what would be at stake in a society in which race didn't regulate access to virtually all strategic resources.

Again, perhaps American society does so poorly at combatting racism because its heart really isn't the struggle. Perhaps Americans know much more about racial thinking and racism than about how to reduce it because (so many of) our hearts aren't in that fight either. I'm reminded of an argument proposed to me two decades ago by the late biomedical scholar and activist Joanne Lukomnik. She suggested that we know so much about depression not simply because our knowledge of depression as a brain disorder has progressed, but because government support for research dramatically shifted away from an understanding of suffering as a function of the social conditions under which people exist—the hallmark of social policy from Roosevelt to Johnson—to an understanding of suffering as a function of the individual's biomedical endowment and errant neurotransmitter uptake—the hallmark of social policy of the past four decades. Unhappiness is a problem with our inborn natures not with the nature of our society. Ironically, the same impulse to ground differences—from susceptibility to diseases to expectations of educational achievement—in an individual's biomedical constitution encourages precisely the sort of racialism that so readily delivers stereotyping and prejudice.