Priming Race

Abstract

Recent research shows that racial categorization can be reduced by contexts in which race does not predict how people interact and get along—a manipulation with little to no effect on sex and age. This suggests that our minds attend to race as an implicit cue to how people are likely to get along. However, the underlying mechanism of how these contexts reduce race is not yet known. Is race not encoded? Or, is race encoded, but then inhibited? The present study arbitrates between these possibilities. Results demonstrate that the reduction in racial categorization is happening at recall. Participants are still encoding targets’ race, but this information is locked away or inhibited. This clarifies how the mind switches away from previously relevant, but now irrelevant, social cues: it does not immediately abandon them, rather, it encodes them but inhibits their use.

Keywords

race coalitional psychology evolutionary psychology social categorization

The Psychology of Coalitions and Alliances

From an evolutionary perspective, the ability to perceive and engage with social groups is made possible by cognitive adaptations designed around coalitional interactions over evolutionary time (e.g., Byrne & Whiten, 1988; Pietraszewski, 2013). Coalitions are sets of three or more individuals who cooperate over time, often in competition with other such sets; are central to the daily lives of humans and other primates; and are powerful determinants of important and fitness-relevant outcomes (e.g., Manson & Wrangham, 1991; von Rueden, Gurven, & Kaplan, 2008).

One function of coalitional adaptations is to predict patterns of social interaction before they occur (Harcourt, 1988; Pietraszewski & German, 2013). This requires monitoring the world for coordination, cooperation, and competition behaviors and attending to any cues in the environment—such as location, dress, behavior, and so on—that happen to correlate with these behaviors and therefore help predict them ahead of time. To generate the right inferences at the right time, coalitional adaptations must also retrieve and activate those cues that are most likely to be relevant for understanding and predicting behavior within a particular situation (Pietraszewski, Cosmides, & Tooby, 2014).

Evidence for each function has been found: People are spontaneously and implicitly categorized by patterns of coordination and cooperation, and when external cues (such as clothing differences or badges) happen to correlate with these patterns, they also become a strong basis for categorization (Pietraszewski et al., 2014; Pietraszewski, Curry, Petersen, Cosmides, & Tooby, 2015). The activation of coalitional categories (which are the mental representations of the cues that help predict patterns of coordination, cooperation, and competition) is also sensitive to manipulations of contextual relevance, such that when experimental stimuli cue the relevance of a particular coalition category, categorization by that category is upregulated and the nonrelevant coalition category downregulated (Pietraszewski et al., 2014).

Racial Categorization

Evidence suggests that the mind implicitly treats the social category [race] as an instance of a coalitional category and that the spontaneous and apparently automatic categorization of others by their race is a reducible by-product of coalition detection, rather than being an unalterable product of cognitive adaptations for attending to race, as once thought (Cosmides, Tooby, & Kurzban, 2003; Hamilton, Stroessner, & Driscoll, 1994; Kurzban, Tooby, & Cosmides, 2001; Messick & Mackie, 1989; Pietraszewski et al., 2014).¹

Experimental evidence shows that people are spontaneously and implicitly categorized by their race when no alternative coalitional information is provided, and that racial categorization does not change in response to manipulations that do not convey coalitional information. These include priming race, priming a dimension that crosscuts race, manipulating contextual relevance (i.e., showing people discussing race relations or some other topic), and explicit instructions to either attend or not attend to race (e.g., Bennett & Sani, 2003; Hewstone, Hantzi, & Johnson, 1991; Stangor, Lynch, Duan, & Glass, 1992; Susskind, 2007; Taylor, Fiske, Etcoff, & Ruderman, 1978).

However, when an experimental context does contain coalitional information, such that an alternative coalition dimension is presented and race no longer correlates who is allied with whom, spontaneous and implicit categorization by race is reduced, both in cooperative contexts (Pietraszewski et al., 2014; Pietraszewski & Schwartz, 2014b) and, most recently, in political contexts (Pietraszewski et al., 2015). Moreover, other chronically activated categories, such as sex, age, or accent, are not effected by these very same manipulations. Racial categorization is also upregulated and downregulated by the same contextual relevance manipulations that upregulate and downregulate other coalitional categories (Pietraszewski et al., 2014; see also Cabecinhas & Amâncio, 1999; Maddox & Chase, 2004; Pietraszewski & Schwartz, 2014a, 2014b; van Bavel & Cunningham, 2009; Weisman, Johnson, & Shutts, 2014; Wilson, See, Bernstein, Hugenberg, & Chartier, 2014 for complementary evidence).

The coalitional account of racial categorization is an account of both the social context and the information processing that leads to racial representation (see Tajfel, 1981): In cultures where physical features become associated with patterns of association, cooperation, and competition through historical and sociological processes (Sidanius & Pratto, 1999; Telles, 2004), coalitional adaptations will encode and store these physical features as probabilistic cues of social interaction, boost their experienced salience; cause them to be encoded, stored, and retrieved more readily; and become the basis of person perception and categorization (Cosmides et al., 2003; Pietraszewski et al., 2014).

Consistent with this, accumulating evidence shows that (1) racial perception, categorization, and information storage processes all respond to social rather than to biological or visual interventions and (2) biological and visible properties are not sufficient on their own to produce racial perceptions without additional social and linguistic inputs (e.g., Hirschfeld, 1996; Peery & Bodenhausen, 2008; Sack, 2005; Sidanius & Pratto, 1999; Stangor et al., 1992; van Bavel & Cunningham, 2009; Wilson et al., 2014).² These results also converge with sociological, historical, and genetic analyses of race—that perceptions of race are grounded in social experiences not in biological reality or visual salience (e.g., Graves, 2001; Tishkoff & Kidd, 2004).

Current Studies

Although prior studies show that racial categorization is reduced, the mechanism of the reduction is not presently known. There are two possible ways that race could be reduced by these manipulations: (1) at the point of encoding, such that when participants see that race is not predictive of coalitional patterns, racial category information is not taken in as strongly or (2) at the point of recall, such that participants still encode targets’ race, but this category information is inhibited at the point of recall (i.e., when the dependent measure is being collected, which in the real world would map onto making inferences and generating expectations). Although the proposal that racial categorization is a by-product of coalitional categorization does not in itself directly speak to this issue, knowing which of these two is happening is important because it reveals how the mind handles switching away from a previously relevant, but now irrelevant, social cue: Does it ignore it completely or does it attend to it but then inhibit it?

To test between these two possibilities—that the change in race is happening either at encoding or at recall—the current study reverses the coalitional manipulation after the encoding phase but before the collection of the dependent measure. If the coalitional manipulation had caused participants to not encode race in the first place, then this additional information cannot not have any effect, because there is no racial category information in participants’ minds to re-prime and thus raise. However, if the coalitional manipulation causes participants to encode but inhibit race, then this information should have a reversal effect, raising racial categorization back up to near baseline levels (i.e., to levels of categorization when no crosscutting alliance information is provided).

Method

Categorization was measured using the Who Said What? memory confusion paradigm, which unobtrusively measures whether a shared dimension is implicitly noticed and remembered by looking at patterns of memory errors. The paradigm features three phases: (1) an initial presentation phase, during which speakers are shown making statements, (2) a 1-min distractor task, and (3) a recall phase in which all of the speakers seen previously are presented in a randomized array and participants try to assign each statement to the correct speaker (see Pietraszewski et al., 2014, for details).

Three between-subject conditions were compared: a noncoalitional baseline, a nonprime coalitional condition, and a race prime coalitional condition.³ In the noncoalitional baseline, different race targets were presented in a neutral, noncoalitional context. This provided a measurement of racial categorization for the particular stimuli used (photos, statements lengths, etc.) in the absence of any coalitional information. The nonprime and race prime coalitional conditions both featured the different race targets within a coalitional context, such that race was crossed with membership in one of two different charity groups, and group membership was marked with shared shirt colors (this manipulation reduces race categorization and has no effect on sex categorization; Pietraszewski et al., 2014). The presentation and recall phases of both of the conditions were identical. The only difference was the 1-min distractor task. In the nonprime coalitional condition, the distractor task was neutral. Participants were shown an outline map of the United States and asked to think of as many state capitals as possible (this was also the distractor task in the noncoalitional baseline). In the race prime coalitional condition, the distractor task was changed to prime the coalitional relevance of race. Participants were shown a collection of racially charged photos and instructed to think of “all the racial conflict and inequality you have been exposed to or are aware of”⁴ (see Figure 1).

Figure 1.
Paradigm design (left to right): Participants first see a sequence of target photos and statements (encoding phase), a 1-min distractor task (depicted on darkened base), and then an array of all of the face photos they had seen previously and are asked to attribute each statement to the correct speaker (recall phase). The only difference between the between-subject prime and nonprime conditions was the 1-min distractor task. In the nonprime condition, this was a neutral task: to think of the capitals of the 50 U.S. states (accompanied by map photo). In the race prime condition: to think of racial conflict and inequality (accompanied by racial conflict photo). Male targets are depicted.

This manipulation of the distractor task will diagnose how race is being reduced. If categorization by race remains low after being primed in the distractor task, race is being reduced at encoding. If categorization by race increases back to baseline levels, however, race is being reduced at recall.

Methodological details of the noncoalitional baseline and nonprime conditions can be found in Pietraszewski et al., 2014 (studies 1 and 5). The methodological details of the race prime condition are described below.

Participants

Sixty-five undergraduate students participated in the race prime conditions (32 females, 33 males; mean age = 20.4 years, SD = 1.49), either for research credit in introductory psychology, or anthropology classes, or for pay (US$12).

Design

There were two between-subject race prime conditions. Male participants viewed all-male targets and female participants all-female targets. Race was crossed with charity group membership, which was marked by shirt color, and the statements presented during the recall phase contained no coalition diagnostic information (to prevent strategic guessing when shirt colors are present⁵). These conditions were chosen because they were expected to produce some of the lowest levels of categorization by race and because the two hypotheses being tested predict either no change or an increase in categorization by race. Therefore, starting with low levels of racial categorization provides the clearest and most sensitive test between the two. Sample sizes were the same as in Pietraszewski, Cosmides, and Tooby (2014). No pilot testing was conducted.

Material and Procedure

Materials and procedure were identical to the coalitions-irrelevant ⁶ conditions of Pietraszewski et al. (2014, study 5). Race was crossed with charity group membership, and group membership was marked by shirt color (see Figure 1).

Results

The results of the noncoalitional baseline, nonprime, and race prime conditions are first reported, then the race prime results are compared with the baseline and nonprime results.⁷ The critical question is whether the race prime result will be more similar to the nonprime result or to the noncoalitional baseline result. Categorization is quantified by the degree to which within-category errors exceed between category errors (see Pietraszewski et al., 2014 for details).

What is the level of categorization by race in the noncoalitional baseline conditions? Strong categorization by race occurred. Male participants viewing male targets in the noncoalitional baseline condition made substantially more within-race errors (M = 6.39, SD = 1.87) than between-race errors, M = 3.91, SD = 1.35, t(31) = 5.13, p < .001, r = .68. Female participants viewing female targets in the noncoalitional baseline condition also made substantially more within-race errors (M = 5.19, SD = 1.81) than between-race errors, M = 3.75, SD = 1.59, t(35) = 3.35, p = .002, r = .49.

What is the level of categorization by race in the nonprimed coalition conditions? Categorization by race was substantially reduced by the coalitional manipulation. Male participants viewing male targets did not significantly categorize targets by their race (within-race errors: M = 5.65, SD = 1.42; between-race errors: M = 5.08, SD = 1.46, t(32) = 1.37, p = .180, r = .24), and female participants viewing female targets did not significantly categorize targets by race (within-race errors: M = 5.48, SD = 1.69; between-race errors: M = 5.13, SD = 1.29, t(51) = 1.00, p = .321, r = .14). In the coalitional condition not involving a prime, there was no significant categorization by race.

What is the level of categorization by race in the race prime conditions? Male participants viewing male targets made substantially more within-race errors (M = 6.62, SD = 1.63) than between-race errors, M = 4.53, SD = 1.65, t(32) = 4.25, p < .001, r = .60. Female participants viewing female targets also made substantially more within-race errors (M = 6.63, SD = 1.86) than between-race errors, M = 5.11, SD = 1.10, t(31) = 3.16, p = .004, r = .49. In the coalitional condition with a race prime, there was significant categorization by race.

The critical question is how does this categorization in the race prime condition compare with the levels found in the noncoalitional baseline and in the nonprime coalitional condition?

How does this level of categorization by race compare with the nonprimed coalition conditions? A linear regression revealed that categorization by race was higher in the race prime condition than in the nonprime condition, for both male participants (race prime: M = 2.09, SD = 2.83; nonprime: M = .58, SD = 2.41, t(64) = 2.34, p = .022, r = .28) and female participants (race prime: M = 1.52, SD = 2.72; nonprime: M = .36, SD = 2.56, t(82) = 1.97, p = .052, r = .21).

How does this level of categorization by race compare with the noncoalitional baseline conditions? A linear regression revealed that categorization by race in the race prime condition was not different from the noncoalitional baseline condition, neither for male participants (race prime: M = 2.09, SD = 2.83; noncoalitional baseline: M = 2.48, SD = 2.74, t(63) = .57, p = .571, r = .07) nor for female participants (race prime: M = 1.52, SD = 2.72; noncoalitional baseline: M = 1.44, SD = 2.59, t(66) = .11, p = .912, r = . 01).

In sum, categorization by race strongly increased in the race prime conditions (for both male and female participants) and was very similar to the levels found in the noncoalitional baselines (see Figure 2).

Figure 2.
Categorization by race and coalition in the nonprime and race prime conditions. Baseline levels of categorization by race (in which race is presented in a neutral, noncoalitional context) is depicted in the gray box. When race was primed after the encoding phase, but before the recall phase (race prime), levels of categorization by race increased back up to the levels found in the noncoalitional baseline (and were not significantly different from those baseline levels). This means racial categorization is encoded but inhibited when crossed with coalition. Categorization by coalition was not affected by the race prime manipulation (no significant difference was found between the nonprime and race prime conditions). Error bars: ± 1 SE

Finally, categorization by coalition, rather than race, can also be compared across the race prime and the nonprime coalition conditions. If the race prime manipulation is only affecting race, and not the orthogonal coalition dimension, then categorization by coalition will not differ across these two conditions.

What is the level of categorization by coalition in the race prime conditions? Male participants viewing male targets made more within-coalition errors (M = 6.15, SD = 1.75) than between-coalition errors, M = 5.00, SD = 1.60, t(32) = 2.27, p = .030, r = .37. Female participants viewing female targets also made more within-coalition errors (M = 6.52, SD = 2.28) than between-coalition errors, M = 5.22, SD = 1.66, t(31) = 1.96, p = .059, r = .33.

What is the level of categorization by coalition in the nonprime coalition conditions? Male participants viewing male targets made more within-coalition errors (M = 6.14, SD = 2.13) than between-coalition errors, M = 4.49, SD = 1.65, t(32) = 2.68, p = .011, r = .43. Female participants viewing female targets also made more within-coalition errors (M = 5.97, SD = 1.99) than between-coalition errors, M = 4.64, SD = 1.80, t(51) = 2.78, p = .008, r = .36.

Does categorization by coalition differ between the race prime and the nonprime coalition conditions? No. A linear regression revealed that categorization by coalition did not differ between the race prime and nonprime conditions for either male or female participants (males: race prime M = 1.15, SD = 2.91; nonprime M = 1.65, SD = 3.54, t(64) = .627, p = .533, r = .08; females: race prime M = 1.30, SD = 3.74; nonprime M = 1.34, SD = 3.46, t(82) = .050, p = .961, r = .01; see Figure 2).

Discussion

Categorization by race increased dramatically when the race prime was introduced after the initial presentation phase. In fact, categorization by race in the race prime conditions was nearly identical to that found in the noncoalitional baseline conditions (see Figure 2). Re-cuing that race is predictive of coalitional alliance after the initial presentation phase dramatically increases the degree to which participants categorize targets by race.

This suggests that coalitional manipulations decrease race by inhibiting or locking away its representation from the rest of the cognitive architecture (or minimally that part of the architecture underwriting statement attribution). In other words, the manipulation is affecting recall, not encoding. This suggests that the mind may simultaneously still encode a previously diagnostic coalitional cue while at the same time inhibiting it.⁸

This makes good design sense for a psychology that faces the task of tracking multiple, changing cues in the world of multiple coalitions: It should be somewhat incredulous that a chronically diagnostic cue will never be diagnostic again when presented with a particular case in which it is not and thus should store but inhibit its use (i.e., promiscuously encode but selectively retrieve). In the current studies, race is not predictive among these particular targets within this particular coalitional context—but it may be for other targets or even among these targets in other contexts.

Therefore, an important direction of future research will be to determine the scope of coalitional manipulations on race. Is race inhibited only for the targets seen in a study or would it also be inhibited for novel individuals? How long does race inhibition last, does it extend to novel situations and contexts? Aside from informing the scope of experimental (and or real-world) reductions, these follow-up studies will also inform how the mind dynamically tracks coalitional cues and how information applied to one person extends to others.

If the mind does not immediately abandon previously diagnostic cues, but inhibits their use, then there also needs to be something like a coalitional cue storage buffer in the mind, in which previously diagnostic cues are encoded and stored. The decay function of such a buffer, and what affects the decay function, will also be fruitful avenues of future research.

In terms of social relevance, these results suggest that in a society in which a particular set of features are perceived as race, a lack of correlation between those features and social interaction patterns will reduce racial categorization, but in a context-specific way—not immediately and absolutely. This need not be a pessimistic finding: This is additional evidence that racial categorization is fundamentally malleable and about social expectations and changes quickly in response to social interaction patterns.

However, a single cross-race cooperation event will not inexorably erase race once and for all. Nor should it: If race is implicitly understood as a social construct, then the averaged sum of experienced social dynamics will determine the strength and perniciousness of racial perception and categorization in each individual mind. Observing cross-race cooperation can powerfully impact and reduce race. Observing cross-race antagonism will reverse this effect and reinforce and strengthen racial representations (as these results show). Racial representations can be changed and doing so will be no harder and no easier than changing patterns of experienced and observed social interaction.

Finally, categorization by the novel (charity) group membership was not significantly reduced by the prime, suggesting that the mind has no trouble tracking two simultaneously predictive and yet orthogonal coalitional cues (computationally, this would mean the systems for tracking coalitional cues are noncompensatory). Although more work is needed to more firmly establish this noncompensatory effect, this is what we would expect to see from cognitive adaptations well designed for tracking multiple, nested coalitional relationships.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Bennett

Sani

(2003). The role of target gender and race in children’s encoding of category-neutral person information. British Journal of Developmental Psychology, 21, 99–112.

Byrne

R. W.

Whiten

(Eds.). (1988). Machiavellian intelligence: Social expertise and the evolution of intellect in monkeys, apes, and humans. Oxford, England: Clarendon Press.

Cabecinhas

Amâncio

(1999). Asymmetries in the perception of other as a function of social position and context. Swiss Journal of Psychology, 58, 40–50.

Cosmides

Tooby

Kurzban

(2003). Perceptions of race. Trends in Cognitive Sciences, 7, 173–179.

Graves

J. L.

(2001). The emperor’s new clothes: Biological theories of race at the millennium. New Brunswick, NJ: Rutgers University Press.

Evans

R. J.

(2004). The coming of the Third Reich. New York, NY: Penguin Books.

Hamilton

D. L.

Stroessner

S. J.

Driscoll

D. M.

(1994). Social cognition and the study of stereotyping. In Devine

P. G.

Hamilton

D. L.

Ostrom

T. M.

(Eds.), Social cognition: I Impact on social psychology (pp. 291–346). San Diego, CA: Academic Press.

Harcourt

A. H.

(1988). Alliances in contests and social intelligence. In Byrne

R. W.

Whiten

(Eds.), Machiavellian intelligence (pp. 132–152). New York, NY: Clarendon Press.

Hewstone

Hantzi

Johnston

(1991). Social categorization and person memory: The pervasiveness of race as an organizing principle. European Journal of Social Psychology, 21, 517–528.

10.

Hirschfeld

L. A.

(1996). Race in the making: Cognition, culture, and the child’s construction of human kinds. Cambridge, MA: MIT Press.

11.

Klauer

K. C.

Hölzenbein

Calanchini

Sherman

J. W.

(2014). How malleable is categorization by race? Evidence for competitive category use in social categorization. Journal of Personality and Social Psychology, 107, 21–40.

12.

Kurzban

Tooby

Cosmides

(2001). Can race be erased? Coalitional computation and social categorization. Proceedings of the National Academy of Sciences, 98, 15387–15392.

13.

Maddox

K. B.

Chase

S. G.

(2004). Manipulating subcategory salience: Exploring the link between skin tone and social perception of blacks. European Journal of Social Psychology, 34, 533–546.

14.

Manson

Wrangham

(1991). Intergroup aggression in chimpanzees and humans. Current Anthropology, 32, 369–390.

15.

Messick

D. M.

Mackie

D. M.

(1989). Intergroup relations. Annual Review of Psychology, 40, 45–81.

16.

Peery

Bodenhausen

G. V.

(2008). Black + White = Black. Psychological Science, 19, 973–977.

17.

Pietraszewski

(2013). What is group psychology? Adaptations for mapping shared intentional stances. In Banaji

Gelman

(Eds), Navigating the social world: What infants, children, and other species can teach us (pp. 253–257). New York, NY: Oxford University Press.

18.

Pietraszewski

Cosmides

Tooby

(2014). The content of our cooperation, not the color of our skin: An alliance detection system regulates categorization by coalition and race, but not sex. PLoS ONE, 9, e88534. doi:10.1371/journal.pone.0088534

19.

Pietraszewski

Curry

Petersen

M. B.

Cosmides

Tooby

(2015). Constituents of political cognition: Race, party politics, and the alliance detection system. Cognition, 140, 24–39.

20.

Pietraszewski

German

(2013). Coalitional psychology on the playground: Reasoning about indirect social consequences in preschoolers and adults. Cognition, 126, 352–363.

21.

Pietraszewski

Schwartz

(2014a). Evidence that accent is a dimension of social categorization, not a byproduct of perceptual salience, familiarity, or ease-of-processing. Evolution and Human Behavior, 35, 43–50.

22.

Pietraszewski

Schwartz

(2014b). Evidence that accent is a dedicated dimension of social categorization, not a byproduct of coalitional categorization. Evolution and Human Behavior, 35, 51–57.

23.

Ratner

K. G.

Kaul

Van Bavel

J. J.

(2013). Is race erased? Decoding race from patterns of neural activity when skin color is not diagnostic of group boundaries. Social Cognitive and Affective Neuroscience, 8, 750–755.

24.

Rosenthal

Rosnow

R. L.

Rubin

D. B.

(2000). Contrasts and effect sizes in behavioral research: A correlational approach. Cambridge, England: Cambridge University Press.

25.

Sack

(2005). The effect of social essentialism on categorization and induction. Dissertation Abstracts International, 66, 3469.

26.

Sidanius

Pratto

(1999). Social dominance: An intergroup theory of social hierarchy and oppression. New York, NY: Cambridge University Press.

27.

Stangor

Lynch

Duan

Glass

(1992). Categorization of individuals on the basis of multiple social features. Journal of Personality and Social Psychology, 62, 207–218.

28.

Susskind

J. E.

(2007). Preadolescents’ categorization of gender and ethnicity at the subgroup level in memory. British Journal of Developmental Psychology, 25, 213–225.

29.

Tajfel

(1981). Social stereotypes and social groups. In Turner

J. C.

Giles

(Eds.), Intergroup behavior (pp. 144–167). Oxford, England: Blackwell.

30.

Taylor

S. E.

Fiske

S. T.

Etcoff

N. L.

Ruderman

A. J.

(1978). Categorical and contextual bases of person memory and stereotyping. Journal of Personality and Social Psychology, 36, 778–793.

31.

Telles

E. E.

(2004). Race in another America: The significance of skin color in Brazil. Princeton, NJ: Princeton University Press.

32.

Tishkoff

S. A.

Kidd

K. K.

(2004). Implications of biogeography of human populations for ‘race’ and medicine. Nature Genetics, 36, 521–527.

33.

von Rueden

Gurven

Kaplan

(2008). The multiple dimensions of male social status in an Amazonian society. Evolution and Human Behavior, 29, 402–415.

34.

van Bavel

J. J.

Cunningham

W. A.

(2009). Self-categorization with a novel mixed-race group moderates automatic social and racial biases. Personality and Social Psychology Bulletin, 35, 321–335.

35.

Weisman

Johnson

M. V.

Shutts

(2014). Young children’s automatic encoding of social categories. Developmental Science. doi:10.1111/desc.12269

36.

Williams

M. J.

Eberhadt

J. L.

(2008). Biological conceptions of race and the motivation to cross racial boundaries. Journal of Personality and Social Psychology, 94, 1033–1047.

37.

Wilson

J. P

See

P. E.

Bernstein

M. J.

Hugenberg

Chartier

(2014). Differences in anticipated interaction drive own group biases in face memory. PLoS ONE, 9, e90668. doi:10.1371/journal.pone.0090668