Article

To Be or Not to Be (Black or Multiracial or White): Cultural Variation in Racial Boundaries

Jacqueline M. Chen1,2, Maria Clara P. de Paula Couto3, Airi M. Sacco4, and Yarrow Dunham5

Abstract

Culture shapes the meaning of race and, consequently, who is placed into which racial categories. Three experiments conducted in the United States and Brazil illustrated the cultural nature of racial categorization. In Experiment 1, a target’s racial ancestry influenced Americans’ categorizations but had no impact on Brazilians’ categorizations. Experiment 2 showed cultural differences in the reliance on two phenotypic cues to race; Brazilians’ categorizations were more strongly determined by skin tone than were Americans’ categorizations, and Americans’ categorizations were more strongly determined by other facial features compared to Brazilians’ categorizations. Experiment 3 demonstrated cultural differences in the motivated use of racial categories. When the racial hierarchy was threatened, only Americans more strictly enforced the Black–White racial boundary. Cultural forces shape the conceptual, perceptual, and ideological construal of racial categories.

Keywords

race, categorization, culture, intergroup relations, face perception

Immigrants to the United States have to complete several forms

to be naturalized, and many report being unsure of how they fit

into the racial/ethnic categories presented on these forms

(Joseph, 2015). A self-identification that many Americans take

for granted causes confusion among others, illustrating the

social nature of racial categories. Our research sheds light on

how racial boundaries are shaped by cultural forces.

Americans frequently essentialize race, treating observed

racial differences as stemming from unobservable but deep

internal properties that are vertically transmitted from parents

to their offspring (Hirschfeld, 1998). Despite these powerful

intuitions, however, determining a person’s race is not always

straightforward. Perceptions of one’s race can be influenced by

irrelevant characteristics (e.g., Freeman, Penner, Saperstein,

Scheutz, & Ambady, 2011; Hugenberg & Bodenhausen,

2004), and perceivers’ attitudes and motivations can influence

how they categorize individuals (e.g., Chen, Moons, Gaither,

Hamilton, & Sherman, 2014). Thus, despite objectivist intui-

tions about race, the ways that people actually racially categor-

ize others depend on the social and motivational context. We

further show that these processes are embedded within a cul-

tural context by conducting three experiments comparing per-

ceivers’ racial categorization processes in the United States

and Brazil. Specifically, we examine cultural differences in the

use of ancestry (Experiment 1) and phenotypic cues (Experi-

ment 2) in racial categorization and then investigate the cultural

specificity of the motivated enforcement of racial boundaries

(Experiment 3).

Both the United States and Brazil have a history of Native

American displacement, European settlement, and African

slavery. However, the two countries adopted different strate-

gies and practices to address racial diversity. While we cannot

do justice to this complex history here, we discuss below our

view that these historical differences have shaped cultural

divergences in race perception today. Our experimental

approach dovetails with sociological research comparing North

and South American racial stratification and ideology at macro-

levels (e.g., Bailey, Saperstein, & Penner, 2014; Telles, 2004,

2014) while also contributing to the growing psychological

literature on social categorization processes.

1 Department of Psychology, University of Utah, Salt Lake City, UT, USA 2 Department of Psychology and Social Behavior, University of California,

Irvine, CA, USA 3 EduLab21, Ayrton Senna Institute, São Paulo, Brazil 4 Department of Psychology, Federal University of Pelotas, Pelotas, Brazil 5 Department of Psychology, Yale University, New Haven, CT, USA

Corresponding Author:

Jacqueline M. Chen, Department of Psychology, University of Utah, Salt Lake

City, UT 84112, USA.

Email: [email protected]

Social Psychological and Personality Science 2018, Vol. 9(7) 763-772 ª The Author(s) 2017 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/1948550617725149 journals.sagepub.com/home/spp

Conventions of American Racial Ideology

Racial categorization in the United States focuses on delineat-

ing boundaries using individuals’ ancestry. For example, the

so-called “one-drop” laws institutionalized hypodescent, the

allocation of mixed ancestry individuals to the lower-status

group by specifying that a person with Black blood was

defined as Black, irrespective of their other ancestries or their

appearance (Davis, 1991). Not only were individuals’ ances-

tries central to determining their race, but society also empha-

sized maintaining racial boundaries by attempting to keep

races “separate but equal.” Thus, historical conventions

enable American perceivers’ assumption that racial groups

are biologically based, discrete, and stable (Banks & Eber-

hardt, 1998; Chen & Hamilton, 2012; Dunham & Olson,

2016; Richeson & Sommers, 2016), an assumption that self-

perpetuates (see Prentice & Miller, 2007; Williams & Eber-

hardt, 2008).

Past studies have provided insight into how American views

of race play out in social perception. White Americans continue

to engage in hypodescent when categorizing racially ambiguous

mixed race faces (Peery & Bodenhausen, 2008) or when consid-

ering how individuals of mixed ancestry should be categorized

(e.g., Ho, Sidanius, Levin, & Banaji, 2011; Sanchez, Good, &

Chavez, 2011; see also Ho, Kteily, & Chen, in press). Yet, it

is only beginning in middle childhood that Americans reliably

associate ancestry with race and exhibit hypodescent in their

categorizations of multiracial targets (Roberts & Gelman,

2015), suggesting that ancestry-based racial categorization pat-

terns are culturally learned. Further, perhaps reflecting the his-

torical effort to subordinate Black–White individuals using

one-drop rules, White Americans who are seeking to preserve

existing racial stratification are especially likely to engage in

hypodescent (Ho, Sidanius, Cuddy, & Banaji, 2013; Krosch &

Amodio, 2014; see also Penner & Saperstein, 2013).

Therefore, previous research clearly suggests that Ameri-

cans are socialized to view race through an essentialist lens that

can be traced back to the country’s historical treatment of race.

We seek to provide illuminating evidence of the cultural nature

of these processes by direct cross-cultural comparison with

race perception in Brazil.

Conventions of Brazilian Racial Ideology

Brazilian racial ideology emphasizes racial miscegenation and

the flexibility of racial categorization. After slavery was

abolished, the government explicitly encouraged interracial

marriage in an attempt to “dilute” Blackness in order to socially

and politically weaken the large African Brazilian population

(Telles, 2004, 2014). Encouraging individual upward mobility

via “self-whitening,” albeit for anti-Black reasons, tacitly

endorses a conceptualization of an individual’s race as flexible

and subject to change, while preserving the subordinate status

of the Black racial group overall. The individual fluidity norm

is in stark contrast with the United States, where miscegena-

tion, far from “diluting” Blackness, would serve to increase the

Black population through the operation of hypodescent. Today,

multiracial people are viewed as an intermediate racial group

between Blacks and Whites (Skidmore, 1993), and the Brazi-

lian census permits people to identify as Multiracial (“Parda”),

Black (“Preta”), or White (“Branca”) (Instituto Brasileiro de

Geografia e Estatı́stica, 2011).

Reflecting a fluid conceptualization of an individual’s race,

there were no formalized rules for racial classifications, nor any

institutionally sanctioned linking of race with ancestry (Telles,

2004). Instead, the push for self-whitening as a mechanism for

upward mobility linked race with socioeconomic status

(Schwartzman, 2007) and appearance (especially skin tone;

Telles, 2004), two attributes that are more malleable than one’s

ancestry. Reflecting the ultimate success of the cultural disso-

ciation between racial appearance and ancestry, Brazilians’

racial appearance and self-identification are only weakly pre-

dictive of their actual amount of African ancestry (Parra

et al., 2003).

To our knowledge, there is no social psychological research

examining Brazilians’ perceptions of race. Sociologists have

argued that Brazilians make racial categorizations on the basis

of appearance, privileging skin tone as the defining feature of

race (Telles, 2014), with little relation to their genetic ancestry

(Santos et al., 2009). Thus, our research seeks to experimen-

tally validate long-standing claims from sociology and provide

the first experimental cross-cultural comparison of race percep-

tion between the United States and Brazil.

Overview of Current Research

Three experiments show that the differences in cultural con-

ventions have powerful psychological consequences, affect-

ing how race is perceived and how racial boundaries are

defended. Experiment 1 examined cultural differences in the

conceptualization of race by manipulating ancestry and pit-

ting it against targets’ appearance. Experiment 2 examined

cultural differences in the perceptual bases of race, specifi-

cally in the use of skin tone versus facial features in racial

categorizations. Experiment 3 investigated the cultural-

embeddedness of motivated race perception by examining

whether the motivated use of racial boundaries functioned dif-

ferently across cultures.

Experiment 1

Experiment 1 investigated cultural differences in individuals’

use of a person’s ancestry versus appearance in racial categor-

ization. We hypothesized that individuals’ categorizations

would reflect cultural differences in the conceptualization of

race. Specifically, we predicted that Americans would categor-

ize targets consistent with their heritage whereas Brazilians

would categorize targets consistent with their appearance. In

addition, we expected to observe hypodescent in the categori-

zation of mixed ancestry targets among Americans but not

among Brazilians.

764 Social Psychological and Personality Science 9(7)

Method

Participants

Americans (n¼ 145; 100 females) participated in exchange for

partial credit for university psychology courses (Mage ¼ 19.93,

SD ¼ 2.16). Brazilians (n¼ 122; 101 females) participated after

being recruited from university psychology courses (Mage =

24.59; SD¼ 3.40). The SOM contains sample size goals, sample

racial demographics, and analyses by participant race for all

three studies.

Materials

The stimulus set consisted of eight faces of Multiracial children

(four female faces) from a larger Brazilian stimulus set (BIC-

Multicolor; Sacco, de Paula Couto, & Koller, 2016). We pre-

tested the faces in both countries (see SOM for details). To

be selected, the stimulus faces had to be considered Multiracial

(as opposed to Black or White) by at least 75% of participants

in both countries.

Survey materials for all three studies were created in English,

translated into Portuguese by a bilingual social psychologist, and

then checked by another bilingual social psychologist. All sur-

veys were programmed in Qualtrics and completed online.

Demographic questions were always at the end of the study, and

participants’ response options for race were determined by the

categories typically available on their country’s census.

Procedure

Participants consented to participate in a study assessing their

social attitudes and beliefs. Participants were randomly assigned

to view one of the targets, whose face was presented with the fol-

lowing background information: “This child was born in the

United States (Brazil). His (her) parents are African American

[vs. One of his (her) parents is African American, and the other

is White vs. His (her) parents are White].” Participants were

asked to categorize the target by race in an open-ended question

(“What race is this child?”).

Design

The study had a 2 (Culture: United States vs. Brazil) � 3

(Ancestry: two Black parents vs. one Black parent and one

White parent vs. two White parents) between-subjects design.

The frequency and type of racial categorizations were the

dependent variables.

Results

Although the majority of participants’ responses (approxi-

mately 76%) fell into the racial categories of Black, Multira-

cial, and White, a substantial proportion of their responses

did not. Americans (but not Brazilians) occasionally generated

alternative racial categorizations (e.g., Indian, Latino). Thus,

we analyzed participants’ categorizations of the target as

Black, Multiracial, White, or other. We tested our predictions

using analysis of variance (ANOVA; below) and multinomial

regression (in SOM), with both analyses reaching the same

conclusions.

We ran a 2 (Culture: United States vs. Brazil) � 3 (Ances-

try: Black vs. Black/White vs. White) � 4 (Racial Categoriza-

tion: Black, Multiracial, White, or Other) mixed model

ANOVA on participants’ categorizations, with the latter factor

being within-subjects. The predicted three-way interaction

between Culture, Ancestry, and Categorization emerged,

F(6, 783) ¼ 9.51, p < .001, Zp 2 ¼ .07 (see Figure 1). We thus

broke down the results separately by culture.

Among Brazilians, there was only a main effect of Categor-

ization, F(3, 357) ¼ 95.90, p < .001, Zp 2 ¼ .45. Regardless of

parents’ race, Brazilians categorized the children predomi-

nantly as Multiracial, M ¼ .72 95% CI [.65, .81], SE ¼ .04.

Multiracial categorizations were significantly more frequent

than Black M ¼ .07 95% CI [.03, .12], SE ¼ .02; White

M ¼ .18 95% CI [.11, .25], SE ¼ .04; or Other M ¼ .00 95% CI [.00, .00], SE ¼ .00 categorizations, all ps < .05. As

expected, Brazilians’ racial categorizations were uninfluenced

by information about the children’s ancestry.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Black Black-White White

Pr op

or ti

on o

f R ac

ia l C

at eg

or iz

at io

ns

Parents' Race

Brazil

White

Mixed

Black

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Black Black-White White

Pr op

or ti

on o

f R ac

ia l C

at eg

or iz

at io

ns

Parents' Race

USA

White

Mixed

Black

Other

Figure 1. Proportion of racial categorizations as a function of parents’ race and country in Experiment 1 (Whiskers denote +1 SE).

Chen et al. 765

Among Americans, there was a main effect of Categoriza-

tion, F(3, 426) ¼ 13.11, p < .001, Zp 2 ¼ .09, that was qualified

by the predicted interaction with Ancestry, F(6, 426) ¼ 20.57,

p < .001, Zp 2 ¼ .23. Americans were more likely to categorize

children as Black when their parents were Black, M ¼ .51 95% CI [.39, .63], SE ¼ .06, as opposed to Black/White, M ¼ .26

95% CI [.14, .37], SE ¼ .06, or White M ¼ .04 95% CI

[–.07, .16], SE ¼ .06, ps < .01. Participants were more likely

to categorize targets as Multiracial when they had only one

Black parent, M ¼ .59 95% CI [.49, .68], SE ¼ .05, as opposed

to two Black parents, [M¼ .04 95% CI [–.06, .14], SE¼ .05, or

two White parents, M ¼ .06 95% CI [–.04, .16], SE ¼ .04, ps <

.001. And participants were more likely to categorize targets as

White when both their parents were White, M ¼ .23 95% CI

[.16, .30], SE ¼ .04, compared to when one parent was White,

M ¼ .00 95% CI [–.07, .07], SE ¼ .03, or both were Black, M

¼ .00 95% CI [–.07, .07], SE¼ .04, ps < .001. Therefore, ances-

try strongly shaped Americans’ racial categorizations of the chil-

dren. The unexpected “Other” categorizations of the targets were

highest for two White parents, M ¼ .62 95% CI [.49, .75], SE¼ .07, ps < .01, and higher two Black parents, M ¼ .40 95% CI [.28, .53], SE ¼ .07, compared to when one parent was

White and the other Black, M ¼ .14 95% CI [.01, .26], SE ¼ .06, p ¼ .004.

Finally, we found evidence of hypodescent such that Amer-

icans were more likely to categorize a child with mixed paren-

tage as Black, M ¼ .26 95% CI [.14, .37], SE ¼ .06, than as

White, M ¼ .00 [–.07, .07], SE ¼ .03, p < .001. Brazilians did

not engage in hypodescent, categorizing a child with mixed

heritage as White, M ¼ .16 95% CI [.04, .27], SE ¼ .06, as

often as Black, M ¼ .07 [–.01, .15], SE ¼ .04, p ¼ .23.

Discussion

This study provides a clear illustration of cultural differences in

how individuals determine another person’s race. Whereas

Brazilians’ categorizations ignored targets’ ancestry and

focused only on appearance, Americans’ categorizations were

heavily influenced by targets’ ancestry and, to a lesser extent,

their appearance. These findings reflect the historical differ-

ences in how the United States and Brazil defined race, the for-

mer in terms of ancestry and hypodescent for multiracial

individuals, the latter in terms of appearance.

The use of the “other” categories, occurring only among

Americans and predominantly in the White parents condition,

shows that ancestry is not the only criterion for race in the

United States. Indeed, Americans’ categorizations were

partially driven by appearance, and this tendency was asym-

metric, such that perceivers most often rejected the ancestral

cue and generated alternative (non-White) categories when tar-

gets’ appearance seemingly did not match the White parentage

information. These findings suggest an interesting possibility—

that different racial categories have different criteria for mem-

bership in the United States. More generally, however, the

documented strong relationship between ancestry and race for

Americans, but not for Brazilians, reveals deep differences in

the factors driving racial categorization in each culture.

Experiment 2

Whereas Experiment 1 documented cultural differences in the

conceptual basis of racial categorization, Experiment 2 sought

a finer grained investigation of the use of two perceptual, that

is, phenotypic, cues (skin tone and other facial features) in

Americans’ and Brazilians’ categorizations. In doing so, we pro-

vided the first experimental investigation of United States-Brazil

differences in the perceptual bases of racial categorization.

Past work suggests that Americans’ racial categorization

will rely on both cues. Americans believe that one’s ancestry

and appearance are closely linked, such that they expect a child

of Black parents to look Black as well (Hirschfeld, 1998). Con-

sistent with this view, in the absence of ancestral information,

American adults’ racial categorizations and social evaluations

of individuals rely on both skin tone and facial features (e.g.,

Stepanova & Strube, 2012). Yet, skin tone is a stronger and

developmentally earlier-emerging predictor of American cate-

gorizations (Dunham, Stepanova, Dotsch, & Todorov, 2015;

Stepanova & Strube, 2012). Thus, we hypothesized that Amer-

icans’ categorizations would use both skin tone and facial fea-

tures, but that they would rely more on skin tone.

With respect to Brazil, macro-level and qualitative analyses

support the dominance of skin color over other phenotypic cues,

including facial features, in lay conceptions of race (Santos et al.,

2009; Telles, 2004, 2014; Travassos & Williams, 2004; but

see Bailey et al., 2014; Banton, 2012). Based on these findings

outside of experimental social psychology, we predicted that

Brazilians would use skin tone more than facial features.

Our research also directly compared the importance of skin

tone and facial features in racial categorizations in the United

States and Brazil. Because both qualitative (e.g., Telles,

2014) and quantitative (e.g., Experiment 1) research across

disciplines argue that lay definitions of race in the United States

focus on ancestry as a primary cue to race and on appearance as

a secondary cue, and because other social sciences indicate that

Brazilians define race primarily in terms of skin tone (e.g.,

Telles, 2014; Travassos & Williams, 2004), we predicted that

Brazilians would use skin tone more strongly than Americans.

Our investigation of between-culture differences in the use of

other facial features was more exploratory, but the sociological

work described above pointing to the centrality of skin color in

Brazil allowed us to cautiously predict that Brazilians would

make less use of these features than Americans.

Method

Participants

One hundred and nine Americans (62 females) were recruited

from Amazon’s Mechanical Turk (Mage ¼ 35.67; SD ¼ 1.25). One hundred twenty-eight Brazilians (53 females,

28 males, and 47 declined to state) were a recruited in a conve-

nience sample (Mage ¼ 30.14; SD ¼ 0.94).

766 Social Psychological and Personality Science 9(7)

Materials and Procedure

Participants learned that they would be viewing faces and cate-

gorizing them by race (“What race is this person?” with Black,

Multiracial, and White response options). Two extensively

validated stimulus sets were used (Dunham et al., 2015; Stepa-

nova & Strube, 2012). The faces varied along two dimensions:

skin tone (very dark to very light) and facial features (very

Afrocentric to very Eurocentric). Both dimensions had 10 lev-

els. Participants were randomly assigned to a stimulus set and

categorized each face by race in random order.

Design

The study had a 2 (Culture: Brazil vs. United States) � 2

(Stimulus set) � 10 (Skin tone: very dark to very light) � 10

(Facial features: very Afrocentric to very Eurocentric) mixed

design, with the latter two factors being within-subjects. The

dependent variable was racial categorization. Including stimu-

lus set as a factor did not change the results, and we collapsed

across this factor.

Results and Discussion

We conducted a fixed-effects multilevel model predicting

categorization (1 ¼ Black, 2 ¼ Multiracial, and 3 ¼ White)

with mean-centered skin tone an