Two Identity Intersections
In ongoing work, we have been examining the extent to which top-down and bottom-up factors bias social perceptions when identities or emotions intersect with one another. Our empirical work has focused primarily on two intersections— the intersection of race and sex and the intersection of sex and emotion. We have examined these intersections for difference types of visual cues—static cues in the face and dynamic cues in the body.
The Intersection of Race and Sex
One domain in which we have conducted a considerable amount of empirical research is at the intersection of race and sex categories in the domain of face perception. We suspected that, like in the facial expression of emotion, facial cues to sex and race categories may covary. We also noted a conspicuous covariation in the stereotypes associated with the categories Black and Male (e.g., aggressive) and the categories Asian and Female (e.g., communal). This is evident in the overlap of stereotype content across the past several decades (see, e.g., Bem, 1974; Devine & Elliot, 1995; Karlins, Coffman, & Walters, 1969; Spence, Helmreich, & Strapp, 1974). Such overlaps may promote integrative priming whereby the perception of one social category (e.g., sex) is facilitated due to its relation to another perceived category (e.g., race; Estes & Jones, 2009).
We reasoned that overlaps in stereotypes or phenotypes (or both) may lead sex categorization to be systematically biased by the race of a face. Specifically, we predicted that “male” categorizations would be more efficient for Black relative to White or Asian faces but that “female” categorizations would be more efficient for Asian relative to White or Black faces. That is, although we expected categorizations to be overwhelmingly accurate, we examined differences in the efficiency of sex category judgments. In a series of studies, we have demonstrated that sex categorization is systematically biased by the race category of a target (Johnson, Freeman, & Pauker, in press), and we tested for both bottom-up and top-down sources of influence.
Our first test of the possibility that race categories may bias sex categorization was straightforward. We reasoned that when a person’s sex category membership was unclear, perceivers might use race category to disambiguate the person’s sex. To test this, we created a set of androgynous faces using commercially available face generating software. Each face was manipulated along a continuum of race that varied from Black to White to Asian. Then, we simply asked participants to categorize each image, as quickly as possible, to be either male or female. The results were telling. Although presumably irrelevant to the sex categorization task at hand, perceivers’ judgments were systematically influenced by the race of each target. We observed a consistent bias to categorize faces to be male, consistent with prior research (e.g., Zarate & Smith, 1990). This tendency was moderated, however, by the race of the face. Consistent with our predictions, we found the highest proportion of “male” categorizations for Black faces but the lowest proportion of “male” categorizations for Asian faces (see Figure 12.1a).
These findings were corroborated in a study in which perceivers made sex category judgments of both male and female faces that varied by race. Not surprisingly, the vast majority of categorizations were accurate, but the response latency for judgments varied as a function of sex and race. Categorizations of men were made more quickly for Black relative to White or Asian faces, but categorizations of women were made more quickly for Asian relative to White or Black faces (see Figure 12.1b). Moreover, analyses of computer mouse trajectories revealed a pronounced deviation toward the incorrect category alternative when participants categorized Black women and Asian men. Importantly, these are the two groups that we suspected might either have facial cues or elicit stereotypes that were conflated with the opposite sex.
While these findings provided evidence that race category biases one’s perception of sex category, the reason for such biases remained unclear. Our results may have been obtained because the stereotypes associated with the category Black are common to the stereotypes for the category Male and because the stereotypes associated with the category Asian are common to the stereotypes for the category
Figure 12.1 Effects of race category for sex categorizations of androgynous faces and men/women. Panel a depicts the proportion of “female” categorizations of androgynous Black, White, and Asian faces. Panel b depicts the response latencies for “male” and “female” categorizations of Black, White, and Asian faces. Adapted from Johnson, K.L., Freeman, J.B., & Pauker, K., Journal of Personality and Social Psychology, in press.
Female. If correct, race categories may have biased sex categorization through a top- down route. Alternately, our results may have been obtained because Black faces are phenotypically similar to Male faces and Asian faces are phenotypically similar to Female faces. If correct, race categories may have biased sex categorization through a bottom-up route. Additional research shed light on each of these possibilities.
First we tested the possibility that shared stereotype content between specific race and sex categories enhanced perceivers’ ability to categorize some faces according to sex (e.g., Black men and Asian women) but impaired their ability to categorize others (e.g., Black women and Asian men). We measured our participants’ degree of association between the categories Black and Male and the categories Asian and Female using a customized Implicit Association Test (IAT). We also recorded sex categorizations, their latencies, and their mouse trajectories for faces that varied by race and sex. Interestingly, when participants held low associations between the categories Black and Male and the categories Asian and Female, their sex category judgments were relatively unaffected by the intersection of sex and race. When participants held high associations between the target categories, in contrast, their sex category judgments showed a pronounced effect for the intersection of sex and race. As before, judgments of Black men and Asian women were more efficient—in terms of both response latency and computer mouse trajectories—relative to judgments of same-sex targets of other races. These findings suggest that shared stereotypes exert a top-down influence on social categorization when identities intersect.
These findings could not speak to the alternative possibility that the cues associated with the categories Black and Male and the categories Asian and Female overlap in their phenotypes. To test this possibility, we obtained objective measures of the gender typicality of both facial photographs and computer-generated faces that varied by race and sex. To the extent that phenotypes overlap, we predicted that the gendered cues on a face would vary by race. Across both categories of stimuli, Black faces were, on average, phenotypically masculine; White faces tended to be phenotypically feminine; and computer-generated Asian faces were phenotypically masculine, but Asian photographs were phenotypically feminine, albeit not significantly so. These findings highlighted a consistent masculine phenotype in the cues in Black faces.
Based on our analyses of facial phenotypes, we reasoned that this phenotypic overlap may also contribute to biases in social categorization, especially for judgments of Black faces. To test this possibility, we statistically controlled for covarying phenotypes and examined the efficiency of categorizations for faces that varied by race and sex. Interestingly, when we controlled for phenotype, the effect of race dropped to nonsignificance for judgments of men but not women. Moreover, controlling for phenotype shed additional light on the unique role of overlapping stereotypes in categorizations as well. Judgments made by participants who held low associations between the categories Black and Male and the categories Asian and Female were unaffected by intersecting race and sex once phenotype was controlled; however, judgments made by participants who held high associations remained tethered to race for female but not male targets, once phenotype was controlled.
Collectively, these findings clarify the role of overlapping stereotypes and phenotypes for sex categorizations when identities intersect. Judgments of men were affected by overlaps in both phenotypes and stereotypes associated with the categories Black and Male; judgments of women were affected primarily by overlaps in the stereotypes associated with the categories Asian and Female.
