From Out-Group Homogeneity to Out-Group Heterogeneity

People often find it more difficult to distinguish ethnic out-group members compared with ethnic in-group members. A functional approach to social cognition suggests that this bias may be eliminated when out-group members display threatening facial expressions. In the present study, 192 White participants viewed Black and White faces displaying either neutral or angry expressions and later attempted to identify previously seen faces. Recognition accuracy for neutral faces showed the outgroup homogeneity bias, but this bias was entirely eliminated for angry Black faces. Indeed, when participants’ cognitive processing capacity was constrained, recognition accuracy was greater for angry Black faces than for angry White faces, demonstrating an out-group heterogeneity bias. People readily confuse individuals from other races and ethnic groups with one another—the ‘‘they all look the same to me’’ phenomenon. This pattern may reflect a more general cognitive bias toward perceiving the membership of other groups as less variable than the membership of one’s own group—the outgroup homogeneity bias (Anthony, Copper, & Mullen, 1992; Ostrom & Sedikides, 1992). This bias is generally interpreted as resulting from constraints on perceptual processing capacity and perceivers’ tendency to allocate limited perceptual resources in a functional way (e.g., Rodin, 1987; Sporer, 2001). Here, we elaborate on this functional approach to the out-group homogeneity bias, reporting a study in which we tested novel predictions about circumstances in which this bias may be eliminated or even reversed. FUNCTIONAL APPROACH TO PERCEPTION OF IN-GROUP AND OUT-GROUP MEMBERS Cognitive resources are famously limited: People cannot attend to, encode, and remember all information available in their social environments (e.g., Todd, Hertwig, & Hoffrage, 2005). Thus, they selectively allocate cognitive resources to processing stimuli likely to have functional implications (Kenrick, Sadalla, & Keefe, 1998). For instance, snakes capture and hold attention more readily than do benign objects (Öhman, Flykt, & Esteves, 2001). In social situations, one attends more closely to some people than to others, favoring individuals whose physical appearance suggests greater benefits to one’s own reproductive fitness or threats to one’s well-being (Maner et al., 2003; Oda, 1997). Whether explained in psychological terms (e.g., needs and goals), economic terms (e.g., costs and benefits), or evolutionary terms (e.g., reproductive fitness), perceptual and cognitive resources are selectively allocated toward individuals who appear to have the most profound functional implications for perceivers. Historically, the interpersonal interactions that mattered most to individuals’ outcomes (e.g., mate selection, reciprocal exchange, and negotiation of status hierarchies) occurred within coalitional groups. Today, these interpersonal interactions continue to occur primarily within cultural and ethnic groups (Fiske, 1992). Contacts with out-group members tend not only to be less frequent, but also to involve group-level rather than individual-level interactions. Thus, we suggest that whereas the costs of allocating cognitive resources to processing in-group members are often offset by the benefits, the functional benefits Address correspondence to Joshua Ackerman, Department of Psychology, Arizona State University, Tempe, AZ 85287-1104, e-mail: [email protected], or to Steven Neuberg, e-mail: steven. [email protected], or to Douglas Kenrick, e-mail: douglas.kenrick@ asu.edu. PSYCHOLOGICAL SCIENCE 836 Volume 17—Number 10 Copyright r 2006 Association for Psychological Science of allocating cognitive resources to individual out-group members are typically outweighed by the costs. BUT WHAT IF THEY ARE ANGRY? Social cognitive processes are typically sensitive to functional aspects of dyadic and group ecology (Haselton, Nettle, & Andrews, 2005; Neuberg, Kenrick, Maner, & Schaller, 2004; Schaller, Park, & Faulkner, 2003). Although individuals’ outcomes may depend more on interactions with in-group members than on interactions with out-group members, there are circumstances in which out-group members may have greater functional importance. One circumstance in which the functional significance of outgroup members increases is when those out-group members are angry. Research has shown that angry facial expressions capture and hold attention (e.g., Fox et al., 2000; Öhman et al., 2001). Two further considerations suggest that perception of anger may lead perceivers to allocate resources toward processing the individuating features of angry target persons. First, anger is an interpersonal emotion; it implies threatening intent held by a specific individual (the person expressing anger) toward another specific individual (the person perceiving the expression). Second, angry expressions, like all emotional expressions, are fleeting; the angry expression may disappear even when the threatening intention persists. Therefore, just as it is functional to attend to people with angry facial expressions (so one can detect and avoid those who intend one harm), it is also functional to encode individuating features of those individuals (so one can detect and avoid them later even when the overt anger cue is gone). This reasoning suggests that angry individuals from both in-groups and out-groups are likely to command an enhanced proportion of available cognitive resources, and such allocation of resources might lead to the attenuation of the out-group homogeneity bias. An additional functional consideration suggests that perceivers may devote even more processing resources to angry outgroup members than to angry in-group members: Angry outgroup members may be judged to pose an even greater potential threat than angry in-group members. Various factors mitigate the actual threat posed by angry in-group members. For example, people more readily empathize with in-group members than with out-group members (Stürmer, Snyder, & Omoto, 2005), and empathy inhibits aggression (Loudin, Loukas, & Robinson, 2003). Moreover, high interdependency makes it more costly to harm in-group members than to harm out-group members. Interactions between groups, in contrast, are more frequently competitive (Wildschut, Pinter, Vevea, Insko, & Schopler, 2003). Thus, anger expressed by an out-group member may be perceived as more threatening than anger expressed by an ingroup member. If so, angry out-group members may receive enhanced scrutiny, and subsequently be better differentiated from one another in memory than are angry in-group members. That is, in the case of perceived anger, there may be an out-group heterogeneity bias. EXPLORATORY EVIDENCE AND THE PRESENT RESEARCH Though designed to test conceptually unrelated hypotheses, two studies from our research program allowed for preliminary exploration of the effects of angry facial expressions on the outgroup homogeneity bias. In one study, 168 White participants sequentially viewed single male faces varying in ethnicity (White, Black) and emotional expression (neutral, angry). Participants were later asked to distinguish these faces from previously unseen foils. A second study presented 206 participants with similar stimuli under more cognitively demanding circumstances: Two faces were presented simultaneously and for a shorter duration. In both of these studies, a significant interaction of ethnicity and facial expression emerged, F(1, 205) 5 48.38, p < .001, prep 5 1.0, Zp 2 1⁄4 :19, and F(1, 167) 5 4.93, p < .05, prep 5 .91, Zp 2 1⁄4 :03, respectively. Results for neutral faces were consistent with the usual out-group homogeneity bias—lower recognition accuracy for neutral Black faces than for neutral White faces. For angry faces, however, the results diverged. In the first study, the usual homogeneity bias was eliminated: Angry Black faces were recognized just as accurately as angry White faces. In the second study, angry Black faces were recognized more accurately than angry White faces; that is, participants exhibited an out-group heterogeneity bias. Results of these studies provided initial evidence that the outgroup homogeneity bias might be eliminated, and perhaps even reversed, for angry facial expressions. However, neither study was specifically designed to test these possibilities. We therefore designed an experiment to test the hypotheses directly and to investigate one factor that may have led to the reversal of the out-group homogeneity bias in the second study—cognitive constraint. Previous research suggests that functional biases are often revealed more strongly when cognitive resources are limited (e.g., Maner et al., 2003; Meissner & Brigham, 2001). Applying this logic, a finding that processing constraints are associated with a reversal (and not merely the mitigation) of the out-group homogeneity bias would be consistent with the notion that angry out-group faces are functionally important enough to demand even more immediate processing than in-group faces.