- Email: [email protected]
Categorization and individuation in the cross-race recognition deficit: Toward a solution to an insidious problem
Journal of Experimental Social Psychology 43 (2007) 334–340 www.elsevier.com/locate/jesp
Categorization and individuation in the cross-race recognition deWcit: Toward a solution to an insidious problem Kurt Hugenberg a,¤, Jennifer Miller b, Heather M. Claypool a a
Department of Psychology, Miami University, USA b University of Cincinnati, USA
Received 30 August 2005; revised 23 February 2006 Available online 24 April 2006
Abstract Recent theory and evidence suggest that the Cross-Race EVect (better recognition for same-race (SR) faces than for cross-race (CR) faces) is due to social-cognitive processes of categorization of out-group members, causing perceivers to attend to category-specifying information of CR faces at the expense of individuating information. Three experiments seek to extend this social-cognitive explanation of the CRE by investigating the extent to which the Cross-Race EVect can be reduced by inducing perceivers to individuate rather than categorize CR faces. In all three experiments, participants who received warning of the Cross-Race EVect prior to encoding, and instructions to individuate out-group members, showed no CRE. Experiment 2 suggests that this elimination of the CRE was not due merely to increased motivation to process all stimuli. This is one of few empirical displays of an elimination of the CRE outside of visual training. Moreover, these results are congenial with Levin’s (2000) feature-selection model, which suggests that the CRE is due to diVerential social cognitions about in-group and out-group members, rather than to diVerences in perceptual expertise. By eliciting individuation of out-group members at encoding, the CRE can be eliminated. © 2006 Elsevier Inc. All rights reserved. Keywords: Cross-Race EVect; Face perception; Categorization; Individuation
Introduction The cross-race recognition deWcit, known more simply as the Cross-Race EVect (CRE), is one of the best-replicated phenomena in face perception (Chance & Goldstein, 1996), and has been of interest to social psychologists for more than half a century (Allport, 1954). Explained simply, the CRE is a tendency to have better recognition accuracy for same-race (SR) faces than for cross-race (CR) faces, an eVect that has been shown to be surprisingly robust across numerous research paradigms (Meissner & Brigham, 2001). From recognition memory tasks in the laboratory to suspect line-ups in police stations, the tendency to have better recognition of SR than CR faces has shown troubling consistency. Although this phenomenon is interesting in its own right, the high degree of interest in the CRE is due, at least in part, to the potentially deleteri*
Corresponding author. E-mail address: [email protected] (K. Hugenberg).
0022-1031/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jesp.2006.02.010
ous consequences of misidentifying members of racial outgroups (Brigham & Malpass, 1985). For the perceiver, consequences of misidentiWcation may include feelings of embarrassment or social opprobrium. For the victim of misidentiWcation, the consequences may be more problematic, leading to feelings of insult or the threat that accompanies the knowledge of being stereotyped. Perhaps more serious still, however, are the potential legal consequences of misidentiWcation. Given the reliance of the criminal justice system on eyewitness identiWcation, and the substantial weight that eyewitnesses have on juridic decisions, understanding the mechanisms underlying the CRE and how to ameliorate these eVects are clearly important goals for social psychological investigation. Mechanisms underlying the Cross-Race EVect: Percept versus concept Despite the robustness of the CRE across experimental contexts, coming to agreement on a satisfying theoretical
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
account for this troubling eVect has proven diYcult. Perhaps the longest standing explanation for the CRE is the perceptual expertise hypothesis (see Meissner & Brigham, 2001). Although there are many diVerent variations of this hypothesis (Ng & Lindsay, 1994), the core argument is that perceivers have diVerential expertise in processing SR versus CR faces, and that this diVerential expertise leads to diVerential recognition accuracy. Given that people typically have substantial contact with individuals of the samerace, extensive expertise is gained at distinguishing between SR faces. The comparatively lesser contact with individuals of other races, however, yields fewer opportunities for distinguishing between CR faces. As such, we are relatively inexpert at distinguishing between CR faces. The speciWc mechanism by which this diVerential expertise elicits diVerential recognition of CR faces is a matter of some debate. For example, in line with popular models of recognition memory (e.g., McClelland & Chappell, 1998), a lack of contact may lead to a lack of expertise with the dimensions on which CR faces actually vary (see McLin & Malpass, 2001). Alternately, lower levels of expertise with CR faces may elicit less holistic and relatively more feature-based processing of CR faces (Rhodes, Brake, Taylor, & Tan, 1989). Mechanisms for the perceptual expertise hypothesis aside, support for this perceptual expertise hypothesis in explaining the CRE has been mixed (Brigham & Malpass, 1985). On the supportive side, practice distinguishing between CR faces can reduce the CRE. For example, Malpass, Lavigueur, and Weldon (1973) found that practice at perceptual discrimination between SR and CR faces in the laboratory can at least temporarily reduce the magnitude of the CRE (see also Elliott, Wills, & Goldstein, 1973; Goldstein & Chance, 1985). More recently, work by Sangrigoli and colleagues (Sangrigoli, Pallier, Argenti, Ventureyra, & de Schonen, 2005) has found that extensive, lifelong training with CR faces can even reverse the direction of the CRE. In their study, individuals of Korean heritage who were adopted as children by Caucasian families in Europe showed a reversal of the CRE by adulthood. That is, despite their Korean heritage, these adopted participants who grew to maturity among Caucasian families showed eVects similar to their adoptive Caucasian parents, Wnding it more diYcult to correctly recognize Asian than White faces. Thus, there is some evidence that the magnitude of the CRE decreases as the amount of expertise with CR faces increases, and with extensive practice can even reverse. However, some studies have failed to Wnd empirical support for the perceptual expertise hypothesis, Wnding no relationship between the amount of expertise with CR faces and the magnitude of the CRE (e.g., Malpass & Kravitz, 1969; Ng & Lindsay, 1994). For example, Ng and Lindsay (1994) easily replicated the typical CRE with White and Asian participants, however, across two studies they found no relationship between the magnitude of the CRE and selfreported contact with members of the relevant racial outgroup. Ng and Lindsay also studied racial minorities among racial majorities, speciWcally White participants
335
living in Singapore and Asian participants living in Canada, who would presumably be forced by dint of their minority status to have extensive experience at distinguishing between CR faces. Even such extensive real-life experience seemed to have no eVect on the magnitude of the CRE. Similarly, attending schools with large numbers of outgroup members has shown to have diVerential eVects on the CRE (see Levin, 2000, for a review), in some cases reducing the eVect (Feinman & Entwhisle, 1976), in other cases, having no eVect (Malpass & Kravitz, 1969), and in still other cases even exacerbating the eVect (Lavarkas, Buri, & Mayzner, 1976). Thus, despite some conWrmatory evidence, the conXicting evidence that real world contact with out-group members does not reliably show a negative relationship with the magnitude of the CRE suggests that something more than mere perceptual expertise may be at play. In response to the sometimes checkered evidence for the expertise hypothesis in explaining the CRE, Levin (1996, 2000) proposed his feature-selection model as an alternate explanatory mechanism for the CRE. This feature-selection model argues that the CRE is due not to diVerential expertise with CR faces per se, but rather to diVerences in social cognitions typically elicited when processing in-group and out-group members (see also Anthony, Copper, & Mullen, 1992). At the core of Levin’s feature-selection hypothesis is the ubiquitous tendency of perceivers to think categorically about out-group members (e.g., Bodenhausen, Macrae, & Hugenberg, 2003). Categorical thinking involves reliance on social categories (e.g., race, sex, age) rather than on individual characteristics of a target. Levin (1996, 2000) argues that this tendency to think categorically about out-group members, but to individuate in-group members, leads to an asymmetrical search for features in SR versus CR faces (see also McLin & Malpass, 2001). The individuation of ingroup members leads perceivers to search for facial features that distinguish one in-group member from another. The tendency to think categorically about out-group members leads to a diVerent search pattern: perceivers search for category-specifying features (e.g., skin tone, brow strength) rather than individuating features. Whereas people encode individuating features in SR faces, they encode race-specifying features at the expense of this individuating information in CR faces. Having only encoded the race-specifying features of CR faces, this leads to real diYculty distinguishing one CR face from another CR face at recognition, resulting in the well-established CRE. Toward a social-cognitive understanding of the Cross-Race EVect The feature-selection model is, at its core, a social-cognitive model of a seemingly perceptual phenomenon, which aligns it with a number of analogous eVects in the social cognition literature. For example, the well-replicated within-category confusions in the “Who said what?” paradigm (Klauer & Wegener, 1998; Taylor, Fiske, EtcoV, & Ruderman, 1978) show a similar pattern, such that
336
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
perceivers tend to confuse the speakers of statements made by members of the same social category (e.g., a statement made by one African American is mistakenly assigned to another African American), suggesting that individuals can encode category specifying information about targets at the expense of individuating information leading to within-category confusions. The out-group homogeneity eVect (e.g., Judd & Park, 1988; Linville, Fischer, & Salovey, 1989), or the tendency to perceive members of out-groups as more homogeneous than members of in-groups, is another robust social cognitive eVect that shows striking similarities to the CRE. Such a social-cognitive explanation of the CRE may not only yield a satisfying theoretical account of this troubling eVect, but it may also oVer a solution to ameliorate the eVect that is radically diVerent than that oVered by a perceptual expertise mechanism. Whereas the only recourse to reduce the CRE under an expertise model seems to be extensive perceptual training, a social-cognitive model oVers a diVerent solution: individuation. If diVerent social cognitions about in-groups and out-groups are responsible for the CRE, then changing these social cognitions may be suYcient to reduce or eliminate the CRE. Although categorical thinking is common when engaging in social cognition about out-groups, prominent theories of stereotyping (e.g., Brewer, 1988; Fiske & Neuberg, 1990) note that thinking categorically about out-group members is by no means inevitable. To the contrary, when social perceivers are aware of their tendency to engage in categorical thinking about out-group members, are motivated to individuate out-group members, and have the cognitive capacity to do so, perceivers may engage in individuation. In contrast to categorization, individuation is the tendency to treat targets not according to category membership, but rather according to unique traits and characteristics. Thus, if the CRE is due to the tendency to categorize CR faces but to individuate SR faces, inducing perceivers to individuate out-group members may eliminate the CRE. Experiments 1a and 1b Experiments 1a and 1b were designed to test the hypothesis that inducing individuation of out-group members during encoding would reduce the CRE. Although Experiments 1a and 1b employed identical procedures, they used diVerent sets of stimulus faces, therefore their results are reported separately. We employed a recognition task commonly used in CRE research, in which participants were asked to attempt to distinguish which Black and White faces had been seen previously in the experiment and which were novel faces. Importantly, before the learning phase, participants were randomly assigned to one of two instruction conditions. In the control condition, designed to replicate past CRE research, participants received no information regarding the CRE. In the individuation instructions condition, however, participants received a description of the CRE and were instructed to attend
closely to the individuating features of CR faces in order to avoid this eVect. We predicted that control participants would show the typical CRE, showing better recognition of SR than CR faces. Of particular interest, however, was whether participants informed of the CRE and instructed to avoid it would be able to do so. If the CRE is due to tendencies to think categorically about out-group members, but to individuate in-group members, providing participants with knowledge of this bias and speciWc instructions to attend to the individuating features of CR faces may reduce or even eliminate the CRE. Method Participants and design Thirty European Americans participated in Experiment 1a; 146 European Americans participated in Experiment 1b. In both experiments, target race (2: Black; White) was manipulated on a within-subjects basis; instructions (2: control; individuation) were manipulated on a betweensubjects basis. Participants were randomly assigned to one of the two levels of instructions. Eight participants in Experiment 1b were excluded from analyses because, during debrieWng, they either reported problems learning the key mappings or admitted to not following instructions, and had error rates of nearly 50%. Thus, all analyses of Experiment 1b were conducted on the remaining 138 participants. Materials and procedure Although the stimulus sets diVered across the two experiments, the stimuli in both consisted of digitized, grayscale pictures of 80 male faces (40 Black; 40 White), 6 £ 4 cm in size, displaying only face and hair. Within each experiment, the 80 faces were randomly assigned to two lists of 40 faces each (20 Black; 20 White). The list of 40 faces seen during the learning phase was counter-balanced on a between-subjects basis.1 After providing informed consent, participants were seated at computers in individual cubicles and were instructed that they would participate in a face recognition experiment, consisting of a learning phase and a recognition phase. All instructions and stimulus presentations were presented by the computer. Participants were Wrst instructed that, during the learning phase, they would see 40 faces on the computer screen, and to attend closely to these faces in order to later recognize them. Participants in the individuation condition also received the following instructions intended to elicit the individuation of outgroup members: Previous research has shown that people reliably show what is known as the Cross-Race EVect (CRE) when learning faces. Basically, people tend to confuse faces that belong to other races. For example, a White learner 1 Preliminary analyses indicated no list eVects in any of the three experiments.
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
will tend to mistake one Black face for another. Now that you know this, we would like you to try especially hard when learning faces in this task that happen to be of a diVerent race. Do your best to try to pay close attention to what diVerentiates one particular face from another face of the same race, especially when that face is not of the same-race as youƒ
After receiving instructions, all participants began the learning phase, during which each of 40 target faces (20 Black; 20 White) were displayed in a randomized sequence on a computer screen; each face was displayed for 3 s. After completing a 5–7 min unrelated distractor task, all participants engaged in the recognition phase. Participants were instructed that they would see a series of faces, some of which they had seen during the learning phase (i.e., old faces) and some of which they had not previously seen (i.e., new faces). Participants were instructed to distinguish which of the faces they had seen during the learning phase (i.e., old faces) and which of the faces were not seen during the learning phase (i.e., new faces) using the ‘a’ key and the ‘5’ key on the keyboard, respectively. The 80 faces observed at the recognition phase included the 40 faces previously seen during learning and 40 new faces (20 Black; 20 White), displayed in a separate random order for each participant. Each face remained on the screen until participants responded, which began the next trial. After completing all tasks, participants were thanked and debriefed. Results and discussion Of primary interest was the extent to which the instruction conditions inXuenced the magnitude of the CRE. Within the signal detection framework, the CRE is observed as lower sensitivity (d⬘) for CR as compared to SR targets. As such, hit rates and false alarm rates were calculated separately for Black and White targets, which were then used to calculate the separate sensitivity (d⬘) parameters for Black and White targets.2 In order to test whether the instructions inXuenced the magnitude of the CRE, the sensitivity scores to Black and White targets in Experiments 1a and 1b were subjected to identical 2 (Target Race) £ 2 (Instructions) mixed-model ANOVAs, with repeated-measures on the Wrst factor. Both ANOVAs yielded nearly identical results. The ANOVAs revealed a main eVect of Target Race in both Experiment 1a, F (1, 28) D 5.27, p D .03, and Experiment 1b, F (1, 136) D 23.06, p < .01, replicating the CRE. As shown in Fig. 1, however, Target Race interacted with Instructions in 2 As is common in signal detection analyses, adjustments to the data were made in all experiments to address the problem of empty cells: 0% was adjusted to 5% and 100% was adjusted to 95%. These adjustments were infrequent, accounting for »2% of cells across all three studies. Alternate adjustments yield nearly identical results.
Black Targets
1.5
White Targets 1.3
1.1
Sensitivity (d' )
Remember, pay very close attention to the faces, especially when they are of a diVerent race than you in order to try to avoid this Cross Race EVect.
337
0.9 Individuation
Control
Instructions
_____________________ 1.5
1.3
1.1
0.9 Individuation
Control
Instructions
Fig. 1. Mean sensitivity (d⬘) as a function of Target Race and Instruction condition for both Experiment 1a (top frame) and Experiment 1b (bottom frame).
both Experiment 1a, F (1, 28) D 3.98, p D .056, and Experiment 1b, F (1, 136) D 5.27, p D .023. Simple eVects tests revealed the typically observed CRE in the control conditions of both Experiment 1a, t (14) D 3.13, p < .01, d D .76, and Experiment 1b, t (70) D 6.04, p < .001, d D .71; however, the individuation instructions were suYcient to eliminate the CRE, in both Experiment 1a, t (14) D .21, p > .8, d D .05, and Experiment 1b, t (66) D 1.53, p > .13, d D .18.3 Across both experiments, we replicated the CRE in the control condition, Wnding lower sensitivity for Black as compared to White faces; however, this better recognition of SR than CR faces depended on the instructions given before learning. When given instructions to attend to the individuating features of CR faces, the CRE was eliminated. Consistent with Levin’s (1996, 2000) feature-selection model, under control conditions, participants seemed to encode individuating features of in-group members, but encode category-specifying features of out-group members. However, when individuals are made aware of their biases about out-group members and are motivated to individuate out-group members, individuation may occur. Indeed, when participants were warned of the CRE and instructed to attend to individuating features of CR faces, they were able to do so, eliminating the CRE. To the authors’ knowledge, this is one of few empirical displays of an elimination of the CRE outside of extensive training (e.g., Malpass et al., 1973, but see Johnson & Fredrickson, 2005), and suggests that the perceptual details 3 Analyses of the bias parameter revealed lower criteria for Black as compared to White targets in all three experiments, ps < .05; however, in no case did Target Race interact with instruction condition. As such, this is not discussed further.
338
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
encoded from CR faces are subject to the well-understood social-cognitive processes of categorization and individuation. Moreover, the current results may be diYcult to explain using mechanisms based purely on perceptual expertise. The instruction manipulation oVered no additional perceptual expertise to participants, but rather made participants aware of the nature of the bias and, as is suggested by the suYciency of the instructions to eliminate the CRE, elicited higher levels of attention to the individuating characteristics of out-group targets. Thus the current results oVer some initial support for the claim that the CRE may be due to a social-cognitive mechanism. Experiment 2 Experiments 1a and 1b indicated that inducing participants to individuate out-group members during encoding may eliminate the CRE. When participants were warned of the CRE and exhorted to attend closely to the individuating features of CR faces, they were able to do so. As has been well-replicated within the stereotyping and impression formation literatures, the individuation of out-group members can occur when social perceivers are aware of their tendency to think categorically about out-group members and are motivated to overcome this tendency. It is possible, however, that a broad motivation to process the faces, absent any speciWc information regarding racial bias, may be suYcient to eliminate the eVect. Thus, there may be nothing particular about the motivation to distinguish one CR face from another CR face, but rather it may be that increased motivation to process all of the targets may lead to similar results. Participants in the individuation conditions of the previous experiments received both more instructions overall as well as more exhortation to process the faces than did control participants. Thus, it is possible that merely inducing a general accuracy motivation may be suYcient to eliminate this eVect, absent any speciWc awareness of the CRE. Experiment 2 was designed to test the alternate explanation that a general accuracy motivation, absent any awareness of the CRE, may be suYcient to eliminate this eVect. SpeciWcally, the procedure of the previous experiments was replicated, with the addition of an instruction condition designed to elicit motivation to process all of the faces, in the absence of speciWc knowledge of the CRE, or instruction to attend to the individuating characteristics of CR faces. If a general accuracy motivation is suYcient to explain the eVects observed in the previous experiments, both the previous instructions designed to elicit the individuation of out-group members and the new instructions designed to elicit more processing in general should eliminate the CRE. If a combination of knowledge of the bias against out-group members and motivation to attend to the individuating features of CR faces are required to induce the individuation of out-group members, a general accuracy motivation should not be suYcient to eliminate the CRE.
Method Participants and design Fifty-Wve European Americans participated in the study. Target Race (2: Black; White) was manipulated on a within-subjects basis; instructions (3: control; individuation; motivation) were manipulated on a between-subjects basis. Three participants were excluded from analyses because, during debrieWng, they admitted to not following instructions, and had error rates of nearly 50%. Thus, all analyses were conducted on the remaining 52 participants. Materials and procedure The materials and procedure employed in Experiment 2 were identical to those of the previous experiments, except as follows. Participants were randomly assigned to one of three instructions conditions. Control and individuation instructions did not diVer from the previous experiments. The general accuracy motivation instructions were designed to increase participants’ motivation to attend closely to all of the stimuli in the learning phase, without providing awareness of any potential racial bias: Previous research has shown that people are often unreliable when it comes to recognizing previously seen faces. Because of this, eyewitness testimonies are often discredited. Now that you know this, we would like you to try especially hard when learning faces in this task. Do your best to try to pay close attention to what diVerentiates one particular face from another faceƒ Remember, try your hardest to pay very close attention to the faces. After receiving instructions, all participants completed learning, distraction, and recognition phases identical to those in the previous experiments. Results and discussion As in the previous experiments, of particular interest was the extent to which the CRE was observed under diVerent instruction conditions. As such, hit and false alarm rates for Black and White faces were used to calculate separate sensitivity parameters for Black and White faces. These sensitivity parameters (d⬘) for Black and White targets were then subjected to a 2 (Target Race) £ 3 (Instruction) mixedmodel ANOVA with repeated-measures on the Wrst factor. Replicating the Wndings of the previous experiments, this ANOVA yielded a main eVect of Target Race, F (1, 49) D 16.06, p < .01. As shown in Fig. 2, however, this main eVect was qualiWed by the predicted Target Race £ Instructions interaction, F (2, 49) D 3.08, p D .055. To investigate the nature of this predicted interaction, the simple eVect of Target Race was tested at each of the three levels of instructions. These analyses revealed that the CRE observed in the control condition, t (18) D 2.75, p D .01, d D .60, was eliminated by the instructions to individuate CR
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
Fig. 2. Mean sensitivity (d⬘) as a function of Target Race and Instruction condition.
faces, t (14) D .25, p D .81, d D .06, replicating the previous experiments. The instructions intended to produce increased processing motivation for all of the target faces, however, were not suYcient to eliminate the CRE, t (17) D 4.48, p < .001, d D 1.01. In order to further explicate the nature of the Target Race £ Instructions interaction, the magnitude of the CRE was compared across the three conditions. As predicted, the individuation instructions (M D .04) elicited a signiWcant decrease in the CRE relative to both the control condition (M D .47), F (1, 49) D 7.12, p D .01, and the general accuracy motivation condition (M D .59), F (1, 49) D 11.36, p < .01. The control and general accuracy condition, however, did not diVer, F (1, 49) D .61, ns. The eVects from Experiment 2 suggest that the results from the previous experiments are not attributable to diVerences in processing motivation alone; the CRE is observed even when participants receive additional instructions to process all faces carefully. Replicating the results of the previous experiments, however, when a speciWc knowledge of the race-based bias is paired with a speciWc motivation to individuate out-group members, participants successfully encode the individuating features of CR targets, eliminating the CRE. General discussion Across three experiments, providing individuals with an awareness of the CRE and motivating them to attend to the individuating features of out-group members led to the elimination of this well-established eVect in face perception. These Wndings Wt well with Levin’s (1996, 2000) feature-selection model, which posits that the CRE is due to diVerential social cognitions toward in-group and outgroup members, leading to an asymmetrical search for features in SR and CR faces. Although people encode the individuating features of SR faces, they tend to encode race-specifying features of CR faces at the expense of individuating information. In line with well-established models of individuation and categorization in the social categorization literature, however, inducing the individuation of out-group members during encoding was suYcient to eliminate categorical thinking, and thus eliminate the CRE.
339
Notably, some models of the CRE that rely on perceptual expertise alone may have some diYculty explaining the current eVects. The individuation instructions do not provide additional perceptual experience in discriminating between CR faces, but yet lead to robust reductions in the magnitude of the CRE. One possibility, however, is that the individuation instructions gave participants expertise-like skills in a deductive manner that are typically only learned inductively via extensive expertise with CR faces. That is, perhaps the individuation instructions focus perceivers on task-relevant features or yield other expertise-related skills, via a mechanism other than accruing actual experience discriminating between CR faces. This is possible, but the skills aVorded by the individuation instructions are clearly quite limited. The individuation instructions employed here oVered neither speciWc strategies to improve performance, nor speciWc dimensions on which CR faces actually diVer. If an expertise mechanism were to explain the current data, it seems reasonable to assume that the expertise-like skill aVorded by our manipulation was simply the awareness of the need to attend to individuating features. Although we are not averse to such an expertise explanation, the focus on individuating features inherent in such an explanation suggests that the social cognitions elicited by in-group and out-group members must also play a central role in the CRE. This is not to say that experience in distinguishing between out-group members plays no role in the CRE. To the contrary, it seems plausible that both suYcient perceptual expertise to distinguish between CR faces and the motivation to individuate out-group members are necessary conditions to successfully discriminate between CR faces. The White participants in the current experiments are embedded in a social context in which they likely have common exposure to Blacks in both the media and their personal lives. We hypothesize that in the absence of any perceptual expertise with the relevant CR faces, even strong motivation and capacity for individuation will likely go awry. Indeed, the dual necessity of perceptual expertise and individuation may account for the mixed literature on expertise eVects in the CRE. As the individuation instructions seemed suYcient to eliminate the CRE, this suggests that our participants came to the experimental context with the perceptual expertise necessary to distinguish between out-group members, but that in the typical cross-race context, they do not employ this expertise. Given the ubiquity of categorical, stereotypic thinking about out-group members, it perhaps should not be a surprise that, unless otherwise instructed or motivated, individuals rely on this tendency to attend to category specifying information in CR faces. Although it is beyond the scope of the current research, in so far as the current results bring the CRE under the aegis of the social-cognitive literature, the current Wndings suggest that other well-established moderators of social categorization and stereotyping (e.g., cognitive load) might have a predictable inXuence on attempts to individuate CR faces. One such extension could
340
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
be to investigate social-cognitive factors that elicit categorical perception of SR faces. For example, conditions that tend to elicit in-group homogeneity (Simon, 1992; Simon & Pettigrew, 1990) might lead to a search for race-specifying features among SR faces, potentially eliminating or even reversing the direction of the CRE. In addition to providing evidence for the social-cognitive basis of the CRE, the current experiments suggest a relatively simple way to reduce the CRE: inform people of the bias and motivate them to individuate out-group members. Although past work has shown that training can eliminate (e.g., Malpass et al., 1973) or even reverse this eVect (Sangrigoli et al., 2005), extensive training may be diYcult to implement as a solution to the CRE; instead, inducing the individuation of out-group members may prove a more tractable solution, at least for individuals who have suYcient expertise to make a distinction. Although more research is required to determine the extent to which social cognitions may modulate the CRE, the current research suggests that even straightforward manipulations of socialcognitive processes may be suYcient to eliminate this otherwise robust bias in face processing. References Allport, G. (1954). The nature of prejudice. Cambridge, MA: Addison-Wesley. Anthony, T., Copper, C., & Mullen, B. (1992). Cross-racial facial identiWcation: A social cognitive integration. Personality and Social Psychology Bulletin, 18, 296–301. Bodenhausen, G. V., Macrae, C. N., & Hugenberg, K. (2003). Social cognition. In I. Weiner (Ed.), Comprehensive handbook of psychology (Vol. 5). New York: Wiley. Brewer, M. B. (1988). A dual process model of impression formation. In R. S. Wyer & T. K. Srull (Eds.), Advances in social cognition (Vol. 1, pp. 1–36). Hillsdale, NJ: Erlbaum. Brigham, J. C., & Malpass, R. S. (1985). The role of experience and contact in the recognition of faces of own- and other-race persons. Journal of Social Issues, 41, 139–155. Chance, J. E., & Goldstein, A. G. (1996). The other-race eVect and eyewitness identiWcation. In S. L. Sporer, R. S. Malpass, & G. Koehnken (Eds.), Psychological issues in eyewitness identiWcation (pp. 153–176). Mahwah, NJ: Erlbaum. Elliott, E. S., Wills, E. J., & Goldstein, A. G. (1973). The eVects of discrimination training on the recognition of white and oriental faces. Bulletin of the Psychonomic Society, 2, 71–73. Feinman, S., & Entwhisle, D. R. (1976). Children’s ability to recognize other children’s faces. Child Development, 47, 506–510. Fiske, S. T., & Neuberg, S. L. (1990). A continuum model of impression formation: From category-based to individuating processes as a function of information, motivation, and attention. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1–74). New York, NY: Academic Press.
Goldstein, A. G., & Chance, J. E. (1985). EVects of training on Japanese face recognition: Reduction of the other-race eVect. Bulletin of the Psychonomic Society, 23, 211–214. Johnson, K. J., & Fredrickson, B. L. (2005). “We all look the same to me”: Positive emotions eliminate the own-race bias in face recognition. Psychological Science, 16, 875–881. Judd, C. M., & Park, B. (1988). Out-group homogeneity: Judgments of variability at the individual and group levels. Journal of Personality and Social Psychology, 54, 778–788. Klauer, K. G., & Wegener, I. (1998). Unraveling social categorization in the “Who said what?” paradigm. Journal of Personality and Social Psychology, 75, 1155–1178. Lavarkas, P. J., Buri, J. R., & Mayzner, M. S. (1976). A perspective on the recognition of other-race faces. Perception & Psychophysics, 20, 475–481. Levin, D. T. (1996). Classifying faces by race: The structure of face categories. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22, 1364–1382. Levin, D. T. (2000). Race as a visual feature: Using visual search and perceptual discrimination tasks to understand face categories and the cross-race recognition deWcit. Journal of Experimental Psychology: General, 129, 559–574. Linville, P. W., Fischer, G. W., & Salovey, P. (1989). Perceived distributions of the characteristics of in-group and out-group members: Empirical evidence and a computer simulation. Journal of Personality and Social Psychology, 57, 165–188. Malpass, R. S., & Kravitz, J. (1969). Recognition for faces of own and other race. Journal of Personality and Social Psychology, 13, 330–334. Malpass, R. S., Lavigueur, H., & Weldon, D. E. (1973). Verbal and visual training in face recognition. Perception and Psychophysics, 14, 283–292. McClelland, J. L., & Chappell, M. (1998). Familiarity breeds diVerentiation: A subjective-likelihood approach to the eVects of experience in recognition memory. Psychological Review, 105, 734–760. McLin, O. H., & Malpass, R. S. (2001). Racial categorization of faces: The ambiguous race face eVect. Psychology, Public Policy, and Law, 7, 98–118. Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigating the own-race bias in memory for faces: A meta-analytic review. Psychology, Public Policy, and Law, 7, 3–35. Ng, W., & Lindsay, R. C. L. (1994). Cross-race facial recognition: Failure of the contact hypothesis. Journal of Cross-Cultural Psychology, 25, 217–232. Rhodes, G., Brake, S., Taylor, K., & Tan, S. (1989). Expertise and conWgural coding in face recognition. British Journal of Psychology, 80, 313–331. Sangrigoli, S., Pallier, C., Argenti, A.-M., Ventureyra, V. A. G., & de Schonen, S. (2005). Reversibility of the other-race eVect in face recognition during childhood. Psychological Science, 16, 440–444. Simon, B. (1992). The perception of ingroup and outgroup homogeneity: Reintroducing the intergroup context. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 3, pp. 1–30). Oxford, England: Wiley. Simon, B., & Pettigrew, T. F. (1990). Social identity and perceived group homogeneity: Evidence for the ingroup homogeneity eVect. European Journal of Social Psychology, 20, 269–286. Taylor, S. E., Fiske, S. T., EtcoV, N. L., & Ruderman, A. J. (1978). Categorical and contextual bases of person memory and stereotyping. Journal of Personality and Social Psychology, 36, 778–793.
Categorization and individuation in the cross-race recognition deWcit: Toward a solution to an insidious problem Kurt Hugenberg a,¤, Jennifer Miller b, Heather M. Claypool a a
Department of Psychology, Miami University, USA b University of Cincinnati, USA
Received 30 August 2005; revised 23 February 2006 Available online 24 April 2006
Abstract Recent theory and evidence suggest that the Cross-Race EVect (better recognition for same-race (SR) faces than for cross-race (CR) faces) is due to social-cognitive processes of categorization of out-group members, causing perceivers to attend to category-specifying information of CR faces at the expense of individuating information. Three experiments seek to extend this social-cognitive explanation of the CRE by investigating the extent to which the Cross-Race EVect can be reduced by inducing perceivers to individuate rather than categorize CR faces. In all three experiments, participants who received warning of the Cross-Race EVect prior to encoding, and instructions to individuate out-group members, showed no CRE. Experiment 2 suggests that this elimination of the CRE was not due merely to increased motivation to process all stimuli. This is one of few empirical displays of an elimination of the CRE outside of visual training. Moreover, these results are congenial with Levin’s (2000) feature-selection model, which suggests that the CRE is due to diVerential social cognitions about in-group and out-group members, rather than to diVerences in perceptual expertise. By eliciting individuation of out-group members at encoding, the CRE can be eliminated. © 2006 Elsevier Inc. All rights reserved. Keywords: Cross-Race EVect; Face perception; Categorization; Individuation
Introduction The cross-race recognition deWcit, known more simply as the Cross-Race EVect (CRE), is one of the best-replicated phenomena in face perception (Chance & Goldstein, 1996), and has been of interest to social psychologists for more than half a century (Allport, 1954). Explained simply, the CRE is a tendency to have better recognition accuracy for same-race (SR) faces than for cross-race (CR) faces, an eVect that has been shown to be surprisingly robust across numerous research paradigms (Meissner & Brigham, 2001). From recognition memory tasks in the laboratory to suspect line-ups in police stations, the tendency to have better recognition of SR than CR faces has shown troubling consistency. Although this phenomenon is interesting in its own right, the high degree of interest in the CRE is due, at least in part, to the potentially deleteri*
Corresponding author. E-mail address: [email protected] (K. Hugenberg).
0022-1031/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jesp.2006.02.010
ous consequences of misidentifying members of racial outgroups (Brigham & Malpass, 1985). For the perceiver, consequences of misidentiWcation may include feelings of embarrassment or social opprobrium. For the victim of misidentiWcation, the consequences may be more problematic, leading to feelings of insult or the threat that accompanies the knowledge of being stereotyped. Perhaps more serious still, however, are the potential legal consequences of misidentiWcation. Given the reliance of the criminal justice system on eyewitness identiWcation, and the substantial weight that eyewitnesses have on juridic decisions, understanding the mechanisms underlying the CRE and how to ameliorate these eVects are clearly important goals for social psychological investigation. Mechanisms underlying the Cross-Race EVect: Percept versus concept Despite the robustness of the CRE across experimental contexts, coming to agreement on a satisfying theoretical
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
account for this troubling eVect has proven diYcult. Perhaps the longest standing explanation for the CRE is the perceptual expertise hypothesis (see Meissner & Brigham, 2001). Although there are many diVerent variations of this hypothesis (Ng & Lindsay, 1994), the core argument is that perceivers have diVerential expertise in processing SR versus CR faces, and that this diVerential expertise leads to diVerential recognition accuracy. Given that people typically have substantial contact with individuals of the samerace, extensive expertise is gained at distinguishing between SR faces. The comparatively lesser contact with individuals of other races, however, yields fewer opportunities for distinguishing between CR faces. As such, we are relatively inexpert at distinguishing between CR faces. The speciWc mechanism by which this diVerential expertise elicits diVerential recognition of CR faces is a matter of some debate. For example, in line with popular models of recognition memory (e.g., McClelland & Chappell, 1998), a lack of contact may lead to a lack of expertise with the dimensions on which CR faces actually vary (see McLin & Malpass, 2001). Alternately, lower levels of expertise with CR faces may elicit less holistic and relatively more feature-based processing of CR faces (Rhodes, Brake, Taylor, & Tan, 1989). Mechanisms for the perceptual expertise hypothesis aside, support for this perceptual expertise hypothesis in explaining the CRE has been mixed (Brigham & Malpass, 1985). On the supportive side, practice distinguishing between CR faces can reduce the CRE. For example, Malpass, Lavigueur, and Weldon (1973) found that practice at perceptual discrimination between SR and CR faces in the laboratory can at least temporarily reduce the magnitude of the CRE (see also Elliott, Wills, & Goldstein, 1973; Goldstein & Chance, 1985). More recently, work by Sangrigoli and colleagues (Sangrigoli, Pallier, Argenti, Ventureyra, & de Schonen, 2005) has found that extensive, lifelong training with CR faces can even reverse the direction of the CRE. In their study, individuals of Korean heritage who were adopted as children by Caucasian families in Europe showed a reversal of the CRE by adulthood. That is, despite their Korean heritage, these adopted participants who grew to maturity among Caucasian families showed eVects similar to their adoptive Caucasian parents, Wnding it more diYcult to correctly recognize Asian than White faces. Thus, there is some evidence that the magnitude of the CRE decreases as the amount of expertise with CR faces increases, and with extensive practice can even reverse. However, some studies have failed to Wnd empirical support for the perceptual expertise hypothesis, Wnding no relationship between the amount of expertise with CR faces and the magnitude of the CRE (e.g., Malpass & Kravitz, 1969; Ng & Lindsay, 1994). For example, Ng and Lindsay (1994) easily replicated the typical CRE with White and Asian participants, however, across two studies they found no relationship between the magnitude of the CRE and selfreported contact with members of the relevant racial outgroup. Ng and Lindsay also studied racial minorities among racial majorities, speciWcally White participants
335
living in Singapore and Asian participants living in Canada, who would presumably be forced by dint of their minority status to have extensive experience at distinguishing between CR faces. Even such extensive real-life experience seemed to have no eVect on the magnitude of the CRE. Similarly, attending schools with large numbers of outgroup members has shown to have diVerential eVects on the CRE (see Levin, 2000, for a review), in some cases reducing the eVect (Feinman & Entwhisle, 1976), in other cases, having no eVect (Malpass & Kravitz, 1969), and in still other cases even exacerbating the eVect (Lavarkas, Buri, & Mayzner, 1976). Thus, despite some conWrmatory evidence, the conXicting evidence that real world contact with out-group members does not reliably show a negative relationship with the magnitude of the CRE suggests that something more than mere perceptual expertise may be at play. In response to the sometimes checkered evidence for the expertise hypothesis in explaining the CRE, Levin (1996, 2000) proposed his feature-selection model as an alternate explanatory mechanism for the CRE. This feature-selection model argues that the CRE is due not to diVerential expertise with CR faces per se, but rather to diVerences in social cognitions typically elicited when processing in-group and out-group members (see also Anthony, Copper, & Mullen, 1992). At the core of Levin’s feature-selection hypothesis is the ubiquitous tendency of perceivers to think categorically about out-group members (e.g., Bodenhausen, Macrae, & Hugenberg, 2003). Categorical thinking involves reliance on social categories (e.g., race, sex, age) rather than on individual characteristics of a target. Levin (1996, 2000) argues that this tendency to think categorically about out-group members, but to individuate in-group members, leads to an asymmetrical search for features in SR versus CR faces (see also McLin & Malpass, 2001). The individuation of ingroup members leads perceivers to search for facial features that distinguish one in-group member from another. The tendency to think categorically about out-group members leads to a diVerent search pattern: perceivers search for category-specifying features (e.g., skin tone, brow strength) rather than individuating features. Whereas people encode individuating features in SR faces, they encode race-specifying features at the expense of this individuating information in CR faces. Having only encoded the race-specifying features of CR faces, this leads to real diYculty distinguishing one CR face from another CR face at recognition, resulting in the well-established CRE. Toward a social-cognitive understanding of the Cross-Race EVect The feature-selection model is, at its core, a social-cognitive model of a seemingly perceptual phenomenon, which aligns it with a number of analogous eVects in the social cognition literature. For example, the well-replicated within-category confusions in the “Who said what?” paradigm (Klauer & Wegener, 1998; Taylor, Fiske, EtcoV, & Ruderman, 1978) show a similar pattern, such that
336
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
perceivers tend to confuse the speakers of statements made by members of the same social category (e.g., a statement made by one African American is mistakenly assigned to another African American), suggesting that individuals can encode category specifying information about targets at the expense of individuating information leading to within-category confusions. The out-group homogeneity eVect (e.g., Judd & Park, 1988; Linville, Fischer, & Salovey, 1989), or the tendency to perceive members of out-groups as more homogeneous than members of in-groups, is another robust social cognitive eVect that shows striking similarities to the CRE. Such a social-cognitive explanation of the CRE may not only yield a satisfying theoretical account of this troubling eVect, but it may also oVer a solution to ameliorate the eVect that is radically diVerent than that oVered by a perceptual expertise mechanism. Whereas the only recourse to reduce the CRE under an expertise model seems to be extensive perceptual training, a social-cognitive model oVers a diVerent solution: individuation. If diVerent social cognitions about in-groups and out-groups are responsible for the CRE, then changing these social cognitions may be suYcient to reduce or eliminate the CRE. Although categorical thinking is common when engaging in social cognition about out-groups, prominent theories of stereotyping (e.g., Brewer, 1988; Fiske & Neuberg, 1990) note that thinking categorically about out-group members is by no means inevitable. To the contrary, when social perceivers are aware of their tendency to engage in categorical thinking about out-group members, are motivated to individuate out-group members, and have the cognitive capacity to do so, perceivers may engage in individuation. In contrast to categorization, individuation is the tendency to treat targets not according to category membership, but rather according to unique traits and characteristics. Thus, if the CRE is due to the tendency to categorize CR faces but to individuate SR faces, inducing perceivers to individuate out-group members may eliminate the CRE. Experiments 1a and 1b Experiments 1a and 1b were designed to test the hypothesis that inducing individuation of out-group members during encoding would reduce the CRE. Although Experiments 1a and 1b employed identical procedures, they used diVerent sets of stimulus faces, therefore their results are reported separately. We employed a recognition task commonly used in CRE research, in which participants were asked to attempt to distinguish which Black and White faces had been seen previously in the experiment and which were novel faces. Importantly, before the learning phase, participants were randomly assigned to one of two instruction conditions. In the control condition, designed to replicate past CRE research, participants received no information regarding the CRE. In the individuation instructions condition, however, participants received a description of the CRE and were instructed to attend
closely to the individuating features of CR faces in order to avoid this eVect. We predicted that control participants would show the typical CRE, showing better recognition of SR than CR faces. Of particular interest, however, was whether participants informed of the CRE and instructed to avoid it would be able to do so. If the CRE is due to tendencies to think categorically about out-group members, but to individuate in-group members, providing participants with knowledge of this bias and speciWc instructions to attend to the individuating features of CR faces may reduce or even eliminate the CRE. Method Participants and design Thirty European Americans participated in Experiment 1a; 146 European Americans participated in Experiment 1b. In both experiments, target race (2: Black; White) was manipulated on a within-subjects basis; instructions (2: control; individuation) were manipulated on a betweensubjects basis. Participants were randomly assigned to one of the two levels of instructions. Eight participants in Experiment 1b were excluded from analyses because, during debrieWng, they either reported problems learning the key mappings or admitted to not following instructions, and had error rates of nearly 50%. Thus, all analyses of Experiment 1b were conducted on the remaining 138 participants. Materials and procedure Although the stimulus sets diVered across the two experiments, the stimuli in both consisted of digitized, grayscale pictures of 80 male faces (40 Black; 40 White), 6 £ 4 cm in size, displaying only face and hair. Within each experiment, the 80 faces were randomly assigned to two lists of 40 faces each (20 Black; 20 White). The list of 40 faces seen during the learning phase was counter-balanced on a between-subjects basis.1 After providing informed consent, participants were seated at computers in individual cubicles and were instructed that they would participate in a face recognition experiment, consisting of a learning phase and a recognition phase. All instructions and stimulus presentations were presented by the computer. Participants were Wrst instructed that, during the learning phase, they would see 40 faces on the computer screen, and to attend closely to these faces in order to later recognize them. Participants in the individuation condition also received the following instructions intended to elicit the individuation of outgroup members: Previous research has shown that people reliably show what is known as the Cross-Race EVect (CRE) when learning faces. Basically, people tend to confuse faces that belong to other races. For example, a White learner 1 Preliminary analyses indicated no list eVects in any of the three experiments.
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
will tend to mistake one Black face for another. Now that you know this, we would like you to try especially hard when learning faces in this task that happen to be of a diVerent race. Do your best to try to pay close attention to what diVerentiates one particular face from another face of the same race, especially when that face is not of the same-race as youƒ
After receiving instructions, all participants began the learning phase, during which each of 40 target faces (20 Black; 20 White) were displayed in a randomized sequence on a computer screen; each face was displayed for 3 s. After completing a 5–7 min unrelated distractor task, all participants engaged in the recognition phase. Participants were instructed that they would see a series of faces, some of which they had seen during the learning phase (i.e., old faces) and some of which they had not previously seen (i.e., new faces). Participants were instructed to distinguish which of the faces they had seen during the learning phase (i.e., old faces) and which of the faces were not seen during the learning phase (i.e., new faces) using the ‘a’ key and the ‘5’ key on the keyboard, respectively. The 80 faces observed at the recognition phase included the 40 faces previously seen during learning and 40 new faces (20 Black; 20 White), displayed in a separate random order for each participant. Each face remained on the screen until participants responded, which began the next trial. After completing all tasks, participants were thanked and debriefed. Results and discussion Of primary interest was the extent to which the instruction conditions inXuenced the magnitude of the CRE. Within the signal detection framework, the CRE is observed as lower sensitivity (d⬘) for CR as compared to SR targets. As such, hit rates and false alarm rates were calculated separately for Black and White targets, which were then used to calculate the separate sensitivity (d⬘) parameters for Black and White targets.2 In order to test whether the instructions inXuenced the magnitude of the CRE, the sensitivity scores to Black and White targets in Experiments 1a and 1b were subjected to identical 2 (Target Race) £ 2 (Instructions) mixed-model ANOVAs, with repeated-measures on the Wrst factor. Both ANOVAs yielded nearly identical results. The ANOVAs revealed a main eVect of Target Race in both Experiment 1a, F (1, 28) D 5.27, p D .03, and Experiment 1b, F (1, 136) D 23.06, p < .01, replicating the CRE. As shown in Fig. 1, however, Target Race interacted with Instructions in 2 As is common in signal detection analyses, adjustments to the data were made in all experiments to address the problem of empty cells: 0% was adjusted to 5% and 100% was adjusted to 95%. These adjustments were infrequent, accounting for »2% of cells across all three studies. Alternate adjustments yield nearly identical results.
Black Targets
1.5
White Targets 1.3
1.1
Sensitivity (d' )
Remember, pay very close attention to the faces, especially when they are of a diVerent race than you in order to try to avoid this Cross Race EVect.
337
0.9 Individuation
Control
Instructions
_____________________ 1.5
1.3
1.1
0.9 Individuation
Control
Instructions
Fig. 1. Mean sensitivity (d⬘) as a function of Target Race and Instruction condition for both Experiment 1a (top frame) and Experiment 1b (bottom frame).
both Experiment 1a, F (1, 28) D 3.98, p D .056, and Experiment 1b, F (1, 136) D 5.27, p D .023. Simple eVects tests revealed the typically observed CRE in the control conditions of both Experiment 1a, t (14) D 3.13, p < .01, d D .76, and Experiment 1b, t (70) D 6.04, p < .001, d D .71; however, the individuation instructions were suYcient to eliminate the CRE, in both Experiment 1a, t (14) D .21, p > .8, d D .05, and Experiment 1b, t (66) D 1.53, p > .13, d D .18.3 Across both experiments, we replicated the CRE in the control condition, Wnding lower sensitivity for Black as compared to White faces; however, this better recognition of SR than CR faces depended on the instructions given before learning. When given instructions to attend to the individuating features of CR faces, the CRE was eliminated. Consistent with Levin’s (1996, 2000) feature-selection model, under control conditions, participants seemed to encode individuating features of in-group members, but encode category-specifying features of out-group members. However, when individuals are made aware of their biases about out-group members and are motivated to individuate out-group members, individuation may occur. Indeed, when participants were warned of the CRE and instructed to attend to individuating features of CR faces, they were able to do so, eliminating the CRE. To the authors’ knowledge, this is one of few empirical displays of an elimination of the CRE outside of extensive training (e.g., Malpass et al., 1973, but see Johnson & Fredrickson, 2005), and suggests that the perceptual details 3 Analyses of the bias parameter revealed lower criteria for Black as compared to White targets in all three experiments, ps < .05; however, in no case did Target Race interact with instruction condition. As such, this is not discussed further.
338
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
encoded from CR faces are subject to the well-understood social-cognitive processes of categorization and individuation. Moreover, the current results may be diYcult to explain using mechanisms based purely on perceptual expertise. The instruction manipulation oVered no additional perceptual expertise to participants, but rather made participants aware of the nature of the bias and, as is suggested by the suYciency of the instructions to eliminate the CRE, elicited higher levels of attention to the individuating characteristics of out-group targets. Thus the current results oVer some initial support for the claim that the CRE may be due to a social-cognitive mechanism. Experiment 2 Experiments 1a and 1b indicated that inducing participants to individuate out-group members during encoding may eliminate the CRE. When participants were warned of the CRE and exhorted to attend closely to the individuating features of CR faces, they were able to do so. As has been well-replicated within the stereotyping and impression formation literatures, the individuation of out-group members can occur when social perceivers are aware of their tendency to think categorically about out-group members and are motivated to overcome this tendency. It is possible, however, that a broad motivation to process the faces, absent any speciWc information regarding racial bias, may be suYcient to eliminate the eVect. Thus, there may be nothing particular about the motivation to distinguish one CR face from another CR face, but rather it may be that increased motivation to process all of the targets may lead to similar results. Participants in the individuation conditions of the previous experiments received both more instructions overall as well as more exhortation to process the faces than did control participants. Thus, it is possible that merely inducing a general accuracy motivation may be suYcient to eliminate this eVect, absent any speciWc awareness of the CRE. Experiment 2 was designed to test the alternate explanation that a general accuracy motivation, absent any awareness of the CRE, may be suYcient to eliminate this eVect. SpeciWcally, the procedure of the previous experiments was replicated, with the addition of an instruction condition designed to elicit motivation to process all of the faces, in the absence of speciWc knowledge of the CRE, or instruction to attend to the individuating characteristics of CR faces. If a general accuracy motivation is suYcient to explain the eVects observed in the previous experiments, both the previous instructions designed to elicit the individuation of out-group members and the new instructions designed to elicit more processing in general should eliminate the CRE. If a combination of knowledge of the bias against out-group members and motivation to attend to the individuating features of CR faces are required to induce the individuation of out-group members, a general accuracy motivation should not be suYcient to eliminate the CRE.
Method Participants and design Fifty-Wve European Americans participated in the study. Target Race (2: Black; White) was manipulated on a within-subjects basis; instructions (3: control; individuation; motivation) were manipulated on a between-subjects basis. Three participants were excluded from analyses because, during debrieWng, they admitted to not following instructions, and had error rates of nearly 50%. Thus, all analyses were conducted on the remaining 52 participants. Materials and procedure The materials and procedure employed in Experiment 2 were identical to those of the previous experiments, except as follows. Participants were randomly assigned to one of three instructions conditions. Control and individuation instructions did not diVer from the previous experiments. The general accuracy motivation instructions were designed to increase participants’ motivation to attend closely to all of the stimuli in the learning phase, without providing awareness of any potential racial bias: Previous research has shown that people are often unreliable when it comes to recognizing previously seen faces. Because of this, eyewitness testimonies are often discredited. Now that you know this, we would like you to try especially hard when learning faces in this task. Do your best to try to pay close attention to what diVerentiates one particular face from another faceƒ Remember, try your hardest to pay very close attention to the faces. After receiving instructions, all participants completed learning, distraction, and recognition phases identical to those in the previous experiments. Results and discussion As in the previous experiments, of particular interest was the extent to which the CRE was observed under diVerent instruction conditions. As such, hit and false alarm rates for Black and White faces were used to calculate separate sensitivity parameters for Black and White faces. These sensitivity parameters (d⬘) for Black and White targets were then subjected to a 2 (Target Race) £ 3 (Instruction) mixedmodel ANOVA with repeated-measures on the Wrst factor. Replicating the Wndings of the previous experiments, this ANOVA yielded a main eVect of Target Race, F (1, 49) D 16.06, p < .01. As shown in Fig. 2, however, this main eVect was qualiWed by the predicted Target Race £ Instructions interaction, F (2, 49) D 3.08, p D .055. To investigate the nature of this predicted interaction, the simple eVect of Target Race was tested at each of the three levels of instructions. These analyses revealed that the CRE observed in the control condition, t (18) D 2.75, p D .01, d D .60, was eliminated by the instructions to individuate CR
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
Fig. 2. Mean sensitivity (d⬘) as a function of Target Race and Instruction condition.
faces, t (14) D .25, p D .81, d D .06, replicating the previous experiments. The instructions intended to produce increased processing motivation for all of the target faces, however, were not suYcient to eliminate the CRE, t (17) D 4.48, p < .001, d D 1.01. In order to further explicate the nature of the Target Race £ Instructions interaction, the magnitude of the CRE was compared across the three conditions. As predicted, the individuation instructions (M D .04) elicited a signiWcant decrease in the CRE relative to both the control condition (M D .47), F (1, 49) D 7.12, p D .01, and the general accuracy motivation condition (M D .59), F (1, 49) D 11.36, p < .01. The control and general accuracy condition, however, did not diVer, F (1, 49) D .61, ns. The eVects from Experiment 2 suggest that the results from the previous experiments are not attributable to diVerences in processing motivation alone; the CRE is observed even when participants receive additional instructions to process all faces carefully. Replicating the results of the previous experiments, however, when a speciWc knowledge of the race-based bias is paired with a speciWc motivation to individuate out-group members, participants successfully encode the individuating features of CR targets, eliminating the CRE. General discussion Across three experiments, providing individuals with an awareness of the CRE and motivating them to attend to the individuating features of out-group members led to the elimination of this well-established eVect in face perception. These Wndings Wt well with Levin’s (1996, 2000) feature-selection model, which posits that the CRE is due to diVerential social cognitions toward in-group and outgroup members, leading to an asymmetrical search for features in SR and CR faces. Although people encode the individuating features of SR faces, they tend to encode race-specifying features of CR faces at the expense of individuating information. In line with well-established models of individuation and categorization in the social categorization literature, however, inducing the individuation of out-group members during encoding was suYcient to eliminate categorical thinking, and thus eliminate the CRE.
339
Notably, some models of the CRE that rely on perceptual expertise alone may have some diYculty explaining the current eVects. The individuation instructions do not provide additional perceptual experience in discriminating between CR faces, but yet lead to robust reductions in the magnitude of the CRE. One possibility, however, is that the individuation instructions gave participants expertise-like skills in a deductive manner that are typically only learned inductively via extensive expertise with CR faces. That is, perhaps the individuation instructions focus perceivers on task-relevant features or yield other expertise-related skills, via a mechanism other than accruing actual experience discriminating between CR faces. This is possible, but the skills aVorded by the individuation instructions are clearly quite limited. The individuation instructions employed here oVered neither speciWc strategies to improve performance, nor speciWc dimensions on which CR faces actually diVer. If an expertise mechanism were to explain the current data, it seems reasonable to assume that the expertise-like skill aVorded by our manipulation was simply the awareness of the need to attend to individuating features. Although we are not averse to such an expertise explanation, the focus on individuating features inherent in such an explanation suggests that the social cognitions elicited by in-group and out-group members must also play a central role in the CRE. This is not to say that experience in distinguishing between out-group members plays no role in the CRE. To the contrary, it seems plausible that both suYcient perceptual expertise to distinguish between CR faces and the motivation to individuate out-group members are necessary conditions to successfully discriminate between CR faces. The White participants in the current experiments are embedded in a social context in which they likely have common exposure to Blacks in both the media and their personal lives. We hypothesize that in the absence of any perceptual expertise with the relevant CR faces, even strong motivation and capacity for individuation will likely go awry. Indeed, the dual necessity of perceptual expertise and individuation may account for the mixed literature on expertise eVects in the CRE. As the individuation instructions seemed suYcient to eliminate the CRE, this suggests that our participants came to the experimental context with the perceptual expertise necessary to distinguish between out-group members, but that in the typical cross-race context, they do not employ this expertise. Given the ubiquity of categorical, stereotypic thinking about out-group members, it perhaps should not be a surprise that, unless otherwise instructed or motivated, individuals rely on this tendency to attend to category specifying information in CR faces. Although it is beyond the scope of the current research, in so far as the current results bring the CRE under the aegis of the social-cognitive literature, the current Wndings suggest that other well-established moderators of social categorization and stereotyping (e.g., cognitive load) might have a predictable inXuence on attempts to individuate CR faces. One such extension could
340
K. Hugenberg et al. / Journal of Experimental Social Psychology 43 (2007) 334–340
be to investigate social-cognitive factors that elicit categorical perception of SR faces. For example, conditions that tend to elicit in-group homogeneity (Simon, 1992; Simon & Pettigrew, 1990) might lead to a search for race-specifying features among SR faces, potentially eliminating or even reversing the direction of the CRE. In addition to providing evidence for the social-cognitive basis of the CRE, the current experiments suggest a relatively simple way to reduce the CRE: inform people of the bias and motivate them to individuate out-group members. Although past work has shown that training can eliminate (e.g., Malpass et al., 1973) or even reverse this eVect (Sangrigoli et al., 2005), extensive training may be diYcult to implement as a solution to the CRE; instead, inducing the individuation of out-group members may prove a more tractable solution, at least for individuals who have suYcient expertise to make a distinction. Although more research is required to determine the extent to which social cognitions may modulate the CRE, the current research suggests that even straightforward manipulations of socialcognitive processes may be suYcient to eliminate this otherwise robust bias in face processing. References Allport, G. (1954). The nature of prejudice. Cambridge, MA: Addison-Wesley. Anthony, T., Copper, C., & Mullen, B. (1992). Cross-racial facial identiWcation: A social cognitive integration. Personality and Social Psychology Bulletin, 18, 296–301. Bodenhausen, G. V., Macrae, C. N., & Hugenberg, K. (2003). Social cognition. In I. Weiner (Ed.), Comprehensive handbook of psychology (Vol. 5). New York: Wiley. Brewer, M. B. (1988). A dual process model of impression formation. In R. S. Wyer & T. K. Srull (Eds.), Advances in social cognition (Vol. 1, pp. 1–36). Hillsdale, NJ: Erlbaum. Brigham, J. C., & Malpass, R. S. (1985). The role of experience and contact in the recognition of faces of own- and other-race persons. Journal of Social Issues, 41, 139–155. Chance, J. E., & Goldstein, A. G. (1996). The other-race eVect and eyewitness identiWcation. In S. L. Sporer, R. S. Malpass, & G. Koehnken (Eds.), Psychological issues in eyewitness identiWcation (pp. 153–176). Mahwah, NJ: Erlbaum. Elliott, E. S., Wills, E. J., & Goldstein, A. G. (1973). The eVects of discrimination training on the recognition of white and oriental faces. Bulletin of the Psychonomic Society, 2, 71–73. Feinman, S., & Entwhisle, D. R. (1976). Children’s ability to recognize other children’s faces. Child Development, 47, 506–510. Fiske, S. T., & Neuberg, S. L. (1990). A continuum model of impression formation: From category-based to individuating processes as a function of information, motivation, and attention. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1–74). New York, NY: Academic Press.
Goldstein, A. G., & Chance, J. E. (1985). EVects of training on Japanese face recognition: Reduction of the other-race eVect. Bulletin of the Psychonomic Society, 23, 211–214. Johnson, K. J., & Fredrickson, B. L. (2005). “We all look the same to me”: Positive emotions eliminate the own-race bias in face recognition. Psychological Science, 16, 875–881. Judd, C. M., & Park, B. (1988). Out-group homogeneity: Judgments of variability at the individual and group levels. Journal of Personality and Social Psychology, 54, 778–788. Klauer, K. G., & Wegener, I. (1998). Unraveling social categorization in the “Who said what?” paradigm. Journal of Personality and Social Psychology, 75, 1155–1178. Lavarkas, P. J., Buri, J. R., & Mayzner, M. S. (1976). A perspective on the recognition of other-race faces. Perception & Psychophysics, 20, 475–481. Levin, D. T. (1996). Classifying faces by race: The structure of face categories. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22, 1364–1382. Levin, D. T. (2000). Race as a visual feature: Using visual search and perceptual discrimination tasks to understand face categories and the cross-race recognition deWcit. Journal of Experimental Psychology: General, 129, 559–574. Linville, P. W., Fischer, G. W., & Salovey, P. (1989). Perceived distributions of the characteristics of in-group and out-group members: Empirical evidence and a computer simulation. Journal of Personality and Social Psychology, 57, 165–188. Malpass, R. S., & Kravitz, J. (1969). Recognition for faces of own and other race. Journal of Personality and Social Psychology, 13, 330–334. Malpass, R. S., Lavigueur, H., & Weldon, D. E. (1973). Verbal and visual training in face recognition. Perception and Psychophysics, 14, 283–292. McClelland, J. L., & Chappell, M. (1998). Familiarity breeds diVerentiation: A subjective-likelihood approach to the eVects of experience in recognition memory. Psychological Review, 105, 734–760. McLin, O. H., & Malpass, R. S. (2001). Racial categorization of faces: The ambiguous race face eVect. Psychology, Public Policy, and Law, 7, 98–118. Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigating the own-race bias in memory for faces: A meta-analytic review. Psychology, Public Policy, and Law, 7, 3–35. Ng, W., & Lindsay, R. C. L. (1994). Cross-race facial recognition: Failure of the contact hypothesis. Journal of Cross-Cultural Psychology, 25, 217–232. Rhodes, G., Brake, S., Taylor, K., & Tan, S. (1989). Expertise and conWgural coding in face recognition. British Journal of Psychology, 80, 313–331. Sangrigoli, S., Pallier, C., Argenti, A.-M., Ventureyra, V. A. G., & de Schonen, S. (2005). Reversibility of the other-race eVect in face recognition during childhood. Psychological Science, 16, 440–444. Simon, B. (1992). The perception of ingroup and outgroup homogeneity: Reintroducing the intergroup context. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 3, pp. 1–30). Oxford, England: Wiley. Simon, B., & Pettigrew, T. F. (1990). Social identity and perceived group homogeneity: Evidence for the ingroup homogeneity eVect. European Journal of Social Psychology, 20, 269–286. Taylor, S. E., Fiske, S. T., EtcoV, N. L., & Ruderman, A. J. (1978). Categorical and contextual bases of person memory and stereotyping. Journal of Personality and Social Psychology, 36, 778–793.