Current Directions in Psychological Science http://cdp.sagepub.com/ Developmental Origins of the Other-Race Effect Gizelle Anzures, Paul C. Quinn, Olivier Pascalis, Alan M. Slater, James W. Tanaka and Kang Lee Current Directions in Psychological Science 2013 22: 173 DOI: 10.1177/0963721412474459 The online version of this article can be found at: http://cdp.sagepub.com/content/22/3/173 Published by: http://www.sagepublications.com On behalf of: Association for Psychological Science Additional services and information for Current Directions in Psychological Science can be found at: Email Alerts: http://cdp.sagepub.com/cgi/alerts Subscriptions: http://cdp.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav >> Version of Record - Jun 4, 2013 What is This? Downloaded from cdp.sagepub.com at UNIV NEBRASKA LIBRARIES on August 22, 2013 474459 research-article2013 CDPXXX10.1177/0963721412474459Anzures et al.Developmental Origins of the ORE Developmental Origins of the Other-Race Effect Current Directions in Psychological Science 22(3) 173­–178 © The Author(s) 2013 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0963721412474459 cdps.sagepub.com Gizelle Anzures1, Paul C. Quinn2, Olivier Pascalis3, Alan M. Slater4, James W. Tanaka5, and Kang Lee6 1 Centre for Brain and Cognitive Development, Birkbeck, University of London; 2Department of Psychology, University of Delaware; 3Laboratoire de Psychologie et Neurocognition, Université Pierre Mendes France; 4School of Psychology, University of Exeter; 5Department of Psychology, University of Victoria; 6Institute of Child Study, University of Toronto Abstract The other-race effect (ORE) in face recognition refers to better recognition memory for faces of one’s own race than faces of another race—a common phenomenon among individuals living in primarily mono-racial societies. In this article, we review findings suggesting that early visual and sociocultural experiences shape one’s processing of familiar and unfamiliar race classes and give rise to the ORE within the 1st year of life. However, despite its early development, the ORE can be prevented, attenuated, and even reversed given experience with a novel race class. Social implications of the ORE are discussed in relation to development of race-based preferences for social partners and racial prejudices. Keywords other-race effect, perceptual narrowing, face perception, face recognition, face processing In primarily mono-racial societies, most people frequently encounter individuals from their own race, whereas encounters with individuals from other races are relatively infrequent. Such asymmetry in exposure to own- and other-race faces presents an ideal “natural experiment” for understanding how specific input from the environment shapes people’s abilities to process familiar and unfamiliar categories of visual objects and their attitudes about them. Indeed, extensive evidence shows that adults are better at recognizing faces of their own race than those of another unfamiliar race (Meissner & Brigham, 2001). In this article, we review findings regarding the developmental origins of this so-called other-race effect (ORE) and the experience-dependent recalibration of one’s visual representation of faces when exposed to a novel race class. We also discuss early visual preferences and recognition biases in relation to the development of racial preferences and prejudices. own race than faces of another race (Bar-Haim, Ziv, Lamy, & Hodes, 2006; Kelly, Liu, et al., 2007; Kelly et al., 2005). However, 3-month-olds with experience with individuals from their own race as well as those from another race show no visual preference for faces belonging to either group (Bar-Haim et al., 2006). Thus, an early visual preference for a race class is shaped by the familiar race class in one’s environment. Such early visual preferences may help in shaping the development of a differentiation and recognition advantage for the familiar race class that also begins to emerge at around 3 months of age. Face recognition in infants is most commonly studied by familiarizing infants with an image of a face and then measuring their visual preferences for a novel face when paired with the familiarized Visual Experience and Expertise for Own-Race Faces Corresponding Authors: Gizelle Anzures, Centre for Brain and Cognitive Development, Henry Wellcome Building, School of Psychology, Birkbeck, University of London, Malet St., London, WC1E 7HX, United Kingdom E-mail: g.anzures@bbc.ac.uk Newborns show no racial preferences (Kelly et al., 2005), but by 3 months of age, infants with experience primarily with own-race individuals prefer to look at faces of their Kang Lee, Dr. Eric Jackman Institute of Child Study, University of Toronto, 45 Walmer Rd., Toronto, Ontario, M5R 2X2, Canada E-mail: kang.lee@utoronto.ca Downloaded from cdp.sagepub.com at UNIV NEBRASKA LIBRARIES on August 22, 2013 Anzures et al. 174 face. An own-race bias in differentiating between small differences in facial identity can be observed in 3-montholds (Hayden, Bhatt, Joseph, & Tanaka, 2007). However, the ORE in 3-month-olds can be easily eliminated via brief exposure to photos of other-race faces (Sangrigoli & de Schonen, 2004). Indeed, 4-month-olds demonstrate holistic processing of both own- and other-race faces (Ferguson, Kulkofsky, Cashon, & Casasola, 2009). However, although 6-month-olds can still recognize faces from certain unfamiliar racial groups, they fail to do so with faces from other racial groups (Kelly et al., 2009; Kelly, Quinn, et al., 2007). By 8 months of age, infants process own-race, but not other-race, faces holistically (Ferguson et al., 2009); furthermore, at 9 months of age, infants show a robust ability to recognize own-race faces while showing poor recognition of all other-race faces (Kelly et al., 2009; Kelly, Quinn, et al., 2007). After infancy, although recognition of both own- and other-race faces improves with age, the ORE persists into childhood and adulthood with continued asymmetry in own- and other-race experience (Goodman et al., 2007; Pezdek, Blandon-Gitlin, & Moore, 2003; Walker & Hewstone, 2006). Overall, it appears that at birth, one’s visual representation of faces is broadly tuned with no preexisting racebased visual preference or recognition bias. However, continued experience with own-race individuals and lack of experience with other-race individuals during the 1st year of life leads to a perceptual narrowing process: One’s visual representation of faces becomes fine-tuned to differentiate between and recognize own-race faces (Scott, Pascalis, & Nelson, 2007). Experience with ownrace individuals likely develops a visual representation of faces based on many dimensions necessary to differentiate between own-race faces—dimensions that may not be ideal in differentiating between faces from another race (Valentine, 1991). Considering infants’ developing visual capabilities, it remains to be seen whether gradual improvements in visual acuity and contrast sensitivity influence the emergence and development of the ORE in face recognition. Improved visual capabilities would allow infants to develop increasingly refined visual representations of faces based on fine-detailed dimensions most crucial in differentiating between own-race individuals. Studies in which researchers examine how developmental changes in perception interact with experiential factors would provide a fuller account of the development of the ORE in face recognition. Sociocultural Experience and Expertise for Own-Race Faces In addition to the importance of visual experience, sociocultural conventions in interacting with social partners appear to shape infants’ visual scanning of faces. Consistent with the Western cultural convention in maintaining eye contact during social interactions (Argyle & Cook, 1976), Caucasian 6- to 10-month-olds show an age-related increase in their visual scanning of the eye region of Caucasian faces (Wheeler et al., 2011). However, they show no such change in their scanning of African American faces. In contrast, consistent with the Eastern cultural convention in limiting direct eye contact (Li, 2004), Asian 4- to 9-month-olds in China do not show an increase in the scanning of own-race eye regions (Liu et al., 2011). Instead, they focus on the central/nasal regions of ownrace faces, but their focus on the central/nasal regions of other-race Caucasian faces declines with age. Thus, infants learn to adopt cultural conventions in social gaze interactions, but they only do so for the familiar race class in their environment. In the case of Caucasian infants, only their face scanning of own-race faces becomes more similar to the face scanning pattern shown by Caucasian adults (Blais, Jack, Scheepers, Fiset, & Caldara, 2008). Likewise, in the case of Chinese infants, only their face scanning of own-race faces maintains a face scanning pattern similar to that shown by Asian adults (Blais et al., 2008; Fu, Hu, Wang, Quinn, & Lee, 2012). The similar developmental timeline in infants’ differential scanning of own- and other-race faces and the emerging ORE suggest that the two behaviors may be related. Indeed, among North American 3-month-olds raised by mono-racial parents, greater scanning frequency between the eye and mouth regions of own-race faces is associated with a greater likelihood of recognizing own-race faces (Gaither, Pauker, & Johnson, 2012). It is interesting to note that among biracial 3-month-olds with Asian mothers, reduced scanning between the eye and mouth regions of Asian faces is associated with a greater likelihood of recognizing Asian faces (Gaither et al., 2012). In future studies, researchers should investigate how face scanning may be related to own- and other-race face recognition among older infants who show a more robust ORE. Perhaps the ORE in older infants is exacerbated by their initial categorization of faces as own- or other-race, which may subsequently instigate differential visual scanning and processing of faces—at the subordinate level of identity for own-race faces versus at the more global level of race for other-race faces (see Ge et al., 2009). Indeed, by 9 months of age, infants visually categorize faces by race, and they also show a robust ORE (Anzures, Quinn, Pascalis, Slater, & Lee, 2010). In contrast, 6-montholds do not appear to consistently categorize faces by race (Anzures et al., 2010), and their ORE in recognition does not seem as robust relative to older infants (Kelly et al., 2009; Kelly, Quinn, et al., 2007). In older children, Downloaded from cdp.sagepub.com at UNIV NEBRASKA LIBRARIES on August 22, 2013 Developmental Origins of the ORE 175 categorization appears to reinforce differential processing of items from familiar/unfamiliar race classes. Topdown influences in cognitive processing are evident in 2- to 6-year-olds who show better recognition of 50% own-race and 50% other-race morphed faces when such faces are depicted as belonging to one’s own-race rather than another race (Shutts & Kinzler, 2007). A similar, albeit more perceptually based, cognitive process may also be occurring in infants showing a robust ORE in the latter half of the 1st year of life. Other-Race Experience and Attenuation of the ORE Although expertise in recognizing the familiar race class begins to develop early, face processing expertise remains plastic and is continuously fine-tuned by experience. Researchers have shown that experience with other-race faces can prevent, attenuate, or even reverse the typical ORE in face recognition. Sangrigoli and de Schonen (2004) found that although the ORE was already present in Caucasian 3-month-olds, it was eliminated after only 2 minutes of visual experience with photographs of three different Asian faces. In contrast, Heron-Delaney et al. (2011) found comparable recognition of Caucasian and Asian faces among Caucasian 6-month-olds. However, proficiency in recognizing Asian faces was maintained at 9 months of age only if, starting at 6 months of age, infants were given approximately 70 minutes of visual experience with photographs of individually named Asian faces (Heron-Delaney et al., 2011). Among older infants already showing the ORE, greater visual experience with other-race faces is needed to attenuate the decline in other-race face recognition. Caucasian 8- to 10-month-olds who show difficulty in recognizing other-race faces begin to show above-chance recognition of novel Asian faces after approximately 100 to 155 minutes of visual experience with dynamic videos of individually named Asian faces (Anzures et al., 2012). It remains to be seen whether a faster attenuation of the decline in other-race face recognition during infancy would be observed if the other race became the ethnic majority rather than the minority. However, a few researchers have examined similar cases among child adoptees. In one study, Asian children from China or Vietnam adopted into Caucasian families in Europe between 2 and 26 months of age showed comparable recognition of Asian and Caucasian faces at 6 to 14 years of age (de Heering, de Liedekerke, Deboni, & Rossion, 2010). In contrast, Asian adults from Korea who were adopted into Caucasian families in Europe between 3 and 9 years of age showed better recognition of otherrace Caucasian faces than own-race Asian faces (Sangrigoli, Pallier, Argenti, Ventureyra, & de Schonen, 2005). Thus, approximately 20 years of primary experience with another race is sufficient to reverse the typical own-race bias in face recognition. To date, research confirms that proficiency in otherrace face recognition can develop following relatively brief visual exposure to photographs or videos of otherrace faces during infancy (Anzures et al., 2012; Sangrigoli & de Schonen, 2004) and adulthood (Goldstein & Chance, 1985; Hills & Lewis, 2006; Rhodes, Locke, Ewing, & Evangelista, 2009; Tanaka & Pierce, 2009). However, live social interactions provide a rich representation of faces through varied viewpoints and varied emotional expressions embedded in meaningful social contexts. It has yet to be investigated whether live social interactions with other-race individuals lead to a faster attenuation of the ORE with longer lasting effects on other-race face recognition in both infants and adults. In addition to visual experience, emphasis on individuating between other-race faces appears to be an important factor in developing proficiency in other-race face recognition among adults (Goldstein & Chance, 1985; Hills & Lewis, 2006; Rhodes et al., 2009; Tanaka & Pierce, 2009). In contrast, emphasis on racial categories leads to a smaller attenuation of the ORE in adults (Tanaka & Pierce, 2009). Similarly, 6-month-olds maintain the ability to individuate between monkey faces at 9 months of age only if they are given visual experience with images of monkeys with different names rather than the category label of “monkey” or no label (Scott & Monesson, 2009). Thus, visual experience alone may not be sufficient to attenuate a bias in face recognition. Attention to visual and naming cues that highlight that other-race faces differ from one another appears to be crucial in learning to process other-race faces at the individual level of identity rather than at the categorical level of race. The ORE and Racial Preferences and Prejudices One might wonder whether early visual preferences and recognition biases for own-race individuals would influence interactions with social partners. Evidence to date suggests that infants do not discriminate among social partners on the basis of race. Ten-month-olds accept toys equally from own- and other-race individuals (Kinzler & Spelke, 2011). Similarly, 2-year-olds give toys equally to own- and other-race individuals (Kinzler & Spelke, 2011), and 3- to 4-year-olds show no bias for own- or other-race peers when choosing a playmate or friend (Abel & Sahinkaya, 1962). However, by 5 years of age, children tend to choose own-race over other-race individuals as playmates or friends (Abel & Sahinkaya, 1962; Kinzler, Shutts, DeJesus, & Spelke, 2009; Kinzler & Spelke, 2011). Downloaded from cdp.sagepub.com at UNIV NEBRASKA LIBRARIES on August 22, 2013 Anzures et al. 176 Although 5-year-olds prefer own-race social partners when they are provided only with visual cues (i.e., images of faces), they tend to choose other-race children portrayed as speaking their native language in their native accent over own-race children speaking their native language with a foreign accent (Kinzler et al., 2009). Thus, children’s social interactions are influenced by their evaluations of in-group/out-group membership, with race being only one of several factors used in their evaluation. Children between 5 and 7 years of age show more positive evaluations (e.g., clean, good, friendly) of ownrace children and show more negative evaluations (e.g., dirty, naughty, unfriendly) of other-race children, whereas younger children show no such biases (Aboud, 2003). However, more contact with other-race individuals is related to less prejudiced attitudes in adults (see also Meissner & Brigham, 2001, for a review) and children (Binder et al., 2009). In addition, mere visual exposure to photographs of other-race faces increases adults’ ratings of likeability for novel faces from the same other-race group (Zebrowitz, White, & Wieneke, 2008). Although there is no direct association between the ORE and selfreports of racial attitudes (see Meissner & Brigham, 2001 for a review), a reduction in the ORE via training in individuating between other-race faces is associated with a reduction in implicit racial biases (Lebrecht, Pierce, Tarr, & Tanaka, 2009). Conclusions and Future Directions The early development of the ORE is shaped and maintained by asymmetries in experience with own- and other-race individuals. This asymmetry in experience likely influences later race-based social preferences and attitudes that emerge at around 5 years of age. The developmental gap between the ORE in face recognition and race-based social preferences and attitudes is likely due to the cognitive demands involved in forming in-group/ out-group categories that are more conceptually based than earlier perceptually based race categories. Nevertheless, experience with other-race individuals can broaden one’s face recognition expertise and improve attitudes about other-race individuals. However, if otherrace experience ceases, the long-lasting effects of such experience on face recognition expertise and social attitudes remain unclear. Timing of exposure may play a crucial role, with earlier exposure leading to longer lasting effects in face recognition. Indeed, primary experience with own-race individuals for only 2 to 26 months after birth followed by a lack of experience with ownrace individuals is sufficient to nonetheless maintain proficiency in recognizing own-race faces at 6 to 14 years of age (de Heering et al., 2010). Furthermore, temporary experience with other-race individuals may require minimal subsequent reexposure to regain advantages in other-race face recognition. In the case of social attitudes, early childhood experience with other-race social partners combined with positive views of other racial groups in the social environment may help to prevent the development of prejudiced attitudes (Bigler & Liben, 2007). Acknowledgments This research was supported by National Institutes of Health (NIH) Grant R01 HD046526. This research was also supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) grant and a National Natural Science Foundation of China (NSFC) grant to Kang Lee, and an NSERC Postdoctoral Fellowship to Gizelle Anzures. Recommended Reading Anzures, G., Kelly, D. J., Pascalis, O., Quinn, P. C., Slater, A. M., de Viviés, X., & Lee, K. (in press). Own- and other-race face identity recognition in children: The effects of pose and feature composition. Developmental Psychology. Influence of face pose and feature composition on the other-race effect in children, and stability of the size of the other-race effect during childhood. Balas, B. (2012). Bayesian face recognition and perceptual narrowing in face-space. Developmental Science, 15, 579–588. Face recognition and development of the other-race effect in a Bayesian framework. Hugenberg, K., Young, S. G., Bernstein, M. J., & Sacco, D. F. (2010). The categorization-individuation model: An integrative account of the other-race recognition deficit. Pyscho­ logical Review, 117, 1168–1187. Reviews theories accounting for the ORE, and examines the influence of categorization, motivated individuation, and perceptual experience on the ORE. Macchi Cassia, V., Kuefner, D., Picozzi, M., & Vescovo, E. (2009). Early experience predicts later plasticity for face processing. Psychological Science, 20, 853–859. Role of early experience on children’s and adults’ recognition of infant faces. Quinn, P. C., Anzures, G., Lee, K., Pascalis, O., Slater, A., & Tanaka, J. W. (2013). On the developmental origins of differential responding to social category information. In M. R. Banaji & S. A. Gelman (Eds.), Navigating the social world: What infants, children, and other species can teach us (pp. 286–291). New York: Oxford University Press. Influence of differential experience with face categories (e.g., gender and race) on infants’ categorization and recognition of faces. Declaration of Conflicting Interests The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article. References Abel, H., & Sahinkaya, R. (1962). Emergence of sex and race friendship preferences. Child Development, 33, 939–943. Downloaded from cdp.sagepub.com at UNIV NEBRASKA LIBRARIES on August 22, 2013 Developmental Origins of the ORE 177 Aboud, F. E. (2003). The formation of in-group favouritism and out-group prejudice in young children: Are they distinct attitudes? Developmental Psychology, 39, 48–60. Anzures, G., Quinn, P. C., Pascalis, O., Slater, A. M., & Lee, K. (2010). Categorization, categorical perception, and asymmetry in infants’ representation of face race. Developmental Science, 13, 553–564. Anzures, G., Wheeler, A., Quinn, P. C., Pascalis, O., Slater, A. M., Heron-Delaney, M., . . . Lee, K. (2012). Brief daily exposures to Asian females reverses perceptual narrowing for Asian faces in Caucasian infants. Journal of Experimental Child Psychology, 112, 484–495. Argyle, M., & Cook, M. (1976). Gaze and mutual gaze. Oxford, England: Cambridge University Press. Bar-Haim, Y., Ziv, T., Lamy, D., & Hodes, R. M. (2006). Nature and nurture in own-race face processing. Psychological Science, 17, 159–163. Bigler, R. S., & Liben, L. S. (2007). Developmental intergroup theory: Explaining and reducing children’s social stereotyping and prejudice. Current Directions in Psychological Science, 16, 162–166. Binder, J., Zagefka, H., Brown, R., Funke, F., Kessler, T., & Mummendey, A. (2009). Does contact reduce prejudice or does prejudice reduce contact? A longitudinal test of the contact hypothesis among majority and minority groups in three European countries. Journal of Personality and Social Psychology, 96, 843–856. Blais, C., Jack, R. E., Scheepers, C., Fiset, D., & Caldara, R. (2008). Culture shapes how we look at faces. PLoS ONE, 3, e3022. doi:10.1371/journal.pone.0003022 de Heering, A., de Liedekerke, C., Deboni, M., & Rossion, B. (2010). The role of experience during childhood in shaping the other-race effect. Developmental Science, 13, 181–187. Ferguson, K. T., Kulkofsky, S., Cashon, C. H., & Casasola, M. (2009). The development of specialized processing of ownrace faces in infancy. Infancy, 14, 263–284. Fu, G., Hu, C. S., Wang, Q., Quinn, P. C., & Lee, K. (2012). Adults scan own- and other-race faces differently. PLoS ONE, 7, e37688. doi:10.1371/journal.pone.0037688 Gaither, S. E., Pauker, K., & Johnson, S. P. (2012). Biracial and monoracial infant own-race face perception: An eye tracking study. Developmental Science, 15, 775–782. Ge, L., Zhang, H., Wang, Z., Quinn, P. C., Pascalis, O., Kelly, D., . . . Lee, K. (2009). Two faces of the other-race effect: Recognition and categorisation of Caucasian and Chinese faces. Perception, 38, 1199–1210. Goldstein, A. G., & Chance, J. E. (1985). Effects of training on Japanese face recognition: Reduction of the other-race effect. Bulletin of the Psychonomic Society, 23, 211–214. Goodman, G. S., Sayfan, L., Lee, J. S., Sandhei, M., Walle-Olsen, A., Magnussen, S., . . . Arredondo, P. (2007). The development of memory for own- and other-race faces. Journal of Experimental Child Psychology, 98, 233–242. Hayden, A., Bhatt, R. S., Joseph, J. E., & Tanaka, J. W. (2007). The other-race effect in infancy: Evidence using a morphing technique. Infancy, 12, 95–104. Heron-Delaney, M., Anzures, G., Herbert, J. S., Quinn, P. C., Slater, A. M., Tanaka, J. W., . . . Pascalis, O. (2011). Perceptual training prevents the emergence of the other race effect during infancy. PLoS ONE, 6(5), e19858. doi:10.1371/ journal.pone.0019858 Hills, P. J., & Lewis, M. B. (2006). Reducing the own-race bias in face recognition by shifting attention. Quarterly Journal of Experimental Psychology, 59, 996–1002. Kelly, D. J., Liu, S., Ge, L., Quinn, P. C., Slater, A. M., Lee, K., . . . Pascalis, O. (2007). Cross-race preferences for samerace faces extend beyond the African versus Caucasian contrast in 3-month-old infants. Infancy, 11, 87–95. Kelly, D. J., Liu, S., Lee, K., Quinn, P. C., Pascalis, O., Slater, A. M., & Ge, L. (2009). Development of the other-race effect during infancy: Evidence toward universality? Journal of Experimental Child Psychology, 104, 105–114. Kelly, D. J., Quinn, P. C., Slater, A. M., Lee, K., Ge, L., & Pascalis, O. (2007). The other-race effect develops during infancy. Psychological Science, 18, 1084–1089. Kelly, D. J., Quinn, P. C., Slater, A. M., Lee, K., Gibson, A., Smith, M., . . . Pascalis, O. (2005). Three-month-olds, but not newborns, prefer own-race faces. Developmental Science, 8, F31–F36. Kinzler, K. D., Shutts, K., DeJesus, J., & Spelke, E. S. (2009). Accent trumps race in guiding children’s social preferences. Social Cognition, 27, 623–634. Kinzler, K. D., & Spelke, E. S. (2011). Do infants show social preferences for people differing in race? Cognition, 119, 1–9. Lebrecht, S., Pierce, L. J., Tarr, M. J., & Tanaka, J. W. (2009). Perceptual other-race training reduces implicit racial bias. PLoS ONE, 4, e4215. doi:10.1371/journal.pone.0004215 Li, H. Z. (2004). Culture and gaze direction in conversation. RASK, 20, 3–26. Liu, S., Quinn, P. C., Wheeler, A., Xiao, N., Ge, L., & Lee, K. (2011). Similarity and difference in the processing of sameand other-race faces as revealed by eye tracking in 4- to 9-month-olds. Journal of Experimental Child Psychology, 108, 180–189. Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigating the own-race bias in memory for faces: A metaanalytic review. Psychology, Public Policy, and Law, 7, 3–35. Pezdek, K., Blandon-Gitlin, I., & Moore, C. (2003). Children’s face recognition memory: More evidence for the cross-race effect. Journal of Applied Psychology, 88, 760–763. Rhodes, G., Locke, V., Ewing, L., & Evangelista, E. (2009). Race coding and the other-race effect in face recognition. Perception, 38, 232–241. Sangrigoli, S., & de Schonen, S. (2004). Recognition of own-race and other-race faces by three-month-old infants. Journal of Child Psychology and Psychiatry, 45, 1219–1227. Sangrigoli, S., Pallier, C., Argenti, A. M., Ventureyra, V. A. G., & de Schonen, S. (2005). Reversibility of the other-race effect in face recognition during childhood. Psychological Science, 16, 440–444. Scott, L. S., & Monesson, A. (2009). The origin of biases in face perception. Psychological Science, 20, 676–680. Scott, L. S., Pascalis, O., & Nelson, C. A. (2007). A domain-general theory of the development of perceptual discrimination. Downloaded from cdp.sagepub.com at UNIV NEBRASKA LIBRARIES on August 22, 2013 Anzures et al. 178 Current Directions in Psychological Science, 16, 197– 201. Shutts, K., & Kinzler, K. D. (2007). An ambiguous-race illusion in children’s face memory. Psychological Science, 18, 763–767. Tanaka, J. W., & Pierce, L. J. (2009). The neural plasticity of other-race face recognition. Cognitive, Affective, & Behavioral Neuroscience, 9, 122–131. Valentine, T. (1991). A unified account of the effects of distinctiveness, inversion, and race in face recognition. Quarterly Journal of Experimental Psychology, 43, 161–204. Walker, P. M., & Hewstone, M. (2006). A perceptual discrimination investigation of the own-race effect and intergroup experience. Applied Cognitive Psychology, 20, 461–475. Wheeler, A., Anzures, G., Quinn, P. C., Pascalis, O., Omrin, D. S., & Lee, K. (2011). Caucasian infants scan ownand other-race faces differently. PLoS ONE, 6, e18621. doi:10.1371/journal.pone.0018621 Zebrowitz, L. A., White, B., & Wieneke, K. (2008). Mere exposure and racial prejudice: Exposure to other-race faces increases liking for strangers of that race. Social Cognition, 26, 259–275. Downloaded from cdp.sagepub.com at UNIV NEBRASKA LIBRARIES on August 22, 2013 PS YC HOLOGICA L SC IENCE Research Article The Development of Implicit Attitudes Evidence of Race Evaluations From Ages 6 and 10 and Adulthood Andrew Scott Baron and Mahzarin R. Banaji Harvard University ABSTRACT—To understand the origin and development of implicit attitudes, we measured race attitudes in White American 6-year-olds, 10-year-olds, and adults by first developing a child-oriented version of the Implicit Association Test (Child IAT). Remarkably, implicit pro-White/ anti-Black bias was evident even in the youngest group, with self-reported attitudes revealing bias in the same direction. In 10-year-olds and adults, the same magnitude of implicit race bias was observed, although self-reported race attitudes became substantially less biased in older children and vanished entirely in adults, who self-reported equally favorable attitudes toward Whites and Blacks. These data are the first to show an asymmetry in the development of implicit and explicit race attitudes, with explicit attitudes becoming more egalitarian and implicit attitudes remaining stable and favoring the in-group across development. We offer a tentative suggestion that mean levels of implicit and explicit attitudes diverge around age 10. How early in development are implicit attitudes toward social groups formed? What is the developmental pattern of the relationship between such attitudes and those that are consciously expressed? When does the dissociation between the two observed in adults emerge in young children? In this article, we report the first evidence of the development of implicit and explicit attitudes toward social and nonsocial groups using three age groups. The presence of implicit forms of attitudes in adults has been well demonstrated, as has the ability to use such at- Address correspondence to Andrew Scott Baron or Mahzarin R. Banaji, Department of Psychology, Harvard University, 33 Kirkland St., Cambridge, MA 02138, e-mail: barona@wjh.harvard.edu or mahzarin_banaji@harvard.edu. Volume 17—Number 1 titudes to predict a wide range of behaviors, including friendliness toward out-groups, selection for a job, and allocation of resources (see Poehlman, Uhlmann, Greenwald, & Banaji, 2005, for a review). Understanding the development of implicit attitudes in young children is imperative given the important role intergroup attitudes play throughout life. Moreover, investigating the nature of implicit social cognition in children provides an opportunity to understand the social-cognitive mechanisms that are universal and the cultural processes that mark the development of these attitudes and preferences. Creating a modified, child-friendly version of the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998), first introduced here as the Child IAT, we measured implicit race attitudes in white North American middle-class children. We selected race as the social category because of evidence that North American children achieve an adultlike concept of this category by age 5 (Hirschfeld, 1996, 2001). In a series of studies, Hirschfeld showed that children as young as 4 do not rely on perceptual information alone when categorizing people. Instead, children appear to essentialize racial kinds, regarding race as a property that is fixed at birth and resistant to change across time and surface features, and even believe it to be predictive of nonobvious properties. In other words, children’s concept of race may be commensurate with that of adults (cf. Allport, 1954). In the present study, we investigated whether kindergartners (5- and 6-year-olds) have implicit attitudes toward race categories soon after the age at which they are expected to have achieved a mature representation of the concept of race. Aboud (1988) showed that self-reports at this age reveal evaluative assessments, or attitudes, associated with racial categories. White North American children begin to report negative explicit attitudes toward out-group members as early as age 3; such attitudes begin to decline by age 7, until they disappear around Copyright r 2006 Association for Psychological Science 53 Development of Implicit Attitudes age 12. What is unknown is how the parallel development of automatic (implicit) associations of good and bad attributes with racial categories unfolds. We tracked implicit race attitudes also in 10-year-olds, as well as adults, to view the developmental progression of such attitudes cross-sectionally. Much has been learned about adults’ implicit attitudes using the IAT (Banaji, 2001; Greenwald et al., 1998; Lane & Banaji, 2004; Nosek, Banaji, & Greenwald, 2002); therefore, this sample also provided a benchmark for testing the new child version of the IAT. To allow more confident interpretation of the results, we also included a measure of implicit attitudes toward nonsocial categories (insects and flowers). Because flowers are known to elicit more positive implicit attitudes than insects in most people (Greenwald et al., 1998), if the insect-flower Child IAT revealed the expected attitude effect, a potential null result on the race test among children could be interpreted as a genuine lack of race bias, rather than a failure of the new measure to detect an effect. METHOD Participants The sample consisted of 79 participants (39 males, 40 females): 27 kindergartners (mean age 5 6 years 1 month; 14 males, 13 females); 30 fifth graders (mean age 5 10 years 2 months; 15 males, 15 females); and 22 adults (mean age 5 19 years; 10 males, 12 females). Participants were recruited from a predominantly middle-class European American community. Children were tested in an elementary school in a Boston suburb; adults were tested in a laboratory at Harvard University. Procedure The IAT The IAT measures the relative strength of association between a target concept (e.g., race: African American and European American) and an attribute concept (e.g., evaluation: words with good meanings and words with bad meanings). The IAT is a response latency measure that rests on an assumption it shares with other measures of associative strength—that the more strongly two concepts have come to be associated with one another, the faster and more accurately they can be paired together (see Banaji, 2001, for a comparison with other measures). In a typical procedure used with adults, participants first practice classifying stimuli in terms of a target concept such as race or gender. For example, pictures of Black and White Americans, appearing one at a time in the middle of the screen, are classified using two keys (typically the ‘‘E’’ and ‘‘I’’ keys) on a computer keyboard. Participants press one key in response to all pictures of Black Americans and press the other key in response to all pictures of White Americans. Trials advance only following correct responses, to encourage low error rates. Participants next practice classifying stimuli in terms of an attribute concept that has two categories. For example, if eval- 54 uation is the attribute dimension, words with good or bad meaning (e.g., love, joy, friend, hate, vomit, bomb) appear one at a time in the middle of the screen, and participants press one key in response to words with a good meaning and press the other key in response to words with a bad meaning. These single-dimension tasks serve to familiarize participants with the target and attribute dimensions and the stimulus set. In the next block of trials, the strength of the association between the target concept (e.g., race) and the attribute concept (e.g., evaluation) is measured. These trials require categorizing the four classes of items using two keys, with one target and one attribute category sharing each response key. Participants are presented with a total of 60 trials (20 practice trials, followed by 40 critical trials) in which they view faces of African Americans and European Americans and good and bad attribute words in equal numbers (15 trials of each stimulus type). Stimuli are presented one at a time. In one block of trials, target concept A is paired with attribute concept A (e.g., ‘‘When you see a Black face or a good word, press the ‘E’ key’’), and target concept B is paired with attribute concept B (e.g., ‘‘When you see a White face or a bad word, press the ‘I’ key’’). Then, the target concepts switch location, such that target concept B is paired with attribute A (e.g., White face and good word), and target concept A is paired with attribute B (e.g., Black face and bad word). The assumption is that the stronger these associations, the faster and more accurately participants will respond in the second block compared with the first. Readers interested in sampling this task may visit www.implicit. harvard.edu. A response latency is recorded for each trial by measuring the time from the onset of the stimulus until a response (correct or incorrect) is entered. Each trial advances following a correct response, and there is a 1-s intertrial interval. The order of targetattribute pairings is counterbalanced between subjects so that order of blocks does not interfere with interpretation of the result. We made several modifications to the standard IAT so that it would be suitable for use with children. The IAT typically uses faces to denote race. We used pictures of Black and White children’s faces. Because of the variability in reading level among children, we substituted voice recordings of good and bad words for printed words. Recordings of the attribute words were made by an adult female and were presented auditorally through speakers built into the computer monitor. Thus, participants were instructed to press one button when they heard a good word and to press the other button when they heard a bad word. For the same reason, all instructions were spoken by the experimenter. Response latencies to all stimuli, pictures and auditory stimuli, were recorded, as were errors in classification. Response latencies for the attribute words were recorded after the full words were spoken. Eight target stimuli were used for each Child IAT. The insectflower test included four pictures of insects and four pictures of Volume 17—Number 1 Andrew Scott Baron and Mahzarin R. Banaji flowers, and the race test included four pictures of European American children and four pictures of African American children. The eight attribute stimuli consisted of four words capturing a good concept (good, nice, fun, happy) and four capturing a bad concept (bad, mean, yucky, mad ); these eight stimuli were used in both Child IATs. We chose words that appear frequently in young children’s vocabulary. Children were introduced to the task as a ‘‘computer game’’ in which they would see pictures and hear words and would have to press a button in response to each. Although all participants were tested individually, the experimenter remained in the room with child participants but not with adults. For the children, motor responses were facilitated by using two large JellyBeans buttons (3-in. diameter) instead of the ‘‘E’’ and ‘‘I’’ computer keys traditionally used with adults. All other aspects of the procedure were identical for adults and children. The insectflower Child IAT was administered first, followed by the race Child IAT. Explicit Attitude Measure: Self-Reported Preference Following the Child IAT, participants viewed a series of paired pictures, presented side-by-side, and provided forced-choice preference judgments. The pairs consisted of same-race children, different-race children (i.e., one White child and one Black child), insects, flowers, and insect-flower pairs (i.e., one insect and one flower). On critical trials, a picture of a Black child and a picture of a White child were paired, and participants indicated whom they preferred. The pictures used in the explicit attitude measure were the same pictures used in the implicit attitude measure. Unlike in the Child IAT, participants were encouraged to take their time and to deliberate over their responses. RESULTS AND DISCUSSION We analyzed the implicit attitude measure following standard protocol for the improved scoring algorithm recommended by Greenwald, Nosek, and Banaji (2003). Two participants in the 6year-old group were unable to complete the race Child IAT; they were included only in analyses of the insect-flower attitude data. For each subject, an IAT score in the form of a measure termed D, a variant of Cohen’s d (see Greenwald et al., 2003), was computed by calculating the difference between the mean response latencies for the two double-categorization blocks within each Child IAT and dividing that difference by its associated pooled standard deviation. Because of a difference in response latency as a function of type of stimulus presentation (pictures vs. spoken words) within each double-categorization block, we calculated separate IAT effects for responses to target stimuli and for responses to attribute stimuli and then averaged them to produce one score for each of the combined blocks. A multivariate analysis of variance (MANOVA) revealed no significant Volume 17—Number 1 main effects of age or order (White 1 Good/Black 1 Bad first or White 1 Bad/Black 1 Good first) on the implicit measure of attitude. Additionally, no significant age-by-order interaction was observed (all ps > .2). 6-Year-Olds Insect-Flower Attitudes Not only were the youngest children in the study able to complete the Child IAT, but an implicit attitude was clearly detected. Six-year-olds were significantly faster to respond to insect 1 bad/flower 1 good trials than insect 1 good/flower 1 bad trials (mean difference 5 109 ms), D 5 0.22, SD 5 0.40, t(26) 5 2.86, p < .01. Although boys showed this preference for flowers over insects to a lesser extent than did girls, the gender difference was not statistically significant. Similarly, 6-year-olds self-reported a clear preference for flowers over insects (77% of the time, participants chose a flower over an insect), t(23) 5 3.24, p < .01. This explicit attitude effect was driven largely by females; females reported such a preference on 96% of the trials, but males preferred flowers on 43% of the trials, t(22) 5 4.02, p < .01. The presence of a gender difference in self-reported attitude, but not in implicit attitude, suggests that by age 6, children’s consciously expressed attitudes may be more exaggerated along gender lines than implicit attitudes for the same attitude objects. Race Attitudes As Figure 1 shows, the 6-year-olds had already developed implicit pro-White/anti-Black associations, observed in faster responding on White 1 good/Black 1 bad trials than Black 1 good/White 1 bad trials (mean difference 5 79 ms). The average IAT effect was significant, D 5 0.22, SD 5 0.24, t(24) 5 4.48, p < .001. These data are the first to reveal the emergence Fig. 1. Implicit race preference in the three age groups. A positive value of D indicates a preference for Whites relative to Blacks. 55 Development of Implicit Attitudes Similarly, 10-year-olds also revealed an explicit preference for Whites over Blacks (68% of the time, they chose the White child over the Black child), t(29) 5 4.13, p < .01, but this preference was significantly more muted than that reported by 6-year-olds (68% vs. 84%, respectively), t(50) 5 2.27, p 5 .027. In other words, although 6- and 10-year-olds showed the same magnitude of implicit race bias, by age 10 children’s selfreported preference for their own group was significantly reduced (see Figs. 1 and 2). Adults Fig. 2. Explicit race preference in the three age groups. of implicit attitudes toward social groups in children as young as 6 years of age. Six-year-olds’ explicit race attitudes were consonant with their implicit attitudes. They self-reported a strong preference for photographic images of White compared with Black children (84% of the time, a picture of a White child was selected over that of a Black child), t(21) 5 6.38, p < .01 (see Fig. 2). Both males and females reported a preference for Whites over Blacks, but there was a significant gender difference (93% vs. 70%, respectively), t(20) 5 2.38, p 5 .03. Insect-Flower Attitudes Replicating the result from many studies using the standard IAT, adults were faster to respond to flower 1 good trials than to insect 1 good trials on the Child IAT (mean difference 5 138 ms), D 5 0.49, SD 5 0.46, t(21) 5 4.98, p < .001. Similarly, adults self-reported a strong preference for flowers over insects (86% of the time, participants chose insects over flowers), t(21) 5 5.43, p < .01, with no gender difference observed. Race Attitudes Adults showed the same implicit pro-White/anti-Black response bias on the race Child IAT as child participants did (mean difference 5 89 ms), D 5 0.22, SD 5 0.41, t(21) 5 2.50, p 5 .021. However, adults self-reported an equal preference for White and Black targets (46% of the time, participants chose the White child over the Black child), t(21) 5 0.672, p 5 .51 (see Figs. 1 and 2). GENERAL DISCUSSION 10-Year-Olds Insect-Flower Attitudes Like 6-year-olds, 10-year-olds were faster to respond to flower 1 good/insect 1 bad trials than to insect 1 good/flower 1 bad trials (mean difference 5 117 ms), D 5 0.30, SD 5 0.50, t(29) 5 3.30, p < .01. Ten-year-olds showed the same pattern of preference on the explicit task as on the Child IAT, choosing flowers over insects 67% of the time, t(29) 5 2.14, p 5 .04. As with the 6-year-olds, a gender difference in reported preference emerged; females were more likely than males to choose flowers over insects (88% vs. 45%, respectively), t(28) 5 3.19, p < .01. Race Attitudes Ten-year-olds were faster to respond on White 1 good/Black 1 bad trials than on Black 1 good/White 1 bad trials (mean difference 5 80 ms), D 5 0.22, SD 5 0.26, t(29) 5 4.58, p < .001. Ten-year-olds and 6-year-olds did not differ in mean levels of implicit race attitudes, which suggests that these attitudes remain stable during the elementary-school years. 56 Taken together, these data show the early emergence of implicit attitudes toward both nonsocial (flower vs. insect) and social (Black vs. White) categories. By age 6, children appear to have formed detectable implicit attitudes toward social groups. Moreover, these attitudes did not vary across the three age groups studied here. Yet for self-reported race attitudes, a quite distinct pattern emerges. An early and strong preference for members of one’s own social group subsides by age 10 and levels off to an equal preference for the in-group and out-group by adulthood. That this dissociation between implicit and explicit attitudes was not observed at an earlier age raises the question of whether or not such implicit-explicit dissociations are even possible in younger children, whose conscious and less conscious attitudes may be more unified in valence than is the case for older children and adults. Note, however, that on the insect-flower test, 6-year-old boys implicitly preferred insects to flowers, but explicitly showed no preference. That such a dissociation was observed suggests that implicit and explicit attitudes need not be congruent at this young age. Volume 17—Number 1 Andrew Scott Baron and Mahzarin R. Banaji What is one to make of these first findings on the development of race attitudes, and especially the dissociation between patterns of implicit and explicit attitudes across age? Should the data be interpreted as revealing general implicit in-group preference (i.e., any group of children tested would show an effect favoring their own group) or an effect that is peculiar to a dominant group’s implicit preference, and therefore not likely to be mimicked by members of minority groups? Although this issue cannot be definitively resolved here, we do offer a few observations from previous research on adults and children. First, substantial data on adult Black Americans (n > 5,000) indicate that, on average, they lack an implicit in-group preference, instead showing no bias in favor of one or the other racial group, even though they report strong in-group liking on selfreport measures (Nosek et al., 2002). Second, Baron, Shusterman, Bordeaux, and Banaji (2004) measured race attitudes in 12- to 14-year-old Black Americans who lived and attended school in Bronx, New York, and replicated the pattern found for Black adults. In other words, at least by age 13, young Black Americans do not show the in-group preference that has come to be the hallmark of White Americans, close to 80% of whom show some degree of in-group preference on the IAT. To date, we have interpreted the relative lack of in-group bias in adult Black Americans as revealing a culturally driven modulation of the default in-group bias. Group membership pushes in the direction of in-group positivity, but that positivity is modulated by the countervailing force of the evaluation of the group in the eyes of the broader culture. That evaluation then ‘‘becomes’’ the implicit attitude of group members. The best next step for research on this issue would be to test a sample of Black American children, matched to the present sample in age, but coming from a predominantly Black community. If Black 6-yearolds reveal the same pattern as the White 6-year-olds in this study, showing strong preference for their own group, this would provide support for the idea that in-group bias is the default, with shifts even by age 10 reflecting an internalization of the attitudes of the larger culture. However, if the obtained result reveals that Black 6-year-olds show an effect that resembles that of adolescent and adult Black Americans (i.e., no preference for the in-group over the out-group), this would suggest that by age 6, the typical in-group preference is modulated by knowledge of the group’s standing in the more broadly based sociocultural hierarchy. Dunham, Baron, and Banaji (2004) reported that Hispanic children as young as 5 show an in-group preference for Hispanic over Black, but show no preference for Hispanic over White, which suggests that implicit intergroup attitudes are learned quite early, and that children who come from disadvantaged groups experience the lower attitudinal status of their own group. In a recent article, Olsson, Ebert, Banaji, and Phelps (2005) reported that both Black and White adult Americans show quicker extinction to fear conditioning involving own-race faces than to fear conditioning involving other-race faces. Olsson Volume 17—Number 1 et al. took this finding as indicating that group membership plays a robust role in attitudes, at least those that involve classical conditioning as the learning mechanism. The factor that mediated the slower extinction to out-group fear was romantic contact—participants who had had romantic relationships with outgroup members were less likely than others to show this persistence of fear learning toward out-group members. Analyses of the tenacity and plasticity of intergroup attitudes across the life span will be crucial in building a proper understanding of the origins of prejudice. What about the role of familiarity in producing the obtained effects? There is little doubt that familiarity plays a role in attitude development—what is familiar is more liked than what is unfamiliar (Cutting, 2003; Zajonc, 1968), and what is liked becomes more familiar because preference presumably leads to greater seeking of contact. However, Dasgupta, McGhee, Greenwald, and Banaji (2000; also see Dasgupta, Greenwald, & Banaji, 2003) ruled out familiarity as the dominant explanation of IAT effects by showing (a) preference for low-familiarity but positive stimuli over high-familiarity but negative stimuli and (b) preference effects that remain even after statistically controlling for familiarity effects item by item. However, in young children, it is quite possible that attitudes, both implicit and explicit, may indeed rely more on familiarity than on preference, and future tests of this possibility will be important. It will be relatively easy to create studies in which children are familiarized with otherwise novel social groups, so that it will be possible to observe potential changes in implicit attitudes that are uncontaminated by existing knowledge of who is good and less good (Baron, Dunham, & Banaji, 2005). Likewise, field studies in schools with broad diversity in ethnicity, class, culture, and nationality will also provide useful data. The present data demonstrate that implicit attitudes can be measured in children using the Child IAT. There is no doubt that this measure will continue to be improved in subsequent studies, in particular, to make it available for use with younger samples. The basic procedure as described here is available for download by investigators interested in understanding a host of implicit attitudes in young children. The most recent procedures and data-analytic suggestions may be found at www.people.fas. harvard.edu/  banaji. In conclusion, the evidence from this and related studies completed in our laboratory suggests that implicit race attitudes are acquired early and remain relatively stable across development, even though explicit attitudes become more egalitarian. It is around age 10 that the split between mean levels of conscious and less conscious race attitudes first emerges, pointing out the differential sensitivity of these two forms of attitude to the societal demand to be unbiased in race-based evaluation. Acknowledgments—This work was supported by grants and fellowships from the National Institute of Mental Health, the 3rd 57 Development of Implicit Attitudes Millennium Foundation, and the Radcliffe Institute for Advanced Study to Mahzarin R. Banaji. We wish to thank Yarrow Dunham, Elizabeth Spelke, and Kristina Olson for comments on an earlier draft and members of the Social Cognition Laboratory and the Laboratory for Development Studies at Harvard for their support. We also thank Roy Ruhling and Jessica Massa for their assistance in producing the manuscript. Finally, we wish to extend our gratitude to Janis Baron for facilitating this data collection. REFERENCES Aboud, F.E. (1988). Children and prejudice. Cambridge, MA: Basil Blackwell. Allport, G. (1954). The nature of prejudice. Cambridge, MA: AddisonWesley. Banaji, M.R. (2001). Implicit attitudes can be measured. In H.L. Roediger, III, J.S. Nairne, I. Neath, & A. Surprenant (Eds.), The nature of remembering: Essays in honor of Robert G. Crowder (pp. 117–150). Washington, DC: American Psychological Association. Baron, A.S., Dunham, Y., & Banaji, M.R. (2005, October). Children’s representations of social groups. Poster presented at the biennial meeting of the Cognitive Development Society, San Diego, CA. Baron, A.S., Shusterman, A., Bordeaux, A., & Banaji, M.R. (2004, January). Implicit race attitudes in African-American and Hispanic children. Poster presented at the annual meeting of the Society for Personality and Social Psychology, Austin, TX. Cutting, J.E. (2003). Gustave Caillebotte, French Impressionism, and mere exposure. Psychonomic Bulletin & Review, 10, 319–343. Dasgupta, N., Greenwald, A.G., & Banaji, M.R. (2003). The first ontological challenge to the IAT: Attitude or mere familiarity? Psychological Inquiry, 14, 238–243. Dasgupta, N., McGhee, D.E., Greenwald, A.G., & Banaji, M.R. (2000). Automatic preference for White Americans: Eliminating the familiarity explanation. Journal of Experimental Social Psychology, 36, 316–328. 58 Dunham, Y., Baron, A.S., & Banaji, M.R. (2004, September). Exploring the relationship between self-identity, self-esteem, and intergroup attitudes in Hispanic-American children. Poster presented at the annual meeting of the New England Social Psychological Association, Storrs, CT. Greenwald, A.G., McGhee, D.E., & Schwartz, J.K.L. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464– 1480. Greenwald, A.G., Nosek, B.A., & Banaji, M.R. (2003). Understanding and using the Implicit Association Test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85, 197– 216. Hirschfeld, L.A. (1996). Race in the making: Cognition, culture, and the child’s construction of human kinds. Cambridge, MA: MIT Press. Hirschfeld, L.A. (2001). On a folk theory of society: Children, evolution, and mental representations of social groups. Personality and Social Psychology Review, 5(2), 107–117. Lane, K.A., & Banaji, M.R. (2004). Evaluative group status and implicit attitudes toward the ingroup. In R.K. Ohme & M. Jarymowica (Eds.), Natura automatyzmow (pp. 25–30). Warsaw, Poland: WIP PAN & SWPS. Nosek, B.A., Banaji, M.R., & Greenwald, A.G. (2002). Harvesting intergroup attitudes and stereotypes from a demonstration website. Group Dynamics, 6, 101–115. Olsson, A., Ebert, J.P., Banaji, M.R., & Phelps, E.A. (2005). The role of social groups in the persistence of learned fear. Science, 309, 785–787. Poehlman, T.A., Uhlmann, E., Greenwald, A.G., & Banaji, M.R. (2005). Understanding and using the Implicit Association Test: III. Metaanalysis of predictive validity. Unpublished manuscript, Yale University, New Haven, CT. Zajonc, R.B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 9(2), 1–27. (RECEIVED 11/29/04; REVISION ACCEPTED 7/14/05; FINAL MATERIALS RECEIVED 7/27/05) Volume 17—Number 1 INTERNATIONAL JOURNAL OF PSYCHOLOGY, 2009, 44 (1), 29–39 The development of intergroup bias in childhood: How social norms can shape children’s racial behaviours Maria Benedicta Monteiro Dalila Xavier de França ISCTE, Department of Social and Organizational Psychology, and Research Centre for Social Research and Intervention, Lisboa, Portugal Universidade Federal de Sergipe, Aracajú, Brasil Ricardo Rodrigues ISCTE, Department of Social and Organizational Psychology, and Research Centre for Social Research and Intervention, Lisboa, Portugal T he present research examined the developmental course of racial behaviours in childhood. It tested the hypothesis that White children’s expressions of racial prejudice do not necessarily decline in middle childhood due to the development of particular cognitive skills but that instead, as argued by the socio-normative approach, children older than seven will go on expressing prejudiced attitudes under appropriate conditions. This would be explained by the presence of an anti-racism norm, along with the existence of values promoting equal rights, which impede blatant expressions of racism. In the first study 283 White children aged 6–7 and 9–10 years performed a task of resource allocation to White and Black target children in conditions of high (White interviewer was present) or low (White interviewer was absent) salience of the anti-racist norm. The 6- to 7-yearold children discriminated against the Black target in both conditions whereas older children discriminated against the Black child only when the anti-racist norm was not salient. In Study 2, 187 White children aged 6–7 and 9–10 years performed the same resource allocation task in conditions of explicit activation of similarity vs dissimilarity or egalitarian vs merit-based norms regarding race relations. Supporting the hypothesis of the role of racist or anti-racist norms on the expression of intergroup discrimination, results have again shown that 6- to 7year-old children discriminated against the Black target in both conditions while older children presented significantly different prejudiced/nonprejudiced behaviours consistent with the activated norms. These results were discussed in terms of the need for a reanalysis of the assumptions and research results of the cognitivedevelopmental theory and of further developments in the socio-normative approach regarding the development of prejudice in childhood. L a présente recherche a examiné le cours du développement des comportements raciaux pendant l’enfance. L’étude a testé l’hypothèse que les expressions de préjugés raciaux des enfants blancs ne diminuent pas nécessairement au milieu de l’enfance à cause du développement d’habiletés cognitives particulières mais plutôt, tel qu’avancé par l’approche socio-normative, sous des conditions appropriées, les enfants de plus de 7 ans continueront à exprimer des préjugés. Ceci serait expliqué par la présence d’une norme anti-raciste ainsi que l’existence de valeurs qui promouvoient les droits égaux, ce qui empêche les expressions flagrantes de racisme. Dans la première étude, 283 enfants blancs âgés de 6–7 et de 9–10 ans ont accompli une tâche d’allocation de ressources à des enfants-cibles blancs et noirs dans des conditions de saillance élevée (un interviewer blanc était présent) ou faible (un interviewer blanc était absent) de la norme anti-raciste. Les enfants de 6–7 ans ont Correspondence should be addressed to Maria Benedicta Monteiro, Department of Social and Organizational Psychology, ISCTE, University Institute of Lisbon, Av. Forças Armadas, Edif. ISCTE, 1649-026, Lisboa, Portugal (E-mail: mbbm@iscte.pt). This article was supported by two research grants awarded to the first and third authors (POCTI/PSI/41970/2001 and SFRH/BD/ 16834/2004) by the Portuguese Foundation for Science and Technology and by a grant from CAPES, Brazil, awarded to the second author. The authors are very grateful to Claudia Dalbert (IJP Editor-in-Chief), Jorge Vala (IJP Consulting Editor), Rodrigo Brito, and three anonymous reviewers for their helpful comments on a previous version of this paper. Much appreciation is extended to the teachers and children for their participation. # 2008 International Union of Psychological Science http://www.psypress.com/ijp DOI: 10.1080/00207590802057910 30 MONTEIRO, DE FRANÇA, RODRIGUES discriminé contre la cible noire dans les deux conditions alors que les enfants plus âgés ont discriminé contre l’enfant noir seulement lorsque la norme anti-raciste n’était pas saillante. Dans la deuxième étude, 187 enfants blancs âgés de 6–7 ans et de 9–10 ans ont accompli la même tâche d’allocation de ressources dans des conditions d’activation explicite de similarité vs dissimilarité ou égalitaire vs normes basées sur le mérite en ce qui concerne les relations inter-raciales. En appuyant l’hypothèse du rôle des normes racistes ou anti-racistes dans l’expression de la discrimination inter-groupe, encore une fois les résultats ont indiqué que les enfants âgés de 6–7 ans ont discriminé contre la cible noire dans les deux conditions alors que les enfants plus âgés ont présenté des préjugés/ des non-préjugés en accord avec les normes activées. Ces résultats ont été discutés en termes du besoin d’une reanalyse des hypothèses de base et des résultats de recherche de la théorie cognitivo-développementale et, aussi, en termes de plus de développements de l’approche socio-normative en ce qui concerne le développement des préjugés pendant l’enfance. E l presente estudio investiga el proceso de desarrollo de conductas racistas en la infancia. Se examinó la hipótesis de que las expresiones de prejuicio racial en niños de raza blanca no disminuyen necesariamente durante la niñez media debido al desarrollo de habilidades cognitivas especı́ficas, sino más bien como argumenta la aproximación socio-normativa, bajo determinadas condiciones, niños por encima de los siete años continuarán presentando actitudes de prejuicio. Esto se puede explicar a través de la existencia de una norma antiracista, ası́ como debido a la existencia de valores que promueven la igualdad de derechos, los cuales impiden una expresión abierta de racismo. En el primer estudio 283 niños de raza blanca entre 6–7 y 9–10 años de edad desarrollaron una tarea de adjudicación de recursos en niños de raza blanca y negra en condiciones de alta (entrevistador blanco estuvo presente) y baja saliencia (entrevistador blanco estuvo ausente) de la norma antiracista. Mientras los niños entre 6–7 años discriminaron a los niños de raza negra en ambas condiciones, los niños de mayor edad discriminaron a los niños de raza negra sólo cuando la norma antiracista no fue activada. En un segundo estudio completaron 187 niños de raza blanca entre 6–7 y 9–10 años de edad la misma tarea de adjudicación de recursos en condiciones de activación explı́cita de similitud vs. diferencia o igualdad vs. normas basadas en el mérito respecto de relaciones raciales. La hipótesis del rol de normas racistas y antiracistas en la expresión de discriminación intergrupal fue confirmada. Los resultados mostraron nuevamente que niños entre 6–7 años de edad discriminaron a niños de raza negra en ambas condiciones, mientras que los niños mayores mostraron una diferencia significativa respecto de las conductas prejuiciosas/no prejuiciosas que eran consecuentes con las normas activadoras. Estos resultados se discutieron en término de las necesidades de un reanálisis de las suposiciones teóricas y de los resultados de la investigación en el campo de la teorı́a del desarrollo cognitivo, ası́ como de las nuevas posiciones de la aproximación socio-normativa respecto del desarrollo de prejuicio en la infancia. Keywords: Child development; Intergroup bias; Normative control; Racial discrimination; Social norms. Research over the past 20 years shows that expressions of racism toward disadvantaged ethnic minorities by White persons have become more indirect. This adjustment seems to be related to the presence of an anti-racist norm, along with the existence of values promoting equal rights, which repress blatant expressions of racism (Gaertner & Dovidio, 1986; Pettigrew & Meertens, 1995). Research has also found that consequences of racial prejudice did not significantly decrease as discrimination maintains the same pervasive and negative effects (Dovidio, Kawakami, & Beach, 2001). Most of these studies have been undertaken with adults and only recently have a few analysed the effects of anti-racist norms on children’s prejudiced behaviours. This limitation can largely be due to the general idea in the mainstream social development literature that intergroup bias in early childhood is related more to limitations in children’s cognitive capacities than to the learning and internalization of social norms (Aboud, 1988). The cognitive-developmental theory states that the prejudiced behaviour that children display in middle childhood, as well as the change that occurs during that period, can be explained by the cognitive capacities pertinent to each of the child’s developmental stages (Aboud, 1988; Bigler & Liben, 1993; Doyle & Aboud, 1995). In Aboud’s view, several concrete operational capabilities that emerge in middle childhood, namely conservation, reconciliation of different perspectives, multiple classification, and attention to individual differences within groups contribute to break down children’s over-use of exaggerated homogenous characteristics and thus to reduce prejudice (Aboud & Amato, 2001).Consistent with the cognitive-developmental theory (CDT), there is some evidence that White children in late INTERGROUP BIAS IN CHILDHOOD childhood show fewer negative attitudes toward other groups than younger children do (Aboud & Skerry, 1984; Doyle, Beaudet, & Aboud, 1988; for a review, see Aboud, 2005). Inconsistent with the CDT, however, other research has shown that the developmental sequence described by the theory can hardly account for children’s development of intergroup bias. Specifically, a number of studies have shown that beyond 6 or 7 years, as well as through the period of adolescence, White children and youngsters continue to display intergroup bias (e.g., Abrams, 1989; Katz, Sohn, & Zalk, 1975; Lawrence, 1991; Rutland, 1999). For instance, Katz et al., using a sample of sixth-grade White American children, found no age differences on subtle indicators of intergroup bias against their Black colleagues. They also found a decline in children’s negative assessment scores in which elements of social desirability were obvious. Furthermore, Lawrence reported that White American children aged 6 to 9 years interpreted pictures depicting ambiguous situations involving two White children more positively than the same ambiguous pictures involving two Black children. Using similar ambiguous situation tasks, McGlothlin, Killen, and Edmonds (2005) also found implicit intergroup bias in White American children aged 6–7 and 9–10 years old. In order to shed light on these contradictory results, more recent research within the socionormative approach (Crandall, Eshleman, & O’Brien, 2002; Milner, 1996; Rutland, 2004; Sechrist, Stangor, & Killen, 2005; Sherif, 1936) has explored the role of norm awareness and normative pressure on White children’s expressions of intergroup bias and intergroup-biased behaviour. According to this approach, as they grow older, White children would be more strongly constrained by parents, teachers, and society in general to comply with the prevailing anti-racist norm, namely in public situations, while keeping more or less private prejudiced beliefs and feelings that result from the dominant influences of their more significant in-groups (Sherif, 1936). In support of this approach, Killen, Lee-Kim, McGlothlin, and Stangor (2002) have shown that both younger and older White children were aware of the anti-racist norm that prevents blatant expressions of prejudiced behaviour. Furthermore, Rutland, Cameron, Milne, and McGeorge (2005) have shown that both younger and older White children aged less than 10 years could be externally motivated to control their prejudiced 31 behaviour under high public self-focus, while they simultaneously showed implicit intergroup bias. According to these authors, because of the process of norm internalization that would occur in middle childhood (Abrams, Rutland, & Cameron, 2003; Ruble, Alvarez, Bachman, & Cameron, 2004), and due to self-presentation concerns (Banerjee, 2002; Katz et al., 1975; Lawrence, 1991; Levy & Troise, 2001), older children’s public behaviour actually seems to become less biased. In the same vein, França and Monteiro (2004) manipulated normative salience using the experimenter’s presence (high salience) versus absence (low salience) during a task involving money allocation to an in-group (White) and an out-group (Black) target. Younger children displayed intergroup bias in both conditions while older children only expressed bias in the low salience condition. The authors concluded that, for the older children, the presence of the interviewer created a normative context where intergroup bias was the undesirable behaviour. For the younger children, however, this normative context was not sufficient to influence their consistent pro ingroup bias. The inhibiting effect of the presence of an experimenter on the expression of anti-normative behaviour has already been observed in a number of studies. Specifically, 8- to 9-year-old girls were found to display significantly less anti-normative aggressive behaviour in the presence of an experimenter than in their absence (Lipscomb, 1972), and White research participants’ reduced racial bias in Internet-based studies was found to be due to the presence of the experimenter rather than to conducting the experiment on the laboratory environment (Evans, Garcia, Garcia, & Baron, 2003). Additionally, Rutland et al. (2005) showed that when children were under a high public self-focus condition, they expressed less intergroup bias. However, to our knowledge no study examining the development of ethnic intergroup bias has either manipulated normative pressure as the presence vs absence of the experimenter or has explicitly activated the anti-racist vs racist norms. Accordingly, in Study 1 we hypothesized that 6- to 7-year-old children would display in-group bias regardless of the presence or absence of the experimenter, whereas 9- to 10-year-old children would display the same in-group bias only when the experimenter was absent. In Study 2 it was expected that younger children’s intergroup bias would not be changed by the activation of the racist or anti-racist norms whereas older children 32 MONTEIRO, DE FRANÇA, RODRIGUES would comply with the norm orientation provided by both manipulations. STUDY 1 Method Participants and design Participants were 283 White Portuguese children (125 first graders, aged 6–7 years; 54.2% female; and 158 fourth graders, aged 9–10 years, 53.8% female) attending 15 primary schools of mixed ethnicity (30–40% Black; 60–70% White) in the suburban area of Lisbon. All children were primarily from working-class backgrounds and all were given parental permission to participate in the study. The design was a 2 (age: 6–7, 9–10) 6 2 (interviewer: present, absent) between-subject factorial plan with intergroup bias as the dependent variable. Procedure In order to replicate França and Monteiro’s (2004) study with White Brazilian children, the same procedure was adopted. Each child was individually interviewed at school by a White female interviewer. The interviewer’s presence was believed to make the anti-racist norm salient. Conversely, her absence was believed to create a favourable context for children’s intergroupbiased behaviour to be expressed. The experimental task followed a helping paradigm. Children were asked to distribute nine 1 Euro-coins to two same-sex (White and Black) target children and to put them into two allegedly locked money-boxes on which the target children’s photos were attached. In the interviewer-absence condition, the interviewer also told the child to keep doing the task while she left the room to drink some water. from one of their caretakers, in a subsample of 201 children (101 aged 6–7 years; 50.5% female; and 100 aged 9–10 years, 55% female; 102 previously assigned to the ‘‘interviewer present’’ condition; 99 assigned to the ‘‘interviewer absent’’ condition), the interviewer proceeded with the following questions after the money allocation task: ‘‘How much money did you give to each child? Why?’’ and ‘‘How much money would your father (mother, other caretaker) give to each child? Why?’’ Children’s exact answers were recorded, regardless of their actual previous allocation behaviour. Results Intergroup bias. In order to test our hypotheses, a 2 (age: 6–7, 9–10) 6 2 (sex: male, female) 6 2 (interviewer: present, absent) ANOVA was performed with the intergroup bias index as dependent variable. As there were no main or interaction effects involving sex, data were collapsed across this variable in further analyses. ANOVA results revealed a main effect for Age, F(1, 282) 5 5.81, p,.05, g2 5 .02, indicating that younger children displayed more intergroup bias (M 5 0.46, SD 5 1.29) than older children (M 5 0.09, SD 5 1.27). This effect was qualified by the Age 6 Interviewer interaction (see Figure 1), F(1, 282) 5 4.18, p,.05, g2 5 .02, indicating that younger children’s intergroup bias did not depend on interviewer’s presence vs absence, F(1, 124) 5 0.56, ns (interviewer present: M 5 0.55, SD 5 1.35; interviewer absent: M 5 0.37, SD 5 1.22), whereas in older children it did, F(1, 157) 5 5.07, p,.05 (interviewer present: M 5 2 0.13, SD 5 1.15; interviewer absent: M 5 0.32, SD 5 1.34). Moreover, t-tests of means against the scale midpoint (0) showed that younger children displayed intergroup bias in both conditions: Measures Dependent variable. An intergroup bias index was computed by subtracting the money given to the Black child from the money given to the White child (2 9 5 maximum out-group favouritism; + 9 5 maximum in-group favouritism). The content of the anti-racist norm. In order to identify children’s justifications for their reported in-group or out-group favouring behaviour, as well as the allocation behaviour they expected Figure 1. Children’s intergroup bias by age and interviewer conditions. INTERGROUP BIAS IN CHILDHOOD 33 merit-based understanding of the situation. Moreover, younger children who reported favouring the in-group did not use the merit justification, and those who reported favouring the out-group did not use the similarity and blatant racism justifications. As the younger age group, older children who reported favouring the out-group did not utilize the blatant racism justification (see Table 1). Chi-square tests indicated that children’s justifications for caretaker’s expected behaviour were, irrespective of age, significantly related to whether they expected their caretaker to favour the ingroup or the out-group: 6–7 years old, x2(8, N 5 101) 5 57.95, p,.001; 9–10 years old, x2(8, N 5 100) 5 114.27, p,.001. Specifically, the results revealed that expected in-group favouring behaviour was mainly justified by similarity and blatant racism explanations, but not by merit (see Table 2). Consistent with findings regarding ownbehaviour justifications, both younger and older children primarily explained their caretakers’ outgroup favouring behaviour in terms of merit, and not by similarity or blatant racism. interviewer present: t(65) 5 3.28, p,.01; interviewer absent: t(58) 5 2.36, p,.05. Older children, however, only favoured the in-group in the interviewer absent condition: interviewer absent: t(81) 5 2.14, p,.05; interviewer present: t(75) 5 21.00, ns. The content of the anti-racist norm. A content analysis was performed on the explanations children gave to their own and their caretakers’ expected intergroup behaviour. This analysis provided three main response categories, largely independent of age and subject (own/caretaker), accounting for 56% of the total answers for ownbehaviour justifications and for 54% of the total answers for caretaker’s expected behaviour justifications. The three justifications were ‘‘perceived similarity’’ (‘‘because he/she looks like me’’), ‘‘blatant racism’’ (because I don’t like him/her— the Black target) and ‘‘merit’’ (‘‘because he/she deserved it more’’). To test whether the distribution of the observed frequencies, within each agegroup, was consistent with an association between children’s reported allocation behaviour and their justifications for that behaviour, a Chi-square was computed for each age-group. Children’s justifications were associated with their reported intergroup allocation behaviour for both 6–7, x2(8, N 5 101) 5 29.17, p,.001, and 9–10 years old, x2(8, N 5 100) 5 37.77, p,.001. To interpret this association, the cells with adjusted standardized residuals above 2 (i.e., observed frequency higher than expected) and below 22 (i.e., observed frequency lower than expected) were analysed. Results showed that for both age groups, in-group favouring behaviour was consistently justified by blatant racism, while out-group favouring behaviour was attributed to a Discussion The results of this study showed that whereas younger children were not sensitive to the presence of the White interviewer as a cue for activating the anti-racist norm, older children seemed to selfregulate their behaviour according to that cue, by only displaying a biased behaviour against the black child when the experimenter was absent. Accordingly we can conclude that children’s age is important, not because intergroup bias declines with age, as stated by CDT, but because with age expression of bias can be better self-monitored TABLE 1 Frequencies of participants’ justifications for reported resource allocation behaviour Justifications Similarity Blatant racism Merit Other Doesn’t answer/ Doesn’t know Total 6–7 years old 9–10 years old Reported allocation behaviour Reported allocation behaviour Favoured in-group Favoured out-group Don’t know/ Nonsense Favoured in-group Favoured out-group Don’t know/ Nonsense 9 14 ( + ) 4(2) 5 17 2(2) 1(2) 12 ( + + ) 5 8 10 ( + ) 4 2 3 5 2 5(+) 9 5 5 1 0(2) 37 ( + ) 13 10 0 0 0 5 8(+) 49 28 24 26 61 13 + indicates that an adjusted standardized residual (asr) above 2 was observed, meaning that the observed frequency for that cell was above its expected frequency; + + asr .4. 2 indicates that an asr below 2 was obtained, meaning that the observed frequency for that cell was below its expected frequency. 34 MONTEIRO, DE FRANÇA, RODRIGUES TABLE 2 Frequencies of participants’ justifications for expected parents’ resource allocation behaviour Justifications Similarity Blatant racism Merit Other Doesn’t answer/ Doesn’t know Total 6–7 years old 9–10 years old Expected allocation behaviour Expected allocation behaviour Favoured in-group Favoured out-group 12 ( + ) 10 ( + ) 1(2) 4 14 1 1 16 3 5 41 26 Doesn’t know/ Nonsense (2) (2) (+ + +) (2) 5 2 1(2) 7 19 ( + ) 34 Favoured in-group 8 7 4 6 1 (+ +) (+ +) (2) (2) 26 Favoured out-group 1 1 45 6 4 57 (2) (2) (+ + +) (2) Doesn’t know/ Nonsense 0 0 0(2 2) 1 16 ( + + + ) 17 + indicates that an adjusted standardized residual (asr) above 2 was observed, meaning that the observed frequency for that cell was above its expected frequency; + + asr .4; + + + asr .6. 2 indicates that an asr below 2 was obtained, meaning that the observed frequency for that cell was below its expected frequency; 2 2 asr ,4. according to different levels of normative pressure present in the context. Consistent with Killen and colleagues’ work (2002), the second relevant finding of this study was that children used different justifications for their and others’ in-group and out-group favouring behaviours: Blatant racism and perceived similarity with the in-group target were the prevalent basis for justifying in-group favouring behaviour, while the merit motive seemed to underlie out-group favouritism. Thus, in a second study these norms were directly manipulated as cues for children’s behaviour (instead of the presence/absence of the experimenter). It was expected that younger children’s intergroup bias would not be changed by the activation of these norms whereas older children would comply with the norm orientation provided by the manipulations. STUDY 2 Method Participants and design One hundred and eighty-seven White Portuguese children (91 first-graders aged 6–7 years, 51.6% female; 96 fourth graders aged 9– 10 years, 56.3% female) participated in this study. Children attended 12 primary schools of mixed ethnicity (30–40% Black Portuguese, 60–70% White Portuguese) in the suburban area of Lisbon. The children were from primarily working-class backgrounds and all were given parental permission to participate in the study. The experimental design was a 2 (age: 6–7, 9–10) 6 2 (activated norm: anti-racist, racist) 6 3 (norm type: similarity-nationality, similarity-humanity, merit) between-subject factorial plan, again with intergroup bias as the dependent variable. Procedure Each child was individually interviewed at school by a White female interviewer. The interviewer gave the child the same instructions as described in Study 1 regarding the task of allocating money to the two target children. Before allowing the child to start the task the experimenter introduced the norm manipulations. The norm of similarity was operationalized in two ways: nationality-based similarity (Blacks and Whites are Portuguese) and humanity-based similarity (Blacks and Whites are persons). The norm of merit was operationalized with the assumption of ethnic asymmetry (White persons earn more money than Black persons but both deserve the same). In the three norm-type conditions an antiracist norm versus a racist norm was also manipulated. In all conditions the interviewer first told the child: ‘‘In Portugal there are many whiteskinned persons and there are others with a darker skin. One group is called the Whites and the other is the Blacks.’’ Then the interviewer proceeded with the antiracist/racist (in brackets) norm manipulation. In the nationality-based condition, she said: ‘‘But their skin colour doesn’t matter, as they all live and work in Portugal and all are Portuguese (But their skin colour is very important, as it shows that White persons are Portuguese and Black persons are not). And that is how it must be’’. In the INTERGROUP BIAS IN CHILDHOOD humanity-based condition, the interviewer told the child: ‘‘(…) but their skin colour doesn’t matter. White persons are very similar to Black persons because they are all human beings (But their skin colour is very important. White persons are very different from Black persons and we prefer people who are more similar to us). And that is how it must be’’. In the merit-based condition, after the first statement, the interviewer told the child: ‘‘(…) White persons have better houses and more toys for their children because they earn more money than Black persons. But both deserve the same things because they both work hard and need the money to live well (But they do not deserve the same things because those who work harder must take home more money). And that is how it must be’’. After checking for the child’s correct understanding of the manipulation content, the experimenter asked him/her to perform the money allocation task (for the procedure see Study 1), during which she turned her back to the child. This procedure was designed to control for the experimenter’s presence effect. Before leaving, children were thanked and debriefed. Special debriefing procedures were run in order to counteract the potential negative effects of the racist manipulations. Children in this experimental condition were invited to comment on what they had been told and the conversation proceeded until the child by himself refused the racist assumptions. Measures Manipulation checks. After the allocation task the manipulation checks for the effect of the antiracist/racist norms were introduced. Children were asked three filler questions and two specific check questions, one racist and one anti-racist, modified according to the type of manipulated norm. The manipulation checks for the racist norm were as follows: ‘‘White persons are Portuguese and Black persons are not’’ (similarity-nationality); ‘‘We prefer people who are more similar to us’’ (similarity-humanity) and ‘‘Those who work harder must earn more money’’ (merit). The antiracist manipulation checks were as follows: ‘‘Both Black and White people are Portuguese’’ (similarity-nationality); ‘‘Both Blacks and Whites are human beings’’ (similarity-humanity), and ‘‘Both Whites and Blacks work hard and deserve money to live well’’ (merit). Children answered the five statements on a 3-point ladder scale (1 5 I think it is not at all true to 3 5 I think it is true). 35 Dependent variable. An intergroup bias index was computed in the same way as in Study 1. Results Preliminary analyses. Exploratory data analyses were performed on the children’s bias index ( 2 9 to + 9) to make sure that it met the distributional requirements of the ANOVA. Data were also examined for sex effects. As sex did not reveal any main or interaction effects, data were collapsed across sexes in further analyses. Manipulation check. Responses to the manipulation check questions were analysed in a 2 (age: 6–7 vs. 9–10 years old) 6 2 (activated norm: racist, anti-racist) 6 3 (norm type: similarity-nationality, similarity-humanity, merit) 6 2 (type of check question: anti-racist, racist) MANOVA with the last factor within participants. The anti-racist check question was reversed so that the scale interpretation could be consistent with the racist check question. Accordingly, higher values on both racist and anti-racist check questions indicate that children respectively agreed more with the racist and less with the anti-racist check questions. The results revealed a main effect of Age, F(1, 175) 5 4.26, p,.05, g2 5 .02, indicating that older children agreed with the racist (and disagreed with the anti-racist) check questions, (9–10 years old: M 5 1.73, SD 5 0.47) significantly less than younger children (6–7 years-old: M 5 1.91, SD 5 0.50). Moreover, a main effect of Activated Norm was also found, F(1, 175) 5 8.05, p,.01, g2 5 .04, showing that the norm activation (racist vs antiracist) was successful. Specifically, participants in the racist norm activated condition agreed more with the check questions (M 5 1.91, SD 5 0.47) than participants assigned to the anti-racist norm activated condition, (M 5 1.72, SD 5 0.50). Finally, an interaction effect of Age 6 Activated Norm was also found, F(1, 175) 5 9.04, p,.01, g2 5 .05. The analysis of simple effects within each age group indicated that for younger participants the activated norm manipulation did not affect children’s responses on the check questions, F(1, 90) 5 0.09, ns, g2 5 .00, with these being equal to the scale midpoint on both conditions: racist (M 5 1.92, SD 5 0.48), t(44) 5 21.10, ns; anti-racist (M 5 1.89, SD 5 0.53), t(45) 5 21.40, ns; for older ones the expected significant effect was found, F(1, 95) 5 15.62, p,.01, g2 5 .14. Older participants in the racist activated norm condition agreed with check questions significantly more than those on 36 MONTEIRO, DE FRANÇA, RODRIGUES the anti-racist activated norm condition: racist (M 5 1.91, SD 5 0.46), t(45) 5 21.27, ns; anti-racist (M 5 1.56, SD 5 0.41), t(49) 5 27.55, p,.001. Intergroup bias. Results revealed a significant effect of Age, F(1, 186) 5 14.49, p,.001, g2 5 .08. As predicted, younger participants (M 5 0.47, SD 5 1.29) were more biased than older participants (M 5 20.29, SD 5 1.15). According to our main hypothesis the age effect was qualified by an Age 6 Activated Norm interaction effect, F(1, 187) 5 4.09, p,.05, g2 5 .02. Simple main effect analyses revealed that younger children exhibited intergroup bias regardless of the norm condition, F(1, 90) 5 0.49, ns, g2 5 .01; racist (M 5 0.37, SD 5 1.44), t(45) 5 2.90, p,.05; anti-racist (M 5 0.57, SD 5 1.07), t(45) 5 3.03, p,.01. In contrast, older children were affected by the norm manipulation, F(1, 95) 5 4.20, p,.05, g2 5 .04, even displaying out-group favouritism in the anti-racist norm condition (M 5 2 0.56, SD 5 1.01), t(49) 5 23.91, p,.001. Differently from the hypothesis, in the racist norm condition older children did not display intergroup bias (M 5 20.09, SD 5 1.24), t(46) 5 20.47, ns. Besides, a main effect for the Norm type was also present, F(2, 187) 5 3.17, p,.05, g2 5 .04. Comparison of the cell means using Duncan’s multiple range test revealed that participants’ ingroup bias in the similarity-humanity norm (M 5 0.45, SD 5 0.90) was higher than in the similaritynationality (M 5 2 0.08, SD 5 1.53) condition, and in the merit norm condition (M 5 20.11, SD 5 1.26). In fact, only in the similarity-humanity condition did the mean index score differ from zero, t(39) 5 3.15, p,.05. Considering that only in the similarity-humanity condition children displayed a biased behaviour, a further ANOVA of 2 (age) 6 2 (activated norm) for that condition (n 5 40; 52.5% female) was carried out on the bias index. Results showed a main effect of Age, F(1, 39) 5 17.78, p,.001, g2 5 .33, meaning that younger children were more biased (M 5 0.90; SD 5 0.45) than older ones (M 5 0.00; SD 5 1.03). A main effect of Activated Norm in the expected direction was also found, F(1, 39) 5 10.76, p,.01, g2 5 .23, showing that children were only biased in the racist norm condition (M 5 0.80; SD 5 0.62), since in the anti-racist norm condition they did not express bias (M 5 0.10, SD 5 1.02). More important, the expected Age 6 Norm interaction effect, F(1, 39) 5 5.48, p,.05, g2 5 .13, was also found (see Figure 2). Simple main effects analyses showed that younger children exhibited a biased behaviour regardless of the norm condition, F(1, Figure 2. Children’s intergroup bias by age and norm activation in the similarity-humanity condition. 19) 5 1.00, ns, g2 5 .05: racist (M 5 1.00, SD 5 0.00), t(9) 5 3.90, p,.001; anti-racist (M 5 0.80, SD 5 0.63), t(9) 5 3.30, p,.01. Older children, however, complied both with the anti-racist norm [even presenting out-group favouritism, F(1, 19) 5 10.13, p,.01, g2 5 .36] (M 5 20.60, SD 5 0.84), t(9) 5 2 2.25, p,.05, and with the racist norm (M 5 0.60, SD 5 0.84), t(9) 5 2.25, p,.05, by displaying clear intergroup bias. Discussion In this study the two norms that most children had used in Study 1 to explain their own/ caretaker money allocation behaviour to the Black and White target children were manipulated: perceived intergroup similarity vs dissimilarity and intergroup merit vs equality. Overall, the main findings were consistent with the hypotheses: Whereas younger children exhibited a biased behaviour regardless of the activation of anti-racist or racist norms (Aboud, 1988; Nesdale, Maass, Durkin, & Griffiths, 2005), older children complied with the anti-racist, but not with the racist, norm. Considering that a main effect for the type of norm indicated that, in fact, children’s biased behaviour was only biased in the similarity-humanity condition, a separate treatment of that condition allowed a more clear-cut picture of results: Older children’s behaviour was in line with the intergroup similarity norm (both in the racist and in the anti-racist conditions) whereas, again, younger children consistently displayed a biased behaviour, regardless of the racist or anti-racist nature of the activated norm. This result seems to underline the fact that different norms can have different effects on the regulation of older children’s behaviour, particularly when norms that facilitate the expression of intergroup bias are used. INTERGROUP BIAS IN CHILDHOOD GENERAL DISCUSSION The present research examined the effect of social norms on the expression or suppression of White children’s intergroup biased behaviours in two age groups. Norm pressure was manipulated either by the presence or absence of the experimenter during the children’s performance of the task (Study 1) or by the verbal activation of anti-racist vs racist norms (Study 2). In line with the hypotheses, we found that both the experimenter’s presence and the activation of an interracial similarity norm did not affect younger children’s intergroup bias but suppressed older children’s biased behaviour. Contrary to the internalization hypothesis (Rutland et al., 2005), older children also displayed bias when the amount of normative pressure was significantly reduced, either by removing the interviewer or by activating a discrimination norm. Moreover, the fact that Study 2 replicated the results of Study 1 is important, as it shows that the mere presence of an in-group adult (Study 1) can be as powerful as a direct verbal activation of a norm (Study 2) to either legitimate or prohibit older children’s expression of intergroup bias. Besides assessing children’s intergroup behaviour, we intended to uncover the normative explanations underlying their reported behaviours and the expected behaviours of their caretakers. Results showed that in-group favouring behaviour was primarily explained by perceptions of similarity between self and the in-group target child and, interestingly, that the use of this explanation didn’t decrease with age, as it would be expected by the cognitivedevelopmental theory (Doyle & Aboud, 1995). Moreover, a significant percentage justified their bias in terms of disliking the out-group member or the out-group as a whole. This result suggests that intergroup bias may be closer to intergroup prejudice than was initially assumed (Rutland, 2004), and supports the idea that the focus of children’s racial attitudes (in-group vs outgroup) is probably more status- and contextrelated than development-dependent. Finally, reported and expected out-group favouring behaviours were consistently explained through the merit of the out-group target child. The merit motive has been found to be a pervasive source of subtle in-group favouritism rather than one for out-group favouritism in White young adults (e.g., Lima & Vala, 2002). Its use by White children suggests that they are aware of its positive social meaning. More important, 37 preferring this external motive to the more internal similarity motive suggests that subtle prejudice can also be at work in this situation. How do White children handle such contradictory norms in their daily lives? Some authors suggest that the developmental path of prejudice may be better understood if the importance of children’s ability to self-regulate their expression of intergroup bias is acknowledged (Rutland, 2004). We suggest that this ability can account for children’s compliance with different social norms according to the normative meaning and pressure that are fuelled into the behavioural contexts. We also suggest that the most stable and active norms regarding interethnic behaviour for both age-groups are intergroup distinctiveness and in-group favouring (Nesdale, 2004; Tajfel, 1982). However, while older children concurrently face the opposite and equally strong norm that prohibits the interethnic bias and can manage the hierarchical use of both norms (Kohlberg, 1963) according to context normative and self-presentation demands, younger children are less able to do this. Thus, although younger children are aware of the anti-racism norm, a substantial normative pressure would be needed for them to skip the in-group favouring norm a...