Abstract
Do situational cues to individuals’ social identities shift the way they look at objects? Do such shifts hinge on the structure of individuals’ self-concept? We hypothesized individuals with integrated identities would exhibit attentional biases congruent with identity cues (assimilative response), whereas those with nonintegrated identities would exhibit attentional biases incongruent with identity cues (contrastive response). Dual identity participants (Asian Americans, Study 1; female lawyers, Study 2) were exposed to identity primes and then asked to focus on central, focal objects in a stimulus display. Among participants with high identity integration, American (Study 1) or lawyer priming (Study 2) shifted attention toward focal objects (assimilative response). Among participants with low identity integration, Asian (Study 1) or female priming (Study 2) shifted attention toward focal objects (contrastive response). Dual identity integration moderates responses to identity cues in attentional focus. Implications for identity structure, object perception, and task performance are discussed.
People from Western cultures tend to view objects as separate from their context, whereas people from East Asian cultures tend to view objects as related to their context (Doherty, Tsuji, & Phillips, 2008; Masuda & Nisbett, 2001). These cultural differences in attentional focus or object perception have been associated with differences in Western and Eastern self-construals (Kuhnen & Oyserman, 2002). Just as Westerners construe themselves as separate and autonomous from others (Heine, Lehman, Markus, & Kitayama, 1999), they visually separate objects from their context. Just as Easterners construe themselves as interdependent with close others (Markus & Kitayama, 1991; Triandis, 1995), they attend to the relationships between objects and their environment. Accordingly, Westerners are better at visual tasks that demand context independent judgments, such as judging the absolute size of objects. Easterners are better at visual tasks that demand context-sensitive judgments, such as judging the relative size of objects (Kitayama, Duffy, Kawamura, & Larsen, 2003).
Besides culture, gender and occupation affect context sensitivity in object perception. Females, who have a more interdependent self-construal than men (Kashima & Hardie, 2000) perform worse on tasks that involve visually separating objects from their context (Phillips, Chapman, & Berry, 2004). Likewise, some occupations place less emphasis on attention to relationships and context than others. Phillips, Chapman, and Berry(2004) found that workers in technical fields that emphasize attention to focal objects (e.g., IT, programming) perform better on tasks that involve visually separating objects from their context, compared to nontechnical workers.
Research has found that a consequence of legal training is a disproportional reliance on objective thought and a de-emphasis on interpersonal relations in problem solving (Daikoff, 1997). For example, female law students see themselves as more autonomous and independent than do female students in other disciplines (Coplin & Williams, 1978). A core task for lawyers is reviewing documents; lawyers are frequently required to scrutinize written clauses to ensure they are not ambiguous or vulnerable to subsequent challenge (Sturm, 1997). Such tasks require attending more to focal objects and less to their context.
Attention to focal objects versus their context has implications for task performance. People who attend more to the context and relationships among the objects perform better at tasks that involve memory for the spatial location of objects (Kuhnen & Oyserman, 2002). Applied to work settings, lawyers who attend more to relationships and context could be better at recalling the location of specific clauses in a document. Police who attend more to focal objects may be more capable of tracking a suspect in a crowd.
In situations providing cues that make social identities salient, such as imagery, language, or symbols associated with the identity, people often exhibit identity-congruent behaviors (Rhee, Uleman, Lee, & Roman, 1995; Verkuyten & Poulisi, 2002; Weber & Morris, 2010). For example, Asian Americans become more likely to describe themselves in terms of individual rights after exposure to American cues, whereas they become more likely to reference collective duties after exposure to Asian cues (Hong, Ip, Chiu, Morris, & Menon, 2001). These reflect assimilative responses to identity cues. Identity cues are thought to raise the cognitive accessibility of knowledge structures associated with the identity, such as norms, increasing the likelihood that they guide behavior (Hong, Morris, Chiu, & Benet-Martinez, 2000).
However, identity cues can also elicit contrastive responses; some studies observe that identity cues evoke a contrastive shift in behavior away from the norms of the cued identity (e.g., Yang & Bond, 1980). The direction of responses to identity cues—assimilation or contrast—appears to be moderated by individual differences in dual identity structure. Individuals can represent their two identities as interconnected and compatible (high identity integration) or as divided and conflicting (low identity integration; Benet-Martinez, Leu, Lee, & Morris, 2002). Individuals with integrated dual identities tend to respond assimilatively to identity cues, whereas those with nonintegrated dual identities tend to respond contrastively. For example, when primed with images of American versus Asian culture, Asian Americans with integrated cultural identities attribute causality to the individual actor; those with nonintegrated cultural identities attribute causality to the situation (Benet-Martinez et al., 2002). Similarly, when queried about their gender versus occupational identity, businesswomen with integrated gender-occupational identities display higher interpersonal orientation; those with nonintegrated gender-occupational identities display lower interpersonal orientation (Sacharin, Lee, & Gonzalez, 2009).
It is intriguing that people with nonintegrated dual identities respond contrastively to cues of their own identities (Benet-Martinez et al., 2002). Studies of bicultural individuals (Mok & Morris, 2010a, 2011b) proposed it could reflect motivation to defend the noncued identity. For individuals with nonintegrated or divided dual identities, they may be inclined to associate situations that emphasize one of their identities with excluding their other identity. Individuals with integrated or interconnected dual identities are unlikely to associate these situations with leaving out another identity. Contrasting against an identity cue may be a self-protective strategy among people with low identity integration to prevent leaving out the noncued identity. This conceptualization might apply to other individuals (e.g., career women; Sacharin et al., 2009) and explain their divergent responses to identity cues.
Growing evidence suggests that dual identity integration moderates responses to identity cues in social perception (Benet-Martinez et al., 2002; Mok & Morris, 2011a). Are such identity dynamics confined to social perception biases or could they govern perception of nonsocial objects also? While past research indicates that attentional biases in object perception can be shifted by situational cues (Kuhnen & Oyserman, 2002), it has not explored the interaction of identity cues and identity structure. We propose that attentional bias to focal objects versus their context is responsive to identity cues and dual identity integration. Individuals with high identity integration should assimilate to identity cues in their attentional focus, whereas those with low identity integration should contrast to identity cues.
We studied the object perception of Asian Americans (Study 1) and female lawyers (Study 2) after identity priming. We hypothesized divergent responses to identity primes as a function of dual identity integration. In Study 1, attentional bias to focal objects after American (vs. Asian) priming would suggest an assimilative response. In Study 2, attentional bias to focal objects after lawyer (vs. female) priming would suggest an assimilative response.
Study 1
We studied Asian Americans. We hypothesized that participants with integrated Asian and American identities would attend more to focal objects after American versus Asian priming (assimilative response). Conversely, participants with nonintegrated Asian and American identities would attend less to focal objects (and more to their context) after American versus Asian priming (contrastive response).
Method
Participants
We recruited 55 Asian Americans from New York City online forums. To be eligible, participants were East Asian and self-identified with both East Asian and American culture. Two participants did not follow directions so they were dropped from the analysis. The final analysis included 53 participants (17 men; mean age = 26.72, SD = 7.77; mean years in the United States = 23.72, SD = 7.01). On a scale of 1 (very weak) to 7 (very strong), participants rated their identification with East Asian (M = 5.04, SD = 1.24) and American culture (M = 5.15, SD = 1.28). No sex differences emerged on the independent or dependent measures so it is not considered further.
Materials and Procedure
Participants received a weblink with instructions to complete the study in one sitting and in a quiet and private location. Participants were randomly assigned to Asian or American priming in a picture evaluation task, similar to Mok and Morris (2009). Participants read: “You are looking at some pictures in magazines. List two things that come to mind as you look at each picture. For example, you can think about the feelings, places, people, or memories that the picture elicits.” Participants viewed four pictures (e.g., Asian children, fried rice in the Asian condition; White children, mashed potatoes in the American condition; see Figure 1). Then they received the object perception task.
A, Asian primes. B, American primes.
We used the task developed by Phillips and his colleagues (Doherty et al., 2008; Phillips et al., 2004) previously used to study biases in object perception as a function of cultural background and occupation. The task comprised multiple trials. In each trial, participants viewed two arrays of circles presented side by side. Each array featured 9 circles arranged in a 3 × 3 format. Participants were asked to focus on the center circle in each array. Their task was to discern whether the center circle in the left or right array was larger in size. They were told to click on the larger center circle with their mouse; the clicked circle turned green. Then they proceeded to the next question.
The task included easy and difficult trials. In the easy trials, the larger of the two center circles (by 2%) was surrounded by 8 circles smaller than itself. The smaller center circle in the adjacent array was surrounded by 8 circles larger than itself. Focal objects appear larger (smaller) if surrounding objects are smaller (larger), an effect referred to as the Ebbinghaus illusion. In the easy trials, the size of the surrounding circles (or context) made it easy for participants to make a correct response.
In the difficult trials, the two center circles differed in size by 2%, 6%, 10%, 14%, or 18%. Unlike the easy trials, the larger of the two center circles was surrounded by eight circles larger than itself. The smaller center circle in the adjacent array was surrounded by 8 circles smaller than itself. Hence, the size of the surrounding circles (or context) could lead participants to make an incorrect response. If participants attend more to focal objects, however, they should be less affected or distracted by the size of the surrounding circles and more accurate in judging which of the two center circles is larger (see Figure 2 for example questions).
Two example questions in the object perception task. In both questions, the larger center circle (by 2%) appears in the right array. In example (a), the size difference between the two center circles is easy to see because the larger center circle is surrounded by small circles; these were filler trials. In example (b), the size difference is hard to see because the larger center circle is surrounded by larger circles than itself and the smaller center circle is surrounded by small circles. This makes the center circle in the left array appear larger; these were the critical trials. Participants with an attentional bias to focal objects (vs. their context) should perform better in the task.
Participants performed two practice questions and then the recorded session of two blocks, each with 24 trials. Whether the larger center circle was on the left or right was counterbalanced across trials. To minimize potential fatigue in the task, we reduced the total number of trials by half compared to past studies (Doherty et al., 2008), yet we did not change the proportion of easy versus difficult trials. Each block had 4 easy trials and 20 difficult trials, which were presented in a predetermined mixed order (e.g., the easy trials did not appear in sequence). The difficult trials were the critical trials whereas the easy trials were fillers. Recent work showed that in a task with correct answers, individuals with low identity integration did not exhibit a contrasting response with easy, more obvious trials (Mok & Morris, 2010b). We recorded the response time for each trial (M = 4.16 sec, SD = 3.29) as an indirect measure of task effort. Longer response times imply greater effort.
Next, we assessed the degree to which individuals integrated their cultural identities. Participants rated on a scale of 1 (strongly disagree) to 7 (strongly agree), four items drawn from Benet-Martinez and Haritatos' (2005) research of bicultural identity integration (“I feel conflicted between the American and Asian ways of doing things,” “I feel like someone moving between two cultures,” “I feel caught between the Asian and American cultures,” and “I don’t feel trapped between the Asian and American cultures”). An identity integration score was formed by reversing the first three items and averaging them with the last item; higher scores reflect higher identity integration (α = .81; M = 4.83, SD = 1.10). Identity integration was uncorrelated with Asian or American identification (r = .14, p = .32, and r = .05, p = .71, respectively), similar to prior work (Mok & Morris, 2010b). This lack of correlation may reflect the restriction in range to the upper levels of identity strength; had we recruited individuals less strongly identified with either culture, we may have observed that identity integration was positively correlated with identity strength (e.g., Benet-Martinez & Haritatos, 2005). 1 Lastly, participants completed a demographic survey.
Results and Discussion
We derived a perception accuracy score by summing the number of correct responses across the difficult trials (M = 11.85, SD = 8.16). Higher scores reflect attentional bias to focal objects (lower context sensitivity). We submitted the accuracy score to a 2 (Prime: Asian vs. American) × Identity integration (mean-centered) Generalized Linear Model (GLM), including the interaction between Prime and Identity integration. The interaction was significant, F(1, 49) = 15.17, p < .001, η p 2 = .24; see Figure 3A. As predicted, individuals with high identity integration (1 SD above the mean) were more accurate after American versus Asian priming (M = 16.81 vs. M = 8.57), F(1, 49) = 8.42, p < .01, η p 2 = .15, suggesting attention to focal objects increased (assimilative response). Conversely, individuals with low identity integration (1 SD below the mean) were less accurate after American versus Asian priming (M = 7.83 vs. M = 15.43), F(1, 49) = 7.02, p = .01, η p 2 = .13, suggesting context sensitivity increased (contrastive response). Controlling for response times did not change the pattern of findings. Hence, the different responses of high and low identity integration participants do not appear reducible to task effort.
Accuracy score across priming conditions for high and low identity integration participants. A, Integration between cultural identities (Study 1). B, Integration between gender and occupational identities (Study 2). Note. High scores imply more attention to focal objects or low context sensitivity. High and low identity integration is plotted at 1 standard deviation (SD) above and below the mean.
Past research suggests that identity strength can moderate responses to identity cues. Individuals with strong identification assimilate to identity cues; those with weak identification contrast to identity cues (e.g., LeBoeuf, Shafir, & Bayuk, 2010). To highlight the role of identity structure in the observed responses, independent of identity strength, we submitted the accuracy score to a GLM model that included four predictors—prime, identity integration, Asian identification, and American identification; all main effects, 2-way interactions, 3-way interactions, and the 4-way interaction was included. Identity integration and the identity strength measures were mean centered. The analysis showed that the moderating influence of identity integration on the accuracy score did not depend on the strength of Asian or American identification (Prime × Identity Integration × Asian Identification, p = .38; Prime × Identity Integration × American Identification, p = .43). The interaction of Prime and Identity Integration remained significant (p < .01). This suggests that identity structure uniquely affects assimilative or contrastive responses to identity primes beyond strength of the identifications.
The current design casts a cautionary note about the effects of identity cues on individuals with high and low identity integration. Without a control condition, it is difficult to assess whether individuals with high identity integration assimilate to cues of either of their identities, whereas those with low identity integration contrast. Hence, we recruited a separate group of 91 Asian Americans who were not culturally primed. They were recruited subsequently in the same manner as the current study and completed the perception task without identity priming. Although running a post hoc control condition risks the possibility that participants differ from the priming groups on key variables, the control sample did not differ from the priming sample on identity integration.
We submitted the accuracy score to a 3 (Prime: Asian vs. none vs. American) × Identity Integration (mean-centered) GLM. The interaction between Prime and Identity Integration was significant, F(2, 138) = 10.19, p < .001, η p 2 = .13. We computed the means for the priming versus control conditions at high and low levels of identity integration (centered at 1 SD above and below the mean, respectively). For participants with high identity integration, accuracy was higher in the American versus control condition, F(1, 112) = 4.23, p < .05, η p 2 = .04; accuracy was lower in the Asian versus control condition, F(1, 112) = 6.07, p < .05, η p 2 = .05 (M American = 17.16 vs. M control = 12.91 vs. M Asian = 8.38). This suggests assimilative responses. For participants with low identity integration, accuracy was lower in the American versus control condition, F(1, 115) = 4.09, p < .05, η p 2 = .04; accuracy was higher in the Asian versus control condition, F(1, 115) = 3.83, p = .05, η p 2 = .03 (M American = 7.83 vs. M control = 11.68 vs. M Asian = 15.43). This suggests contrastive responses. The results support our interpretation that individuals with high identity integration respond assimilatively to identity cues in object perception, whereas those with low identity integration respond contrastively.
We found initial evidence that dual identity integration moderates the effects of identity primes on object perception. Study 2 sought to replicate this pattern by examining gender and occupational identity integration.
Study 2
We studied female lawyers. We hypothesized that participants with integrated female-lawyer identities would attend more to focal objects after lawyer versus female priming (assimilative response). Conversely, participants with nonintegrated female-lawyer identities would attend less to focal objects (and more to their context) after lawyer versus female priming (contrastive response).
Method
Participants
Fifty-one female students at Columbia Law School were recruited through campus fliers. Two participants were excluded because they did not follow directions, leaving 49 participants in the final analysis (mean age = 25.14, SD = 2.10). Participants had at least 1 year of legal education (M = 2.20 years, SD = .69). Participants' ethnic background varied (32 White and 17 Asian). To control for potential cultural effects on the results (Asians are more context sensitive than Westerners), the analysis controlled for ethnicity.
Materials and Procedure
Participants received a weblink as in Study 1. Participants were randomly assigned to the female or lawyer prime condition. In the female condition, they were asked to think about themselves as a woman and write 8 aspects that are descriptive of them as a woman. In the lawyer condition, they were asked to think about themselves as a lawyer and write 8 aspects that are descriptive of them as a lawyer. All participants then completed the same object perception task as in study 1. An accuracy score for difficult trials was computed as in study 1. The response time for each trial, reflecting task effort, was 5.46 sec (SD = 5.92), comparable to Study 1, t(100) = 1.39, p = .17.
Next, participants rated on a scale of 1 (not at all positive) to 4 (very positive) how positive they felt about being a woman (M = 3.80, SD = .41) or lawyer (M = 3.53, SD = .58). Participants were also asked whether they felt their female and lawyer sides were integrated. On a scale of 1 (strongly disagree) to 7 (strongly agree), they rated four items (“Succeeding as a lawyer involves the same side of myself as succeeding as a woman,” “I feel torn between the expectations of being a woman and the expectations of my profession,” “My self-concept seamlessly blends my professional identity with my identity as a woman,” and “I do not feel any tension between my goals as a woman and my goals as a lawyer"). 2 We formed an identity integration score by averaging ratings across the four items after reverse scoring the second item; higher scores reflect higher integration between gender and occupational identities (α = .61, M = 4.36, SD = 1.08). 3 Identity integration was uncorrelated with positivity toward being a woman or lawyer (r = .16, p = .28 and r = .13, p = .39, respectively).
Results and Discussion
We computed a perception accuracy score by summing the number of correct responses across the difficult trials (M = 19.90, SD = 7.37), as in the prior study. Higher scores reflect attentional bias to focal objects (lower context sensitivity). We submitted the accuracy score to a 2 (Prime: female vs. lawyer) × Identity Integration (mean-centered) GLM, including the interaction between Prime and Identity Integration. Ethnicity was included as a covariate. Results showed the predicted interaction of Prime and Identity Integration was significant, F(1, 44) = 8.00, p < .01, η p 2 = .15; no other effects emerged. As displayed in Figure 3B, participants with high identity integration (1 SD above the mean) were more accurate (attentive to focal objects) after lawyer versus female priming (assimilative response; M = 23.47 vs. M = 18.28). The pattern was marginally significant, F(1, 44) = 3.14, p = .08, η p 2 = .07. Participants with low identity integration (1 SD below the mean) were less accurate (more context sensitive) after lawyer versus female priming (contrastive response; M = 17.09 vs. M = 24.27), F(1, 44) = 5.53, p < .05, η p 2 = .10. 4
Additional analysis that controlled for response time did not change the pattern of findings, suggesting task effort does not explain the results. Also, the moderating influence of identity integration on the accuracy score did not depend on positivity toward being a woman or lawyer (Prime × Identity Integration × Woman-Positivity, p = .62; Prime × Identity Integration × Lawyer-Positivity, p = .26). The interaction of Prime and Identity Integration remained significant (p < .01). These results indicate that identity structure uniquely shapes assimilative or contrastive responses to identity primes beyond positive views about the identities. 5
General Discussion
Ways of looking at objects shift with identity cues and identity integration. When primed with images of American versus Asian culture, Asian Americans with integrated cultural identities attend more to focal objects, whereas those with nonintegrated cultural identities attend less to focal objects (and more to their context; Study 1). When queried about their lawyer versus female identity, female lawyers with integrated gender-occupational identities attend more to focal objects, whereas those with nonintegrated gender-occupational identities attend less to focal objects (and more to their context; Study 2). Evidence from the control group without identity priming in Study 1 supports our interpretation that individuals with high identity integration assimilate to identity cues in their attentional focus, whereas those with low identity integration contrast against identity cues.
Theoretical Implications
We show that identity cues and identity integration interact to influence perception of nonsocial, besides social stimuli (Benet-Martinez et al., 2002). The interaction effect applies to visual attention, besides conceptual attention (Benet-Martinez et al., 2002). Moreover, we extend research on dual identity integration to other identity domains. Past studies focused on the effects of cultural identity integration on responses to identity cues (Benet-Martinez et al., 2002). The current research probes the effects of gender and occupational identity integration (Sacharin et al., 2009). Our research with female lawyers implies that the interplay between identity cues and dual identity integration generalizes to different identity domains (e.g., female engineers, Cheng, Sanchez-Burks, & Lee, 2008).
Identity integration appears to be a key predictor of divergent responses to identity cues, as it showed effects beyond identity strength (Study 1) or positivity toward the identities (Study 2). Research implies that positivity toward a social category can be a proxy for identification strength (Luhtanen & Crocker, 1992). Thus, our findings converge with evidence that identity structure shapes responses to identity cues independent of the strength of identifications (Mok & Morris, 2010a).
We extend research on the effects of culture, gender, and occupation on object perception. Whereas past research implies that context sensitivity in object perception is relatively stable within individuals (Hedden, Ketay, Aron, Markus, & Gabrielli, 2008; Masuda & Nisbett, 2001; Phillips et al., 2004), we show it can momentarily shift with identity cues. Moreover, the direction of shift—assimilation or contrast—depends on individuals’ identity structure. Our findings also show that lawyer cues affect how female lawyers construe nonsocial objects, besides social objects (Coplin & Williams, 1978).
Practical Implications
Focus of attention affects work performance. Some tasks reward a narrow focus on objects (e.g., auditing); others reward attention to holistic patterns (e.g., fashion design). Our findings suggest that in Westernized workplaces, individuals with integrated Western and Asian identities may be more inclined to isolate objects from their context; those with nonintegrated Western and Asian identities may be more attuned to relationships among the objects (e.g., their description of events could include more information about the setting). Likewise, female lawyers with integrated gender-occupational identities may be more effective in legal work that requires narrow attention to objects (e.g. editing contracts); those with nonintegrated gender-occupational identities may be more effective in recalling the location of specific clauses in a document.
Directions for Future Research
More research should examine the mechanism underlying divergent (contrastive) responses to identity primes as a function of dual identity integration. Studies of biculturals (Mok & Morris, 2010a, 2011b) propose the contrastive response reflects an identity-protection strategy. Whereas individuals with integrated/interconnected dual identities could assimilate to identity cues without leaving behind their other identity, individuals with nonintegrated/divided dual identities would feel that following an identity cue excludes their other identity. The contrastive response could reflect resisting identity cues in order to prevent excluding the noncued identity. Future research should test this conceptualization and with identity integration in other domains.
Research could explore the extent to which context sensitivity in viewing nonsocial versus social objects are related. For example, studies could examine whether the effects of identity priming on context sensitivity in object perception (tapped by the current perception task) mediate the effects on judging others' behavior in terms of situational factors. This would cast light on the link between perceptions of nonsocial versus social stimuli.
Conclusion
The present findings reveal the importance of studying identity structure to gain a richer understanding of how individuals respond to identity cues. We found that people with integrated dual identities respond assimilatively to identity cues in object perception, whereas people with nonintegrated dual identities respond contrastively. Attention to focal objects versus their context has implications for task accuracy in identity-related situations.
Footnotes
Acknowledgements
We thank C.Y. Chiu and Gary Li for comments on an earlier draft and Hedan Zeng for data collection assistance.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The current research was supported by faculty research funds to Michael Morris from Columbia Business School and its program on Social Intelligence.