NOVEL SELF-CATEGORIZATION OVERRIDES RACIAL BIAS:

A MULTI-LEVEL APPROACH TO INTERGROUP PERCEPTION AND EVALUATION

By

Jay Joseph Van Bavel

A thesis submitted in conformity with the requirements

For the degree of Doctor of Philosophy

Graduate Department of Psychology

In the University of Toronto

© Copyright by Jay Joseph Van Bavel, 2008

ii

Novel self-categorization overrides racial bias:

A multi-level approach to intergroup perception and evaluation

Jay Joseph Van Bavel

Doctor of Philosophy, 2008

Graduate Department of Psychology

University of Toronto

Abstract

People engage in a constant and reflexive process of categorizing others according to their

race, gender, age or other salient social category. Decades of research have shown that social

categorization often elicits stereotypes, prejudice, and discrimination. Social perception is

complicated by the fact that people have multiple social identities and self-categorization

with these identities can shift from one situation to another, coloring perceptions and

evaluations of the self and others. This dissertation provides evidence that self-categorization

with a novel group can override ostensible stable and pervasive racial biases in memory and

evaluation and examines the neural substrates that mediate these processes. Experiment 1

shows that self-categorization with a novel mixed-race group elicited liking for ingroup

members, regardless of race. This preference for ingroup members was mediated by the

orbitofrontal cortex – a region of the brain linked to subjective valuation. Participants in

novel groups also had greater fusiform and amygdala activity to novel ingroup members,

suggesting that these regions are sensitive to the current self-categorization rather than

features associated with race. Experiment 2 shows that preferences for ingroup members are

iii

evoked rapidly and spontaneously, regardless of race, indicating that ingroup bias can

override automatic racial bias. Experiment 3 provides evidence that preferences for ingroup

members are driven by ingroup bias rather than outgroup derogation. Experiment 4 shows

that self-categorization increases memory for ingroup members eliminating the own-race

memory bias. Experiment 5 provides direct evidence that fusiform activity to ingroup

members is associated with superior memory for ingroup members. This study also shows

greater amygdala activity to Black than White faces who are unaffiliated with either the

ingroup or outgroup, suggesting that social categorization is flexible, shifting from group

membership to race within a given social context. These five experiments illustrate that

social perception and evaluation are sensitive to the current self-categorization – however

minimal – and characterized by ingroup bias. This research also offers a relatively simple

approach for erasing several pervasive racial biases. This multi-level approach extends

several theories of intergroup perception and evaluation by making explicit links between

self-categorization, neural processes, and social perception and evaluation.

iv

Acknowledgements

I am indebted to the many friends and colleagues who supported me throughout the

dissertation process. Foremost, I thank my mentor, friend and collaborator Dr. William

Cunningham for countless hours of guidance and support. This dissertation and my growth as

a scientist have doubtless been shaped by your thirst for discovery and commitment to

excellence. I wish to thank my intrepid committee, Drs. Alison Chasteen and Jordan

Peterson, for interrupting their trips to Memphis, TN and Albuquerque, NM to chat about

prejudice and social cognitive neuroscience. Finally, thanks to Drs. Mickey Inzlicht, Adam

Anderson and Susan Fiske for their insightful comments on this dissertation.

Although it would be nearly impossible to credit everyone who contributed to this

dissertation in one form or another, several people deserve special thanks for going above

and beyond the call of duty during my tenure in graduate school, including Drs. Dominic

Packer, Marilynn Brewer, Ken & Karen Dion, Ken DeMarree, Russ Fazio, Ken Fujita,

members of Social Cognitive and Affective Neuroscience Lab (Amanda, Ashley, Ingrid,

Nathan, Norman, and Samantha), and Mike McCaslin. I also thank Dr. Marvin Chun and two

anonymous reviewers for their comments on the experiment described in Chapter 3, and Dr.

Laurie Rudman and four anonymous reviewers for their comments on the experiments

described in Chapters 4 and 5. Most important, thanks to Sonja Shute; her love made each

failure softer and each success sweeter.

This dissertation was supported by a Canada Graduate Scholarship from the Social

Sciences and Humanities Research Council of Canada and two Ontario Graduate

Scholarships.

v

Table of Contents

List of Tables…………………………………………………………………………… viii

List of Figures……………………………………………………………..…………… ix

List of Appendices………………………………………………………...…………… xii

Chapter 1: Social Perception and Evaluation……………………………….………..… 1

The Nature of Prejudice………………………………………………..…….… 2

Social Identity and Self-Categorization………………………………..…….… 7

Models of Evaluation………………….………………………………..……… 8

Chapter 2: A Social Cognitive Neuroscience Approach…………..…………………… 12

Social Perception and Categorization…………………………………..……… 14

Social Evaluation……………………………………………………………..… 16

Automatic and Controlled Processing………………………………………..… 18

Reconstrual…………………………………………………………………..… 21

Overview of the Current Research…………………………………………….. 23

Chapter 3: The Neural Substrates of Ingroup Bias…………………………………..… 26

Experiment 1……………………………………….………………………..… 27

Method………………………………………………………………………… 28

Results……………………………………………………………….………… 31

Discussion……………………………………………………………………… 36

Chapter 4: The Contextual Sensitivity of Intergroup Evaluation……………………… 40

Experiment 2…………………………………………………………………… 42

Method……………………………………………………………….………… 43

Results………………………………………………………………..………… 45

vi

Discussion……………………………………………………………………… 48

Chapter 5: Ingroup Bias versus Outgroup Derogation…………………………………. 51

Experiment 3………………………………………………..……………… ..… 51

Methods………………………………………………………………………… 53

Results……………………………………………………………….………… 54

Discussion…………………………………………………………….…...…… 56

Chapter 6: Self-Categorization and Intergroup Memory……………………………… 59

Experiment 4…………………………………………………………………… 60

Method……………………………………………………………….………… 61

Results………………………………………………………………..………… 62

Discussion…………………………………………………………….……...… 64

Chapter 7: Neural Substrates of Social Perception and Evaluation………………….… 66

Experiment 5…………………………………………………………………… 67

Method……………………………………………………………….………… 69

Results………………………………………………………………………..… 72

Discussion…………………………………………………………….……...… 77

Chapter 8: General Discussion………………………………………………………… 79

Flexibility in Social Perception and Evaluation….…………………..……...… 79

The Primacy of the Ingroup?…………………………………………………... 82

Social Categorization………………………………………………………...… 84

Social Cognitive Neuroscience……………………………………………….... 86

Reducing Prejudice ………………………………………………………….… 88

Future Directions……………………………………………………………..… 92

vii

Conclusion.…………………………………………………………………..… 96

References……………………………………………………………………………… 97

Footnotes…………………………………………………………………….……….… 118

Tables………………………………………………………………………..……….… 122

Figures………………………………………………………………………………..… 126

Appendices…………………………………………………………………………...… 142

viii

List of Tables

Table 1. Descriptive statistics for in-scanner ratings. Means and standard deviations

(SD) are provided for accuracy and reaction times of responses to Black-

ingroup, Black-outgroup, White-ingroup, and White-outgroup. Accuracy = the

proportion of trails with correct response during the two-second face

presentation; Reaction Time = time in ms between the presentation of the face

and the response. Excludes all trials where the reaction time ≤ 300 ms

(Experiment 1)……………………………………………………………………

122

Table 2. Brain activity as a function of group membership (ingroup vs. outgroup) and

race (Black vs. White) of the faces. Full brain analyses (p ≤ .001) and region of

interest analyses (p ≤ .01) are based on threshold activity in 10 or more

contiguous voxels. Brodmann Areas (BA) and MNI coordinates (x, y, z) of

activation are provided (Experiment 1).……………………………………....…

123

Table 3. Brain activity as a function the interaction between group membership

(ingroup vs. outgroup) and race (Black vs. White) of the faces. Full brain

analyses (p ≤ .001) and region of interest analyses (p ≤ .01) are based on

threshold activity in 10 or more contiguous voxels. Brodmann Areas (BA) and

MNI coordinates (x, y, z) of activation are provided (Experiment 1)……………

124

Table 4. Brain activity as a function of race (Black vs. White) of the unaffiliated

control faces. Full brain analyses (p ≤ .001) and region of interest analyses (p ≤

.05) are based on threshold activity in 10 or more contiguous voxels. Brodmann

Areas (BA) and MNI coordinates (x, y, z) of activation are provided

(Experiment 5).……………………………...……………………………………

125

ix

List of Figures

Figure 1. The explicit (group) and implicit (race) categorization tasks during fMRI.

There were two explicit and two implicit categorization blocks in each of six

runs. Each block started with a directions screen followed by twelve randomly

presented faces. Faces presented within each block were separated by fixation

crosses. After the completion of each block, directions for the next block

appeared (Experiment 1).……………..………………………………….………

126

Figure 2. The effect of Group membership (Ingroup, Outgroup) on self-reported

liking on a 6-point scale (1 = dislike to 6 = like). Ratings are centered on the

scale midpoint (3.5) such that more positive scores represent liking and more

negative scores represent disliking. Error bars show standard errors

(Experiment 1).…………………………………………………………..………

127

Figure 3. Maps of brain activity stronger to ingroup than outgroup faces in the (A)

fusiform (coronal view; y = -48) and (B) amygdala (coronal view; y = 0)

(Experiment 1).……………………………………………………………..…….

128

Figure 4. (A) Map of brain activity stronger to ingroup than outgroup faces in the

OFC (sagittal view; x = -24) and (B) the correlation between OFC (ingroup-

outgroup) and mean ingroup bias (ingroup-outgroup) on self-reported liking

(Experiment 1).…………………………………………………………………..

129

Figure 5. Individual differences in OFC activity (ingroup - outgroup) mediate the

effect of group membership on individual differences in self-reported ingroup

bias (ingroup - outgroup) (Experiment 1).………………………………………

130

Figure 6. The effect of prime Race (Black, White) and Group membership (Ingroup,

x

Outgroup) on response accuracy to positive and negative words (0.5 = chance

responding). Higher scores represent greater accuracy and lower scores

represent less accuracy. (A) Outgroup members show the standard pattern of

racial bias, whereas (B) Ingroup members are evaluated positively, regardless

of race. Error bars show standard errors (Experiment 2).…………………..……

131

Figure 7. The effect of Race (Black, White) and Group membership (Ingroup,

Outgroup) on self-reported liking on a 6-point scale (1 = dislike to 6 = like).

Error bars show standard errors (Experiment 2).…………………………...……

132

Figure 8. The effect of prime Race (Black, White) and Group membership (Ingroup,

Outgroup, Unaffiliated) on response accuracy to positive and negative words

(0.5 = chance responding). (A) Outgroup members and (B) unaffiliated faces

show the standard pattern of racial bias, while (C) Ingroup members are

evaluated positively, regardless of race. Error bars show standard errors

(Experiment 3).……………………………………………………………...……

133

Figure 9. The effect of Race (Black, White) and Group membership (Ingroup,

Outgroup, Unaffiliated) on self-reported liking on a 6-point scale (1 = dislike to

6 = like). Error bars show standard errors (Experiment 3).………………………

134

Figure 10. The effect of Race (Black, White) on accuracy to faces during a memory

task. Higher scores represent greater accuracy and lower scores represent less

accuracy (0.33 = chance responding). Error bars show standard errors

(Experiment 4).……...………………………….……………………….……….

135

Figure 11. The effect of Group membership (Ingroup, Outgroup, Unaffiliated) on

accuracy to faces during a memory task. Higher scores represent greater

xi

accuracy and lower scores represent less accuracy (0.33 = chance responding).

Error bars show standard errors (Experiment 4).………………………..……….

136

Figure 12. The effect of Group membership (Ingroup, Outgroup) on self-reported

liking on a 6-point scale (1 = dislike to 6 = like) for faces that were correctly

and incorrectly identified during the memory task. Error bars show standard

errors (Experiment 4).………...………………………….………………………

137

Figure 13. The effect of Race (Black, White) and Group membership (Ingroup,

Outgroup, Unaffiliated) on accuracy to faces during a memory task. Higher

scores represent greater accuracy and lower scores represent less accuracy (0.33

= chance responding). Error bars show standard errors (Experiment 5)……...….

138

Figure 14. The effect of Race (Black, White) and Group membership (Ingroup,

Outgroup, Unaffiliated) on self-reported liking on a 6-point scale (1 = dislike to

6 = like). Error bars show standard errors (Experiment 5).………………………

139

Figure 15. Mean reaction times (ms) to Ingroup and Outgroup faces paired with

positive and negative words. Lower scores represent stronger associations

between group membership (Ingroup, Outgroup) and valence stimuli (i.e.,

between ingroup faces and positive words). Error bars show standard errors

(Experiment 5).………….………………...………………….………………….

140

Figure 16. (A) Map of brain activity stronger to Ingroup than Outgroup faces (axial

view; x = -17), fusiform gyrus is region in yellow in the bottom-right corner and

(B) the correlation between fusiform activity (ingroup-outgroup) and own-

group memory bias (ingroup-outgroup) (Experiment 5).………………………..

141

xii

List of Appendices

Appendix A: Core Concepts and Definitions……………………………………..…… 142

1

Chapter 1: Social Perception and Evaluation

Albert Einstein famously observed that “few people are capable of expressing with

equanimity opinions which differ from the prejudices of their social environment.” Prejudice

has marked much of human history, from the feud between the Hatfields and McCoys to the

pogroms of the Second World War. The pervasive and destructive nature of social prejudice

has motivated social scientists to understand and ultimately reduce prejudice. Following this

tradition, the current dissertation provides of series of experiments designed to explore how

social categorization can both elicit and erase intergroup biases in perception and

evaluation.1

From a functional perspective, an essential facet of human cognition is the ability to

quickly divide the world into different categories of objects, events and people. The process

of categorization sorts stimuli on the basis of similarity in a spontaneous and reflexive

fashion allowing individuals to efficiently process an otherwise overwhelming amount of

incoming information and generalize existing knowledge to new stimuli (Bruner, 1957). To

simplify the challenges of social living people engage in a constant and reflexive process of

categorizing others according to their race, gender, age or other salient social category

(Brewer, 1988; Fiske & Neuberg, 1990). The moment people are categorized they are

associated with information about the social category at the cost of considering the entire

constellation of their unique characteristics, compromising accuracy for efficiency.2 By

construing others on the basis of a salient social category, perceivers can make use of a host

of information about that social category, including cultural stereotypes and personal biases.

2

In doing so, social categorization can elicit prejudiced perceptions, thoughts, and evaluations

(Allport, 1954).

The Nature of Prejudice

A host of social, cultural, and developmental factors sew the seeds for prejudice. In

contemporary society exposure to stereotypes and biases is remarkably pervasive. People

inherit biases from family (Aboud, 1988), peers (Bagley & Verma, 1979), and the media

(Jones, 1997) through an unrelenting stream of information about various social groups.

Moreover, these biases are woven into our culture and social institutions, from the number of

minority teachers in the classroom to the representation of females in corporate board rooms.

The power of social learning (Bandura, 1977) leaves few immune to the influence of these

biases. Over a lifetime constant exposure to stereotypes and prejudices generates deeply

entrenched associations (Staats & Staats, 1958) that color the way people see, feel and act

toward others. Indeed, racial biases develop early in life and remain stable across the lifespan

despite the acquisition of more egalitarian attitudes with age (Baron & Banaji, 2006).

Dating to the early 20th

century, scholars began to notice a growing tension between

racial bias and widely help philosophical beliefs in liberty, equality, and fraternity.

Economist Gunnar Myrdal (Myrdal, 1944) described this “ever-raging conflict” as the

American Dilemma. Different from the bigots or old-fashioned racists who feel wholly

justified in their prejudices, conflicted forms of prejudice became an increasingly common

affliction among North Americans who pay for their prejudices with guilt or compunction

(Allport, 1954).

The discrepancy between overt expressions of prejudice and underlying bias has been

magnified by social and legislative injunctions against formal or explicit forms of prejudice

3

(e.g., Brown vs. Board of Education, 1954). Although expressions of prejudice and blatant

forms of discrimination have declined over the past several decades (Bobo, 2001) the deeply

rooted biases often remain. In fact, racial prejudice and discrimination are evident when they

are measured through unobtrusive or subtle means (Crosby, Bromley, & Saxe, 1980), the

social norms against the expression of prejudice are ambiguous (Gaertner & Dovidio, 1977),

or there is competition over limited resources (Levine & Campbell, 1972). These bias are

especially destructive when people can justify discrimination on an ideological basis (Jost &

Banaji, 1994; Sidanius & Pratto, 1999) or perceive a real or symbolic threat (Stephan &

Stephan, 2000).3 These biases are even expressed (usually in a more subtle fashion) by

people who explicitly endorse egalitarian values, including those who genuinely believe they

are non-prejudiced (Gaertner & Dovidio, 1986; Pettigrew & Meertens, 1995).

The persistence of racial bias in the face of egalitarian values led researchers to propose

a distinction between automatic and controlled processes in social prejudice. According to

this approach, associations with race and other social categories are often so well learned that

they are automatically activated upon encountering members of these groups (Devine, 1989;

Fazio, Jackson, Dunton, & Williams, 1995; Greenwald, McGhee, & Schwartz, 1998). For

example, Devine (1989) proposed that the activation of stereotypes occurs without intention,

effort, or conscious control, and regardless of personal prejudice toward a group, making it a

virtually unavoidable aspect of intergroup perception. These initial insights were bolstered by

the development of several implicit measures (Fazio et al., 1995; Greenwald et al., 1998;

Payne, Cheng, Govorun, & Stewart, 2005) that have documented the presence of these biases

in people who report egalitarian values (Cunningham, Nezlek, & Banaji, 2004). These

measures revealed that the majority of North Americans appear to have at least some

4

automatic racial bias against Blacks (Nosek, Banaji, & Greenwald, 2002) suggesting that

these biases are remarkably pervasive. Moreover, people with stronger automatic racial bias

display more subtle, non-verbal discrimination in interracial interactions (Dovidio,

Kawakami, & Gaertner, 2002; McConnell & Leibold, 2001).

People are remarkably adept at dividing up the world into us and them, even in the

absence of any factors typically posited to account for intergroup bias, such as stereotypes,

prior contact with ingroup or outgroup members and competition over resources. This simple

act of categorizing oneself as a member of a group happens in every culture (Brown, 1991 )

and evokes a range of affective, cognitive, and behavioral preferences for ingroup members

(Tajfel, 1970). A series of classic minimal group studies randomly assigned participants to

groups on the basis of unimportant and arbitrary distinctions, such as whether they tend to

overestimate or underestimate the number of dots on a screen, and found that they

preferentially allocated money to fellow ingroup (e.g., under-estimators) compared to

outgroup members (e.g., over-estimators) (Tajfel, Billig, Bundy, & Flament, 1971). This

pattern of minimal intergroup bias has been replicated dozens of times and has been

characterized as ingroup bias (see Brewer, 1979 for a review), rather than outgroup

derogation. Moreover, minimal intergroup bias is not limited to conscious judgments and

behavior, it also leads to preferences for ingroup over outgroup members on automatic

attitude measures (Ashburn-Nardo, Voils, & Monteith, 2001; Otten & Wentura, 1999; see

also Perdue, Dovidio, Gurtman, & Tyler, 1990). Thus, automatic intergroup biases can

emerge on the basis of simple intergroup distinctions in the absence of well-learned

stereotypes and prejudices.

5

The past half-century of research on prejudice has painted a troubling picture. Most

North Americans appear to hold biases toward racial and other minority groups that are

triggered by ordinary cognitive processes, like categorization. Race, in particular, affects

categorization within milliseconds (Ito & Urland, 2003) and appears to be highly salient and

difficult to suppress (Gergen, 1967; Park & Rothbart, 1982). This has lead several

researchers to conclude that race is automatically encoded (Hewstone, Hantzi, & Johnston,

1991; Stangor, Lynch, Duan, & Glass, 1992) whereas other social categories (religion,

occupation, etc.) may be easier to suppress. Further, attempts to suppress racial bias often

backfire, leading to mental exhaustion (Richeson & Shelton, 2003), the increased use of

stereotypes (Macrae, Bodenhausen, Milne, & Jetten, 1994), or worse – unfriendly interracial

interactions (Norton, Sommers, Apfelbaum, Pura, & Ariely, 2006). Even when racial bias is

temporarily suppressed or modulated, it often remains lurking just below the surface, until

the right context or justification triggers a reappearance. It is therefore unsurprising that

studies continue to reveal racial bias in real-world contexts, including discrimination in

hiring (Bertrand & Mullainathan, 2003) and mortgage lending (M. A. Turner et al., 2002).

The issue of racial bias provides a formidable opponent for researchers and policy makers

concerned with reducing prejudice and discrimination.

One of the reasons race may serve as such a powerful trigger for social prejudice and

discrimination is because it provides a cue to group membership. In social environments

where there is less than complete racial integration – whether in the past or present – race or

ethnicity may provide a visually salient cue to group membership (Cosmides, Tooby, &

Kurzban, 2003; Sidanius & Pratto, 1999). When race co-varies with the categorization of

others into us versus them, race may become a particularly salient and stable basis for social

6

perception and evaluation, incorporating the well-learned stereotypic and evaluative

associations with the human propensity for ingroup favoritism (Brewer, 1979). Alternatively,

when race is unrelated to group membership, other arbitrary intergroup dimensions other

than race may form the basis of social perception and evaluation (Kurzban, Tooby, &

Cosmides, 2001; Sidanius & Pratto, 1999). A series of studies by Kurzban and colleagues

(2001) recently examined this possibility using a classic memory confusion paradigm

designed to reveal how people categorize others (Taylor, Fiske, Etcoff, & Ruderman, 1978).

In these experiments participants were asked to form impressions of eight individuals. They

saw a series of statements, each of which was paired with a photo of the individual who said

it and were asked in a subsequent task to recall which statements were made by each

individual. The individuals varied independently on race (half were White and half were

Black) and group membership (half were members of one group and half were members of

another). Consistent with previous research (Stangor et al., 1992), participants made more

within-race than between-race errors during recall. That is, they were more likely to

misattribute statements from one Black individual to another, rather than a White individual,

indicating that they were using race to encode the statements. Participants also made more

within-group than between-group errors, indicating that they were also using group

membership to encode the statements. Further, when group membership was made visually

salient by showing pictures of one group wearing yellow shirts and the other group wearing

grey shirts, the effect of group membership on encoding became larger than race, suggesting

that participants were using group membership more than race to categorize individuals.

More importantly, this visually salient group distinction actually decreased race-based

encoding.

7

The most arresting aspect of this research is that very brief exposure to these intergroup

alliances was sufficient to elicit categorization according to group membership and make this

a more potent social category than race, a category marked by years of exposure. This

demonstrated that categorization along racial lines may be malleable to a certain type of

social context: one in which race is irrelevant to group membership. This research also raised

a number of important questions. First, would the effects of group membership extend to

evaluations? Categorization elicits evaluation, and perceiving people according to their group

membership should reduce racial bias. Further, mere membership in an arbitrary group is

sufficient to increase evaluative and behavioral preferences for ingroup members (Brewer,

1979); presumably people who are actually assigned to one of the groups should be

especially prone to using group membership as a cue for categorization rather than race, and

possibly show a preference for ingroup members, regardless of race. Second, were these

changes in categorization specific to Black or White faces? The experiments by Kurzban and

colleagues (2001) only examined the main effects of race and group membership on

encoding. It is possible that group membership only affected the categorization of White

faces, which would limit the social implications of this research. Much of the research cited

above illustrates prejudice and discrimination to Blacks, making the more pertinent issue

whether group membership erases racial biases against visual minorities (i.e., Blacks). This

dissertation addresses these questions.

Social Identity and Self-Categorization

“Man in his totality is a dynamic complex of ideas, forces, and possibilities.

According to the motivations and relations of life and its changes, he makes of

himself a differentiated and clearly defined phenomenon. As an economic and

political man, as a family member, and as a representative of an occupation he

is, as it were, an elaboration constructed ad hoc” – Simmel (Simmel & Wolf,

1950)

8

The research on minimal groups highlights the context dependent nature of self-

categorization and social perception. As Georg Simmel observed, people have many dynamic

and overlapping social identities, and their self-categorization with any of these identities –

however minimal – can shift from one situation to another (Tajfel, 1982). When one of these

social identities is salient people are more likely to perceive themselves and others as

interchangeable exemplars of a social category rather than unique individuals. This self-

categorization, in turn, colors social perceptions and evaluations of the self and others in line

with contents of the current self-categorization (J. C. Turner, Hogg, Oakes, Reicher, &

Wetherell, 1987; J. C. Turner, Oakes, Haslam, & McGarty, 1994). This perspective suggests

that social perception and evaluation of complex social stimuli are context-dependent and

specific to the current self-categorization. Indeed, most studies on multiply-categorizable

social stimuli suggest they are evaluated according to the most important or salient social

category (Macrae, Bodenhausen, & Milne, 1995; Mitchell, Nosek, & Banaji, 2003; Mullen,

Migdal, & Hewstone, 2001; Urban & Miller, 1998). The purpose of the present research is to

examine whether social perception and evaluation reflect the current self-categorization

rather than more visually salient social categories, even when this self-categorization is

relatively arbitrary. More specifically, this dissertation explores whether self-categorization

with a novel mixed-race group modifies people’s evaluation, memory, and neural processing

of others in terms of their current self-categorization rather than race.

Models of Evaluation

The past few decades of research in social psychology have shown that social

categorization and evaluation occur automatically (Bargh, Chaiken, Govender, & Pratto,

9

1992; Fazio, Sanbonmatsu, Powell, & Kardes, 1986), providing a platform for judgments and

behavior. Although these automatic evaluations are often functional, they can also include

inaccurate or prejudiced responses which conflict with more controlled judgments and values

(Devine, 1989). These insights lead to a distinction between automatic perceptions and

evaluations that occur rapidly and without intent, and slower evaluations that reflect current

goals and motivations (see Chaiken & Trope, 1999). While automatic and controlled aspects

of evaluation are often in agreement, they may also be dissociated (Greenwald & Banaji,

1995; T. D. Wilson, Samuel, & Schooler, 2000), especially when people have the motivation

and opportunity to modulate their initial evaluations (Fazio & Towles-Schwen, 1999).

These dual process models highlight the fact that evaluations are multifaceted and

complex. Yet, few of these models account for the interaction between automatic and

controlled processes that give rise to different evaluations and the role context plays in

shaping the most automatic aspects of evaluative processing (e.g., Gawronski &

Bodenhausen, 2006; Strack & Deutsch, 2004). In contrast, the Iterative Reprocessing Model

(Cunningham & Zelazo, 2007; Cunningham, Zelazo, Packer, & Van Bavel, 2007) proposes a

multi-level framework based on recent advances in social psychology and cognitive

neuroscience for understanding how multiple processes interact during evaluation. The IR

model proposes that current evaluations are constructed from relatively stable attitude

representations (a subset of which are active at any given time) stored in memory through

iterative and interactive evaluative processing. A fundamental assumption underlying the IR

model is that brain systems are organized hierarchically, such that automatic processes

influence and are influenced by more controlled processes. Whereas automatic processes

provide relatively coarse perceptual and evaluative information, with additional iterations

10

more controlled processes can interact with automatic processes and provide more nuanced

or contextually appropriate evaluations. These higher-order controlled processes (mediated

by the prefrontal cortex) can incorporate goals and context in the current evaluation, and set

the stage for automatic construals of stimuli consistent with these goals or contextual

information.

The IR Model provides a framework for understanding why automatic evaluations,

which are based on ostensibly stable underlying associations, appear sensitive to the

motivational and social context (see Blair, 2002 for review). Automatic racial bias, for

example, is reduced when White participants are placed in a subordinate role relative to a

Black partner (Richeson & Ambady, 2003), when they are exposed to admired Black

exemplars (Dasgupta & Greenwald, 2001), or when they are in the presence of a Black

experimenter (Lowery, Hardin, & Sinclair, 2001). Similarly, categorizing complex social

stimuli in different ways moderates the activation of underlying automatic attitudes, leading

to different evaluations: categorizing Black athletes and White politicians according to race

activates an automatic preference for White politicians; however, categorizing the same

individuals according to occupation activates an automatic preference for Black athletes

(Mitchell et al., 2003). These studies illustrate how ostensibly stable associations can give

rise to contextually sensitive automatic evaluations of social categories.

The IR Model provides a framework to understand how self-categorization can

impact intergroup perception and evaluation. Drawing on the distinction between evaluations

and attitudes, the current self-categorization can be constructed from relatively stable

contents of a given identity (a subset of which are active at any given time) stored in

memory. This allows for the “inherently variable, fluid, and context dependent” nature of

11

self-categorization (J. C. Turner et al., 1994) while retaining the a set of stable personal and

social identities. More important, the IR Model proposes that motivations or context can set

the stage for automatic construals and evaluations of stimuli. In this way, social contexts can

trigger the activation of a particular social identity that can, in turn, elicit certain perceptions

and evaluations. Finally, the IR Model proposes that multiple competing identities can give

rise to dissociations between automatic evaluations and higher-order goals, leading to the

modulation of lower-order processes to generate evaluations congruent with current goals

and values.

This dissertation examines these processes across multiple levels-of-analysis, linking

the effects of self-categorization and social identity on social perception and evaluation to

brain function. Specifically, we assigned participants to minimal, mixed-race groups to test

whether social perception and evaluation reflect the current salient social identity – however

minimal – when another social category is visually and social salient. We examined these

effects on race because it affects categorization within milliseconds (Ito & Urland, 2003),

appears to be highly salient and difficult to suppress (Park & Rothbart, 1982) and is

associated with presumably stable and pervasive racial biases in memory and evaluation.

Previous social identity research has focused on what it accomplishes (computation), whereas

relatively little research has examined how identity modulates information processing

(algorithmic) and the brain regions responsible (implementation) – critical levels-of-analysis

for understanding complex information processing systems (Marr, 1982). This dissertation

represents the first steps in developing a multi-level model of social identity and self-

categorization that includes computational, algorithmic and implementational aspects of

information processing.

12

Chapter 2: A Social Cognitive Neuroscience Approach

One of the most promising developments in the social and cognitive sciences is the

increased integration across multiple-levels of analysis. In particular, the emergence of social

cognitive neuroscience holds the promise of understanding human social behavior by

investigating the affective and cognitive operations of the human brain (Cacioppo, Berntson,

Sheridan, & McClintock, 2000; Heatherton, Macrae, & Kelley, 2004; Ochsner & Lieberman,

2001). This approach is based on the assumption that complex, multi-faceted issues like

social prejudice cannot be fully understood by either a strictly social or biological approach.

By exploring the links between brain and behavior, social cognitive neuroscience can help

break complex social phenomenon like social prejudice into component processes, and

identify the operating characteristics of these components and how they work in concert to

solve the complex task of successful social navigation (Cunningham & Johnson, 2007). This

approach offers the benefit of generating more general, process-focused models of human

cognition and provides important constraints from adjacent levels-of-analysis.

Although social cognitive neuroscience is a relatively young field it has already

generated theoretical breakthroughs in several central domains of social psychology.

Research on social rejection and ostracism, for example, has been recently examined using

tools from cognitive neuroscience (Eisenberger & Lieberman, 2004; Eisenberger, Lieberman,

& Williams, 2003; G. MacDonald & Leary, 2005). Linking across levels of analysis revealed

diverse evidence showing that social pain triggers brain and other physiological systems

known to process physical pain. Finding that emotional pain is processed by mechanisms

used for physical pain, lead to the novel hypothesis that analgesic drugs should diminish

13

responses to social rejection. As predicted, participants randomly assigned to take

acetaminophen (vs. a placebo) for three weeks reported significantly lower daily hurt feelings

than those taking a placebo, an effect that grew stronger each day to the end of the study

(DeWall, MacDonald, Webster, Tice, & Baumeister, 2007). This line of research illustrates

some benefits of a multilevel approach to issues that fall within the traditional boundaries of

the social sciences.

Similarly, research on the neural bases of prejudice has helped corroborate social

cognitive models and provided novel insights. In particular, social cognitive neuroscience

research on prejudice has illustrated the surprising speed with which people start to control

their racial stereotypes (Amodio et al., 2004) and the complex interactions between automatic

and controlled processing (Cunningham, Johnson et al., 2004). These and other findings fall

outside traditional social psychological theory on these issues and highlight the need for

complex multilevel theories that account for the operating characteristics of the human social

cognitive apparatus. A general conclusion from this research is that perceptions and

evaluations based on social categories (typically race) are related to a host of neural

processes from early visual processing to higher-level aspects of executive function

(Cunningham & Johnson, 2007). This widely distributed pattern of brain activity suggests

that social categorization influences not a single “group perception” module, but rather a

constellation of neural processes that collectively give rise to an array of social biases.

This initial social cognitive neuroscience research on social prejudice provided

important hints for understanding the mechanisms of prejudice and intergroup discrimination

and offered exciting evidence of the automaticity of intergroup perception and evaluation,

and the complex interactions between the component processes that guide behavior. But

14

these discoveries also raise important questions about the precise processes and computations

underlying these patterns of brain activity. A goal of this dissertation is to investigate the role

of specific psychological process (e.g., self-categorization) associated with certain brain

regions using experimental manipulation and convergent behavioral evidence. The

experiments in this dissertation move between levels-of-analysis, using social theory and

behavioral data to generate predictions for brain activity (e.g., Experiments 1 and 5) and

using neuroscience data to generate behavioral prejudices (e.g., Experiment 4), in an iterative

fashion. The ultimate goal of this research is to generate a multi-level model of social identity

and self-categorization.

Social Perception and Categorization

As noted in the first chapter, nearly a half century of empirical research has examined

the way social categories alter the way that people see the social world. One of the most

robust and widely replicated phenomenon in social perception is the fact that people appear

to be better at remembering people from their own race than from other races (Malpass &

Kravitz, 1969)– an effect that has been variably termed the cross-race effect, same-race bias

or own-race bias (ORB). Although the ORB may appear relatively harmless, it can have

serious real-world implications. For example, the ORB can lead an eyewitness in a criminal

trial to misidentify a suspect from another race, which can lead to a wrongful conviction and

even a death sentence for an innocent person (Brigham & Ready, 2005). The effects may be

especially pernicious when paired with racial stereotypes and prejudices (e.g., Seeleman,

1940). Despite the long history of research on the ORB there remains no theoretical

consensus about the source of this bias.

15

The most popular account of ORB is a perceptual expertise model. From this

perspective, a lifetime of experience interacting with own-race family, friends and

acquaintances relative to members of another race produces a specific expertise encoding

and/or recalling own-race faces (e.g., Valentine & Endo, 1992). Accordingly, a recent

functional magnetic resonance imaging (fMRI) study explored the relationship between the

ORB and brain regions sensitive to perceptual expertise (Golby, Gabrieli, Chiao, &

Eberhardt, 2001), namely an area of visual cortex known as the fusiform gyrus. To examine

the role of the fusiform in the ORB, Black and White participants viewed same-race and

other-race faces, as well as objects (radios) during fMRI. Brain activity to the faces was first

contrasted with objects to identify the fusiform face area (FFA), a sub-region of the fusiform

gyrus that responds preferentially to faces relative to other visual stimulus (Kanwisher,

McDermott, & Chun, 1997). Identifying the FFA was important because this region has been

shown to play an important role in perceptual expertise (Gauthier, Tarr, Anderson,

Skudlarski, & Gore, 1999). As expected, the FFA was more sensitive to own-race than other-

race faces for both Black and White participants (see also Lieberman, Hariri, Jarcho,

Eisenberger, & Bookheimer, 2005). Moreover, the degree of same-race bias on a subsequent

memory test, (i.e., superior memory for same-race over other-race faces) correlated with the

degree of same-race bias in the fusiform gyrus. These results suggested that the fusiform may

mediate the effect of perceptual expertise on own-race bias.

It is possible that ORB is the result of self-categorization (Levin, 2000; Sporer, 2001).

According to this view, categorizing others as ingroup or outgroup members may alter the

depth or type of processing that they receive (Bernstein, Young, & Hugenberg, 2007). For

instance, people might view ingroup members as more important and be more likely to

16

process them as individuals, in contrast to less relevant outgroup members who are lumped

together (even perceptually) simply as “them.” Whereas ingroup members are processed as

individuals, extracting information about what makes each person unique, outgroup members

are processed as interchangeable members of a general social category (see also Outgroup

Homogeneity Effect, Ostrom & Sedikides, 1992). Indeed, the fusiform appears to play a role

in individuation (Rhodes, Byatt, Michie, & Puce, 2004), raising the possibility that the ORB

in fusiform activity may actually play a role in individuating racial ingroup members, rather

than responding to a class of stimuli that are well known. These studies highlight the effects

of social and self-categorization on early aspects of social perception. The ways in which

people divide others into meaningful categories may influence the way they see the social

world and affect downstream cognitive and affective processes, and ultimately, behavior.

Social Evaluation

The central focus of social cognitive neuroscience research on social prejudice has

been the affective response people have to race. Although the neural networks involved in an

affective evaluative response are likely distributed (Cunningham et al., 2007), initial research

focused on the amygdala. The amygdala is a small structure in the temporal lobe linked to an

array of social and affective processes, including learning emotional information (for a

review see Phelps, 2006), perceiving emotional faces (Whalen et al., 1998), and directing

attention to important stimuli (Vuilleumier, 2005). The amygdala has also been implicated in

fear conditioning (LeDoux, 2000) and processing negative stimuli (Cunningham, Johnson,

Gatenby, Gore, & Banaji, 2003; Hariri, Tessitore, Mattay, Fera, & Weinberger, 2002), even

during rapid (33ms) – and perhaps non-conscious – presentations (Whalen et al., 1998),

suggesting that this region may play a critical role in rapid and unconscious evaluation of the

17

environment. As described in the previous chapter, racial biases are often automatic, making

the amygdala a promising starting point for understanding the neural processes that underlie

social prejudice.

Initial studies on the neural substrates of social prejudice examined amygdala activity

to faces from different racial groups. The first fMRI study of racial processing presented

blocks of Black and White faces to Black and White participants (Hart et al., 2000) and

found greater amygdala activation to racial outgroup than ingroup faces (i.e., White

participants viewing Black faces and Black participants viewing White faces). Although this

initial finding was qualified by a small sample (N = 8) and relatively weak effects (i.e., the

reported pattern was only observed in the second half of the study), a number of subsequent

studies have found similar effects (e.g., Krendl, Macrae, Kelley, & Heatherton, 2006;

Lieberman et al., 2005; Ronquillo et al., 2007), indicating that the amygdala may play a role

in racial bias.

The role of the amygdala in processing rapid affective stimuli has also lead to

research examining the link between the amygdala and automatic racial bias. In one paper,

Phelps and colleagues (Phelps et al., 2000) examined whether White participants with more

automatic racial bias also had the strongest amygdala response to Black (>White) faces

during fMRI. Although the study did not find overall greater amygdala activation to Black

than White faces, individual differences in amygdala activity to Black-White faces were

correlated with two indirect measures of racial bias; the Implicit Association Test

(Greenwald et al., 1998) and the startle eye-blink (a physiological measure). However, a

direct measure of prejudice, the Modern Racism Scale (McConahay, 1986), was uncorrelated

with amygdala activity. The dissociation between these direct and indirect measures of racial

18

bias and amygdala activity was consistent with the view that the amygdala is involved in

automatic racial bias.

Automatic and Controlled Processing

As described in the previous chapter, the dissociation between automatic and controlled

measures likely stems from different aspects of evaluative processing. Under certain

circumstances, people can control their automatic responses, and generate different or more

nuanced evaluations and judgments in the service of their goals and values (Cunningham &

Zelazo, 2007; Fazio, 1990; Greenwald & Banaji, 1995). The study of prejudice regulation in

social psychology and in social cognitive neuroscience has tended to focus on the inhibition

or suppression of automatic evaluations deemed inappropriate or suboptimal (Devine, 1989;

Petty & Wegener, 1993). Specifically, when people have the motivation and opportunity to

engage more controlled processing, the influence of automatically-activated stereotypes and

prejudice is dramatically reduced (Devine, 1989; Dovidio, Kawakami, Johnson, Johnson, &

Howard, 1997; Fazio et al., 1995). In this view, the automatic activation of prejudiced

representations and biased processes lead to discriminatory behavior unless controlled

processes driven by values, goals and motivations reduce these biases.

Cognitive neuroscience has identified at least two separate, but related, neural systems

involved in controlled processing: a conflict-detection and a regulatory-control system

(Botvinick, Braver, Barch, Carter, & Cohen, 2001; A. W. MacDonald, Cohen, Stenger, &

Carter, 2000). The conflict-detection system monitors current ongoing processing and

provides a signal to other brain regions when incompatible representations are active,

including conflicts between automatic responses and current goals. This signal often

indicates the need for additional processing from the regulatory control system. This later

19

system implements more controlled processing to resolve conflict and direct processing in a

goal-congruent fashion. The conflict-detection system is thought to be mediated by the

anterior cingulate cortex (ACC) and the slower, regulatory-control system is through to be

mediated by regions of anterior and lateral prefrontal cortex (PFC). In the context of

prejudice, automatic racial biases that contrast with egalitarian goals should trigger the

conflict-detection system, which would then recruit the regulatory-control system to modify

biased processing.

This proposed relationship between automatic and controlled processing in racial bias

was recently examined in a fMRI study (Cunningham, Johnson et al., 2004). To isolate

automatic and controlled processing of race, several egalitarian White participants were

presented with Black and White faces for 30ms or 500ms based on the assumption that rapid

subliminal images would elicit automatic (and unconscious) racial processing, such as the

amygdala, whereas the supraliminal would elicit more controlled processing in the ACC and

lateral PFC. As predicted, there was greater amygdala activity following the subliminal Black

than White faces and this differential amygdala activity was highly correlated with racial bias

on the Implicit Association Test (IAT, Greenwald et al., 1998), replicating previous research

(Phelps et al., 2000). In contrast, supraliminal Black (> White) faces were associated with

activity in brain regions involved in conflict-detection and regulatory-control: the ACC and

lateral PFC. Moreover, the reduction in amygdala activity to race between the subliminal and

supraliminal conditions was negatively correlated with ACC and lateral PFC activity,

suggesting that these egalitarian participants were controlling aspects of their automatic

racial bias. Another study found a similar pattern, such that amygdala activity to Black faces

was negatively correlated with lateral PFC faces (Lieberman et al., 2005). Taken together,

20

these results support the idea that people can modify automatic biases via controlled

processing when they have both the motivation and opportunity.

In the short term, the controlled suppression of automatic biases can reduce the

expression of these biases. However, controlled processing has a number of important

limitations. There is extensive experimental evidence, for example, that controlled processes

operate like a limited resource (Baumeister, Bratslavsky, Muraven, & Tice, 1998) due to

metabolic constraints (Gailliot & Baumeister, 2007). Specifically, attempts to suppress or

override automatic or pre-potent responses reduce controlled resources. Thus, participants

with a large discrepancy between the strength of their automatic racial biases and their

current goals will deplete their controlled resources faster than others, leading to the most

discrimination during extended or sequential interracial interactions. Indeed, White

participants with high levels of automatic racial bias on an IAT have the worst cognitive

control on a Stroop task following an interracial interaction (Richeson & Shelton, 2003).

Presumably, participants with the most automatic racial bias had the most bias to control, and

were therefore cognitively depleted.

To examine whether this reduction in control was mediated by the neural systems

described above, the authors conducted a follow-up fMRI study (Richeson et al., 2003).

White participants completed a measure of their automatic racial bias, viewed Black and

White faces during fMRI, interacted with a Black confederate and then performed the Stroop

task. As would be expected if participants were attempting to control prejudice, participants

had heightened activation in the ACC and lateral PFC while viewing Black faces, similar to

the effects in the supraliminal conditions in the study by Cunningham and colleagues (2004).

In addition, PFC activity to Black faces mediated the relationship between automatic racial

21

bias and impaired cognitive control on the Stroop task following an interracial interaction

(Richeson et al., 2003). In other words, individuals with the highest levels of automatic racial

bias also had the highest levels of PFC activity to Black (> White) faces and the highest

levels of cognitive impairment following an interracial interaction, and PFC activity to Black

faces explained the relationship between automatic racial bias and subsequent impairments in

controlled processing.

These patterns of results support the idea that egalitarian participants appear to control their

automatic racial bias and that this regulation depletes controlled processing resources.

This research paints a disheartening picture for intergroup bias; the people who work

the hardest to control their unwanted bias may ironically be the ones who suffer the costs,

and may ultimately express these biases during sustained interactions. In a similar vein,

efforts to suppress stereotypes and prejudice can rebound and actually increase the

accessibility of these biases above baseline levels (Macrae et al., 1994; Wegner, 1994).4

Moreover, the problems with top-down controlled processing are not limited to the

mistreatment or perception of a target, but may lead to unhealthy physiological side effects

(such as high blood pressure) (Gross, 1998; Gross & Thompson, 2007). However, controlling

prejudice and stereotypes can be accomplished through other means. Instead of response-

focused strategies aimed to suppress or inhibit an affective response, antecedent-focused

strategies can shape the initial activation of an affective response by construing the stimulus

or context differently (Gross & Thompson, 2007). Antecedent-focused forms of control are

adaptive, efficient, and strong, since this initial construal of a stimulus can have a cascade of

downstream effects on evaluation and behavior.

Reconstrual

22

Reconstruing members of stigmatized social categories as individuals or as members

of an alternative valued group may provide a powerful alternative to more heavy-handed

controlled processing. Indeed, changing processing goals can alter the way low-level

processes like the amygdala process positive and negative information about people

(Cunningham, Van Bavel, & Johnsen, 2008). One processing goal that may be an especially

powerful means of reducing bias is to individuate people and place less emphasis on their

group membership (Brewer, 1988; Fiske & Neuberg, 1990). Individuating a member of a

stigmatized group may reduce reliance on social category cues and the activation of

stereotypes and prejudices. To test this hypothesis, a recent fMRI study had White

participants process Black and White faces as individuals or as members of a social group

(Wheeler & Fiske, 2005). Consistent with the research described above, when participants

engaged in social categorization (e.g., classifying the faces by age) they had greater

amygdala activity to the Black than White faces. However, when participants were simply

asked to consider the preferences of each individual (deciding whether each person preferred

certain vegetables) they had greater amygdala activity to White than Black faces –

completely reversing the standard amygdala response to race. This supports the view that

people can shape their own evaluative responses by attending to certain pieces of information

and ignoring others.

Categorization itself is a flexible process and leads to the construal of any given person

according to multiple social groups. Moreover, people have many dynamic and overlapping

social identities, and their current self-categorization with any of these identities can

presumably alter they way they perceive and evaluate others (Tajfel, 1982; J. C. Turner et al.,

1987; J. C. Turner et al., 1994). More specifically, this current self-categorization, whether

23

deliberate or not, can presumably shift the construal of others in a manner congruent with the

current salient identity. Accordingly, self-categorization with any group – however minimal

– should lead to the perception and evaluation of complex social stimuli in terms of their

(minimal) group membership, ignoring other, orthogonal category dimensions.

Overview of the Current Research

This dissertation examines the effects of self-categorization on social perception and

evaluation across multiple levels-of-analysis. Here, self-categorization reflects the ability of

participants to rapidly identify with a novel group and to process others according to this

social identity. To examine these novel self-categorizations on intergroup processing

participants were assigned to one of two novel mixed-race teams in each of five experiments.

Making race orthogonal to team membership was important for a number of reasons. First,

including race provided a more stringent test of the self-categorization hypothesis.

Specifically, because group members were visually identical (the faces of each team were

fully counterbalanced and thus identical across participants) while race was highly visually

salient, any effects of self-categorization would need to override an alternative visually

salient social cue with well-learned semantic and evaluative associations. Second, including

race provided a clear test about the role of self-categorization in a number of racial biases,

including fusiform activity and better memory for own-race faces. If these effects are caused

(at least in part) by self-categorization then the current novel self-categorization should elicit

the same ingroup biases. However, if these effects are caused by factors other than self-

categorization (e.g., stereotypes, experience, novelty, prejudice, etc) then race should

continue to elicit these biases. Finally, including race allowed me to examine the effect of

social perception and evaluation in a more complex social context with multiply-

24

categorizable targets. Although everyone in the real world can be categorized and evaluated

according to any number of criteria, the majority of research on intergroup relations has

focused on simple social contexts with a single salient social category (in fact, other

categories are usually experimentally controlled or counterbalanced to intentionally reduce

their influence on psychological processing). Including orthogonal categories allowed me to

examine the use of multiple social categories and the integration and resolution of this

complexity in social perception and evaluation.

This dissertation includes five experiments exploring the effects of a novel self-

categorization and race on intergroup evaluation, memory, and neural processing.

Experiment 1 examines whether brain regions associated with perceptual and evaluative

processing are sensitive to the current self-categorization (with a novel team) rather than

features associated with race. This experiment also explores the neural mediators of ingroup

bias on self-reported evaluations. Experiment 2 examines the effect of self-categorization

with a novel group on automatic construals and evaluations and whether these biases

override race – a social category known to trigger automatic evaluations regardless of

motivational concerns. Experiment 3 examines the specificity of automatic evaluations

within a given a social context. Participants in this experiment evaluated faces from the two

mixed-race teams plus new Black and White faces unaffiliated with either team to determine

whether a novel self-categorization generates ingroup bias or outgroup derogation.

Experiment 4 examines the effect of self-categorization with a novel group on memory and

whether this own-group memory bias overrides the own-race memory bias. Experiment 5

extends the research from Experiments 1-4 by examining the neural mediators of ingroup

biases in automatic evaluation and memory and examining the neural processing of

25

unaffiliated faces within a given social context (by including unaffiliated control faces during

fMRI). These experiments examine the sensitivity of social perception and evaluation to the

current self-categorization – however minimal – and inform multi-level models of social

prejudice.

26

Chapter 3: The Neural Substrates of Ingroup Bias

Ingroup identities shape not only preferences, thoughts, and behavior, but also basic

perceptions of others. As described in the introduction, people are better at recognizing faces

from their own racial or ethnic groups compared to faces from other racial groups (Malpass

& Kravitz, 1969; Sporer, 2001). Extending this research, Golby and colleagues (Golby et al.,

2001) found that Black and White participants showed heightened activity in the fusiform

face area to racial ingroup faces during fMRI. In previous work, this brain region has been

associated with processing and individuating faces (Kanwisher et al., 1997; Rhodes et al.,

2004) and general perceptual expertise (Gauthier et al., 1999). In addition, participants with

the strongest fusiform activity to racial ingroup (> outgroup) faces displayed the greatest

ORB on a subsequent recognition task. Although the authors argued that own-race biases in

fusiform activity were due to superior perceptual expertise for racial ingroup faces, it remains

unclear whether ingroup biases in perception mediated by the fusiform stem from perceptual

expertise with own-race faces, racial self-categorization, or some combination of these

processes.

Although ORB was originally explained in terms of experience with racial ingroup

members (i.e., people have more interactions with members of their own race), there is

increasing evidence that self-categorization with a group can itself lead to more in-depth or

individuated processing of ingroup members (Sporer, 2001). Indeed, the so-called ORB has

been replicated across a variety of non-racial social categories (e.g., Rule, Ambady, Adams,

& Macrae, 2007), including minimal groups, demonstrating that mere categorization with a

group is sufficient to produce greater recognition of ingroup faces when prior exposure to

27

ingroup and outgroup members is equivalent (Bernstein et al., 2007). This suggests that self-

categorization may motivate ingroup biases in social perception (Balcetis & Dunning, 2006),

including more in-depth or individuated processing of ingroup than outgroup members.

The amygdala has also been implicated in social perception and evaluation. Several

studies have found that viewing images of racial outgroup members activates the amygdala

more than racial ingroup members (Hart et al., 2000), and that this difference in amygdala

activity correlates with implicit measures of racial bias (Cunningham, Johnson et al., 2004;

Phelps et al., 2000). These correlations with racial bias, coupled with studies demonstrating a

link between the amygdala and fear conditioning (LeDoux, 1996), have led researchers to

interpret differences in amygdala activation during intergroup perception as evidence of

negativity (including disgust and fear) toward stigmatized groups (e.g., Lasana T. Harris &

Fiske, 2006; Krendl et al., 2006; Lieberman et al., 2005). However, these data are also

consistent with the recent idea that the amygdala may play a role in processing any

motivationally-relevant stimuli (Anderson & Phelps, 2001; Cunningham et al., 2008;

Whalen, 1998). In contexts where race is the most salient social category, the amygdala may

be responsive to members of groups who are stereotypically associated with threat or novelty

(Dubois et al., 1999). When there are other bases for categorization, however, the amygdala

may be responsive to members of groups that are currently relevant, such as ingroup

members (Allport, 1954) in the minimal group contexts.

Experiment 1

The present experiment was designed to examine the role of self-categorization on

neural processing, especially the amygdala and fusiform gyrus. We used a variant of the

28

minimal group paradigm, in which White participants were randomly assigned to a novel

mixed-race team without a history of contact or conflict (see Kurzban et al., 2001, for a

similar paradigm).5 Participants in several previous fMRI studies varied in terms of their

experience with the target groups (e.g., White participants presented with White and Black

faces), making it possible that differences in the familiarity or novelty of certain groups may

have accounted for differences in fusiform (Gauthier et al., 1999) or amygdala (Dubois et al.,

1999) activity, respectively. However, the novel mixed-race teams used in our study were

equated in terms of familiarity and novelty, ruling these variables out as explanations for any

observed patterns of activation. In addition, by crossing race and group membership we were

able to examine the flexibility of social categorization (J. C. Turner et al., 1987); would

categorizing the social environment in terms of novel group memberships cause participants

to process targets in terms of their current salient group identity rather than race? Following

previous research, we hypothesized that ingroup (> outgroup) members would be associated

with greater activity in the fusiform gyri; however, it remained unclear whether fusiform

activity would be limited to more familiar racial ingroup members (Golby et al., 2001) or

whether it would extend to novel ingroup members, consistent with the motivated social

perception of relevant ingroup members (Sporer, 2001). Previous research also suggested

that amygdala activity might be greater to outgroup members, or at least racial outgroup

members (Hart et al., 2000). However, research linking the amygdala to processing

motivational-relevance (Anderson & Phelps, 2001; Cunningham et al., 2008; Whalen, 1998)

led us to propose that novel ingroup members would instead be associated with greater

amygdala activity.

Method

29

Participants

Twenty-two White participants (14 females, mean age = 25) were paid $50 for

completing the study.6 Participants reported no abnormal neurological history and had

normal or corrected-to-normal vision.

Materials and Procedure

Group Assignment. Participants arrived at the imaging center and posed for a digital

facial photograph. Participants were randomly assigned to one of two teams (Leopards or

Tigers) and completed two brief learning tasks (~15 minutes). Race was orthogonal to team

membership; there were six Black and six White males on each team. Faces were randomly

assigned to teams and fully counterbalanced.7 In the first task, participants spent three

minutes memorizing the team membership of 24 faces: 12 members of the Leopards and 12

members of the Tigers. In the second task, participants were presented with each of the 24

faces one at a time and indicated whether the face was a member of the Leopards or Tigers.

Participants also saw and categorized their own face three times (randomly interspersed

within the learned faces) to enhance the self-relevance of their task and identification with

their team. For the first series of trials, participants were reminded with a label on the screen

whether the face was a Leopard or Tiger. In the second series of trials, this label was

removed so that participants needed to rely only on their memory. Following each trial,

feedback indicated if the response was correct.

Categorization Task. Participants completed six runs of four blocks containing 12 trials

for a total of 288 trials during fMRI. On each trial, participants categorized one of the 24

faces in one of two ways (see Figure 1). Participants did not see their own face during fMRI

or the ratings/memory tasks. On explicit trials, participants categorized each face according

30

to team membership (Leopard or Tiger). These trials are termed explicit because participants’

attention was explicitly focused on group membership. On implicit trials, participants

categorized each face according to skin color (Black or White). Team and race labels were

counterbalanced (left vs. right) within runs, creating four randomized blocks within each run.

Each of the 24 faces was categorized twice in each run (once by team membership and once

by race). Direction screens were presented before each block for six seconds to cue the

categorization required for the following block of 12 trials. Each face appeared for two

seconds, during which time participants responded with a button box in their right hand. To

allow for estimation of the hemodynamic signal, fixation crosses appeared between names

for two, four or six seconds (in pseudo-random order).

Scanning Parameters. Participants were scanned using a Siemens 3T scanner.

Functional scanning was prescribed parallel to the AC–PC line and nearly isotropic

functional images were acquired from inferior to superior using a single-shot gradient echo

planar pulse sequence (32 axial slices; 3.5 mm thick; 0.5 mm skip; TE = 25 ms; TR = 2000

ms; in-plane resolution = 3.5 × 3.5 mm; matrix size = 64 × 64; FOV = 224 mm). Following

functional imaging, a high resolution MPRAGE anatomical image (176 sagittal slices; TE =

2.15 ms; TR = 1760 ms; resolution = 1.0 × 1.0 × 1.0 mm) was collected for normalization.

Rating and Memory Tasks. After scanning, participants completed two computerized

questionnaires. First, participants completed a face rating task in which they were told that

“people can often quickly determine who they like or dislike based on subtle facial features

and expressions” and asked to rate each of the 24 faces on a 6-point liking scale (1 = dislike

to 6 = like). Second, participants completed a team memory task in which they reported

Note new labels

+

Note new labels

+

+

+

Stimulus: 2s

Directions: 6s

Fixation: 2s + Jitter

31

whether each face was a Leopard or Tiger. Faces were presented in random order in both

tasks.8

Results

Behavioral Results

Analysis strategy. To assess reactions during fMRI, separate 2 (group: ingroup,

outgroup) × 2 (race: Black, White) × 2 (task: explicit, implicit) repeated-measures analyses

were conducted on reaction time and accuracy after first removing all responses ≤ 300 ms

after stimulus presentation. To assess controlled evaluations and memory, separate 2 (group:

ingroup, outgroup) × 2 (race: Black, White) repeated-measures analyses were conducted on

liking and accuracy, respectively.

fMRI Categorization Speed and Accuracy. As shown in Table 1, participants

responded much faster, F(1, 21) = 113.29, p < .001, and with greater accuracy, F(1,21) =

19.40, p < .001, during the implicit categorization task. Replicating previous research (Levin,

1996), participants were faster to categorize Black (> White) faces during the implicit task,

when they were focused on race, F(1, 21) = 5.05, p < .04. In addition, a Task × Group

interaction, F(1, 21) = 7.73, p = .01, indicated that participants were also faster to recall and

categorize ingroup (> outgroup) faces during the explicit categorization task, when they were

focused on group membership. There were no other main effects or interactions on reaction

time or accuracy (ps > .15).

Ratings and Memory. We analyzed participants’ ratings of accurately recalled faces

(determined from the memory task).9 Replicating previous research (see Brewer, 1979),

participants preferred novel ingroup to outgroup members F(1, 20) = 4.59, p = .04, and this

preference for ingroup members was driven by ingroup liking, M = 0.40, t(20) = 3.23, p <

32

.01, and there was no evidence of outgroup disliking, M = -0.03, t(20) = -0.15, p = .88,

relative to the midpoint (3.5) of the rating scale. In other words, participants liked ingroup

members and were neutral toward outgroup members (see Figure 2). There were no effects of

Race or a Race × Group interaction (ps > .26). These data are consistent with the view that

ingroup bias and outgroup derogation are not necessarily reciprocal – liking an ingroup does

not necessitate disliking an outgroup (Allport, 1954; Brewer, 1999).

fMRI Results

fMRI Preprocessing and Analysis. Data were prepared for analysis using SPM5

(Wellcome Department of Cognitive Neurology, London, UK). Data were corrected for slice

acquisition time and motion, co-registered to structural images, transformed to conform to

the default T1 MNI brain interpolated to 3 × 3 × 3 mm, and smoothed using an 9 mm FWHM

(full-width-half-maximum) kernel. Data were also corrected for artifacts and detrended. Data

were analyzed using the general linear model in SPM5. The BOLD signal was modeled as a

function of a canonical hemodynamic response function and its temporal derivative with a

128 s high-pass filter. First-level images were analyzed at the second-level with a 2 (group:

ingroup, outgroup) × 2 (race: Black, White) × 2 (task: explicit, implicit) repeated-measures

ANOVA. To help reduce type-1 errors all effects from whole-brain analyses were reported as

statistically significant if they exceeded p ≤ .001 (uncorrected) with at least ten contiguous

voxels. All effects from region-of-interest (ROI) analyses (i.e., a priori analyses of the

amygdala and a posteriori ROIs used to test moderation) were reported as statistically

significant if they exceeded p ≤ .01 (uncorrected) with at least ten contiguous voxels. We

also used ROIs to examine correlations. Data from 17 participants who successfully learned

the group membership of the faces (average = 87% correct) and completed the categorization

33

task were included in the fMRI analyses. One participant was excluded for head motion (> 3

mm), one for random responding during implicit categorization (> 3.6 SD below the mean in

accuracy), and three for not learning the teams (less than 71% correct, where 50% = chance).

Novel Group Membership. Consistent with the idea that self-categorization motivates

aspects of social perception and that ingroup members are motivationally-primary in minimal

group contexts, novel ingroup (> outgroup) members were associated with greater activity in

bilateral fusiform gyri (see Table 2 and Figure 3A). These results converge with previous

studies showing greater fusiform activity to racial ingroup (> outgroup) members (Golby et

al., 2001; Lieberman et al., 2005), and extend these findings to novel groups matched on

familiarity. Further, these results are consistent with research suggesting that the fusiform

may play a more general role in individuating stimuli (Rhodes et al., 2004) and behavioral

evidence that mere categorization leads to the individuation of novel ingroup (> outgroup)

members (Bernstein et al., 2007). Whereas previous studies linking ingroup processing and

fusiform activity may be due, at least in part, to familiarity with racial ingroup members

(Golby et al., 2001; Lieberman et al., 2005), the present data suggest that aspects of self-

categorization other than familiarity contribute to this relationship (Sporer, 2001).

Although previous research has suggested that amygdala activity to Black faces may

reflect the processing of negative information, recent models posit a more general role for the

amygdala in processing motivationally-relevant stimuli (Anderson & Phelps, 2001;

Cunningham et al., 2008; Whalen, 1998). Consistent with this latter model, novel ingroup (>

outgroup) members were associated with greater amygdala activity, our a priori ROI (see

Figure 3B). While this result may appear to differ from studies showing greater amygdala

activity to racial outgroup faces (e.g., Cunningham, Johnson et al., 2004; Hart et al., 2000),

34

racial attitudes and stereotypes are likely to be motivationally-relevant in some intergroup

contexts and irrelevant in others. For example, racial outgroup (> ingroup) faces are

associated with amygdala activity when people think of them in terms of their social category

membership, while racial ingroup (> outgroup) faces are associated with amygdala activity

when people think of them as individuals (Wheeler & Fiske, 2005).

The contention that amygdala activity to novel ingroup members may reflect

motivational consequences of belonging to a group rather than negativity is corroborated by

other significant regions from the whole-brain analysis. Novel ingroup (> outgroup) members

were associated with greater activity in the orbitofrontal cortex (OFC; see Figure 4A) and

dorsal striatum (putamen). The OFC plays a key role in linking social and appetitive stimuli

to hedonic experience (Anderson et al., 2003; Kringelbach, 2005) and is sensitive to social

rewards (L. T. Harris, McClure, van den Bos, Cohen, & Fiske, 2007; van den Bos, McClure,

Harris, Fiske, & Cohen, 2007). Interestingly, differences in OFC activity (ingroup-outgroup)

were correlated with individual differences in ratings of ingroup bias (ingroup-outgroup),

such that people who liked their ingroup also had stronger OFC activity (r = .54, p < .03; see

Figure 4B).10

Similarly, the dorsal striatum is active during acts of mutual cooperation

(Rilling et al., 2002) and viewing pictures of loved ones (Bartels & Zeki, 2000). In contrast,

no regions were significantly more active to outgroup than ingroup faces.

Correlations between Neural Activity and Ingroup Bias. We tested whether the neural

activity in response to viewing novel ingroup (>outgroup) members was associated with

individual differences in ingroup bias (i.e., liking ingroup members more than outgroup

members). We created individual differences in brain activity (ingroup - outgroup) in each

ROI (fusiform, amygdala, OFC, and striatum) and correlated them with individual

35

differences in self-reported liking (ingroup - outgroup). Participants with greater OFC

activity to ingroup (> outgroup) members reported a stronger preference for ingroup

(>outgroup) members (r = .54, p < .03).8 As seen in Figure 5, the effects of group

membership on self-reported liking were mediated by OFC activity. Specifically, the effect

of group membership on ingroup bias was significantly reduced (from b = 0.389, p = .05, to b

= -0.131, p = .63) when controlling for increases in OFC activity to ingroup (> outgroup)

members (Sobel test t = 2.22, p < .03).

Racial Group Membership. To further examine the neural components involved in

processing social groups, we compared brain activity to Black and White faces. There was

greater activity in the visual cortex (see Table 2) to Black faces while no regions were more

active during the presentation of White faces.11

These data are consistent with a series of

studies showing that race is largely ignored when it is orthogonal to current group

membership (Kurzban et al., 2001).

Interactions with Race, Group and Task. To examine the automaticity of these

ingroup biases in neural processing we tested whether the brain regions showing ingroup bias

were moderated by race (Black vs. White), attention (explicit vs. implicit) to team

membership, or both. We extracted ROIs from the regions of fusiform gyri, amygdala, OFC,

and dorsal striatum showing ingroup bias (ingroup > outgroup) and compared voxels from

each of these regions for Black and White faces and during the implicit and explicit

categorization tasks. Ingroup biases (ingroup > outgroup) in neural activity were not

moderated by race, categorization task, or a race × task interaction. Whereas previous studies

have shown that activity in these regions can be modulated by explicit processing goals

(Cunningham et al., 2003), the ingroup biases in neural activity reported here do not appear

36

to require explicit attention to team membership, nor do they pertain strictly to Black or

White faces.

We also conducted whole-brain analyses on the Group × Task, Group × Race, and

Group × Task × Race interactions. Although these analyses revealed few higher-order effects

(at p ≤ .001) we have included a list of brain regions sensitive to crossed-categories (i.e., the

Group × Race interaction; see Table 3).

Discussion

This study reveals a constellation of neural activity consistent with models of flexible

self-categorization (J. C. Turner et al., 1987). Viewing novel ingroup members was

associated with greater activation in the fusiform gyri, amygdala, OFC, and dorsal striatum,

relative to novel outgroup members. Moreover, OFC activity mediated the effect of group

membership on self-reported ingroup bias. Importantly, ingroup biases in neural processing

occurred within minutes of team assignment, in the absence of explicit team-based rewards

or punishments, and independent of pre-existing attitudes, stereotypes, or familiarity. Ingroup

biases in neural processing were not moderated by target race or categorization task,

suggesting that they did not require explicit attention to team membership and may have

occurred relatively automatically.

This study provides neural evidence that ingroup members are processed in greater

depth than outgroup members – placing ingroup biases in perception firmly within the realm

of motivated social perception (Balcetis & Dunning, 2006). By virtue of their motivational

significance in a variety of contexts (e.g., economic, psychological, and evolutionary)

ingroup members often warrant greater/deeper processing than outgroup members (Brewer,

1979, 1999). By assigning participants to novel groups and providing equal exposure to

37

ingroup and outgroup faces, our experimental design was able to minimize the role of

familiarity and novelty as causal variables in these neural ingroup biases. The absence of

expertise with the faces also raises the possibility that the fusiform gyri may be associated

with attentional biases (Wojciulik, Kanwisher, & Driver, 1998) toward ingroup members,

greater individuation (Rhodes et al., 2004) of ingroup members, or both.

Whereas earlier studies reported amygdala activity to racial outgroup faces – often

interpreted as reflecting negativity or fear toward outgroup or stigmatized group members

(e.g., Lieberman et al., 2005) – participants in the current study had greater amygdala activity

to novel ingroup members (see also Chiao et al., in press). Interestingly, Wheeler and Fiske

(2005) found both patterns: greater amygdala activity to racial outgroup faces during a social

categorization task (deciding whether a person was young or old) and greater amygdala

activity to racial ingroup faces during an individuation task. Their study captures the

flexibility of amygdala function, which can respond to positive and negative stimuli, stimulus

intensity, and, more generally, the motivational relevance of stimuli (Cunningham et al.,

2008; Whalen, 1998). To this end, the amygdala may be involved in segregating relevant

from irrelevant stimuli in order to enhance perception of important stimuli (Anderson &

Phelps, 2001; Vuilleumier, 2005; Whalen, 1998). This view of amygdala function also offers

an alternative explanation for a previous study showing that both White and Black

participants have greater amygdala activity to Black faces (Lieberman et al., 2005).

Amygdala activity among White participants may reflect attention toward negatively

stereotyped Blacks (Eberhardt, Goff, Purdie, & Davies, 2004), whereas amygdala activity

among Black participants (who generally have a stronger racial identity and may therefore

view Black faces as more relevant (Crocker, Luhtanen, Blaine, & Broadnax, 1994)) may

38

reflect increased processing of ingroup membership. In other words, different psychological

mechanisms may guide the processing of racial ingroup and outgroup members. We propose

that the amygdala activity to ingroup members in the current study may stem from their

motivational relevance and salience in the current group context.

The relevance of different social category memberships also varies according to social

context (J. C. Turner et al., 1987). Assigning people to mixed-race groups may change the

way they construe race, and sensitize perceptual and affective processes to currently-relevant

group memberships. Indeed, people categorize others according to race when it is the salient

social category, but categorize according to other group memberships (and ignore race) when

they are part of a mixed-race group (Kurzban et al., 2001). This is a difference between the

present experiment and previous fMRI studies where race is the most salient difference

between faces. In this experiment, ingroup biases in neural activity were not moderated by

race. However, in contexts where race provides the most salient group distinction, attitudes,

cultural stereotypes (especially threat), and personal values (egalitarianism) may provide the

most relevant motivational guides.

The pattern of ingroup bias in the current experiment extended to self-reported

preferences for ingroup members. Participants with stronger preference for ingroup members

had stronger OFC activity to ingroup relative to outgroup members. This brain-behavior

relationship is consistent with a recent study showing a strong correlation between a similar

region of the OFC and self-reported pleasantness ratings of food (Kringelbach, O'Doherty,

Rolls, & Andrews, 2003) and, more generally, the idea that this region plays a central role in

representing and processing subjective value (Anderson et al., 2003; Kringelbach, 2005). To

our knowledge, this is the first fMRI study to identify the neural mediators of self-reported

39

intergroup biases and it demonstrates an important link between the pervasive preference for

novel ingroup members (Brewer, 1979) and brain regions that process reward and subjective

value (Anderson et al., 2003; Kringelbach, 2005).

Ingroup biases in neural activity did not require explicit attention to team membership.

Although there were differences in task difficulty (judging by faster reaction times and

higher accuracy in the implicit task), neural ingroup biases did not differ across tasks,

suggesting that these biases are relatively automatic. Similarly, Bernstein and colleagues

proposed that “ingroup faces are processed in a default, automatic manner, characterized by

holistic processing” (Bernstein et al., 2007, p. 711). Indeed, several regions activated by

ingroup members respond automatically to social stimuli; for example, the amygdala

responds to subliminal presentations of arousing faces (Whalen et al., 1998) and the fusiform

responds to faces within 100-200ms (Liu, Harris, & Kanwisher, 2002). Future research will

need to use electroencephalography and other techniques to examine the time-course of these

ingroup biases and whether or not they emerge prior to conscious awareness.

40

Chapter 4: The Contextual Sensitivity of Intergroup Evaluation

Consistent with Self-Categorization Theory (J. C. Turner et al., 1987), Experiment 1

provided evidence that the amygdala, fusiform, and OFC are sensitive to novel ingroup

members, regardless of race. Very brief membership in a novel group elicits neural

processing associated with group membership and appears to override race, a category

marked by years of exposure. This experiment offers evidence that social perception along

racial lines may be malleable to a certain type of social context: one in which race is

irrelevant to group membership. As discussed in the introduction, social categorization elicits

evaluation, and perceiving people according to their group membership may therefore reduce

pervasive and ostensibly stable automatic racial bias (Devine, 1989; Nosek et al., 2002).

Consistent with this assertion, previous research has shown that categorizing complex social

stimuli in different ways moderates the activation of underlying automatic attitudes, leading

to different evaluations (Blair, 2002). For example, categorizing Black athletes and White

politicians according to race activates an automatic preference for White politicians;

however, categorizing the same individuals according to occupation activates an automatic

preference for Black athletes (Mitchell et al., 2003). Accordingly, shifting social

categorization from race to group membership may reduce pervasive automatic racial biases

in favor of a preference for ingroup members (Ashburn-Nardo et al., 2001; Brewer, 1979).

Although Experiment 1 offered evidence that participants in a novel group prefer ingroup to

outgroup members, regardless of race, it is important to determine whether these effects

extend to automatic evaluations for at least two reasons. First, participants generally show

automatic racial bias while reporting egalitarian attitudes, making it important to examine the

41

effect of group membership on automatic racial biases. Second, as noted in the introduction,

automatic and controlled racial biases are associated with different forms of racial

discrimination and are therefore important to study independently. Experiment 2 examines

whether people assigned to a novel group have an automatic preference for ingroup

members, regardless of race.

Experiment 2 offers an important test of the sensitivity of spontaneous social perception

and evaluation to a current self-categorization. Previous research has shown the sensitivity of

automatic evaluations to categorization processes: changing category labels alters evaluations

(Mitchell et al., 2003). However, the automatic evaluation of multi-categorizable social

stimuli remains unclear. Recent research has shown that certain implicit attitude measures,

such as the Implicit Association Test (Greenwald et al., 1998), evoke evaluations consistent

with the category labels rather than the unique characteristics of individual stimuli (see Olson

& Fazio, 2003). Therefore, the automatic evaluations of Black athletes and White politicians

may have been driven by attitudes toward the current category labels rather than the

spontaneous construal of these complex stimuli. The current experiment uses a measure of

automatic evaluations without category labels to examine more spontaneous construals. In

the context of mixed-race teams, a visually salient social category with well-learned

evaluative associations, like race, may drive automatic evaluations. However, if automatic

evaluation is truly sensitive to the social context and/or the current self-categorization the

minimal intergroup distinctions should override competing, orthogonal social category cues

like race. Excluding labels also allows for more complex evaluations of multi-categorizable

social stimuli (Crisp & Hewstone, 2007), including the additive effects of race and group

membership (i.e., leading to the most positive evaluations of White-ingroup members) or the

42

evaluation of some targets according to their group membership (e.g., ingroup members) and

others according to their race (e.g., outgroup members). This experiment explores whether

people spontaneously construe and automatically evaluate novel ingroup members according

to their group membership when it is orthogonal to race.

Experiment 2

To examine the sensitivity of automatic evaluation to the current self-categorization

participants were randomly assigned to a novel mixed-race team without a long history of

contact or conflict. In this variant of the minimal group paradigm, participants memorized 12

ingroup and 12 outgroup members and then completed measures of their automatic and

controlled evaluations toward these multi-categorizable targets.12

Using a similar paradigm,

Kurzban and colleagues (2001) found that assigning participants to mixed-race teams led

people to categorize targets according to team membership rather than race. However, they

did not measure evaluations or analyze the potential interaction between team membership

and race. It therefore remains possible that automatically activated racial biases emerge even

when race is orthogonal to team membership, or that race may affect evaluations of ingroup

or outgroup members differently. More generally, research and theory on the evaluation of

multiply-categorizable targets suggests that orthogonal social categories can influence

intergroup evaluations in a variety of ways (for a review see Crisp & Hewstone, 2007;

Deschamps & Doise, 1978): participants’ evaluations of individual faces could stem from

pre-existing racial bias (White > Black), current self-categorization (ingroup > outgroup), the

sum of these categories, or an interaction between race and self-categorization (see Crisp &

Hewstone, 2007 for a review). To examine the spontaneous construals of these multi-

43

categorizable faces participants completed automatic and controlled evaluations of the

individual targets without explicit category labels (Olson & Fazio, 2003).

Methods

Participants.

One hundred and nine University of Toronto undergraduate students (84 females)

successfully completed the study for partial course credit.13

Materials and Procedure

Procedure. Similar to Experiment 1, participants were randomly assigned to one of

two competing groups (the Lions or Tigers) in the experimental condition (N = 72).

However, in the current experiment one-third of participants were randomly assigned to a

control condition (N = 37) in which they merely learned about the groups without being

assigned to one of them. The stimuli and procedure were nearly identical to Experiment 1

with two important differences. First, one-third of the participants were randomly assigned to

a control condition in which they were not assigned to either of the two groups. This allowed

us to examine whether merely learning about the two groups was sufficient to reduce the

construal and evaluation of multi-categorizable stimuli according to race or whether the

inclusion of the self in a group was critical to changing evaluations. Second, participants

completed a measure of their automatic evaluations to examine the spontaneous construal

and evaluation of race and group membership.

Learning. Participants completed two brief learning tasks (~15 minutes). Race was

orthogonal to team membership; there were six Black and six White males on each team. The

24 faces were randomly assigned to each team and fully counterbalanced. In the first learning

task, participants spent three minutes memorizing the team membership of all 24 faces

44

simultaneously: 12 members of Lions and 12 members of the Tigers. In the second learning

task, participants were presented with each of the 24 faces one-at-a-time and indicated

whether each face was a member of the Lions or Tigers. Participants in the experimental

condition (i.e., assigned to a team) also saw and categorized their own face three times during

the second learning task (randomly interspersed within the other 24 faces) to enhance self-

categorization with their team during the learning phase. Participants did not see their own

face again during any other phase of the study. During the first set of trials, each face was

presented with the team name (Lions or Tigers) on the computer monitor to help enhance

learning. During the second set of trails, the team name was removed so that participants

needed to rely only on their memory. Following each trial, feedback indicated if the response

was correct. After learning the faces from the two teams, participants completed measures of

their automatic and controlled evaluations of the faces without explicit group labels, in

counterbalanced order.

Automatic Evaluation Measure. To measure automatic evaluations of the faces

participants completed a response-window priming task on a personal computer

(Cunningham, Preacher, & Banaji, 2001; see Draine & Greenwald, 1998 for details). During

this task participants were instructed to rapidly categorize each word as “good/liked” or

“bad/disliked” (see Olson & Fazio, 2004). Participants were instructed to press “1” when a

good word appeared and “2” when a bad word appeared. Following 24 practice trials,

participants completed three critical blocks with 96 trials in each block. On each trial in these

critical blocks a face from the learning task appeared for 150 ms (followed by a blank screen

for 50 ms) prior to a positive (e.g., love) or negative (e.g., hatred) target word, which

appeared for a 525 ms response window. Participants were instructed to ignore the faces. The

45

dependent measure was the proportion of trials in which each participant correctly

categorized the word as good or bad (i.e., response accuracy) (α = .45). To provide an

estimate of more automatic (i.e., rapid) evaluative processing, all responses that occurred

after 600 ms were coded as incorrect, similar to previous research using response-window

priming.14

Following previous research (Draine & Greenwald, 1998), we assumed that faces

with positive associations would increase accuracy to positive words and decrease accuracy

to negative words.

Controlled Evaluation Measure. To measure more controlled evaluations of the faces

participants rated each face. Participants were told that “people can often quickly determine

who they like or dislike based on subtle facial features and expressions”, and asked to rate

each of the 24 faces on a 6-point liking (1 = dislike to 6 = like) (α = .83). Faces were

presented in random order. We measured evaluations of individual faces to increase the

correspondence between automatic and controlled measures (Fishbein & Ajzen, 1974). By

assessing evaluations to individual faces with automatic and controlled measures we

enhanced confidence that any dissociations between measures were due to the automaticity

of evaluative processing, and not conceptual incongruities (e.g., responding to faces on the

automatic measure and category labels on the controlled measure).

Results

Analysis strategy.

Traditional analysis of implicit attitude measures has tended to focus on mean-level

differences in reaction time or accuracy. However, this approach has the consequence of

reducing hundreds of trials to a single data point per participant leading to a significant loss

of power and meaningful variance. In order to more accurately measure evaluation we used

46

multi-level modeling (Goldstein, 1995). Multi-level modeling allows for the direct analysis

of accuracy on individual trials and is able to overcome violations of independence that occur

as a result of correlated trials within participants. When an assumption of independence is not

satisfied ignoring the dependency among trials can lead to invalid statistical conclusions;

namely the underestimation of standard errors and the overestimation of the significance of

predictors (Cohen, Cohen, West, & Aiken, 2003). We therefore created multi-level models

with trials nested within participants to provide more appropriate estimates of regression

parameters. Multi-level modeling has been used successfully on automatic attitude measures

in previous research (Dunham, Baron, & Banaji, 2006; Nezlek & Cunningham, October,

1998). Multi-level models for all analyses were implemented in the SAS PROC MIXED

procedure (see Singer, 1998). To assess automatic evaluations we conducted 2 (group:

ingroup, outgroup) × 2 (race: Black, White) × 2 (target valence: positive, negative) repeated-

measures analysis on response accuracy for experimental participants, and a 2 (race: Black,

White) × 2 (target valence: positive, negative) analysis for control participants.15

To assess

controlled evaluations we conducted a 2 (group: ingroup, outgroup) × 2 (race: Black, White)

repeated-measures analysis on liking for experimental participants, and an analysis on race

(Black, White) for control participants.

The effects of group and race on automatic evaluations.

Following previous research (e.g., Fazio et al., 1995), we expected control participants

to show automatic racial bias on the response-window priming task. As anticipated, control

participants had more positive associations for White than Black faces, F(1, 36) = 3.42, p =

.07. Simple effects analyses confirmed that participants had relatively positive evaluations

(i.e., higher accuracy to positive than negative words) for White faces, t(36) = 2.37, p < .02,

47

but were relatively neutral to Black faces, t(36) = 0.08, p = .94. Simply learning the mixed-

race faces in a context in which race was orthogonal to group membership evoked the

standard pattern of racial bias: an automatic preference for White over Black faces.

The primary hypothesis of this experiment was that self-categorization with a novel

mixed-race group would eliminate automatic racial bias on the response-window priming

task. As seen in Figure 6, group membership moderated automatic racial bias. Specifically,

the three-way interaction between Group × Race × Valence, F(1, 71) = 5.99, p < .02, was

characterized by automatic racial bias (a preference for White over Black faces) among

outgroup faces, F(1, 71) = 6.40, p = .01, but not ingroup faces, F(1, 71) = 0.87, p = .35.

Simple effects analyses confirmed that participants had more positive evaluations (i.e., higher

accuracy to positive than negative words) of Black-ingroup than Black-outgroup faces, F(1,

71) = 10.77, p < .01.16

These results demonstrate that membership in a novel group can

improve automatic racial evaluations. While White outgroup members were evaluated more

positively than Black outgroup members – mirroring the pattern of results in the control

condition – novel ingroup members were evaluated positively regardless of race.

The effects of group and race on controlled evaluations.

Despite the evidence of automatic racial bias, we expected that controlled evaluations

would be more egalitarian. Replicating previous research, there were no strong racial biases

on ratings of Black and White faces – both in the control, t(36) = -0.69, p = .49, and

experimental, F(1, 71) = 2.98, p < .09, conditions. If anything, experimental participants had

a marginal preference for Black (M = 3.49) over White (M = 3.34) faces (see Figure 7).

These relatively egalitarian racial evaluations were dissociated from the racial bias evidence

on more automatic evaluations. In contrast, we expected ingroup bias would be evident on

48

both automatic and controlled measures, because participants have little reason to modify

their evaluations of novel group members. As expected, experimental participants preferred

novel ingroup (M = 3.52) to outgroup faces (M = 3.31), F(1, 71) = 5.39, p = .02. These

preferences were not qualified by a Race × Group interaction, F(1, 71) = 0.00, p = .95.

Automatic and Controlled Correlations.

To examine the correlations between automatic and controlled evaluations among

experimental participants we computed mean automatic racial bias (White-positive + Black-

negative – White-negative + Black-positive), controlled racial bias (White – Black),

automatic ingroup bias (Ingroup-positive + outgroup-negative – ingroup-negative +

outgroup-positive), and controlled ingroup bias (ingroup – outgroup) scores. There were

modest correlations between automatic and controlled racial bias (r = .17, p = .14) and

ingroup bias (r = .15, p = .22). Although these correlations were in the typical range of raw

automatic-controlled correlations (Cunningham et al., 2001; Hofmann, Gawronski,

Gschwendner, Le, & Schmitt, 2005), they were not statistically significant.

Discussion

These data provide evidence that automatic evaluations are sensitive to a novel self-

categorization within a complex intergroup context. Participants in the control condition,

who were not a member of either mixed-race team, had an automatic preference for White

relative to Black faces (Fazio et al., 1995), suggesting that they were construing faces

according to race – a visually salient social category. In contrast, participants in the

experimental condition did not evaluate multiply-categorizable faces according to a single

salient social category. Instead, their automatic evaluations were a complex interaction

between self-categorization and race. In particular, they showed positive evaluations for

49

White and Black ingroup members, eliminating the standard pattern of automatic racial bias,

while their automatic evaluations of outgroup members continued to reveal racial bias,

similar to the control condition. Taken together, these results suggest that automatic

evaluation was sensitive to the current social context, shifting from evaluations based on race

(control condition) to evaluations that reflected their current self-categorization

(experimental condition).

Controlled evaluations were also sensitive to the current self-categorization:

participants assigned to a mixed-race team reported a preference for novel ingroup members,

regardless of race. Replicating previous research (Devine, 1989), control participants

revealed a dissociation between their automatic and controlled evaluations, showing an

automatic preference for White over Black faces while reporting more egalitarian evaluations

on controlled measures. More unique to this experiment, we found this mean-level

dissociation between automatic and controlled measures while controlling for

correspondence: having participants evaluate individual faces on both measures. In contrast,

preferences for novel ingroup members were evidenced on both measures. These data

suggest that racial bias can be modulated by more controlled processing or through

construing racial minorities as novel ingroup members.

This experiment offers evidence that automatic evaluations are sensitive to the current

self-categorization, leading to automatic evaluations of ingroup members according to their

novel group membership rather than race. These results are fully consistent with self-

categorization theory and other models that highlight the contextual sensitivity of self-

categorization and the impact of these categorizations on social perception and evaluation (J.

C. Turner et al., 1987). This pattern of data highlights the power of self-categorization on

50

evaluation, especially given the visually salient nature of race and the extensive evidence that

racial biases are automatically activated upon the mere presentation of Black names or faces

(Devine, 1989; Fazio et al., 1995). Perhaps the most impressive finding is that self-

categorization with a novel group generates more positive automatic evaluations of Black-

ingroup than Black-outgroup members. The following experiment tested whether this shift in

evaluations was driven by more positive evaluations of Black-ingroup members or more

negative evaluations of Black-outgroup members.

51

Chapter 5: Ingroup Bias versus Outgroup Derogation

Participants assigned to a novel team in Experiment 2 had positive evaluations of

ingroup faces, regardless of race, and evaluated White-outgroup faces more positively that

Black outgroup faces. The current experiment examines whether the automatic preference for

Black-ingroup compared to Black-outgroup members in the previous experiment was driven

by ingroup bias, outgroup derogation, or some combination. One possible explanation for the

change in automatic racial bias in the previous experiment is that self-categorization with a

novel group generated ingroup bias, improving evaluations of Black-ingroup members.

Another possibility is that self-categorization with a novel group generated outgroup

derogation, diminishing evaluations of Black-outgroup members. A third possibility is that

ingroup bias and outgroup derogation combined to generate the difference in evaluations of

Black-ingroup compared to Black-outgroup members. To test these possibilities we

compared evaluations of ingroup and outgroup members to unaffiliated faces (i.e., Black and

White faces that were not a member of either group).

Experiment 3

Experiment 3 was designed to specifically assess the nature of evaluations of multi-

categorizable stimuli. To our knowledge, no published studies have used appropriate controls

to isolate the direction of evaluation in minimal groups (McCaslin & Petty, 2008) or crossed-

category contexts. However, previous minimal group research generally suggests a pattern of

ingroup bias – a preference for ingroup members (Brewer, 1979) – and although participants

readily allocate more rewards to ingroup members, they do not allocate more punishments to

52

outgroup members (Mummendey et al., 1992). Moreover, evaluations of ingroup and

outgroup members are not necessarily reciprocally related (Allport, 1954; Brewer, 1999).

In Experiment 2, participants responded to unaffiliated Black and White faces during

the evaluation tasks to contrast against ingroup and outgroup faces and generate more precise

inferences about the nature of intergroup preferences in complex social contexts. Based on

previous research in this domain, it seemed that self-categorization would improve automatic

evaluations of Black-ingroup members without leading to the derogation of Black-outgroup

members. In fact, the priming data in Experiment 2 suggested that White-ingroup, White-

outgroup and Black-ingroup faces were positively evaluated (i.e., facilitation to positive and

inhibition to negative words) while Black-outgroup members evoked relatively neutral

evaluations. However, a meta-analysis on the evaluation of multi-categorizable targets

(Urban & Miller, 1998) revealed that more positive evaluations of crossed-category targets

(e.g., Black-ingroup members) are usually offset by more negative evaluations of double-

outgroup targets (e.g., Black-outgroup members). This experiment examined the direction of

automatic evaluations within a given intergroup context; namely, whether automatic

evaluations of Black faces were altered by ingroup bias and/or outgroup derogation.

The current experiment was similar to Experiment 2, with the primary difference that

participants evaluated Black and White faces that were unaffiliated with either mixed-race

team and evaluations of these faces were contrasted with ingroup and outgroup members to

identify the nature of preference for Black-ingroup compared to Black-outgroup members.

Participants did not learn these unaffiliated faces during the first phase of the experiment, but

merely saw them during both evaluation tasks, providing a set of Black and White control

faces that reflected baseline evaluations of novel faces. If self-categorization elicits ingroup

53

bias, Black-ingroup faces should be evaluated more positively than Black-unaffiliated faces.

However, if self-categorization elicits outgroup derogation, Black-outgroup faces should be

evaluated more negatively than Black-unaffiliated faces.

Methods

Participants.

One hundred and twenty-six undergraduate students from The Ohio State University

successfully completed the study for partial course credit. Two participants were removed for

using their iPod or cell phone during the study and two were removed for responding on less

than 15% of trials during the priming task, leaving 122 (57 females) participants for analysis.

Materials and Procedure.

Similar to Experiment 2, participants were randomly assigned to one of two competing

groups (the Lions or Tigers) in the experimental condition (N = 81) or a control condition (N

= 41) in which they merely learned about the groups. The current experiment was similar to

Experiment 2, with the primary difference that participants evaluated Black and White faces

unaffiliated with either mixed-race group. Automatic (α = .75) and controlled (α = .87)

evaluations of unaffiliated faces were contrasted with ingroup and outgroup members to

identify the nature of preference for Black-ingroup compared to Black-outgroup members.

Specifically, participants memorized eight faces belonging to the Tigers and eight belonging

to the Lions and four Black and four White faces who were unaffiliated with the Lions or

Tigers appeared for the first time during the evaluation tasks. Participants did not see

unaffiliated faces during the learning phase of the experiment, but saw them during both

evaluation tasks. Again, faces were randomly assigned to group and fully counterbalanced. In

addition, on the priming task half the participants pressed “1” when a good word appeared

54

and “2” when a bad word appeared, and half the participants used the opposite word-number

mappings.17

Analyses were identical to Experiment 2 with the exception that there were three

groups (ingroup, outgroup, unaffiliated) of faces. To assess automatic evaluations we

conducted a 3 (group: ingroup, outgroup, unaffiliated) × 2 (race: Black, White) × 2 (target

valence: positive, negative) repeated-measures analysis on response accuracy for

experimental participants. To assess controlled evaluations we conducted a 3 (group:

ingroup, outgroup, unaffiliated) × 2 (race: Black, White) repeated-measures analysis on

liking for experimental participants. Analyses contrasted unaffiliated faces with ingroup and

outgroup members to identify the nature of preference for Black-ingroup compared to Black-

outgroup members.

Results

The effects of group and race on automatic evaluations.

The primary goal of this experiment was to examine the direction of automatic

evaluation by comparing the evaluations of unaffiliated faces with ingroup and outgroup

members. We anticipated and found that ingroup members were evaluated positively,

regardless of race, and more positively than unaffiliated faces or outgroup members. In other

words, membership in a novel group elicited automatic ingroup bias. As seen in Figure 8, a

marginal Group × Valence interaction, F(1, 80) = 2.93, p = .09, indicated that participants

had more positive automatic evaluations of ingroup members relative to unaffiliated faces,

F(1, 80) = 5.64, p < .02, but had similar evaluations for outgroup members and unaffiliated

faces, F(1, 80) = 2.78, p = .10; if anything, they had a marginal preference for outgroup

members. However, these effects were qualified by a Group × Race × Valence interaction,

55

F(2, 160) = 3.71, p < .03, characterized by automatic racial bias (a preference for White over

Black faces) toward the unaffiliated faces, F(1, 80) = 3.06, p < .08, but not ingroup faces,

F(1, 80) = 0.85, p = .36, which were evaluated positively, regardless of race. In fact,

comparing only ingroup to unaffiliated faces confirmed that these evaluations were

significantly different, F(1, 80) = 7.29, p < .01. Simple effects analyses confirmed that

participants had more positive automatic evaluations (i.e., higher accuracy to positive than

negative words) of Black-ingroup than Black-control faces, F(1, 80) = 12.74, p < .001,

indicating that membership in a novel mixed-race group improved automatic evaluations of

Black faces.18

In contrast, comparing only outgroup to unaffiliated faces confirmed that these

evaluations were not significantly different, F(1, 80) = 0.80, p = .37. These results are

consistent with the hypothesis that ingroup members were evaluated positively, regardless of

race, and that unaffiliated faces and outgroup members were similarly (and apparently

according to race). This provided evidence of ingroup bias in the absence of outgroup

derogation.

The effects of group and race on controlled evaluations.

Following Experiment 1, we anticipated that participants would explicitly report racial

egalitarianism toward all the faces (including unaffiliated) and a preference for novel ingroup

faces relative to outgroup and unaffiliated faces. As seen in Figure 9, participants reported a

preference for ingroup members, F(2, 160) = 6.19, p < .01. Specifically, participants reported

positive automatic evaluations of ingroup members (M = 3.52) relative to control faces (M =

3.29), t(80) = 3.78, p < .001, but nearly identical attitudes toward outgroup members (M =

3.28) and control faces, t(80) = 0.14, p < .88. There was also a non-significant effect of race,

F(1, 80) = 1.72, p < .19, such that participants trended toward a preference for Black over

56

White faces; however, there were no Group × Race interactions, F(2, 160) = 0.23, p = .79. As

expected, participants reported ingroup bias and egalitarian racial attitudes to all groups.

Automatic and Controlled Correlations.

Similar to Experiment 2, the correlations between automatic and controlled racial bias

(r = -.11, p = .32) and ingroup bias (r = .00, p = .97) were not statistically significant.

Discussion

The current experiment offers evidence of the direction of automatic and controlled

evaluations in a novel group context. Replicating and extending the result from Experiment

2, automatic and controlled evaluations were characterized by ingroup bias relative to

unaffiliated faces. While previous minimal group studies have found data consistent with

ingroup bias (i.e., a relative preference for ingroup members) (Brewer, 1979), this is one of

the first experiments to show that novel group membership evokes ingroup bias rather than

outgroup derogation relative to a proper control group (McCaslin & Petty, 2008). In

particular, automatic evaluations of Black-ingroup members improved relative to unaffiliated

(and outgroup) Black faces. Importantly, improved automatic evaluations of Black-ingroup

faces were not offset by more negative evaluations of Black-outgroup members relative to

unaffiliated Black faces. These results suggest that ingroup membership increased positivity

toward ingroup members without increasing negativity toward outgroup members.

Interestingly, participants had an automatic preference for unaffiliated White relative to

Black faces, suggesting that they were construing and evaluating these faces according to

race. Automatic evaluations of outgroup faces in Experiment 2 and 3 were similar in this

respect, showing the standard pattern of racial bias (White > Black). These results suggest

that automatic evaluations may be sensitive to different social categories for different faces;

57

namely the current self-categorization of ingroup members and the race of outgroup members

and unaffiliated faces. Moreover, these evaluations occurred while participants were

randomly presented with a series of individual faces, raising the possibility that automatic

evaluations can shift between group membership and race on a case-by-case basis within a

stable intergroup context.

This pattern of data suggests that people may be spontaneously and rapidly construing

and evaluating complex social stimuli according to different social categories within a given

context. In the current intergroup context, stimuli can be evaluated according to at least two

social categories: group membership and race. In particular, ingroup and outgroup members

can be evaluated using either of those categories while unaffiliated faces are likely to be

evaluated only by race. The amygdala results in Experiment 1 suggest that ingroup members

may be especially important in this intergroup context, generating ingroup biases in

perception and evaluation. However, the pattern of racial bias toward outgroup members

suggests that they were evaluated according to race – mirroring evaluations of unaffiliated

faces.

Processing ingroup members according to their group membership and outgroup

members and unaffiliated faces according to their race may occur for a number of reasons.

One possibility is that outgroup members are treated as less relevant or receive less attention

to their group membership, and are thus processed based on more superficial visually-salient

social category cues, such as race. A second possibility is that outgroup members are

associated with relatively neutral evaluations and thus race-based association provide the

basis for outgroup evaluations. These two possibilities imply a hierarchical stage-based

model of intergroup perception and evaluation in which ingroup membership is processed

58

first and if individual faces do not belong to the ingroup they are processed according to their

race, whether due to an absence of attention toward or evaluative associations with outgroup

faces. A third possibility is that the current intergroup context may trigger positive automatic

evaluations of any ingroup member, including novel ingroup members and White faces. In

other words, any group associated with the self (whether racial or novel) elicits positive

automatic evaluations. Although the current research does not speak to the specific

mechanism behind this effect, the evidence does suggest that race is used to automatically

evaluate outgroup members and unaffiliated faces.

59

Chapter 6: Self-Categorization and Intergroup Memory

An assumption implicit in this multi-level approach is that changes in neural processing

mediate perception and evaluation. For example, the fusiform activity to novel ingroup

members in Experiment 1 was assumed to mediate aspects of social perception consistent

with the current self-categorization. We specifically suggested that the fusiform may play a

role in individuating ingroup members which may explain the correlation between fusiform

and superior memory for individual own-race faces in previous research (Golby et al., 2001).

If so, participants assigned to a novel group should show superior memory for individual

ingroup members consistent with the fusiform activity in Experiment 1 (ingroup > outgroup).

Experiment 4 was conducted to directly examine the effect of a novel self-categorization on

memory.

As described in the introduction, people are better at remembering faces from their own

racial or ethnic groups compared to faces from other racial groups (Malpass & Kravitz, 1969)

– and this biased social perception may stem from self-categorization with one’s racial group

rather than aspects of the social category per se, such as experience processing own-race

faces. Accordingly, if self-categorization shifts from race to a novel group identity, the ORB

should diminish. Thus, a novel self-categorization should increase memory for novel ingroup

compared to outgroup members and erase any own-race biases in memory.

One means by which self-categorization might affect evaluation is by altering social

perception. The first three experiments offered indirect evidence to this effect: neural

processing and evaluation of multi-categorizable targets elicited a host of ingroup biases

toward novel ingroup members, consistent with the current self-categorization. One means

60

by which self-categorization might alter evaluation is by increasing memory of ingroup

members relative to outgroup members. This aspect of individuation should increase ingroup

preferences for a number of reasons, including mere familiarity (Zajonc, 1968), personalized

processing (Brewer, 1988), and more motivated aspects of evaluations, including the desire

to enhance the evaluation of ingroup members (Tajfel, 1982). Accordingly, we proposed that

superior memory for ingroup members would increase ingroup bias. In addition, it seemed

likely that ingroup memory would affect controlled rather than automatic evaluations

because personalization (Brewer, 1988) and motivated enhancement (Crandall & Eshleman,

2003) are assumed to require more controlled processing.

Experiment 4

Experiment 4 was designed to determine whether self-categorization would increase

memory for novel ingroup members, regardless of race, and whether this would moderate

ingroup biases in evaluation. Based on previous research (Bernstein et al., 2007), we

expected that novel ingroup members would be better remembered on a subsequent recall

task. Increased memory for ingroup members is also expected to improve controlled

evaluations of ingroup members for at least three reasons. First, familiarity generally leads to

liking (Zajonc, 1968).19

Second, remembering individuals may stem from more individuated

encoding (Sporer, 2001) and individuation is presumably associated with more positive

evaluations, especially compared to categorization with a stereotyped social group. Third,

people are motivated to enhance evaluations of ingroup members (Tajfel, 1982) and

remembering team membership of individuals is a critical prerequisite for evaluative

enhancement. It was less clear, however, whether memory would also affect both automatic

61

evaluations. Although learning the faces was likely necessary for generating preferences, it

seemed likely that automatic evaluations occurring within the first few hundred milliseconds

would precede conscious recognition of team membership for these novel faces.

Methods

Participants.

One hundred and eleven undergraduate students from Experiment 3 successfully

completed the face memory task for partial course credit.20

Materials and Procedure

Procedure. The experimental procedure was identical to Experiment 3, with the

addition of the face memory task. Participants in the experimental (N = 75) and control (N =

36) condition completed the face memory task at the end of the session. Randomly assigning

some participants to a control condition in which they were not assigned to either of the two

groups allowed us to examine whether merely learning about the two groups was sufficient to

reduce the ORB or whether the inclusion of the self in a group was the critical to changing

intergroup memory.

Memory Task. To measure memory participants reported the team membership for

each face. Participants saw each of the 24 faces and were asked to indicate with a button

press whether each face was a member of the (a) Lions, (b) Tigers, or (c) neither team. Faces

were presented in random order. Response accuracy was recorded.21

Analysis strategy. Multi-level modeling allowed us to examine the effect of memory on

automatic and controlled evaluations by combining accuracy on the memory task with

responses on the automatic and controlled evaluation tasks. Specifically, the accuracy for

each individual face for each participant on the memory task was used as a separate regressor

62

to predict trial-by-trial responses to each face for each participant on the automatic and

controlled evaluations tasks completed in Experiment 3. To account for the fact that some

participants had perfect memory for some cells of faces (e.g., Black-ingroup) we did not

model the random effects of the independent variables in these models. Different from

Experiment 3, all analyses excluded unaffiliated faces because they were not seen during the

learning task. To assess memory we conducted a 2 (group: ingroup, outgroup) × 2 (race:

Black, White) repeated-measures analysis on accuracy for experimental participants, and a 2

(race: Black, White) analysis for control participants. To assess the effect of memory on

automatic evaluations we conducted a 2 (memory: accurate, inaccurate) × 2 (Group: ingroup,

outgroup) × 2 (target valence: positive, negative) repeated-measures analysis on response

accuracy for experimental participants. To assess the effect of memory on controlled

evaluations we conducted a 2 (memory: accurate, inaccurate) × 2 (group: ingroup, outgroup)

repeated-measures analysis on liking for experimental participants.

Results

The effects of group and race on memory.

Following research on the ORB (Malpass & Kravitz, 1969), we anticipated that

participants in the control condition would report superior memory for White compared to

Black faces. As seen in Figure 10, participants had superior memory for White (M = .67)

than Black (M = .60) faces, t(35) = 1.69, p < .05 (one-tailed).22

Thus, learning mixed-race

faces in a context in which race was orthogonal to group membership evoked ORB: superior

memory for White over Black faces.

More important, however, was the effect of race and group membership on memory

among experimental participants. Following the pattern of fusiform activity in Experiment 1

63

and recent research by Bernstein and colleagues (2007), we anticipated that participants

would have superior memory for novel ingroup compared to outgroup faces. As seen in

Figure 11, participants had superior memory for novel ingroup (M = .73) than outgroup (M =

.68) faces, F(1, 74) = 3.96, p = .05. Moreover, there was no effect of Race, F(1, 74) = 0.35, p

= .56, and these effects were not qualified by a Group × Race interaction, F(1, 74) = 1.67, p =

.20. These results show that intergroup memory is sensitive to the current self-categorization,

regardless of race. Novel ingroup members were remembered better than outgroup members

and there was no evidence of an ORB.

The effects of memory on automatic evaluations.

We anticipated that automatic evaluations would not be moderated by memory. Indeed,

memory did not moderate any aspect of automatic evaluations, including group membership

(ps > .28).

The effects of group and race on controlled evaluations.

There were a number of reasons that memory might increase ingroup bias, including

familiarity, personalization, and motivated enhancement. Accordingly, we anticipated that

increased memory for ingroup members would be associated with greater ingroup bias.

Indeed, a Memory × Group interaction, F(1, 1118) = 3.98, p < .05, indicated that participants

had more positive automatic evaluations of ingroup members they remembered (M = 3.58)

compared to ingroup members they did not remember (M = 3.34), t(524) = 2.07, p < .04 (see

Figure 12), but had similar evaluations for outgroup members they remembered (M = 3.39)

and did not remember (M = 3.22), t(524) = 1.63, p > .10. These data support the view that

better memory for novel ingroup members is associated with greater ingroup bias in more

controlled evaluations. Figure 12 also shows a clear ingroup preference over outgroups only

64

for correctly remembered faces, t(773) = 17.26, p < .0001, compared to no differences for

incorrectly remembered faces, t(283) = 0.01, p > .92. If participants did not remember who a

person was, they did not show ingroup bias.

Discussion

Experiment 4 indicated that self-categorization with a novel group could increase

memory for ingroup members, regardless of their race, and this increase in memory

moderated reported preferences for ingroup members. In contrast, memory had no effect on

automatic evaluations of group membership. This experiment extends previous research

(Bernstein et al., 2007) by showing that self-categorization with a novel mixed-race team can

improve memory for Black-ingroup members, providing a novel approach to override one of

the most pervasive racial biases in social psychology (Meissner & Brigham, 2001): superior

memory for own-race faces. We have subsequently replicated this pattern of own-group bias

several times (Van Bavel & Cunningham, 2008), suggesting that the effect of self-

categorization on memory is fairly robust. These follow-up studies have also found that own-

group bias is moderated by social identification, such that participants who are highly

identified with their novel group show more own-group memory bias than less identified

participants.

In the current research, ingroup members who were remembered were associated with

the most positive ratings whereas memory was unrelated to ratings of outgroup members.

The moderation of ingroup bias by memory suggests that evaluative ingroup bias may stem

from the motivated enhancement of ingroup members (Tajfel, 1982) or personalization

(Brewer, 1988), which is assumed to be a slower, more controlled process. In contrast,

evidence that outgroup evaluations are unaffected by memory suggests that participants feel

65

little motivation to derogate outgroup members or that outgroup members either receive more

superficial, less personalized attention. In either event, these results rule out the possibility

that the lack of outgroup derogation in Experiment 3 was simply the result of an inferior

memory for individual outgroup members. In other words, this analysis rules out the

potential confound between ingroup biases in memory and evaluation, and provides evidence

that memory for outgroup members does little to generate outgroup derogation. To our

knowledge, this experiment was the first to examine the interface between memory and

evaluation in a minimal group context, linking ingroup biases in perception and evaluation.

66

Chapter 7: Neural Substrates of Social Perception and Evaluation

The first four experiments illustrate the power of a novel self-categorization to alter

social perception and evaluation of people according to their group membership and override

ostensibly pervasive effects of race. Although successful social perception in some contexts

may be as simple as quickly recognizing who is with us or against us, in other contexts, it

may also require the ability to categorize some people according to one dimension and other

people on a different dimension. Basketball players, for example, must quickly differentiate

teammates from opponents, regardless of other irrelevant social category information (e.g.,

Ruscher, Fiske, Miki, & Van Manen, 1991), but also correctly identify the referees.

Consistent with this example, the results of Experiment 3 suggested that Black and White

faces unaffiliated with the ingroup or outgroup were automatically construed and evaluated

according to their race: participants had an automatic preference for unaffiliated White

compared to Black faces (see Figure 8c). Moreover, these evaluations occurred while

participants were randomly presented with a series of individual faces, raising the possibility

that automatic evaluations can shift between group membership and race on a case-by-case

basis within a stable intergroup context.

People may construe ingroup members according to their group membership and

unaffiliated faces according to their race for a number of reasons. On one hand, the current

intergroup context triggers self-categorization with the novel ingroup, increasing attention to

the team membership of ingroup and outgroup members. Indeed, learning these faces during

the first phase of these experiments likely enhanced the accessibility of team membership

during subsequent tasks (Fazio et al., 1986; Roskos-Ewoldsen & Fazio, 1992). On the other

67

hand, Black and White unaffiliated faces in Experiment 3 were not a member of either team,

making race the most (visually) salient social category cue for processing these faces. As

noted in Experiment 3, the race of these unaffiliated faces may have elicited automatic racial

bias (White > Black) for a number of reasons, including the well-learned evaluative

associations people have toward these racial categories or the possibility that participants

self-categorized with White faces, increasing their positivity toward racial ingroup members.

Although Experiment 3 does not speak to the source of this racial bias, it does suggest that

team membership is the most relevant dimension for processing ingroup faces and that race is

the most relevant dimension for processing unaffiliated faces.

Experiment 5

Assigning people to mixed-race groups appeared to change the way participants’

construed race, sensitizing perceptual and affective processes to their current-relevant group

membership. Recall that the amygdala was more active to novel ingroup (> outgroup)

members in Experiment 1, regardless of race. This was consistent with models that posit a

more general role for the amygdala in processing motivationally-relevant stimuli (e.g.,

Cunningham et al., 2008) such as ingroup members (Allport, 1954). However, these results

differed from previous research showing that Black (> White) faces were associated with

amygdala activity. One possible explanation for this discrepancy is that race was the only

salient social category in previous studies (Hart et al., 2000; Lieberman et al., 2005) but was

trumped by the salience of group membership when participants were assigned to mixed-race

teams. Experiment 3 suggested that participants assigned to a mixed-race team may still use

race to construe and evaluate faces, especially if they are unaffiliated with the ingroup or

68

outgroup. Accordingly, the amygdala may be sensitive both to novel ingroup members and

unaffiliated Black faces within the same social context. The current experiment examines this

hypothesis. The primary goal of Experiment 5 was to examine the sensitivity of neural

processing – especially the amygdala – to the ingroup members and Black unaffiliated faces.

A secondary goal of Experiment 5 was to directly examine the relationship between

ingroup biases in social perception and evaluation and the neural processing of group

membership. An important element of a multi-level approach is directly linking neural

processing to behavioral evidence to help elucidate the underlying process. The current

experiment examined the neural mediators of ingroup biases in automatic evaluation and

memory. Self-categorization with a novel group led to greater fusiform activity to ingroup

members in Experiment 1 and better memory for ingroup members in Experiment 4. As

described earlier, fusiform activity to own-race faces has been shown to correlate with

superior own-race memory (Golby et al., 2001), indicating that neural processing in the

fusiform is linked to memory for individual group members. Although these initial results

were taken as evidence that ORB is associated with expert processing, Experiments 1 and 4

suggest that self-categorization elicits ingroup biases in fusiform activity and memory,

respectively, possibly by increasing the individuation of ingroup members (Rhodes et al.,

2004; Sporer, 2001). The current experiment directly examined the link between ingroup

biases in fusiform activity and memory with the assumption that fusiform activity would

mediate the effects of self-categorization on ingroup biases in memory.23

In addition, we examined the link between automatic evaluations and neural processing.

In Experiment 1, participants with stronger self-reported preference for ingroup members had

stronger OFC activity to ingroup relative to outgroup members; a region involved in

69

representing and processing subjective value (Kringelbach, 2005). Consistent with previous

research (Phelps et al., 2000), there was no correlation between the amygdala and self-

reported evaluations. As noted in the introduction, several studies have reported a correlation

between automatic measures of racial bias and amygdala activity (Cunningham, Johnson et

al., 2004; Phelps et al., 2000). The current experiment examined whether more positive

automatic evaluations (i.e., ingroup bias) would correlate with amygdala activity to ingroup

members.24

Participants completed two single-category Implicit Association Tests (SC-IAT)

(Karpinski & Steinman, 2006) to independently measure their automatic evaluations of

ingroup and outgroup members and examine the relationship between these evaluations and

amygdala activity.

Method

Participants

Nineteen White participants (11 females, mean age = 20.1) were paid $40 for

completing the study. Participants reported no abnormal neurological history and had normal

or corrected-to-normal vision.

Materials and Procedure

Group Assignment. Similar to Experiment 1, participants were randomly assigned to

one of two competing groups (the Leopards or Tigers). The procedure was nearly identical to

Experiment 1 with three important differences. First, participants only memorized eight faces

belonging to the Leopards and eight belonging to the Tigers (similar to Experiment 3).

Second, four Black and four White faces that were unaffiliated with the Leopards or Tigers

appeared for the first time during fMRI. Again, faces were randomly assigned to group and

fully counterbalanced. Third, participants completed a different categorization task during

70

scanning, a different memory task following scanning, and two new automatic evaluation

measures following scanning. These new tasks are described below.

Categorization Task. Participants completed five runs of four blocks containing 12

trials for a total of 240 trials during fMRI. On each trial, participants categorized one of the

24 faces in one of two ways. On Ingroup trials, participants responded only if the face was an

ingroup member. On Outgroup trials, participants responded only if the face was an outgroup

member. Ingroup and Outgroup blocks were counterbalanced within runs, creating four

randomized blocks within each run. Each of the 24 faces was categorized twice in each run

(once in the Ingroup task and once in the Outgroup task). Direction screens were presented

before each block for six seconds to cue the categorization required for the following block

of 12 trials. Each face appeared for two seconds, during which time participants responded

with a button box in their right hand. To allow for estimation of the hemodynamic signal,

fixation crosses appeared between names for two, four or six seconds (in pseudo-random

order).

Scanning Parameters. Participants were scanned using a Siemens 3T scanner.

Functional scanning was prescribed parallel to the AC–PC line and nearly isotropic

functional images were acquired from inferior to superior using a single-shot gradient echo

planar pulse sequence (32 axial slices; 3.5 mm thick; 0.5 mm skip; TE = 25 ms; TR = 2000

ms; in-plane resolution = 3.5 × 3.5 mm; matrix size = 64 × 64; FOV = 224 mm). Following

functional imaging, a high resolution MPRAGE anatomical image (176 sagittal slices; TE =

2.15 ms; TR = 1760 ms; resolution = 1.0 × 1.0 × 1.0 mm) was collected for normalization.

Rating Task. After scanning, participants completed several computerized

questionnaires. First, participants completed a face rating task in which they were told that

71

“people can often quickly determine who they like or dislike based on subtle facial features

and expressions” and asked to rate each of the 24 faces on a 6-point liking scale (1 = dislike

to 6 = like). Faces were presented in random order.

Memory Task. To measure memory participants saw each of the 24 faces and were

asked to indicate with a button press whether each face was a member of the (a) Leopards,

(b) Tigers, or (c) neither team. Response accuracy was recorded. Faces were presented in

random order.

Automatic Evaluation Task. To measure automatic evaluations of the faces

participants completed two Single Category Implicit Association Tests (Karpinski &

Steinman, 2006) on a personal computer. One SC-IAT assessed their automatic evaluations

of ingroup members and the other SC-IAT assessed their automatic evaluations of outgroup

members. The SC-IAT is an adaptation of the IAT (Greenwald et al., 1998) designed to

measures associations toward a single category of stimuli. The primary advantage of the SC-

IAT over the traditional IAT is that there is a single stimulus category to evaluate allowing

for a more independent estimate of evaluations toward category. Intergroup biases in the

traditional IAT may stem from positive evaluations of one category (e.g., ingroup) or

negative evaluations of the other category (e.g., outgroup), or both. Each SC-IAT consisted

of two blocks where each consisted of 20 practice trials that were not included in the analysis

followed by 60 critical trials. On the positive block participants were instructed to press “f”

when a good word or a target face appeared, and press “j” when a bad word or unrelated face

appeared. On the negative block participants were instructed to press “j” when a bad word or

a target face appeared, and press “f” when a good word or unrelated face appeared. Faces

were presented in random order in both blocks and the order of SC-IATs was

72

counterbalanced. On the ingroup SC-IAT target faces were ingroup members and unrelated

faces were outgroup members or unaffiliated faces. On the outgroup SC-IAT target faces

were outgroup members and unrelated faces were ingroup members or unaffiliated faces. The

dependent measure was the average reaction time during critical trials in the positive and

negative blocks. Following previous research on the IAT (Greenwald et al., 1998) all

responses that occurred less than 300 ms following stimulus presentation were deleted and

reaction times were logged to normalize the distribution (note: results are presented using

raw reaction time in milliseconds to ease interpretation). It was assumed that more positive

associations toward a target group (e.g., ingroup) would lead to faster reaction times during

the positive block and slower reaction times during the negative block, and visa versa.

Results

Behavioral Results

Analysis strategy. To assess reactions during fMRI, we conducted a 2 (group: ingroup,

outgroup) × 2 (race: Black, White) × 2 (task: Ingroup, Outgroup) repeated-measures analyses

on reaction time.25

To assess controlled evaluations and memory, we conducted separate 3

(group: ingroup, outgroup, unaffiliated) × 2 (race: Black, White) repeated-measures analyses

on liking and accuracy, respectively. To assess automatic evaluations, we conducted 2

(group: ingroup, outgroup) × 2 (block: positive, negative) repeated-measures analysis on

logged reaction time.

fMRI Categorization Speed and Accuracy. We examined reaction time (in

milliseconds) during fMRI. Similar to Experiment 1, participants were faster to categorize

ingroup (M = 1272) compared to outgroup (M = 1350) members, F(1, 1460) = 8.43, p < .01.

They were also faster to categorize White (M = 1285) compared to Black (M = 1337) faces,

73

F(1, 1460) = 3.68, p < .06. However, a Group × Task interaction, F(1, 1460) = 42.84, p <

.0001, indicated that participants were faster to categorize ingroup members during the

Ingroup task (M = 1170) than during the Outgroup task (M = 1373) and faster to categorize

outgroup members during the Outgroup task (M = 1269) than during the Ingroup task (M =

1431). This pattern of results indicated that participants were fastest when the face fit the

current task – a correct detection.

Memory. It was assumed that participants would have superior memory for novel

ingroup than outgroup faces, consistent with the results in Experiment 4. As anticipated, a

Group × Race interaction, F(2, 36) = 3.12, p < .06, indicated that participants had better

memory for White (M = 0.78) and Black (M = 0.82) ingroup members, regardless of race,

and for White (M = 0.79) compared to Black (M = 0.67) outgroup members and White (M =

0.55) compared to Black (M = 0.36) unaffiliated faces (see Figure 13). These results show

that ingroup members were remembered best, regardless of race, but that outgroup members

and unaffiliated faces showed the standard pattern of ORB, mirroring the pattern of

automatic evaluations in Experiment 3.

Controlled Evaluations. The results in Experiment 3 lead us to predict that participants

would report racial egalitarianism toward all the faces (including unaffiliated) and a

preference for novel ingroup faces relative to outgroup and unaffiliated faces. As seen in

Figure 14, participants reported a preference for ingroup members, F(2, 36) = 13.91, p <

.001. Specifically, participants reported positive controlled evaluations of ingroup members

(M = 3.93) relative to control faces (M = 3.08) who were rated similar to outgroup members

(M = 3.37. There was also a non-significant effect of race, F(1, 18) = 0.61, p < .44, such that

participants trended toward a preference for Black over White faces; however, there were no

74

Group × Race interactions, F(2, 36) = 1.26, p = .29. As expected, participants reported

ingroup bias and egalitarian racial attitudes to all groups.

Automatic Evaluations. Following the results from Experiment 3, we expected

participants to show positive evaluations of ingroup members on the Ingroup SC-IAT and

neutral evaluations of outgroup members on the Outgroup SC-IAT. As seen in Figure 15,

participants had more positive evaluations of ingroup compared to outgroup members.

Specifically, the Group × Valence interaction, F(1, 16) = 19.43, p < .001, was characterized

by automatic ingroup bias (faster reaction times between positive + ingroup), F(1, 17) =

26.79, p < .0001, but not outgroup derogation (faster reaction times between negative +

outgroup), F(1, 16) = .34, p = .57. These results replicate the previous priming data showing

an automatic ingroup bias and relatively neutral evaluations of outgroup faces.26

fMRI Results

fMRI Preprocessing and Analysis. Data were prepared for analysis using SPM5

(Wellcome Department of Cognitive Neurology, London, UK). Data were corrected for slice

acquisition time and motion, co-registered to structural images, transformed to conform to

the default T1 MNI brain interpolated to 3 × 3 × 3 mm, and smoothed using an 9 mm FWHM

(full-width-half-maximum) kernel. Data were also corrected for artifacts and detrended. Data

were analyzed using the general linear model in SPM5. The BOLD signal was modeled as a

function of a canonical hemodynamic response function and its temporal derivative with a

128 s high-pass filter. First-level images were analyzed at the second-level with a 2 (group:

ingroup, outgroup) × 2 (race: Black, White) × 2 (task: explicit, implicit) repeated-measures

ANOVA. All effects from whole-brain analyses were reported as statistically significant if

they exceeded p ≤ .001 (uncorrected) with at least ten contiguous voxels. To conduct a priori

75

analyses of the amygdala all voxels were extracted from the left and right amygdala using an

anatomical mask. All effects from ROI analyses (i.e., a priori analyses of the amygdala and a

posteriori ROIs used to test moderation) were reported as statistically significant if they

exceeded p ≤ .05 (uncorrected) with at least ten contiguous voxels. We also used ROIs to

examine correlations. Data from all 19 participants were included in the fMRI analyses.

Amygdala. The primary goal of the current experiment was to examine the sensitivity

of the amygdala to the group membership (i.e., ingroup > outgroup) of affiliated faces and

the race (i.e., Black > White) of unaffiliated faces. Following the results from Experiment 1,

we predicted that the amygdala would be more active to novel ingroup than outgroup

members, regardless of race. As anticipated, novel ingroup members were associated with

greater amygdala activity (MNI coordinates: 33, 0, -24) than outgroup members, F(1, 18) =

8.88, p < .001 (one-tailed). In other words, the amygdala was sensitive to the group

membership of faces, showing greater activity to ingroup members.

These results differed from previous research showing that Black (> White) faces were

associated with amygdala activity (e.g., Cunningham, Johnson et al., 2004; Hart et al., 2000),

leading us to propose that the amygdala may respond to race when it is the most relevant or

salient social category. The results from Experiment 3 suggest that people assigned to a

mixed-race group construe ingroup members according to the group membership and

unaffiliated faces according to their race. We therefore anticipated that unaffiliated faces

would elicit greater activity to Black than White unaffiliated faces. Indeed, we found that

Black faces (M = .30) were associated with greater amygdala activity than White faces (M =

.18), F(1, 18) = 4.06, p < .03 (one-tailed).27

There was no effect of race among ingroup or

outgroup faces (ps > .21) and the effect of race on amygdala activity was not moderated by

76

task (p > .15). Together, these results are consistent with models that posit a more general

role for the amygdala in processing motivationally-relevant stimuli (e.g., Cunningham et al.,

2008) such as ingroup members (Allport, 1954), but also Black unaffiliated faces for whom

race is the only salient social category cue.

Fusiform and Memory. The secondary goal of the current research was to directly

examine the relationship between ingroup biases in social perception and evaluation and the

neural processing of group membership. Self-categorization with a novel group led to greater

fusiform activity to ingroup members in Experiment 1 and better memory for ingroup

members in Experiment 5. We therefore anticipated and found a positive correlation between

ingroup biases in fusiform activity (ingroup-outgroup) and individual differences in memory

(ingroup-outgroup). As seen in Figure 16, people who had superior memory for ingroup

members also had greater fusiform activity (MNI coordinates: 48, -42, -15) to ingroup

members (r = .61, p < .01). These results support the view that self-categorization elicits

ingroup biases in fusiform activity and memory, possibly by increasing the individuation of

ingroup members (Rhodes et al., 2004; Sporer, 2001).

Amygdala and Automatic Evaluations. In addition, participants completed two SC-

IATs (Karpinski & Steinman, 2006) to measures their automatic evaluations of ingroup and

outgroup members, respectively, and examine the link between automatic evaluations and

neural processing. Several studies reviewed in the introduction have found a correlation

between automatic racial biases and amygdala activity (Cunningham, Johnson et al., 2004;

Phelps et al., 2000). The current experiment examined whether automatic ingroup bias would

correlate with amygdala activity to ingroup members. Amygdala activity to ingroup or

77

outgroup members was not correlated with automatic evaluations toward ingroup and

outgroup members (rs < .35).

Discussion

The current experiment provided evidence that the amygdala responds both to ingroup

compared to outgroup members and unaffiliated Black compared to White faces. These

results replicate and extend Experiment 1 and 3, by showing that people process group

membership when it is relevant and race when it the only salient social category cue in a

fashion consistent with models that posit a more general role for the amygdala in processing

motivationally-relevant stimuli (e.g., Cunningham et al., 2008). That is, ingroup members

(Allport, 1954) in the current intergroup context and Black faces when race is the only salient

social category cue. This pattern of amygdala activity is also consistent with the idea that the

amygdala may respond to race in contexts where it is the most salient social category, which

was generally the case in previous research (Hart et al., 2000; Lieberman et al., 2005) but not

Experiment 1.

This experiment offered direct evidence of a link between ingroup biases in social

perception and evaluation and the neural processing of group membership. As predicted,

participants with the largest ingroup bias in fusiform activity (ingroup-outgroup) also had the

largest ingroup bias in memory (ingroup-outgroup). Together with Experiments 1 and 5,

these data support the view that self-categorization can elicit ingroup biases in fusiform

activity and memory. As described earlier, fusiform activity to own-race faces has been

similarly shown to correlate with superior own-race memory (Golby et al., 2001). Yet, this

latter study was initially interpreted as evidence that the fusiform was responding to own-

race faces because it plays a key role in expert processing (Gauthier et al., 1999). In the

78

current experiments, both ingroup and outgroup faces were relatively novel making it

unlikely that a lifetime of exposure is necessary to generate ingroup biases in memory of

fusiform activity. Instead, the current data suggest that self-categorization may play an

important role in these biases, possibly by increasing the individuation of ingroup members

(Rhodes et al., 2004; Sporer, 2001).

79

Chapter 8: General Discussion

The central finding from the current dissertation is that self-categorization with a novel

group systematically alters social perception and evaluation. Data from five experiments

suggests that when people are assigned to a mixed-race group they develop positive

evaluations and superior memory for ingroup members, regardless of race. These ingroup

biases in perception and evaluation are mediated by a shift in neural processing in the

amygdala, fusiform, and OFC, which appear to mediate several of these behavioural biases.

Specifically, Experiment 1 provided evidence that OFC activity mediated reported

preferences for ingroup members and Experiment 5 provided evidence that fusiform activity

mediated superior memory for ingroup members. Although the amygdala was not correlated

with evaluations or memory, it appears to play a role in tracking the motivational relevance

(Cunningham et al., 2008; Whalen, 1998), showing sensitivity to ingroup members, but also

to Black faces unaffiliated with either mixed-race group. Similarly, unaffiliated Black and

White faces also elicit automatic racial biases (Experiment 3) and the ORB (Experiment 5).

These results suggest that people unaffiliated with the ingroup or outgroup were

automatically construed and evaluated according to their race and automatic construals may

shift between group membership and race on a case-by-case basis within a stable intergroup

context.

Flexibility in Social Perception and Evaluation

Humans belong to many dynamic and overlapping social groups and the importance of

any given social category (e.g., age, race, gender, and occupation) can shift from one

situation to another. In such a complex and dynamic social world, a central challenge for

80

adaptive human behaviour is the flexible and appropriate categorization and evaluation of

others. As we noted earlier, basketball players need to differentiate teammates from

opponents while retaining the ability to process referees and fans according to an entirely

different set of criteria. Social perception may therefore include the ability to rapidly and

spontaneously categorize some people according to one dimension and other people on a

different dimension. Consistent with this perspective, Experiments 3 and 5 suggest that

people unaffiliated with the ingroup or outgroup are construed and evaluated according to

their race whereas ingroup members are evaluated and processed according to their group

members and apparently regardless of race. Moreover, these effects occurred while

participants were randomly presented with a series of individual faces. It therefore seems that

intergroup perceptions and evaluations spontaneously shift from one salient categorization to

another (and back again) within a single intergroup context – what we call flexibility.

In the current intergroup context, stimuli can be evaluated according to at least two

social categories: group membership and race. In particular, ingroup and outgroup members

can be evaluated using either of those categories while unaffiliated faces are likely to be

evaluated only by race. All five experiments offer evidence that people in novel groups

process ingroup members according to their group membership, and both experiments that

included unaffiliated faces offer evidence that they were processed according to their race.

The current intergroup context likely triggers self-categorization with the novel ingroup,

increasing attention to the group membership of ingroup and outgroup members. The

amygdala results in Experiment 1 and 5 suggest that ingroup members may be especially

important in this intergroup context, generating ingroup biases in perception and evaluation.

81

In contrast, the Black and White unaffiliated faces were not a member of either team, making

race the most (visually) salient social category cue for processing these faces.

More intriguing is the pattern of racial bias toward outgroup members which was

similar to the pattern for unaffiliated faces and participants in a control condition. As noted

earlier, construing ingroup members according to their group membership and outgroup

members and unaffiliated faces according to their race may occur for a number of reasons.

One possibility is that outgroup members are perceived as less relevant or receive less

attention to their group membership, and are thus processed based on more superficial

visually-salient social category cues, such as race. A second possibility is that outgroup

members are associated with relatively neutral evaluations leaving race-based association to

provide the basis for outgroup evaluations. These two possibilities imply a hierarchical stage-

based model of intergroup perception and evaluation in which ingroup membership is

processed first and if a given face does not belong to the ingroup it is processed according to

race, regardless of whether this is due to an absence of attention toward outgroup faces or an

absence of strong evaluative associations toward outgroup faces. A third possibility is that

something about the intergroup context may trigger positive automatic evaluations of any

ingroup member, including novel ingroup members and White faces. In other words, any

group associated with the self (whether racial or novel) may elicit positive automatic

evaluations. This latter possibility assumes that White outgroup and unaffiliated faces elicit

automatic racial bias (White > Black) because participants self-categorize with White faces,

increasing their positivity toward racial ingroup members whereas the other possibilities

allow that well-learned evaluative associations toward these racial categories may also drive

racial bias. Although the current research does not speak to the specific mechanism behind

82

this effect, the evidence does suggest that race is used to automatically evaluate outgroup

members and unaffiliated faces.

These findings advance the literature on intergroup evaluation in a number of ways. As

noted earlier, previous research on automatic evaluations focused on relatively simple social

categories or examined the automatic evaluations of multiply-categorizable targets with

measures that use explicit category labels (e.g., Mitchell et al., 2003), finding that evaluations

are largely consistent with the most salient category. The present data, however, suggest that

people spontaneously and rapidly evaluate others according to a blend of contextually-

relevant self-categorizations and pre-existing associations toward salient social categories,

and may shift between these categories in a flexible fashion.

The Primacy of the Ingroup?

“Although we could not perceive our own in-groups excepting as they contrast to

out-groups, still the in-groups are psychologically primary.... Hostility toward out-

groups helps strengthen our sense of belonging, but it is not required.” – Allport

(1954, p. 42)

The importance of ingroup members for (intra)group cooperation, reproduction, and

survival is now well established (Correll & Park, 2005; D. S. Wilson & Sober, 1994). As

Allport (1954) anticipated, individuals who are able to accurately identify, value, and

cooperate with ingroup members enjoy numerous functional benefits including the

fulfillment of basic psychological needs. Accordingly, intergroup biases and conflict most

frequently arise – at least initially – from this apparently innocuous form of ingroup bias

(Brewer, 1999). Although the value of group membership and identity is widely

acknowledged among social psychologists, research on social prejudice has nevertheless

focused heavily on outgroup derogation, perhaps due to an implicit understanding that this

83

form of prejudice is more pernicious. The focus on outgroup derogation has left unanswered

a number of basic questions about the nature of intergroup biases in perception, evaluation,

and behavior, including the direction and nature of bias in novel group contexts.

The current research found clear and repeated evidence of ingroup biases on a host of

perception and evaluation tasks. Across multiple experiments, ingroup members were

associated with more positive evaluations on both automatic (Experiments 2, 3, 5) and

controlled (Experiments 1-5) measures of evaluation, memory (Experiments 4-5), and neural

processing (Experiments 1, 5). These patterns reflected both a preference for ingroup

members relative to outgroup members, the midpoint of various measures, and most

importantly, unaffiliated control faces. The inclusion of controls faces in these studies

showed that ingroup biases were not offset by corresponding outgroup derogation: control

faces were evaluated similar to outgroup members. Although previous minimal group

research generally suggests a pattern of ingroup bias (Brewer, 1979), no published

experiments have used appropriate controls to isolate the nature of ingroup biases in minimal

group contexts (McCaslin & Petty, 2008). The current set of experiments help address this

gap and offers evidence that ingroup biases in perception and evaluation are broad and

associated in important ways. For example, Experiment 4 shows that these ingroup biases not

only entail superior memory for novel ingroup members, but that better memory for ingroup

members is associated with the most positive reported evaluations of ingroup members. In

contrast, memory for outgroup members was unrelated to evaluation. Although it remains

unclear whether the link between ingroup bias in memory and evaluation reflected the more

deliberate personalization (Brewer, 1988) of ingroup members or the motivated enhancement

of ingroup members (Tajfel, 1982), it does suggest that either process is specific to ingroup

84

bias. Moreover, evidence of ingroup bias in the absence of outgroup derogation is consistent

with the suggestion that ingroup and outgroup evaluations are not necessarily reciprocally

related (Allport, 1954; Brewer, 1999).

Social Categorization

A central tenet of Self Categorization Theory is that the activation and application of

any given social identity is dependent on the current intergroup context (J. C. Turner et al.,

1987; J. C. Turner et al., 1994). It follows that assignment to a novel group should increase

self-categorization and drive social perception and evaluation in a fashion consistent with this

novel group membership, overriding the use of other social identities and social categories

until the context changes. However, the majority of research on crossed-categories has

examined the effects of crossing two social categories (see Crisp & Hewstone, 2007) or

novel groups (see Vanbeselaere, 1991) on intergroup bias, making it unclear whether self-

categorization is actually powerful enough to override other salient or well-learned social

categories. The current research provides evidence that mere categorization with a novel

group is sufficient to override the perception and evaluation of ingroup members according

to their race (Kurzban et al., 2001). In other words, a relatively unimportant self-

categorization can interact with, or even override, automatic and controlled evaluations and

memory for a visually salient and important social category like race. In fact, priming any

alternative social identity (e.g., occupation) should reduce or override racial biases in

memory, and perhaps more generally in social perception.

A central assumption in the current research is that social perception and evaluation are

closely linked. However, several recent papers have found an effect of crossed-categories on

categorization, but not evaluation (e.g., Vescio, Judd, & Kwan, 2004) leading researchers to

85

question the link between social categorization and intergroup bias (Park & Judd, 2005). It is

likely the case that initial perceptual processing does not correspond perfectly with

subsequent evaluative processing. However, the current research is premised on the idea that

initial perceptual processing triggers evaluative associations and additional component

processes that come online and continue to operate on representations. With more time, more

component processes come online, and begin to integrate initial evaluative associations with

motivates and personal values (Cunningham et al., 2007). In light of this framework, the

current research offers several insights for ongoing research on categorization and ingroup

bias. First, the current research used measures of evaluation without explicit category labels,

allowing multiple social categories (e.g., group and race) to drive evaluations. This

minimized the extraneous influence of specific category labels on evaluation and allowed for

spontaneous construals to affect evaluation. This approach revealed that participants

spontaneously evaluated ingroup members according to their group membership, but also

that race was used during evaluation in a rather complex fashion. Therefore, the current

crossed-category context did alter automatic evaluations. Second, the current research found

that more controlled evaluations were not influenced by race highlighting the fact that

automatic and controlled evaluations can be dissociated, especially when additional

component processes alter initial evaluations to suit motivational concerns. Taken together,

the current research suggests that categorization and evaluation will be closely linked when

they match in terms of measurement (allowing construals to influence evaluation) and time

(i.e., the number of component processes engaged in categorization and evaluation is

similar), especially evaluations of social groups prone to motivated suppression or inhibition,

like race (Crandall & Eshleman, 2003; Dunton & Fazio, 1997; Plant & Devine, 1998).

86

An important implication of the current research is that many of the effects of race

previously attributed to stereotypes, prejudice, familiarity, etc., such as the ORB (Malpass &

Kravitz, 1969), may stem from aspects of self-categorization. These racial biases in social

perception and evaluation may be elicited most reliably and perhaps exacerbated when race

is a cue to group membership. Indeed, some researchers (Kurzban et al., 2001) have argued

that race should only guide social categorization when it is a reliable cue to current group

membership. This may explain why certain racial biases seem so pervasive – North

American society emphasizes racial distinctions in many domains, and because people live

within this context, racial self-categorizations may be repeatedly primed in everyday life.

Social Cognitive Neuroscience

This multi-level approach extends several theories of intergroup perception and

evaluation by making explicit links between self-categorization, neural processes, and social

perception and evaluation. The current set of experiments are generally consistent with at

least two general models of evaluation: the MODE model (Fazio, 1990) and the IR model

(Cunningham & Zelazo, 2007; Cunningham et al., 2007). Both models predict and explain a

number of key findings in the current research, including the dissociation between automatic

and controlled racial biases, the relationship between ingroup biases in memory and liking

(Experiment 4), the effects of construal on downstream processing, and contextual flexibility

of intergroup evaluation. The current research is, however, informed most by the IR mode of

evaluation.

The MODE model describes the key role of accessibility in evaluation (Fazio, 1990).

In the current research the accessibility of group-based evaluations can stem from the act of

self-categorization, but also from the act of learning and rehearsing the membership of group

87

members during the first phase of the study (Roskos-Ewoldsen & Fazio, 1992). According to

this perspective, the mere rehearsal of team membership may increase the accessibility of this

category relative to race which should in turn reduce racial bias following the learning task,

even among participants who are not a member of either team. To test this possibility we

included control participants in several experiments who learned team membership (which

was orthogonal to race). The results of Experiment 2 and 4 indicated that simply learning

about team membership was not sufficient to eliminate racial bias: control participants had

more positive automatic evaluations of White compared to Black faces and superior memory

for White compared Black faces. These results suggest that changes in social perception and

evaluation stem from self-categorization – including the “self” in a novel group – rather than

the accessibility of team membership induced during learning.

As we noted in the introduction, the IR model (Cunningham et al., 2007) shares a

number of core features with Self Categorization Theory (J. C. Turner et al., 1987; J. C.

Turner et al., 1994) and provides a useful multi-level framework for understanding how a

novel self-categorization should effect social perception and evaluation and the brain regions

that should mediate this relationship. For example, IR specifies the role of affective

processing in the amygdala (low-level motivational significance) and OFC (integrating

context to generate subjective valuation) and how motivations can alter the automatic

construal and evaluation of stimuli. Building on the IR Model, we propose that the contents

of several relatively stable social identities are stored in memory and that any current self-

categorization is constructed from a subset of these contents. This allows for the “inherently

variable, fluid, and context dependent” nature of self-categorization (J. C. Turner et al., 1994)

while retaining the a set of stable personal and social identities. It also allows for the

88

perception and evaluation of unaffiliated faces according to race (see Experiments 3 and 5) in

a salient intergroup context where race is irrelevant to group membership. Thus, while self-

categorization with a novel mixed race group was sufficient to reduce automatic racial bias

toward ingroup members, the effects of race on automatic evaluation (Experiment 3) and

neural processing (Experiment 5) were elicited for unaffiliated faces. This indicates that the

underlying contents associated with race (such as aspects of White racial identity) were not

completely erased – or even fully suppressed – by self-categorization with a novel group.

The current research is situated at the interface of IR and SCT, expanding IR by

linking aspects of self-categorization and social perception to evaluation and expanding SCT

to multi-levels of analysis. This research also offers insights about the algorithms and neural

implementation of self-categorization on social perception and evaluation. The experiments

in this dissertation serve as initial steps toward generating a multi-level model of social

identity and self-categorization.

Reducing prejudice

“He drew a circle that shut me out

Heretic, rebel, a thing to flout.

But Love and I had the wit to win:

We drew a circle that took him in!” - Markham (1936)

For more than half a century social psychologists have sought to understand and

eliminate prejudice. Much of this work has examined overt expressions of prejudice and acts

of discrimination. However, increasing evidence in the past few decades has identified a

discrepancy between reported prejudice and other acts of discrimination (Crosby et al., 1980)

and shown that many people still have automatic negative responses toward various minority

groups (Nosek et al., 2002). As noted earlier, these automatic biases predict discrimination,

89

including negative non-verbal behavior and biased hiring decisions towards racial and other

social groups (Dovidio et al., 2002; Dovidio et al., 1997). The current research offers a

simple shift in self-categorization can swiftly and effectively reduce racial biases in social

perception and evaluation.

By drawing a racially-diverse circle – that is, re-categorizing Black and White targets

as members of a common ingroup – participants generated positive evaluations of ingroup

members, regardless of their race. These experiments join a growing body of research that

has manipulated aspects of self-categorization to reduce intergroup bias. In the present

research, participants assigned to a novel group evaluated both Black and White ingroup

members positively, eliminating the standard pattern of automatic racial bias (Cunningham et

al., 2001; Fazio et al., 1995). In contrast, participants who merely saw two mixed-race groups

showed the standard pattern of automatic racial bias (see Experiments 2 and 3), highlighting

the power of self-categorization and social identification to shape automatic evaluation.

Taken together, these results raise the possibility that participants were either generating a

superordinate social identity that included all Black and White targets (except for outgroup

members) (Gaertner, Rust, Dovidio, Bachman, & Anastasio, 1996) or that the crossed-

categorization of novel group membership and race lead to this reduction in racial bias (Crisp

& Hewstone, 2007). According to the Common Ingroup Identity Model (Gaertner &

Dovidio, 2000), common ingroup identities are superordinate identities that subsume all

lower-level identities. For example, when Black and White people are categorized as humans

it should lead people to evaluate all novel Black and White people positively (insofar as

humans are evaluated positively). This conceptualization is captured in Nobel Peace Laureate

John Hume’s famous assertion that “our common humanity transcends our differences”. In

90

the current research, if participants were generating a superordinate identity they should have

shown little or no racial bias toward any Black individual who was not an outgroup member.

However, participants in Experiment 3 revealed automatic racial bias toward Black faces

who were unaffiliated with the ingroup or outgroup (Experiment 5 revealed a similar effect

on neural processing), suggesting that participants were not generating a superordinate social

identity that extended beyond the current salient ingroup to include all potential members of

these social categories. This raises the possibility that aspects of crossed-categorization may

have accounted for the reduction in racial bias.

In a recent review of the crossed-categorization research, Crisp and Hewstone (2007)

argue that there are two routes to reduced ingroup bias in crossed-category contexts:

differentiation and decategorization. Differentiation involves the generation of a shared

ingroup identity that brings outgroup members (i.e., Blacks) closer to the ingroup resulting in

reduced bias. Decategorization involves a shift toward more individuated processing (Fiske

& Neuberg, 1990), consistent with the own-group memory bias in Experiments 4 and 5 and

greater activity in the fusiform gyrus – a brain region involved in individuation – to ingroup

members in Experiments 1 and 5. Taken together, these experiments raise the possibility that

a shared ingroup identity – however minimal – may lead individuals to feel positive about

and individuate ingroup members.

Self-categorization in the current set of experiments increased positive evaluations of

novel ingroup members without increasing negativity toward outgroup members. This stands

in contrast to a recent meta-analysis which found that crossed-categorization tends to

increase evaluations of the double-ingroup while leading to a corresponding decrease in

evaluations of the double-outgroup (Mullen et al., 2001). Accordingly, the current research

91

provides exciting evidence that assigning people to mixed-race teams may provide a socially

constructive mechanism for attenuating racial bias. It is important to note, however, that

while automatic evaluations may be sensitive to the current self-categorization leading to a

reduction in racial bias, the underlying attitudes may remain relatively static (Cunningham et

al., 2007), producing racial biases to unaffiliated faces. At the very least, these data suggest

that automatic evaluations of complex social stimuli are not necessarily dominated by any

particular social category – even one as socially and visually salient as race.

Most prejudice reduction strategies have focused on the suppression of automatically-

activated evaluations and stereotypes (Monteith, Sherman, & Devine, 1998; Plant & Devine,

1998). However, response suppression is a narrow and inefficient form of emotion regulation

(Gross & Thompson, 2007) and may lead to increased stereotype accessibility (Macrae et al.,

1994) or interpersonal discrimination (Richeson & Shelton, 2003). The current research

suggests an alternative, antecedent-focused route to prejudice reduction: namely, changing

the ways that others are automatically construed. Changing automatic processing may be

especially important, because evaluation is dependent on information from early processes

(Cunningham et al., 2007). Small biases during the initial stages of perceptual and evaluation

processing can have dramatic downstream effects. Although evaluation and behavior may

feel controlled and deliberate, it is heavily informed by these initial automatic and non-

conscious processes. The current experiments illustrate the power of a novel self-

categorization to alter automatic components of social perception and evaluation and

ultimately override ostensibly pervasive effects of race.

The social benefits of mixed-race group membership are offset, however, by the caveat

that self-categorization leads to ingroup biases in evaluation. Although ingroup bias toward a

92

novel group may seem like a fair tradeoff for more pernicious racial biases, it is worth

revisiting the effects of minimal group membership on more overt indices of intergroup

discrimination (Tajfel et al., 1971). In many contexts, such as hiring or voting, any

differential preference for one group over another may lead to the same pattern of behavioral

discrimination, whether it is driven by ingroup bias or outgroup derogation. Therefore, self-

categorization with a novel group may offer a simple and promising approach to reduce

racial bias but it must be carefully weighed against the possibility of spawning new forms of

intergroup bias.

Future Directions

The current research raises a number of interesting questions. Although the different

results between the control and experimental conditions in Experiments 2-4 suggests that the

self played an important role in shifting automatic evaluations from race to group

membership (see Rudman, 2004 for a more detailed discussion of the role of the self in

automatic evaluations), it is unclear whether this pattern of ingroup bias stems from

evaluative conditioning during the learning paradigm (where participants categorized their

own photo and the faces of ingroup and outgroup members) (Gawronski & Bodenhausen,

2006; Walther & Trasselli, 2003 ), self-anchoring (Otten & Epstude, 2006), or some other

psychological process. Future research should also explore whether the current pattern of

results extend to other implicit attitude measures (Fazio et al., 1995; Payne et al., 2005) and

social categories, including prejudices based on age, gender, nationality, and religion. Indeed,

Kurzban and colleagues (Kurzban et al., 2001) argue that the effects of certain social

categories (i.e., age and gender) on social perception should be less malleable than race.

93

The paradigm employed in all five studies has several features that may have increased

the overall salience of race. For example, half of each team is Black, a mix that over-

represents Blacks relative to the general (or student) population at the testing sites. Further,

aspects of the tasks may have called attention to race, including the implicit categorization

task in Experiment 1. On one hand, any factor that increased the salience of race makes the

ingroup bias – regardless of race – all the more impressive. It indicates that very minimal

self-categorizations can override the effects of race, even when race is made hyper-salient by

contextual factors and explicit attention. On the other hand, it renders the racial bias toward

unaffiliated faces less impressive, raising the possibility that these factors may be necessary

to make race salient and generate these effects.28

Future research should explore this issue.

Motivated Social Perception.

An exciting implication from the current research is that self-categorization may

motivate ingroup biases in social perception (Balcetis & Dunning, 2006), including more in-

depth or individuated processing of ingroup than outgroup members. Indeed, Brewer’s Dual

Process Model of Person Perception (Brewer, 1988) suggests that relevance and self-

involvement can lead to the personalization of others, processing them as individuals while

using their social category information to generate judgments and evaluations. Although this

interpretation of our data is consistent with recent theory and research on amygdala

(Anderson & Phelps, 2001; Cunningham et al., 2008; Whalen, 1998) and OFC function

(Kringelbach, 2005) and ingroup biases in memory (Bernstein et al., 2007), it is a notable

deviation from current theory on the function of the fusiform (Palmeri & Gauthier, 2004). If

these processes truly stem from motivated aspects of social identity (Tajfel, 1982) then

participants with the strongest identification with their novel ingroup should show the largest

94

ingroup biases in brain activity and social perception. Further, the need for distinctiveness,

social identity affirmation, social ostracism, and other factors that modify the need for social

connectedness should modify these ingroup biases in social perception and evaluation.

The Time-course of Intergroup Perception

An assumption across the current set of experiments is that self-categorization is

altering the construal of individual faces, shifting attention to ingroup membership rather

than race. The nature of the specific brain regions involved (i.e., amygdala and fusiform) and

evidence from automatic evaluations suggest that ingroup members may be differentiated

automatically during very early aspects of processing (Bernstein et al., 2007). In particular,

several of the regions showing sensitivity to ingroup members are known to respond

relatively automatically; the amygdala responds to subliminal presentations of arousing faces

(Whalen et al., 1998) and the fusiform responds to faces within the first 100-200 ms

following presentation (Liu et al., 2002). Moreover, these ingroup biases in neural activity

did not require explicit attention to group membership (Experiment 1), or more specifically,

to ingroup members (Experiment 5). Although this pattern of results supports the view that

self-categorization may alter the construal of faces during the earliest phase of social

perception, temporal resolution of fMRI (in the range of seconds) is insufficient to identify

rapid changes in construal. Future research will need to use electroencephalography (EEG),

magnetoencephalography (MEG), and other techniques to determine whether self-

categorization overrides racial bias (Dickter & Bartholow, 2007; Ito & Urland, 2003) during

these early stages of perceptual processing by changing the construal of faces according to

their group membership. If a novel self-categorization can change very early intergroup

95

processing racial and other biases may be eliminated before they impact subsequent

perceptions, evaluations, or behaviour.

Attention, Encoding, and Retrieval

The current research has been relatively silent to the source of these ingroup biases in

social perception and evaluation. Although the current results are described in terms of in

vivo biases they may stem from attentional biases during the learning phase. Future research

will need to examine the allocation of attention and nature of encoding during the initial

learning phase and explore how these biases affect subsequent processing of ingroup and

outgroup members. One possibility is that people allocate more attention to ingroup members

during encoding (e.g., eye gaze is focused longer on ingroup faces) which leads to increased

processing of these faces and superior memory (Ruscher et al., 1991). Surprisingly, the

encoding aspect of group learning has been largely ignored in previous minimal group

research despite the obvious role it plays in real world group interactions.

Discrimination

As noted earlier, automatic biases are associated with various forms of discrimination,

including negative non-verbal behavior and biased hiring decisions towards racial and other

minorities (Dovidio et al., 2002; Dovidio et al., 1997). An assumption implicit in the current

research is that changes in evaluation will ultimately affect behavior (Fishbein & Ajzen,

1975). We therefore expect that self-categorization with a novel group should produce

positive, pro-social interactions with ingroup members and reduce the subtle forms of

discrimination associated with automatic racial bias, at least for ingroup members. According

to the example above, simply assigning people to a game of pick-up basketball should swiftly

change the perceptions, evaluations, and behavior toward teammates, leading to positive

96

social interactions, regardless of race or creed.29

This may be especially true of the non-

verbal forms of behavior that characterize normal, spontaneous social interactions.

Conclusion

In an era of increasing globalization, social and economic harmony depends on the

ability of people to cooperate with others from a variety of ethnic, geographic, and religious

backgrounds. In such a complex and dynamic social world, a central challenge for adaptive

human behavior is the flexible and appropriate categorization and evaluation of others. The

current research illustrates that self-categorization with a social group – however minimal –

can dramatically alter social perception and evaluation, and override ostensibly pervasive

racial biases. By using a multi-level approach this research also elucidates the neural

substrates that underlie ingroup biases in social perception and evaluation. Humans are a

fundamentally social species and understanding the neural component processes that underlie

intergroup perception and evaluation promises important insights into how people navigate

their complex social worlds. In the words of E.O. Wilson (1998), “(i)f social scientists

choose to select rigorous theory as their ultimate goal, as have the natural scientists, they will

succeed to the extent they traverse broad stretches of time and space. That means nothing less

than aligning their explanations with those of the natural sciences.”

97

References

Aboud, F. E. (1988). Children and prejudice. New York: Blackwell.

Allport, G. W. (1954). The Nature of Prejudice. Reading, MA: Addison Wesley.

Amodio, D. M., Harmon-Jones, E., Devine, P. G., Curtin, J. J., Hartley, S. L., & Covert, A.

E. (2004). Neural signals for the detection of unintentional race bias. Psychological

Science, 15(2), 88-93.

Anderson, A. K., Christoff, K., Stappen, I., Panitz, D., Ghahremani, D. G., Glover, G., et al.

(2003). Dissociated neural representations of intensity and valence in human

olfaction. Nature Neuroscience, 6, 196-202.

Anderson, A. K., & Phelps, E. A. (2001). Lesions of the human amygdala impair enhanced

perception of emotionally salient events. Nature, 411, 305-309.

Ashburn-Nardo, L., Voils, C. I., & Monteith, M. J. (2001). Implicit associations as the seeds

of intergroup bias: How easily do they take root? Journal of Personality and Social

Psychology, 81, 789-799.

Bagley, C., & Verma, G. (1979). Racial prejudice: The individual and society. Farnborough,

U.K.: Saxon House.

Balcetis, E., & Dunning, D. (2006). See what you want to see: Motivational influences on

visual perception. Journal of Personality and Social Psychology, 91(4), 612-625.

Bandura, A. (1977). Social Learning Theory. New York: General Learning Press.

Bargh, J. A. (1994). The Four Horsemen of automaticity: Awareness, efficiency, intention,

and control in social cognition (2 ed.). Hillsdale, NJ: Erlbaum.

98

Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the automatic

attitude activation effect. Journal of Personality & Social Psychology, 62(6), 893-

912.

Baron, A. S., & Banaji, M. R. (2006). The Development of Implicit Attitudes: Evidence of

Race Evaluations from Ages 6 and 10 and Adulthood. Psychological Science, 17(1),

53-58.

Bartels, A., & Zeki, S. (2000). The neural basis of romantic love. Neuroreport, 11, 3829-

3834.

Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998). Ego depletion: Is the

active self a limited resource? Journal of Personality and Social Psychology, 74(5),

1252-1265.

Bernstein, M., Young, S., & Hugenberg, K. (2007). The Cross-Category Effect: Mere social

categorization is sufficient to elicit an own-group bias in face recognition.

Psychological Science, 18, 709-712.

Bertrand, M., & Mullainathan, S. (2003). Are Emily and Greg more employable than Lakisha

and Jamal?: A field experiment on labor market discrimination, Working paper series

WP 03-22: Massachusetts Institute of Technology Department of Economics.

Blair, I. V. (2002). The malleability of automatic stereotypes and prejudice. Personality &

Social Psychology Review, 6(3), 242-261.

Bobo, L. (2001). Racial attitudes and relations at the close of the twentieth century. In N. J.

Smelser, W. J. Wilson & F. Mitchell (Eds.), America Becoming: Racial Trends and

Their Consequences (pp. 264-301). Washington, D.C.: National Academy Press.

99

Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict

monitoring and cognitive control. Psychological Review, 108(3), 624-652.

Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive-

motivational analysis. Psychological Bulletin, 86(2), 307-324.

Brewer, M. B. (1988). A dual process model of impression formation. In T. Srull & R.Wyer

(Eds.), Advances in Social Cognition (Vol. 1): Earlbaum.

Brewer, M. B. (1999). The psychology of prejudice: Ingroup love or outgroup hate? Journal

of Social Issues, 55, 429-444.

Brigham, J. C., & Ready, D. J. (2005). Own-race bias in lineup construction. Law and

Human Behavior, 9(4), 415-424.

Brown, D. E. (1991 ). Human Universals. New York: McGraw-Hill.

Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64(2), 123-152.

Cacioppo, J. T., Berntson, G. G., Sheridan, J. F., & McClintock, M. K. (2000). Multilevel

integrative analyses of human behavior: Social neuroscience and the complementing

nature of social and biological approaches. Psychological Bulletin, 126, 829-843.

Chaiken, S., & Trope, Y. (1999). Dual-process theories in social psychology. New York:

Guilford Press.

Chiao, J. Y., Iidaka, T., Gordon, H. L., Nogawa, J., Bar, M., Aminoff, E., et al. (in press).

Cultural specificity in amygdala response to fear faces. Journal of Cognitive

Neuroscience.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied Regression/Correlation

Analysis for the Behavioral Sciences. Mahwah, NJ: Erlbaum.

100

Correll, J., & Park, B. (2005). A model of the ingroup as a social resource. Personality and

Social Psychology Review, 9, 341-359.

Cosmides, L., Tooby, J., & Kurzban, R. (2003). Perceptions of race. Trends in Cognitive

Sciences, 7(4), 173-179.

Crandall, C. S., & Eshleman, A. (2003). A justification-suppression of the expression and

experience of prejudice. Psychological Bulletin, 129(3), 414-446.

Crisp, R. J., & Hewstone, M. (2007). Multiple social categorization. In M. P. Zanna (Ed.),

Advances in Experimental Social Psychology (Vol. 39, pp. 163-254). Orlando, FL:

Academic Press.

Crocker, J., Luhtanen, R., Blaine, B., & Broadnax, S. (1994). Collective self-esteem and

psychological well-being among White, Black, and Asian college students.

Personality and Social Psychology Bulletin, 20, 503-513.

Crosby, F., Bromley, S., & Saxe, L. (1980). Recent unobtrusive studies of Black and White

discrimination and prejudice: A literature review. Psychological Bulletin, 87, 546-

563.

Cunningham, W. A., & Johnson, M. K. (2007). Attitudes and evaluation: Toward a

component process framework. In E. Harmon-Jones & P. Winkielman (Eds.), Social

neuroscience: Integrating biological and psychological explanations of social

behavior (pp. 227-245). New York: The Guilford Press.

Cunningham, W. A., Johnson, M. K., Gatenby, J. C., Gore, J. C., & Banaji, M. R. (2003).

Neural components of social evaluation. Journal of Personality and Social

Psychology, 85, 639-649.

101

Cunningham, W. A., Johnson, M. K., Raye, C. L., Gatenby, J. C., Gore, J. C., & Banaji, M.

R. (2004). Separable neural components in the processing of black and white faces.

Psychological Science, 15, 806-813.

Cunningham, W. A., Nezlek, J. B., & Banaji, M. R. (2004). Implicit and explicit

ethnocentrism: Revisiting the ideologies of prejudice. Personality and Social

Psychology Bulletin, 30, 1332-1346.

Cunningham, W. A., Preacher, K. J., & Banaji, M. R. (2001). Implicit attitude measures:

Consistency, stability, and convergent validity. Psychological Science, 12, 163-170.

Cunningham, W. A., Van Bavel, J. J., & Johnsen, I. R. (2008). Affective flexibility:

Evaluative processing goals shape amygdala activity. Psychological Science, 19(2),

152-160.

Cunningham, W. A., & Zelazo, P. D. (2007). Attitudes and evaluations: A social cognitive

neuroscience perspective. Trends in Cognitive Sciences, 11(3), 97-104.

Cunningham, W. A., Zelazo, P. D., Packer, D. J., & Van Bavel, J. J. (2007). The iterative

reprocessing model: A multi-level framework for attitudes and evaluation. Social

Cognition, 25, 736-760.

Dasgupta, N., & Greenwald, A. G. (2001). Exposure to admired group members reduces

automatic intergroup bias. Journal of Personality and Social Psychology, 81, 800-

814.

Deschamps, J. C., & Doise, W. (Eds.). (1978). Crossed category memberships in intergroup

relations. Cambridge: Cambridge University Press.

Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled components.

Journal of Personality and Social Psychology, 56(1), 5-18.

102

Devine, P. G., Plant, E. A., Amodio, D. M., Harmon-Jones, E., & Vance, S. L. (2002). The

regulation of explicit and implicit race bias: The role of motivations to respond

without prejudice. Journal of Personality & Social Psychology, 82(5), 835-848.

DeWall, C. N., MacDonald, G., Webster, G. D., Tice, D. M., & Baumeister, R. F. (2007).

Acetaminophen reduces psychological hurt feelings over time.Unpublished

manuscript.

Dickter, C. L., & Bartholow, B. D. (2007). Racial ingroup and outgroup attention biases

revealed by event-related brain potentials. Social Cognitive and Affective

Neuroscience, 2(3), 189-198.

Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and explicit prejudice and

interracial interaction. Journal of Personality & Social Psychology, 82(1), 62-68.

Dovidio, J. F., Kawakami, K., Johnson, C., Johnson, B., & Howard, A. (1997). On the nature

of prejudice: Automatic and controlled processes. Journal of Experimental Social

Psychology, 33, 510-540.

Draine, S. C., & Greenwald, A. G. (1998). Replicable unconscious semantic priming.

Journal of Experimental Psychology: General, 127, 286-303.

Dubois, S., Roission, B., Schiltz, C., Bodart, J. M., Michel, C., Bruyer, R., et al. (1999).

Effect of familiarity on the processing of human faces. NeuroImage, 9, 278-289.

Dunham, Y., Baron, A. S., & Banaji, M. R. (2006). From American city to Japanese village:

A cross-cultural investigation of implicit race attitudes. Child Development, 77(5),

1268-1281.

103

Dunton, B. C., & Fazio, R. H. (1997). An individual difference measure of motivation to

control prejudiced reactions. Personality & Social Psychology Bulletin, 23(2), 316-

326.

Eagly, A. H., & Chaiken, S. (1993). Psychology of Attitudes. Fort Worth, TX: Harcourt,

Brace, Jovanovich.

Eberhardt, J. L., Goff, P. A., Purdie, V. J., & Davies, P. G. (2004). Seeing Black: Race,

Crime, and Visual Processing. Journal of Personality & Social Psychology, 87(6),

876-893.

Eisenberger, N. I., & Lieberman, M. D. (2004). Why rejection hurts: A common neural alarm

system for physical and social pain. Trends in Cognitive Sciences, 8, 294-300.

Eisenberger, N. I., Lieberman, M. D., & Williams, K. D. (2003). Does rejection hurt? An

fMRI study of social exclusion. Science, 302, 290-292.

Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: The MODE

model as an integrative framework. In M. P. Zanna (Ed.), Advances in experimental

social psychology (Vol. 23, pp. 75-109). New York: Academic Press.

Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). Variability in automatic

activation as an unobtrusive measure of racial attitudes: A bona fide pipeline? Journal

of Personality and Social Psychology, 69(6), 1013-1027.

Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the automatic

activation of attitudes. Journal of Personality & Social Psychology, 50(2), 229-238.

Fazio, R. H., & Towles-Schwen, T. (1999). The MODE model of attitude-behavior

processes. In S. Chaiken & Y. Trope (Eds.), Dual Process Theories in Social

Psychology (pp. 97-116). New York: Guilford Press.

104

Fishbein, M., & Ajzen, I. (1974). Attitudes toward objects as predictors of single and

multiple behavioral criteria. Psychological Review, 81, 59-47.

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to

theory and research. Reading, MA: Addison-Wesley.

Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from category-

based to individuating processes: Influences of information and motivation on

attention and interpretation. In M. P. Zanna (Ed.), Advances in experimental social

psychology (Vol. 23, pp. 1-74). New York: Academic Press.

Forster, J., Liberman, N., & Friedman, R. S. (2007). Seven principels of goal activation: A

systematic approach to distinguishing goal priming from priming of non-goal

constructs. Personality and Social Psychology Review, 11(3), 211-233.

Gaertner, S. L., & Dovidio, J. F. (1977). The subtlety of white racism, arousal and helping

behavior. Journal of Personality and Social Psychology, 35, 691-707.

Gaertner, S. L., & Dovidio, J. F. (1986). The aversive form of racism. In J. F. D. S. L.

Gaertner (Ed.), Prejudice, Discrimination, and Racism. Orlando, FL: Academic

Press.

Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias: The Common Ingroup

Identity Model. Philadelphia, PA: Psychology Press.

Gailliot, M. T., & Baumeister, R. F. (2007). The physiology of willpower: Linking blood

glucose to self-control. Personality and Social Psychology Review, 11, 303-327.

Gauthier, I., Tarr, M. J., Anderson, A. W., Skudlarski, P., & Gore, J. C. (1999). Activation of

the middle fusiform 'face area' increases with expertise in recognizing novel objects.

Nature Neuroscience, 2, 568–573.

105

Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in

evaluation: An integrative review of implicit and explicit attitude change.

Psychological Bulletin, 132(5), 692-731.

Gergen, K. J. (1967). The significance of skin color in human relations. Daedulus, 96, 390-

406.

Golby, A. J., Gabrieli, J. D. E., Chiao, J. Y., & Eberhardt, J. L. (2001). Differential fusiform

responses to same- and other-race faces. Nature Neuroscience, 4, 845-850.

Goldstein, H. (1995). Multilevel Statistical Models (Second ed.). London: Arnold.

Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem,

and stereotypes. Psychological Review, 102(1), 4-27.

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual

differences in implicit cognition: The Implicit Association Test. Journal of

Personality and Social Psychology, 74, 1464-1480.

Gross, J. J. (1998). Antecedent- and response-focused emotion regulation: Divergent

consequences for experience, expression, and physiology. Journal of Personality and

Social Psychology, 74, 224-237.

Gross, J. J., & Thompson, R. A. (2007). Emotion regulation: Conceptual foundations. In J. J.

Gross (Ed.), Handbook of emotion regulation (pp. 3-24). New York: Guilford Press.

Hariri, A. R., Tessitore, A., Mattay, V. S., Fera, F., & Weinberger, D. R. (2002). The

amygdala response to emotional stimuli: A comparison of faces and scenes.

Neuroimage, 17, 317-323.

Harris, L. T., & Fiske, S. T. (2006). Dehumanizing the lowest of the low: Neuroimaging

responses to extreme out-groups. Psychological Science, 17, 847-853.

106

Harris, L. T., McClure, S. M., van den Bos, W., Cohen, J. D., & Fiske, S. T. (2007). Regions

of MPFC differentially tuned to social and non-social affective evaluation. Cognitive

and Behavioral Neuroscience, 7, 309-316.

Hart, A. J., Whalen, P. J., Shin, L. M., McInerney, S. C., Fischer, H., & Rauch, S. L. (2000).

Differential response in the human amygdala to racial outgroup versus ingroup face

stimuli. Neuroreport, 11, 2351-2355.

Heatherton, T. F., Macrae, C. N., & Kelley, W. M. (2004). What the social brain sciences can

tell us about the self. Current Directions in Psychological Science, 13(5), 190-193.

Hewstone, M., Hantzi, A., & Johnston, L. (1991). Social categorization and person memory:

The pervasiveness of race as an organizing principle. European Journal of Social

Psychology, 21, 517-528.

Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M. (2005). A meta-

analysis on the correlation between the implicit association test and explicit self-

report measures. Personality and Social Psychology Bulletin, 31(10), 1369-1385.

Ito, T. A., & Urland, G. R. (2003). Race and gender on the brain: Electrocortical measures of

attention to the race and gender of multiply categorizable individuals. Journal of

Personality and Social Psychology, 85(4), 616-626.

Johnson, M. K., & Hirst, W. (1993). MEM: Memory subsystems as processes. In S. E. G.

A.F. Collins, M.A. Conway, & P.E. Morris (Ed.), Theories of Memory (pp. 241-286).

East Sussex, England: Erlbaum.

Jones, J. M. (1997). Prejudice and racism. USA: McGraw Hill Companies, Inc.

Jost, J. T., & Banaji, M. R. (1994). The role of stereotyping in system-justification and the

production of false consciousness. British Journal of Social Psychology, 33, 1-27.

107

Kanwisher, N., McDermott, J., & Chun, M. (1997). The Fusiform Face Area: A module in

human extrastriate cortex specialized for the perception of faces. Journal of

Neuroscience, 17, 4302-4311.

Karpinski, A., & Steinman, R. B. (2006). The Single Category Implicit Association Test as a

Measure of Implicit Social Cognition. Journal of Personality and Social Psychology,

91(1), 16-32.

Krendl, A. C., Macrae, C. N., Kelley, W. M., & Heatherton, T. F. (2006). The good, the bad,

and the ugly: An fMRI Investigation of the functional anatomic correlates of stigma.

Social Neuroscience, 1, 5–15.

Kringelbach, M. L. (2005). The human orbitofrontal cortex: Linking reward to hedonic

experience. Nature Reviews Neuroscience, 6, 691-702.

Kringelbach, M. L., O'Doherty, J., Rolls, E. T., & Andrews, C. (2003). Activation of the

human orbitofrontal cortex to a liquid food stimulus is correlated with its subjective

pleasantness. Cerebral Cortex, 13, 1064-1071.

Kurzban, R., Tooby, J., & Cosmides, L. (2001). Can race be erased? Coalitional computation

and social categorization. Proceedings of the National Academy of Sciences, 98,

15387-15392.

LeDoux, J. E. (1996). The emotional brain: The mysterious underpinnings of emotional life.

New York; Toronto: Simon & Schuster.

LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience, 23, 155-

184.

Levin, D. T. (1996). Classifying faces by race: The structure of face categories. Journal of

Experimental Psychology: Learning, Memory, and Cognition, 22(6), 1364-1382.

108

Levin, D. T. (2000). Race as a visual feature: Using visual search and perceptual

discrimination tasks to understand face categories and the cross-race recognition

deficit. Journal of Experimental Psychology: General, 129, 559-574.

Levine, R. A., & Campbell, D. T. (1972). Ethnocentricism: theories of conflict, ethnic

attitudes and group behavior. New York: Wiley.

Lieberman, M. D., Hariri, A., Jarcho, J. M., Eisenberger, N. I., & Bookheimer, S. Y. (2005).

An fMRI investigation of race-related amygdala activity in African-American and

Caucasian-American individuals. Nature Neuroscience, 8, 720-722.

Liu, J., Harris, A., & Kanwisher, N. (2002). Stages of processing in face perception: An

MEG study. Nature Neuroscience, 5, 910-916.

Lowery, B. S., Hardin, C. D., & Sinclair, S. (2001). Social influence effects on automatic

racial prejudice. Journal of Personality & Social Psychology, 81(5), 842-855.

MacDonald, A. W., Cohen, J. D., Stenger, V. A., & Carter, C. S. (2000). Dissociating the

Role of the Dorsolateral Prefrontal and Anterior Cingulate Cortex in Cognitive

Control. Science, 288, 1835-1838.

MacDonald, G., & Leary, M. R. (2005). Why does social exclusion hurt? The relationship

between social and physical pain. Psychological Bulletin, 131(2), 202-223.

Macrae, C. N., Bodenhausen, G. V., & Milne, A. B. (1995). The dissection of selection in

person perception: Inhibitory processes in social stereotyping. Journal of Personality

& Social Psychology, 69(3), 397-407.

Macrae, C. N., Bodenhausen, G. V., Milne, A. B., & Jetten, J. (1994). Out of mind but back

in sight: Stereotypes on the rebound. Journal of Personality and Social Psychology,

67(5), 808-817.

109

Malpass, R. S., & Kravitz, J. (1969). Recognition for faces of own and other "race". Journal

of Personality and Social Psychology, 13, 330-334.

Markham, E. (1936). Outwitted. In The best loved poems of the American people (pp. 67):

Doubleday.

Marr, D. (1982). Vision. San Francisco, CA: W. H. Freeman.

McCaslin, M. J., & Petty, R. E. (2008). Is outgroup hate fundamental? Investigating

intergroup bias in the Minimal Group Paradigm. Paper presented at the Annual

conference of the Society of Personality and Social Psychology

McConahay, J. B. (1986). Modern racism, ambivelance, and the modern racism scale. In J. F.

Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination and racism (pp. 91- 126).

New York: Acedemic.

McConnell, A. R., & Leibold, J. M. (2001). Relations among the implicit association test,

discriminatory behavior, and explicit measures of racial attitudes. Journal of

Experimental Social Psychology, 37, 435-442.

Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigating the own-race bias in

memory for faces: A meta-analytic review. Psychology, Public Policy, and Law, 7, 3-

35.

Mitchell, J. P., Nosek, B. N., & Banaji, M. R. (2003). Contextual variations in implicit

evaluation. Journal of Experimental Psychology: General, 132, 455-469.

Monin, B. (2003). The Warm Glow Heuristic: When liking leads to familiarity. Journal of

Personality and Social Psychology, 85(6), 1035-1048.

Monteith, M. J., Sherman, J. W., & Devine, P. G. (1998). Suppression as a stereotype control

strategy. Personality & Social Psychology Review, 2(1), 63-82.

110

Mullen, B., Migdal, M. J., & Hewstone, M. (2001). Crossed categorization versus simple

categoization and intergroup evaluations: A meta-analysis. European Journal of

Social Psychology, 31(6), 721-736.

Mummendey, A., Simon, B., Dietze, C., Grunert, M., Haeger, G., Kessler, S., et al. (1992).

Categorization is not enough: Intergroup discrimination in negative outcome

allocation. Journal of Experimental Social Psychology, 28, 125-144.

Myrdal, G. (1944). An American dilemma: The negro problem and modern democracy. New

York: Harper & Bros.

Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of

inhibitionless spreading activation and limited-capacity attention. Journal of

Experimental Psychology: General, 106(3), 226-254.

Nezlek, J. B., & Cunningham, W. A. (October, 1998). Studying individual differences in the

activation of racial prejudice using the IAT and hierarchical linear modeling. Paper

presented at the Society of Experimental Social Psychology.

Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on

mental processes. Psychological Review, 84(3), 231-259.

Norton, M. I., Sommers, S. R., Apfelbaum, E. P., Pura, N., & Ariely, D. (2006). Color

Blindness and Interracial Interaction: Playing the Political Correctness Game.

Psychological Science, 17(11), 949-953.

Nosek, B. A., Banaji, M. R., & Greenwald, A. G. (2002). Harvesting intergroup attitudes and

stereotypes from a demonstration website. Group Dynamics, 6(1), 101-115.

Ochsner, K. N., & Lieberman, M. D. (2001). The emergence of social cognitive

neuroscience. American Psychologist, 56, 717-734.

111

Olson, M. A., & Fazio, R. H. (2003). Relations between implicit measures of prejudice: What

are we measuring? Psychological Science, 14(6), 636-639.

Olson, M. A., & Fazio, R. H. (2004). Reducing the Influence of Extrapersonal Associations

on the Implicit Association Test: Personalizing the IAT. Journal of Personality &

Social Psychology, 86(5), 653-667.

Ostrom, T. M., & Sedikides, C. (1992). Out-group homogeneity effects in natural and

minimal groups. Psychological Bulletin, 112, 536-552.

Otten, S., & Epstude, K. (2006). Overlapping Mental Representations of Self, Ingroup, and

Outgroup: Unraveling Self-Stereotyping and Self-Anchoring. Personality and Social

Psychology Bulletin, 32(7), 957-969.

Otten, S., & Wentura, D. (1999). About the impact of automaticity in the Minimal Group

Paradigm: Evidence from affective priming tasks. European Journal of Social

Psychology, 29(8), 1049-1071.

Palmeri, T. J., & Gauthier, I. (2004). Visual Object Understanding. Nature Reviews

Neuroscience, 5, 291-303.

Park, B., & Judd, C. M. (2005). Rethinking the link between categorization and prejudice

within the social cognition perspective. Personality and Social Psychology Review,

9(2), 108-130.

Park, B., & Rothbart, M. (1982). Perception of out-group homogeneity and levels of social

categorization: Memory for the subordinate attributes of in-group and out-group

members. Journal of Personality and Social Psychology, 42, 1051-1068.

112

Payne, B. K., Cheng, C. M., Govorun, O., & Stewart, B. D. (2005). An inkblot for attitudes:

Affect misattribution as implicit measurement. Journal of Personality and Social

Psychology, 89(3), 277-293.

Perdue, C. W., Dovidio, J. F., Gurtman, M. B., & Tyler, R. B. (1990). Us and them: Social

categorization and the process of intergroup bias. Journal of Personality and Social

Psychology, 59(3), 475-486.

Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejudice in western Europe.

European Journal of Social Psychology, 25(1), 57-75.

Petty, R. E., & Wegener, D. T. (1993). Flexible correction processes in social judgment:

Correcting for context-induced contrast. Journal of Experimental Social Psychology,

29, 137-165.

Phelps, E. A. (2006). Emotion and cognition: Insights from studies of the human amygdala.

Annual Review of Psychology, 24 (57), 27-53.

Phelps, E. A., O'Connor, K. J., Cunningham, W. A., Funayama, E. S., Gatenby, J. C., Gore,

J. C., et al. (2000). Performance on indirect measures of race evaluation predicts

amygdala activation. Journal of Cognitive Neuroscience, 12, 729-738.

Plant, E. A., & Devine, P. G. (1998). Internal and external motivation to respond without

prejudice. Journal of Personality and Social Psychology, 75(3), 811-832.

Rhodes, G., Byatt, G., Michie, P. T., & Puce, A. (2004). Is the fusiform face area specialized

for faces, individuation, or expert individuation? Journal of Cognitive Neuroscience,

16, 189-203.

Richeson, J. A., & Ambady, N. (2003). Effects of situational power on automatic racial

prejudice. Journal of Experimental Social Psychology, 39, 177-183.

113

Richeson, J. A., Baird, A. A., Gordon, H. L., Heatherton, T. F., Wyland, C. L., Trawalter, S.,

et al. (2003). An fMRI examination of the impact of interracial contact on executive

function. Nature Neuroscience, 6, 1323-1328.

Richeson, J. A., & Shelton, J. N. (2003). When prejudice does not pay: Effects of interracial

contact on executive function. Psychological Science, 14(3), 287-290.

Rilling, J. K., Gutman, D. A., Zeh, T. R., Pagnoni, G., Berns, G. S., & Kilts, C. D. (2002). A

neural basis for social cooperation. Neuron 35, 395-405.

Ronquillo, J., Denson, T. F., Lickel, B., Lu, Z.-L., Nandy, A., & Maddox, K. B. (2007). The

effects of skin tone on race-related amygdala activity: An fMRI investigation. Social

Cognitive and Affective Neuroscience, 2, 39-44.

Roskos-Ewoldsen, D. R., & Fazio, R. H. (1992). On the orienting value of attitudes: Attitude

accessibility as a determinant of an object's attraction of visual attention. Journal of

Personality and Social Psychology, 63(2), 198-211.

Rudman, L. A. (2004). Social justice in our minds, homes, and society: The nature, causes,

and consequences of implicit bias. Social Justice Research, 17, 129-142.

Rule, N. O., Ambady, N., Adams, R. B., Jr., & Macrae, C. N. (2007). Us and them: Memory

advantages in perceptually ambiguous groups. Psychonomic Bulletin & Review,, 14,

687-692.

Ruscher, J. B., Fiske, S. T., Miki, H., & Van Manen, S. (1991). Individuating processes in

competition: Interpersonal versus intergroup. Personality and Social Psychology

Bulletin, 17, 595-605.

Russell, J. A. (1979). Affective space is bipolar. Journal of Personality and Social

Psychology, 37, 345-356.

114

Seeleman, V. (1940). The influence of attitude upon the remembering of pictorial material.

Archives of Psychology, 36(258), 6-69.

Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchy

and oppression. New York: Cambridge University Press.

Simmel, G., & Wolf, K. H. (1950). The Sociology of Georg Simmel. Glencoe, Ill: Free Press.

Singer, J. D. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models,

and individual growth models. Journal of Educational and Behavioral Statistics,

23(4), 323-355.

Sporer, S. L. (2001). Recognizing faces of other ethnic groups: An integration of theories.

Psychology, Public Policy, and Law, 7, 36-97.

Staats, A. M., & Staats, C. K. (1958). Attitudes established by classical conditioning. Journal

of Abnormal and Social Psychology, 57(1), 37-40.

Stangor, C., Lynch, L., Duan, C., & Glass, B. (1992). Categorization of individuals on the

basis of multiple social features. Journal of Personality and Social Psychology, 62,

207-281.

Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory of prejudice. In S.

Oskamp (Ed.), Reducing prejuice and discrimination (pp. 23-46). Hillsdale, N.J.:

Erlbaum.

Strack, F., & Deutsch, R. (2004). Reflective and Impulsive Determinants of Social Behavior.

Personality & Social Psychology Review, 8(3), 220-247.

Tajfel, H. (1970). Experiments in intergroup discrimination. Scientific American, 223, 96-

102.

115

Tajfel, H. (1972). La categorisation sociale (English trans.). In S. Moscovici (Ed.),

Introduction a lapsychologie sociale. Paris: Larouse.

Tajfel, H. (1982). Social identity and intergroup behavior. Cambridge, England: Cambridge

University Press.

Tajfel, H., Billig, M., Bundy, R., & Flament, C. (1971). Social categorization and intergroup

behaviour. European Journal of Social Psychology, 1, 149-178.

Taylor, S. E., Fiske, S. T., Etcoff, N. L., & Ruderman, A. (1978). Categorical and contextual

bases of person memory and stereotyping. Journal of Personality and Social

Psychology, 36, 778-793.

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987).

Rediscovering the social group: A Self-Categorization Theory. Oxford: Basil

Blackwell.

Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and collective:

Cognition and social context. Personality and Social Psychology Bulletin, 20(5), 454-

463.

Turner, M. A., Freiberg, F., Godfrey, E. B., Herbig, C., Levy, D., & Smith, R. E. (2002). A

Paired Testing Study of Mortgage Lending Institutions: Washington, DC: U.S.

Department of Housing and Urban Development, April.

Urban, L. M., & Miller, N. (1998). A theoretical analysis of crossed categorization effects: A

meta-analysis. Journal of Personality and Social Psychology, 74(4), 894-908.

Valentine, T., & Endo, M. (1992). Towards an exemplar model of face processing: The

effects of race and distinctiveness. Quarterly Journal of Experimental Psychology:

Human Experimental Psychology, 44A, 671-703.

116

Van Bavel, J. J., & Cunningham, W. A. (2008). Self-categorization and social identity elicit

ingroup biases in attention and memory. The Ohio State University.

van den Bos, W., McClure, S. M., Harris, L. T., Fiske, S. T., & Cohen, J. D. (2007).

Dissociating affective evaluation and social cognitive processes in ventral medial

prefrontal cortex. Cognitive and Behavioral Neuroscience, 7, 337-346.

Vanbeselaere, N. (1991). The different effects of simple and crossed categorizations: A result

of the category differentiation process or of differential category salience? In W.

Stroebe & M. Hewstone (Eds.), European Review of Social Psychology (Vol. 2, pp.

247-278). Chichester: John Wiley.

Vescio, T. K., Judd, C. M., & Kwan, V. S. Y. (2004). The crossed categorization hypothesis:

Evidence of reductions in the strength of categorization but not intergroup

categorization but not intergroup bias. Journal of Experimental Social Psychology,

40, 478-496.

Vuilleumier, P. (2005). How brains beware: Neural mechanisms of emotional attention.

Trends in Cognitive Sciences, 9, 585-594.

Walther, E., & Trasselli, C. (2003 ). I like her, because I like myself: self-evaluation as a

source of interpersonal attitudes. Experimental Psychology, 50 (4), 239-246.

Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review, 101(1), 34-

52.

Whalen, P. J. (1998). Fear, vigilance and ambiguity: Initial neuroimaging studies of the

human amygdala. Current Directions in Psychological Science, 7, 177-188.

117

Whalen, P. J., Rauch, S. L., Etcoff, N. L., McInerney, S. C., Lee, M., & Jenike, M. A. (1998).

Masked presentations of emotional facial expressions modulate amygdala activity

without explicit knowledge. Journal of Neuroscience, 18, 411-418.

Wheeler, M. E., & Fiske, S. T. (2005). Controlling racial prejudice: Social-cognitive goals

affect amygdala and stereotype activation. Psychological Science, 16(1), 56-63.

Wilson, D. S., & Sober, E. (1994). Reintroducing group selection to the human behavioral

sciences. Behavioral and Brain Sciences, 17, 585-654.

Wilson, E. O. (1998). Consilience: The unity of knowledge. New York: Knopf.

Wilson, T. D., Samuel, L., & Schooler, T. Y. (2000). A model of dual attitudes.

Psychological Review, 107, 101-126.

Wojciulik, E., Kanwisher, N., & Driver, J. (1998). Modulation of activity in the fusiform face

area by covert attention: An fMRI study. Journal of Neurophysiology, 79, 1574-1579.

Wyer, N. A., Sherman, J. W., & Stroessner, S. J. (2000). The roles of motivation and ability

in controlling the consequences of stereotype suppression. Personality and Social

Psychology Bulletin, 26, 13-25.

Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social

Psychology Monographs, 9(2, Pt. 2).

118

Footnotes

1For the purposes of this dissertation, we focus on a specific form of prejudice, namely

evaluations that favor ingroups over other social groups (or their members) that stem from

biased beliefs and/or feelings and lead to discrimination (Allport, 1954). Although we

acknowledge that prejudice can take many forms, including positive evaluations of certain

outgroups, we focus on prejudices that favor ingroups compared to outgroups because they

are more common and detrimental to social justice. Thus although there are certainly

differences between different prejudices, this research assumes that prejudice based on race,

gender, religion, etc. share a common set of underlying psychological processes. The

explication of these processes with one form of prejudice should inform research and theory

on others.

2It was perhaps this observation that led novelist E. B. White to quip “Prejudice is a

great time saver. You can form opinions without having to get the facts.”

3Sufficiently justified prejudice – such as that toward terrorists or enemy soldiers –

elicits discrimination (and even violence) without guilt, shame or regret.

4However, suppression may be successful when perceivers have both the motivation

and ability (Monteith et al., 1998; Wyer, Sherman, & Stroessner, 2000), although it may have

negative effects on subsequent tasks that require controlled processing (Richeson & Shelton,

2003).

5To enhance group processing participants saw faces and read the teams were

competing, unlike the classic MGP.

6No effects were moderated by participants’ gender so it is not discussed further

7There was no main effect of team (Leopards/Tigers) so it is not discussed further.

119

8Rating/memory data for one participant were lost due to computer error.

9Although participants were more accurate at recalling the team membership of White

than Black faces, F(1,20)= 5.43, p = .03, there were no differences between novel ingroup

and outgroup faces, regardless of race (ps > .15).

10Replicating previous research, ingroup bias (ingroup–outgroup) in self-reported

liking was not correlated with amygdala activity (Phelps et al., 2000), or other regions

showing ingroup bias. Among regions showing ingroup bias, the left fusiform was correlated

with the right fusiform (r = .62, p < .01) and putamen (r = .59, p = .01).

11We extracted ROIs from each showing ingroup bias and none were more active to

Black or White faces. A posteriori ROIs were based on voxels from 5mm spheres centered

on the maxima of significant regions.

12Unlike the classic MGP, participants learned the faces and read that the teams were

competing, to develop familiarity with faces for the subsequent tasks and enhance self-

categorization, respectively.

13The reported effects were not moderated by participants’ gender and this variable is

not discussed further.

14Following previous research (Cunningham et al., 2001; Draine & Greenwald, 1998),

response accuracy was used as the dependent measure during response-window priming.

Eligible responses were limited to the first 600 ms because longer response times allow for

more controlled processing (Neely, 1977).

15We could not examine the 2 (condition: control, experimental) × 2 (group: ingroup,

outgroup) × 2 (race: Black, White) × 2 (valence: positive, negative) repeated-measures

120

analysis on response accuracy since participants in the control condition did not have ingroup

or outgroup members.

16Simple effects analyses also revealed that participants had relatively positive

automatic evaluations (i.e., higher accuracy to positive than negative words) for White-

ingroup faces (t(71) = 2.24, p < .03), White-outgroup faces (t(71) = 1.96, p = .05), and Black-

ingroup faces (t(71) = 3.34, p = .001), but were relatively neutral to Black-outgroup members

(t(71) = -0.98, p = .33).

17Because participants in Experiment 1 always pressed “1” when a good word

appeared and “2” when a bad word appeared, right-hand biases in information processing

(Nisbett & Wilson, 1977) could have created the direction of these preferences. We

counterbalanced response options during the priming task in Experiment 2.

18Replicating Experiment 1, a simple effects analysis confirmed that participants also

had more positive automatic evaluations of Black-ingroup than Black-outgroup faces, t(80) =

1.68, p < .05 (one-tailed), indicating that self-categorization improved automatic evaluations

of Black faces. Unlike Experiment 1, however, control participants did not reveal a Race ×

Target Valence interaction on accuracy, F(1, 40) = 1.20, p = .28. In addition, replicating

Experiment 1, they revealed no effect of Race on liking, t(40) = -0.97, p = .34.

19I also acknowledge that liking breeds a feeling of familiarity (Monin, 2003).

20To assess the links between intergroup memory and evaluation we added a face

memory task to Experiment 3. Because this was not part of the original experimental design,

it was added after 11 participants had already completed the experiment.

21Control participants completed the same memory task. The only difference is that

they did not belong to either group.

121

22I used a one-tailed test because of strong a priori hypotheses about the direction of

this effect. Although the own-race bias is a robust phenomenon, it may have been slightly

attenuated by the mixed-race context.

23There was a memory task in Experiment 1 that examined memory for the two groups

(Leopards and Tigers). However, the task only included two groups. Therefore, participants

who memorized all the members of a single group could perform perfectly. This made it an

insensitive measure of memory for ingroup and outgroup members.

24There was no automatic evaluation measure in Experiment 1.

25I did not analyze unaffiliated faces because both tasks could be completed

successfully without responding to unaffiliated faces.

26One ingroup and two outgroup SC-IAT scores were lost due to computer errors.

27This effect was using the anatomical ROI of the right amygdala.

28Importantly, the higher-order interactions observed in experiments 2, 3 and 5 –

especially on the measures of automatic evaluation – showing racial bias toward outgroup

members and unaffiliated faces but not ingroup members mitigates against the possibility that

the overall pattern of effects are due to general self-consciousness about race. Not only are

deliberate inhibitory processes unlikely to affect evaluations in such a selective fashion, but

they are especially unlikely to do so on the response-window priming task we employed in

Experiments 2 and 3.

29The example of a basketball team raises an interesting issue about the content of the

social context. A context associated with racial stereotypes, such as basketball, may itself

alter the salience of certain social identities, potentially increasing (rather than reducing) the

use of pre-existing stereotypes and attitudes.

122

Table 1. Descriptive statistics for in-scanner ratings. Means and standard deviations (SD) are

provided for accuracy and reaction times of responses to Black-ingroup, Black-outgroup,

White-ingroup, and White-outgroup. Accuracy = the proportion of trails with correct

response during the two-second face presentation; Reaction Time = time in ms between the

presentation of the face and the response. Excludes all trials where the reaction time ≤ 300

ms (Experiment 1).

Accuracy and Reaction Time During fMRI Categorization Tasks

Accuracy

Reaction Time

Task Mean SD Mean

SD

Implicit (Race) Task

White-Ingroup .96 .19 896 298

White-Outgroup .95 .22 902 310

Black-Ingroup .94 .24 862 282

Black-Outgroup

.96

.19

849

264

Explicit (Group) Task

White-Ingroup .86 .35 1210 329

White-Outgroup .79 .41 1250 322

Black-Ingroup .79 .41 1233 319

Black-Outgroup

.79

.41

1270

316

123

Table 2. Brain activity as a function of group membership (ingroup vs. outgroup) and race

(Black vs. White) of the faces. Full brain analyses (p ≤ .001) and region of interest analyses

(p ≤ .01) are based on threshold activity in 10 or more contiguous voxels. Brodmann Areas

(BA) and MNI coordinates (x, y, z) of activation are provided (Experiment 1).

Areas of Statistically Significant Blood-Oxygen-Level-Dependent (BOLD) Activation

Brain Region BA Side Voxels t-value x y z

Novel ingroup > outgroup

Amygdala 34 L 10 3.53 -24 3 -15

Orbitofrontal Cortex 47 L 16 5.18 -30 39 -6

Fusiform 37 R 61 5.50 39 -48 -15

Fusiform 37 L 46 4.96 -36 -42 -24

Putamen 48 R 32 5.65 27 -6 15

Inferior Temporal Cortex 37 L 20 4.88 -54 -57 -6

Novel ingroup < outgroup

No regions were significant

White > Black

No regions were significant

Black > White

Angular gyrus 39 L 18 5.34 -36 -54 30

Inferior Occipital Cortex 19 L 85 5.38 -48 -78 -6

Inferior Occipital Cortex 19 R 43 5.10 42 -75 -6

Inferior Occipital Cortex 18 R 10 5.15 33 -93 -9

Inferior Occipital Cortex 18 L 31 5.04 -30 -96 -9

Brain regions with statistically significant activation. All parameters are based on cluster

maxima

124

Table 3. Brain activity as a function the interaction between group membership (ingroup vs.

outgroup) and race (Black vs. White) of the faces. Full brain analyses (p ≤ .001) and region

of interest analyses (p ≤ .01) are based on threshold activity in 10 or more contiguous voxels.

Brodmann Areas (BA) and MNI coordinates (x, y, z) of activation are provided (Experiment

1).

Areas of Statistically Significant Blood-Oxygen-Level-Dependent (BOLD) Activation

Brain Region Side Voxels t-value x y z

congruent (wi + bo) > incongruent (wo + bi)

Rolandic Oper R 44 6.52 42 -36 21

Lingual Gyrus R 20 5.08 21 -93 -6

Paracentral Lobule L 26 4.63 -12 -27 72

Anterior Cingulate L 13 4.63 0 33 6

incongruent (wo + bi) > congruent (wi + bo)

Temporal Inferior R 21 4.81 45 -48 -6

Temporal Inferior R 31 4.53 45 -48 -24

Supramarginal R 14 4.07 45 -39 39

Brain regions with statistically significant activation. All parameters are based on cluster

maxima.

125

Table 4. Brain activity as a function of race (Black vs. White) of the unaffiliated control

faces. Full brain analyses (p ≤ .001) and region of interest analyses (p ≤ .05) are based on

threshold activity in 10 or more contiguous voxels. Brodmann Areas (BA) and MNI

coordinates (x, y, z) of activation are provided (Experiment 5).

Areas of Statistically Significant Blood-Oxygen-Level-Dependent (BOLD) Activation

Brain Region BA Side Voxels t-value x y z

White > Black

Middle Temporal Lobe 37 R 33 4.01 45 -57 -3

Fusiform Gyrus 37 R 23 3.83 30 -60 -9

Superior Parietal Cortex 5 R 50 3.83 15 -54 72

Precuneus 5 L 32 3.65 -15 -60 66

Black > White

Middle Occipital Cortex 18 L 202 5.70 -27 -96 -3

Middle Occipital Cortex 18 R 96 4.25 30 -99 3

Superior Frontal Cortex 11 R 11 3.72 18 51 3

Anterior Cingulate Cortex 32 R 15 3.70 12 39 30

Middle Cingulate Cortex 23 L 22 3.45 -3 -12 33

Amygdala 34 R NA 4.06 Anatomical

Brain regions with statistically significant activation. All parameters are based on cluster

maxima.

126

Figure 1. The explicit (group) and implicit (race) categorization tasks during fMRI. There

were two explicit and two implicit categorization blocks in each of six runs. Each block

started with a directions screen followed by twelve randomly presented faces. Faces

presented within each block were separated by fixation crosses. After the completion of each

block, directions for the next block appeared (Experiment 1).

Note new labels

+

Note new labels

+

+

+

Stimulus: 2s

Directions: 6s

Fixation: 2s + Jitter

Explicit (Group) Task Implicit (Race) Task

127

Figure 2. The effect of Group membership (Ingroup, Outgroup) on self-reported liking on a

6-point scale (1 = dislike to 6 = like). Ratings were centered on the scale midpoint (3.5) such

that more positive scores represent liking and more negative scores represent disliking. Error

bars show standard errors (Experiment 1).

-1

-0.5

0

0.5

1

Ingroup Outgroup

Lik

ing

128

Figure 3. Maps of brain activity stronger to ingroup than outgroup faces in the (A) fusiform

(coronal view; y = -48) and (B) amygdala (coronal view; y = 0) (Experiment 1).

129

Figure 4. (A) Map of brain activity stronger to ingroup than outgroup faces in the OFC

(sagittal view; x = -24) and (B) the correlation between OFC (ingroup-outgroup) and mean

ingroup bias (ingroup-outgroup) on self-reported liking (Experiment 1).

-0.1

0

0.1

0.2

0.3

-2 -1 0 1 2

Self-reported Liking (ingroup - outgroup)

OF

C A

ctiv

ity

(in

gro

up

-ou

tgro

up

) i

130

Figure 5. Individual differences in OFC activity (ingroup - outgroup) mediate the effect of

group membership on individual differences in self-reported ingroup bias (ingroup -

outgroup) (Experiment 1).

Group on OFC t = 5.30, b = 0.109 (SE = 0.021)

OFC on liking t = 2.45, b = 4.788 (SE = 1.956)

Group on liking t = 2.12, b = 0.389 (SE = 0.184)

Group on liking (controlling for OFC) t = -0.49, b = -.131 (SE = 0.267), p = 63.

Sobel Test : t = 2.22, p < .03.

b = 0.109, p < .001 b = 4.788, p < .03

Ingroup -

Outgroup

OFC

Liking

b = 0.389, p = .05

b = -0.131, p = .63

131

Figure 6. The effect of prime Race (Black, White) and Group membership (Ingroup,

Outgroup) on response accuracy to positive and negative words (0.5 = chance responding).

Higher scores represent greater accuracy and lower scores represent less accuracy. (A)

Outgroup members show the standard pattern of racial bias, whereas (B) Ingroup members

are evaluated positively, regardless of race. Error bars show standard errors (Experiment 2).

0.7

0.75

0.8

0.85

0.9

Black WhitePrime

Pro

po

rtio

n c

orr

ect

Positive

Negative

0.7

0.75

0.8

0.85

0.9

Black WhitePrime

Pro

po

rtio

n c

orr

ect

Positive

Negative

132

Figure 7. The effect of Race (Black, White) and Group membership (Ingroup, Outgroup) on

self-reported liking on a 6-point scale (1 = dislike to 6 = like). Error bars show standard

errors (Experiment 2).

3

3.2

3.4

3.6

3.8

Outgroup Ingroup

Rep

ort

ed L

ikin

g (

1-6

)

Black

White

133

Figure 8. The effect of prime Race (Black, White) and Group membership (Ingroup,

Outgroup, Unaffiliated) on response accuracy to positive and negative words (0.5 = chance

responding). (A) Outgroup members and (B) unaffiliated faces show the standard pattern of

racial bias, while (C) Ingroup members are evaluated positively, regardless of race. Error

bars show standard errors (Experiment 3).

0.7

0.75

0.8

0.85

Black White

Outgroup

Pro

port

ion

corr

ect

Positive

Negative

0.7

0.75

0.8

0.85

Black White

Ingroup

Positive

Negative

0.7

0.75

0.8

0.85

Black White

Unaffiliated

Pro

po

rtio

n c

orr

ect

Positive

Negative

134

Figure 9. The effect of Race (Black, White) and Group membership (Ingroup, Outgroup,

Unaffiliated) on self-reported liking on a 6-point scale (1 = dislike to 6 = like). Error bars

show standard errors (Experiment 3).

3

3.2

3.4

3.6

3.8

Outgroup Ingroup Control

Rep

ort

ed L

ikin

g (

1-6

)

Black

White

135

Figure 10. The effect of Race (Black, White) on accuracy to faces during a memory task.

Higher scores represent greater accuracy and lower scores represent less accuracy (0.33 =

chance responding). Error bars show standard errors (Experiment 4).

0.5

0.6

0.7

0.8

Black White

Pro

port

ion

Corr

ect

136

Figure 11. The effect of Group membership (Ingroup, Outgroup, Unaffiliated) on accuracy

to faces during a memory task. Higher scores represent greater accuracy and lower scores

represent less accuracy (0.33 = chance responding). Error bars show standard errors

(Experiment 4).

0.5

0.6

0.7

0.8

Outgroup Ingroup

Pro

po

rtio

n C

orr

ect

137

Figure 12. The effect of Group membership (Ingroup, Outgroup) on self-reported liking on a

6-point scale (1 = dislike to 6 = like) for faces that were correctly and incorrectly identified

during the memory task. Error bars show standard errors (Experiment 4).

3

3.2

3.4

3.6

3.8

Correct Incorrect

Rep

ort

ed L

ikin

g (

1-6

)

Ingroup

Outgroup

138

Figure 13. The effect of Race (Black, White) and Group membership (Ingroup, Outgroup,

Unaffiliated) on accuracy to faces during a memory task. Higher scores represent greater

accuracy and lower scores represent less accuracy (0.33 = chance responding). Error bars

show standard errors (Experiment 5).

139

Figure 14. The effect of Race (Black, White) and Group membership (Ingroup, Outgroup,

Unaffiliated) on self-reported liking on a 6-point scale (1 = dislike to 6 = like). Error bars

show standard errors (Experiment 5).

3

3.2

3.4

3.6

3.8

4

4.2

Outgroup Ingroup Unaffiliated

Rep

ort

ed L

ikin

g (

1-6

)

Black

White

140

Figure 15. Mean reaction times (ms) to Ingroup and Outgroup faces paired with positive and

negative words. Lower scores represent stronger associations between group membership

(Ingroup, Outgroup) and valence stimuli (i.e., between ingroup faces and positive words).

Error bars show standard errors (Experiment 5).

1000

1100

1200

1300

1400

Outgroup Ingroup

Rea

ctio

n T

ime

(ms)

Positive

Negative

141

Figure 16. (A) Map of brain activity stronger to Ingroup than Outgroup faces (axial view; x

= -17), fusiform gyrus is region in yellow in the bottom-right corner and (B) the correlation

between fusiform activity (ingroup-outgroup) and own-group memory bias (ingroup-

outgroup) (Experiment 5).

-1

-0.5

0

0.5

1

1.5

-0.4 -0.2 0 0.2 0.4 0.6

Own-group Memory Bias (ingroup-outgroup)

Fu

sifo

rm

Acti

vit

y (

ing

ro

up

-ou

tgro

up

)

142

Appendix A

Core Concepts and Definitions

These definitions are specific to the use of these concepts in the current dissertation and are

not necessarily intended for a more general application. These definitions are intended to

represent the central, unique aspects of each concept notwithstanding obvious conceptual and

empirical overlap or interactions with the other concepts.

Automatic: Processes that do not require awareness, intention, or control (Bargh, 1994) and

often occur rapidly. In the current research, the primary distinction between automatic and

controlled measures of evaluation is that the former are implicit (i.e., they do not ask

participants to report evaluations). However, the automatic measure used in Experiments 2

and 3 only included responses during the first 600 ms following presentation of a stimulus to

capture the speed associated with automaticity.

Attitude: Relatively stable representations stored in memory the form a psychological

tendency that is expressed through an evaluation with some degree of valence (Eagly &

Chaiken, 1993).

Evaluation: A current affective judgment based constructed from relatively stable attitude

representations (a subset of which are active at any given time) stored in memory

(Cunningham et al., 2007).

Self-categorization: Psychological connections between the self and some class of stimuli

that are different from other classes of stimuli. Self-categorization can include personal (i.e.,

define the individual as unique from others) and social identity (i.e., define the individual in

143

terms of similar characteristics with a social group and different characteristics from other

social groups)(J. C. Turner et al., 1994).

Group: Two or more individuals who share a common social identity and/or perceive

themselves to be members of the same social category.

Social Identity: An individual’s knowledge they belong to a certain group, the significance

of this group, their relationship to the group and its members, and the associations they have

with the group and its members (Tajfel, 1972).

Prejudice: This includes all forms intergroup bias (see below) except automatic bias

(Devine, Plant, Amodio, Harmon-Jones, & Vance, 2002).

Intergroup bias: A differential evaluation or judgment of two or more groups. This usually

includes the connotation that one group is preferred or superior to others on some dimension.

For example, racial bias can be manifested as an evaluative preference for one race over

another (e.g., White > Black).

Ingroup bias: Positive evaluations toward ingroup members (does not imply negative

evaluations toward outgroup members).

Outgroup derogation: Negative evaluations toward outgroup members (does not imply

positive evaluations toward ingroup members).

Motivation Processing associated with the pursuit or attainment of a goal (e.g., the need to

belong). Goals involve value (proportional to the product of expectancy and value), post-

attainment decrements in motivation, and are moderated by equifinality and multifinality

(Forster, Liberman, & Friedman, 2007).

Cognition: A set of basic or processes or computational operations on information (Johnson

& Hirst, 1993). This can include extracting or imbuing information with affective features.

144

Affect: Features of information, including valence (positivity and negativity) and intensity or

arousal (Russell, 1979).