Recognizing faces across continents: The effect of within-race variations on the own-race bias in face recognition

Abstract

People are better at recognizing faces of their own race than faces of other racial groups. This own-race bias (ORB) in face recognition manifests in some studies as a full crossover interaction between race of observer and race of face, but in others the interaction is accompanied by main effects or other complexities. We hypothesized that this may be due in part to unacknowledged within-race variation and the implicit assumption that the terms white and black describe perceptually homogeneous race categories. We therefore tested white and black South Africans on their recognition of black and white American faces and black and white South African faces. Our results showed the expected interaction, but only for South African faces. This finding supports explanations of the ORB that are premised on intergroup contact and perceptual experience and highlights the danger of assuming homogeneity of appearance within groups.

Full text

Previous studies have shown that people recognize
faces of their own race more accurately than they recog-
nize faces of other races, a finding that has become known
as the cross-race effect, or the own-race bias (Malpass &
Kravitz, 1969; see Meissner & Brigham, 2001, for a meta-
analytic review). A number of theoretical explanations for
this effect have been proposed (see Sporer, 2001, for a
review). One of the most widely accepted explanations
for this effect is that poorer recognition of other-race faces
may be rooted in the observer’s perceptual learning and
the amount of contact that he or she has had with people
of other races. An alternative explanation, proposed by
Levin (1996) and MacLin and Malpass (2001), suggests
that racial categorization occurs automatically and early
in the perceptual encounter with faces of another race,
taking attention away from individuating characteristics
of the face.
An own-race recognition bias typically manifests as a
disordinal (full crossover) interaction between race of ob-
server and race of face, so that observers from each group
show superior recognition performance on own-race over
other-race faces (although see Sporer, 2001). Most stud-
ies have reported a significant interaction between race of
observer and race of face that is accompanied by a signifi-
cant main effect either of race of participant or of race of
face (see, e.g., Chiroro & Valentine, 1995; Ng & Lindsay,
1994; Sporer, 1999). For example, Sporer (1999) found
that Turkish participants did not differ in their ability to
recognize Turkish and German faces, whereas German
participants recognized German faces better than they did
Turkish faces. Similarly, Walker and Hewstone (2006)
found an own-race bias among white U.K. residents in their
recognition of South Asian faces, but not for Asian people
living in the U.K. A reversal of the own-race bias effect
has also been reported in the literature, where participants
appeared to recognize other-race faces better than they did
own-race faces. For example, Wright, Boyd, and Tredoux
(2003) found that whereas white South African students
showed the expected own-race recognition advantage for
white faces, black South African students showed a sur-
prising recognition advantage for white faces.
Thus, although the cross-race effect has been con-
firmed in several individual studies, as well as in meta-
analyses (e.g., Antony, Copper, & Mullen, 1992; Both-
1089 Copyright 2008 Psychonomic Society, Inc.
Recognizing faces across continents:
The effect of within-race variations on
the own-race bias in face recognition
PATRICK M. CHIRORO
University of Pretoria, Pretoria, South Africa
COLIN G. TREDOUX
University of Cape Town, Cape Town, South Africa
STEPHANO RADAELLI
University of Pretoria, Pretoria, South Africa
AND
CHRISTIAN A. MEISSNER
University of Texas, El Paso, Texas
People are better at recognizing faces of their own race than faces of other racial groups. This own-race bias
(ORB) in face recognition manifests in some studies as a full crossover interaction between race of observer and
race of face, but in others the interaction is accompanied by main effects or other complexities. We hypothesized
that this may be due in part to unacknowledged within-race variation and the implicit assumption that the terms
white and black describe perceptually homogeneous race categories. We therefore tested white and black South
Africans on their recognition of black and white American faces and black and white South African faces. Our
results showed the expected interaction, but only for South African faces. This finding supports explanations of
the ORB that are premised on intergroup contact and perceptual experience and highlights the danger of assum-
ing homogeneity of appearance within groups.
Psychonomic Bulletin & Review
2008, 15 (6), 1089-1092
doi:10.3758/PBR.15.6.1089
P. Chiroro, patrick.chiroro@gmail.com

1090 CHIRORO, TREDOUX, RADAELLI, AND MEISSNER
responding to poster invitations. Twenty-four were black South
Africans (M  21.05 years, SD  1.56) and 24 were white South
Africans (M  20.34 years, SD  1.67). An equal number of male
and female participants were included in each racial group. Each
participant received R20 (approximately $3) for his or her time.
Apparatus and Stimuli
Four sets of photographs were used: 24 black and 24 white U.S.
faces, and 24 black and 24 white South African faces. The South
African faces were collected by the second author, and the U.S. faces
by the fourth author. The four sets of faces were selected from larger
sets of stimuli on the basis of their (approximately equal) memora-
bility ratings, as provided by 16 independent raters of white South
African descent. (It should be noted that prior studies have demon-
strated strong correlations in interracial perceptions of memorabil-
ity; see, e.g., Meissner, Brigham, & Butz, 2005.) All the faces were
digitally removed from their backgrounds, using RealDraw Pro 3.1,
and were equated for light contrast, using GIMP 2.0. Two different
frontal views of each face in neutral pose were generated by apply-
ing a lighting and color transformation, one for use during study and
the other for use at test. SuperLab Pro v2.0 was used to control expo-
sure duration and interstimulus interval. The participants’ responses
were captured via a Cedrus RB560 response box. All the stimuli
were presented on a 15-in. color monitor connected to a Pentium 4
computer, with resolution set at 1,024  768 pixels. Color resolution
was set at 24 bits.
Design
The study utilized a 2  2  2 mixed design in which race of
participant (black vs. white) varied between participants and race of
face (black vs. white) and country of stimulus origin (South Africa
vs. United States) were manipulated within participants. Each face
served as a target and as a distractor in each of the conditions an
equal number of times.
Procedure
The participants were tested individually in a quiet room. During
study, each face was shown for 3 sec, with an interstimulus interval
of 1.5 sec. The participants were requested to remember the faces
presented to them. At test, the yes–no recognition procedure was
used. During study and at test, black and white faces were presented
randomly (i.e., mixed study and test lists were used). At test, presen-
tation of faces was response terminated; the participants responded
by pressing one of two keys on the Cedrus RB560 response box. The
keys were clearly marked “yes” (for faces seen previously) or “no”
(for new faces). Hits and false positives made by each participant on
each category of faces were recorded. The participants were thanked
and debriefed at the end of the experiment.
RESULTS
Hits and false positives were combined to form signal
detection measures of discrimination accuracy (d) and
response criterion (c), and a three-way split-plot factorial
ANOVA was performed on these measures. Of particular
interest in this study was the three-way interaction involv-
ing country of origin of faces, race of face, and race of
participant. This interaction was statistically significant
for d [F(1,46)  5.24, p  .026, r
2
 .10], but not for c
[F(1,46)  1.98, p  .17, r
2
 .04].
Figure 1 shows the mean d scores and suggests that the
two-way interaction of race of participant and race of face
was significant for the participants when they were tested
with South African faces, but not when they were tested
with U.S. faces. This was confirmed with follow-up tests
of each of the simple interaction effects [F(1,46)  21.37,
well, Brigham, & Malpass; 1989; Meissner & Brigham,
2001; Shapiro & Penrod, 1986), the results have varied
widely between studies. Some of this variation may be ac-
counted for by differences in the way the studies have been
conducted. For example, the meta-analysis by Meissner
and Brigham showed that the own-race bias is moderated
by type and nature of stimulus material used, encoding
time, and delay between study and test. These factors may
explain some of the differences between findings across
studies.
Ecological factors, such as variations in population
distribution and patterns of contact, could explain an ad-
ditional amount of the variance and may, in particular,
explain why interactions are often accompanied by main
effects. In habitats where minority groups are sparse, for
instance, it is feasible to expect that minority groups will
be provided far greater opportunity for learning cross-race
faces than will members of the majority, and that they will
show less recognition bias.
However, we suspect that an additional, largely unac-
knowledged factor—what might be called within-race
face variation—may explain a further amount of the ef-
fect variance.
In particular, the race categories white and black have
been assumed in the literature to be perceptually homo-
geneous, but this may be far from the truth. Faces that
are categorized in the U.S. as black may differ in many
ways from those similarly categorized as black in Africa
or South America, and this may be difficult to discern
without a “trained eye”—particularly difficult, we dare-
say, for cross-race recognition researchers, who tend to
be white Americans and Europeans. Even within each
continent, huge variations appear to exist in the facial
and bodily physiognomy of racial groups. For example,
the Maasai of Kenya look very different from the Zulu
people of South Africa, and both of those ethnic groups
share few similarities with typical black Americans. For
the Maasai or Zulu people, black U.S. faces may indeed be
unfamiliar and constitute an out-group stimulus set. From
a theoretical point of view, these within-race variations in
physiognomy are likely to be mapped onto differences in
intergroup contact, and this should show in face recogni-
tion performance.
Thus, a primary aim of the present experiment was to
test whether people of a common race will exhibit an own-
race bias that is independent of geographic region. To as-
sess this, white and black South Africans were presented
with black and white faces from both South Africa and
the United States. We anticipated that the black and white
South Africans would demonstrate an in-group face rec-
ognition advantage that is geographically specific. Thus,
the black South Africans were expected to show superior
recognition of black South African faces over all other
comparison groups.
METHOD
Participants
Forty-eight undergraduate students at a large, multiracial uni-
versity in South Africa volunteered to participate in the study by

IN-GROUP FACE RECOGNITION ADVANTAGE 1091
stimuli that has been used in several studies demonstrat-
ing an own-group recognition bias with U.S. participants
(e.g., Meissner et al., 2005). Given that this exact set of
study/test stimuli has not been used previously, however,
we concede that a replication of the present experiment
with U.S. participants would appear warranted.
Nevertheless, this demonstration of a regional basis for
the recognition advantage for own-race faces supports ex-
planations that are premised on intergroup contact and
perceptual experience (Brigham & Malpass, 1985; Meiss-
ner & Brigham, 2001). Populations develop perceptual
expertise for faces that they regularly interact with, and
this is probably moderated by the perceived utility of the
interaction (Chiroro & Valentine, 1995; Malpass, 1990).
South African blacks and whites have a heritage of en-
forced segregation (Worden, 2000), and the presence of
an other-race recognition deficit occurs most likely as a
result of their lack of meaningful contact. They also show
a recognition deficit for faces from other geographic re-
gions, regardless of race.
In the end, we wonder whether it is perhaps time that
the concepts of race and, specifically, own-race bias be
retired from this literature. Race cannot be defined with
any precision (Malpass, 1993), and reliance on it may lead
to false conclusions, as demonstrated by the results from
this study. It may also lead to linguistic difficulties, since
recognition advantages have now been established for age
(Wright & Stroud, 2002) and sex (Rehnman & Herlitz,
2006). One solution may be to take the lead from Sporer’s
(2001) face-processing model and refer to the phenom-
enon as the in-group face recognition advantage.
AUTHOR NOTE
The research reported in this article was funded in part by a grant
awarded to the second author by the NRF in South Africa and to the third
author by the NSF in the United States. This financial assistance is grate-
p  .001, r
2
 .31; and F(1,46)  1.19, p  .27, r
2
 .03,
respectively]. Consistent with our predictions, U.S. faces
were treated as an out-group, with performance signifi-
cantly more accurate for South African faces than for U.S.
faces: A linear contrast testing this effect was significant
for both black [F(1,23)  11.13, p  .003, r
2
 .04] and
white [F(1,23)  13.34, p  .002, r
2
 .04] participants.
DISCUSSION
Face recognition researchers have investigated the own-
race recognition bias for almost 40 years, but few have
attempted to provide a definition of race. This is not sur-
prising, since the concept of race is notoriously unclear,
with most biologists asserting that it has no defensible
definition (e.g., Tobias, 1972; see also Sternberg, Grig-
orenko, & Kidd, 2005, who call it a “folk taxonomy”).
Specifically, if black (or white) is used to denote a race
of people, and if there is a true own-race bias in face rec-
ognition, black people living on one continent could be
expected to show a recognition advantage for black faces
from another continent. The results from our study refute
this, showing instead that the recognition advantage for
own-group faces is ethno-geographically specific: White
South Africans showed a recognition advantage for white
South African faces but not for white U.S. faces, and black
South Africans showed a similar recognition advantage
for black South African faces but not for black U.S. faces.
In fact, our results show that white and black South Afri-
cans exhibit a recognition advantage for their own ethnic
group over all three of the comparison groups.
A possible counterexplanation we wish to dismiss is
that the U.S. faces we used may themselves be somewhat
unusual and that U.S. participants might not show the
own-group recognition bias with these faces. The U.S.
faces used in this study were taken from a larger set of
2
1.5
1
d
0.5
0
South African Stimulus Materials
White Participants Black Participants
White faces Black faces
2
1.5
1
d
0.5
0
U.S. Stimulus Materials
White Participants Black Participants
White faces Black faces
Figure 1. Discrimination (d ) of South African and U.S. faces by South African observers. The t bars above each
column represent standard errors.

1092 CHIRORO, TREDOUX, RADAELLI, AND MEISSNER
own- and other-race faces: A dual-process approach. Applied Cogni-
tive Psychology, 19, 545-567.
Ng, W. J., & Lindsay, R. C. L. (1994). Cross-racial facial recognition:
Failure of the contact hypothesis. Journal of Cross-Cultural Psychol-
ogy, 25, 217-232.
Rehnman, J., & Herlitz, A. (2006). Higher face recognition in
girls: Magnified by own-sex and own-ethnicity bias. Memory, 14,
289-296.
Shapiro, P. N., & Penrod, S. (1986). A meta-analysis of facial identifi-
cation studies. Psychological Bulletin, 100, 139-156.
Sporer, S. L. (1999, July). The own-race bias in Germany: Testing
the contact hypothesis with Turks and Germans. Paper presented at
the 4th European Conference on Psychology and Law of the U.S.
Psychology– Law Association and the European Psychology–Law
Association, Dublin.
Sporer, S. L. (2001). Recognizing faces of other ethnic groups: An inte-
gration of theories. Psychology, Public Policy, & Law, 7, 36-97.
Sternberg, R. J., Grigorenko, E. L., & Kidd, K. K. (2005). Intelli-
gence, race and genetics. American Psychologist, 60, 46-59.
Tobias, P. V. (1972). The meaning of race (2nd ed.). Johannesburg:
South African Institute of Race Relations.
Walker, P. M., & Hewstone, M. A. (2006). Perceptual discrimination
investigation of the own-race effect and intergroup experience. Ap-
plied Cognitive Psychology, 20, 461-475.
Worden, N. (2000). The making of modern South Africa. Oxford:
Blackwell.
Wright, D. B., Boyd, C. E., & Tredoux, C. G. (2003). Inter-racial
contact and the own race bias for face recognition in South Africa and
England. Applied Cognitive Psychology, 17, 365-373.
Wright, D. B., & Stroud, J. N. (2002). Age differences in lineup
identification accuracy: People are better with their own age. Law &
Human Behavior, 26, 641-654.
(Manuscript received November 23, 2007;
revision accepted for publication June 3, 2008.)
fully acknowledged. Correspondence concerning this article should be
addressed to P. M. Chiroro, Department of Psychology, University of Pre-
toria, Pretoria 0002, South Africa (e-mail: patrick.chiroro@gmail.com).
REFERENCES
Antony, T., Copper, C., & Mullen, B. (1992). Cross-racial facial
identification: A social cognitive integration. Personality & Social
Psychology Bulletin, 18, 296-301.
Bothwell, R. K., Brigham, J. C., & Malpass, R. S. (1989). Cross-
racial identification. Personality & Social Psychology Bulletin, 15,
19-25.
Brigham, J. C., & Malpass, R. S. (1985). The role of experience and
contact in the recognition of faces of own- and other-race persons.
Journal of Social Issues, 41, 139-155.
Chiroro, P., & Valentine, T. (1995). An investigation of the contact
hypothesis of the own-race bias in face recognition. Quarterly Journal
of Experimental Psychology, 48A, 879-894.
Levin, D. T. (1996). Classifying faces by race: The structure of face
categories. Journal of Experimental Psychology: Learning, Memory,
& Cognition, 22, 1364-1382.
MacLin, O. H., & Malpass, R. S. (2001). Racial categorization of faces:
The ambiguous race face effect. Psychology, Public Policy, & Law,
7, 98-118.
Malpass, R. S. (1990). An excursion into utilitarian analyses, with side
trips. Behavioral Science Research, 24, 1-15.
Malpass, R. S. (1993). “They all look alike to me.” In M. Merrens &
G. Brannigan (Eds.), The undaunted psychologist (pp. 74-88). New
York: McGraw-Hill.
Malpass, R. S., & Kravitz, J. (1969). Recognition for faces of own-
and other-race faces. Journal of Personality & Social Psychology,
13, 330-334.
Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigat-
ing the own-race bias in memory for faces: A meta-analytic review.
Psychology, Public Policy, & Law, 7, 3-35.
Meissner, C. A., Brigham, J. C., & Butz, D. A. (2005). Memory for