This article was downloaded by: [Tufts University]

On: 31 October 2014, At: 11:22

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office:
Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Quarterly Journal of Experimental

Psychology
Publication details, including instructions for authors and subscription
information:
http://www.tandfonline.com/loi/pqje20

Reducing the own-race bias in face

recognition by shifting attention
a a
Peter J. Hills & Michael B. Lewis
a
Cardiff University , Cardiff, UK
Published online: 17 Feb 2007.

To cite this article: Peter J. Hills & Michael B. Lewis (2006) Reducing the own-race bias in face
recognition by shifting attention, The Quarterly Journal of Experimental Psychology, 59:6, 996-1002, DOI:
10.1080/17470210600654750

To link to this article: http://dx.doi.org/10.1080/17470210600654750

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed in this publication
are the opinions and views of the authors, and are not the views of or endorsed by Taylor &
Francis. The accuracy of the Content should not be relied upon and should be independently
verified with primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use
can be found at http://www.tandfonline.com/page/terms-and-conditions
 THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY
2006, 59 (6), 996 – 1002

Short article

Reducing the own-race bias in face recognition by

shifting attention

Peter J. Hills and Michael B. Lewis

Cardiff University, Cardiff, UK
Downloaded by [Tufts University] at 11:22 31 October 2014

The own-race bias (ORB) in face recognition can be interpreted as a failure to generalize expert
perceptual encoding developed for own-race faces to other-race faces. Further, black participants
appear to use different features to describe faces from those used by white participants (Shepherd
& Deregowski, 1981). An experiment is reported where the size of the ORB was assessed using a stan-
dard face recognition procedure. Four groups were tested at two time intervals. One group received a
training regime involving learning to distinguish faces that varied only on their chin, cheeks, nose, and
mouth. Three control groups did not receive this training. The ORB, present prior to training, was
reduced after the critical perceptual training. It is concluded that the ORB is a consequence of a failure
of attention being directed to those features of other race faces that are useful for identification.

Own-race faces are better remembered than faces
of another, less familiar race (e.g., Meissner &
Brigham, 2001). This is the crux of the own-race
bias (ORB: Sporer, 2001) and has been implicated
in many false identifications in eyewitness testi-
mony (e.g., Leippe, 1995). It is a reliable effect
(Chance & Goldstein, 1996), which is arguably
consistent across racial groups (Bothwell,
Brigham, & Malpass, 1989; but see Anthony,
Copper, & Mullen, 1992, for a different result).
According to a meta-analysis conducted by
Meissner and Brigham (2001) people are 2.23
times more likely to recognize an own-race face
than an other-race face in a recognition experi-
ment. There are many theories that have been
developed to explain the ORB. These are summar-
ized in detail by Sporer (2001). From these the-
ories, procedures have been developed to reduce
the ORB in participants.
One of the most frequently discussed methods
for reducing the ORB is with experience (Galper,
1973; Slone, Brigham, & Meissner, 2000).

Correspondence should be addressed to Peter Hills, School of Psychology, Cardiff University, Park Place, Cardiff CF10 3AT,
UK. Email: HillsPJ@Cardiff.ac.uk
This research was supported by grant PTA– 030–2003– 00524 from the ESRC to Peter Hills. The authors would like to thank
Paul Hutchings for providing some of the photographs used for this experiment. Furthermore, the authors would like to thank
Rachael Elward, Mary Halsey, and Claire Jones for some of the data collection. Finally, the authors would like to thank Dan
Wright and an anonymous reviewer for detailed and thought-provoking comments on a previous draft of this paper.

996 # 2006 The Experimental Psychology Society

http://www.psypress.com/qjep DOI:10.1080/17470210600654750

Attempts have been made to reduce the ORB using
training regimes to increase contact with other-race
faces. Malpass, Lavigueur, and Weldon (1973) used
a simple 1-hour visual training paradigm, whereby
white participants were shown black and white
faces and were given feedback for a four-alternative
forced-choice recognition test. Training did
improve recognition accuracy scores, especially if
participants were given shock feedback.
Following on from these studies, Lavrakas,
Buri, and Mayzner (1976) conducted another
training study, in which white participants were
given a concept-learning visual training task. The
training faces were made from an Identi-Kit, and
the two concepts that participants had to learn
Downloaded by [Tufts University] at 11:22 31 October 2014
were either light eyes or a conjunction of dark
eyes and thick lips. Participants were given feed-
back for their judgement. Lavrakas et al. also
tested whether the type of contact was relevant
in the ORB and field dependence/independence
of the participants. Their results indicated that
those white participants who had black friends
had a lower ORB than those who simply
knew more black participants. Moreover, field-
independent white participants were better able
to recognize black faces. Finally, the training on
specific features did reduce the ORB, possibly
due to removing the deleterious effect of
remembering only skin-colour (Lavrakas et al.).
These results are consistent with a recent theo-
retical account of the ORB. Levin (2000) advocates
a feature selection process in the recognition of
faces. By this, participants extract the most diag-
nostic visual feature. As such, in processing an
own-race face the most diagnostic visual feature
could contain any individuating information.
However, processing an other-race face will
involve a race feature as the most diagnostic. As
such, an other-race face is processed according to
a prototype and is therefore not individuated,
thus making it more difficult to recognize.
An alternative explanation based upon Valentine
(1991, expanded by Lewis, 2004), is that every face
is stored within a multidimensional face-space. The
face-space stores faces along relevant dimensions.
Other-race faces do not vary consistently along
these dimensions and as such are grouped close
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (6)
REDUCING THE OWN-RACE BIAS
together farther away from the mean. This makes
them more difficult to distinguish from each
other, but easy to differentiate from own-race faces.
The face-space explanation suggests that other-
race faces are not best described by the dimensions
within face-space. Though these dimensions have
not been specified precisely, it can be suggested
that the reason that other-race faces are not appro-
priately stored on the face-space is that there are
physiognomic differences between black and
white faces.
Ellis, Deregowski, and Shepherd (1975) looked
at the frequencies with which black and white
participants referred to various facial features.
Their results clearly showed a different pattern of
features examined by black participants and
white participants. Face outline, eye size, eye-
brows, chin, and ears were significantly more fre-
quently used by black participants as descriptors,
while white participants used hair colour, hair
texture, and eye colour. Their data also show that
black participants used lips, mouth, and nose
more frequently, especially when describing black
faces. Furthermore, black participants use a more
varied description, using more features than
white participants. This may represent an import-
ant differentiation between the face recognition
abilities of black and white participants or physiog-
nomic differences between black and white faces.
How facial descriptions differ across black and
white faces was looked at in detail by Shepherd
and Deregowski (1981). Using INDSCAL ana-
lyses, physiognomic differences across black and
white faces were discovered. Internal facial features,
with round fat faces, thick lips, and broad noses are
related to black physiognomy. This is contrasted
with features associated with hair, which are
related to white physiognomy. Hair features
explain roughly 85% of the variance in the
INDSCAL solution for white faces, but only 35%
for black faces. The internal lower facial features
explain roughly 75% of the variance in the
INDSCAL solution for black faces, but the same
features only explain 35% of the variance for
white faces. Nevertheless, using a homogenous set
of faces, both black and white participants will use
the same features to discriminate between them.
997
 HILLS AND LEWIS
Taking these results with those of Lavrakas et al.
(1976) indicates that training white participants to
look at specific features will indeed improve the rec-
ognition of black faces if those features are the most
diagnostic. The current experiment used a percep-
tual learning experimental procedure in which
white participants were trained using white faces,
which can be discriminated upon only by lower
facial features (chin, nose, mouth, and cheeks). It
was hypothesized that this training should sub-
sequently reduce the ORB. To control for greater
exposure to faces, three control groups were
included. One group received no training whatso-
ever. A second group received noncritical-feature
training. In this condition, white participants were
Downloaded by [Tufts University] at 11:22 31 October 2014
given the same level of training, except that the
faces used for training differed only on features not
diagnostic to black faces. A final control group was
presented with the same faces as those in the
crucial feature-critical trained condition, except
that the task was extraneous to the face. In other
words, this control tested between simple presen-
tation of faces and actual attention paid to the faces.
METHOD
Participants
A total of 124 Cardiff University undergraduates
were randomly divided into one of four training
groups: feature critical, no training, noncritical
feature, and colour blob discrimination. All par-
ticipants had normal or corrected vision. The
data from 2 participants were removed since they
had been raised in Africa, and they were replaced
by additional participants.
Apparatus and stimuli
All stimuli were presented on a high-resolution
colour monitor, displayed using DirectRTTM
Research Software (EmpirisoftTM). Participant
responses were recorded on a standard keyboard.
1
Development of the MacBrain Face Stimulus Set was overseen by Nim Tottenham and supported by the John D. and Catherine
T. MacArthur Foundation Research Network on Early Experience and Brain Development. Please contact Nim Tottenham at
tott0006@tc.umn.edu for more information concerning the stimulus set.
998 THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (6)
Participants viewed the monitor at a distance of
60 cm.
Two face sets were used to assess the ORB
before and after training. These were the
MacBrain Face Stimulus Set1 and a face set
collected by Paul Hutchings at Cardiff University.
Each face had two pictures taken of it, which
were both neutral in expression and frontal views.
One of these pictures was used in the presentation
phase, while the other face was used in the recog-
nition phase of the recognition paradigm. This
was to avoid pictorial recognition. All stimuli had
a resolution of 72 dpi and were 150 mm by 200 mm.
Three different training sets of faces were con-
structed using a face composite computer package
FACESTM (InterquestTM ). This package uses 16
features to make a face. Within each feature a
choice of between 10 and 100 features can be
made. For each set, 40 different target faces were
created. From these targets, 10 lookalikes were
created, which were identical in certain features
depending on condition.
For the feature-critical faces, all the faces had
similar hair style and colour. The lookalikes only
differed according to “lower” facial features: the
chin, nose, mouth, and cheeks. They were identi-
cal to the targets according to “white” facial fea-
tures. The coloured blob discrimination
condition used these faces; however, each face
had a coloured blob placed on it in a random
place. The noncritical-feature faces were created
in the same way as the feature-critical faces,
except that all the lookalikes were identical in
the lower regions: chin, nose, mouth, and cheeks.
Only the eyes and hair styles differed. An
example of each of the faces is shown in Figure 1.
As an additional control, all participants filled
out the Modern Racism Scale (McConahay, 1986).
Design
Participants’ ORB level was assessed before and
after training. The ORB was tested in a simple

Downloaded by [Tufts University] at 11:22 31 October 2014
Figure 1. Examples of the face stimuli used in the training session.
The left face is the target face. The right face is a lookalike. Feature-
critical faces are shown in a: they differ only on nose, chin, mouth,
and cheeks. Noncritical-feature faces are shown in b: they differ
only in hair style, eyes, and eyebrows. Coloured blob faces are
shown in c: they are the same as those in the feature-critical
condition, except that the coloured blobs differ in hue (in figure
blue blob is shown as white, orange blob is shown as black).
yes/no recognition paradigm, whereby half the
target faces were black, and half were white. The
face sets used for the recognition paradigm were
counterbalanced, so that half the participants had
the MacBrain Face Stimulus Set for the pretrain-
ing recognition paradigm task, and the other half
had Paul Hutchings’ set. In each recognition
paradigm task, the targets were counterbalanced
with distractors, and the order of presentation
was randomized.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (6)
REDUCING THE OWN-RACE BIAS
Procedure
This experiment involved three phases. The first
phase was the face recognition test, the second
was the training, and the third was the second
face recognition test.
The standard old/new recognition tests con-
sisted of participants being shown 20 faces (10
black and 10 white). Participants were asked to
assess the faces for attractiveness on a 0– 9 scale.
Each face was presented for 2 s. As a distractor
task, participants filled out a questionnaire,
which took 3 minutes. After the questionnaire,
the participants were given the recognition
phase, whereby they were presented 40 faces (20
black, 20 white of which 20 were targets, and 20
were distractors). They were asked for a yes or
no response as to whether they had seen the face
previously.
The training phase consisted of a standard per-
ceptual learning paradigm (e.g., Hall, 1980). The
participants were preexposed to the target face,
and the target face was assigned a name. The par-
ticipants then were presented with 20 further faces
sequentially. For each face, participants had to
state whether the face matched the target face
(or colour in the colour blob discrimination con-
dition). A total of 10 target faces made up one
training block. The training blocks got progress-
ively harder, using more similar faces. There
were a total of four training blocks, lasting a
total of 60 minutes. Those participants in the
no-training condition did not undergo this phase
of the experiment. Instead, they were asked to
leave the experimental booth and to return after
60 minutes.
The final recognition test was identical to the
first phase of the experiment, but using a different
questionnaire and a different face set.
RESULTS
All scores on the Modern Racism Scale were
between 0 and 4 implying that no participants
held racist views, and therefore this was not con-
sidered any further. The signal detection theory
999
 HILLS AND LEWIS

measure d 0 was calculated using the MacMillan DISCUSSION

and Creelman (1991) method. The ORB for
each participant was calculated according to the By altering the focus of attention of white partici-
formula: pants, the ORB has been reduced in this exper-
iment. Participants were trained, in a perceptual
learning paradigm, to focus on lower facial features
0
ORB ¼ (dW
dB0 )=(dW
0
þ dB0 ) (1) (i.e., those used by black participants to describe
black faces). Since the ORB was not reduced
0 when the training faces were presented without
where d W is the discrimation performance on white
deliberate attention being paid to them, the results
faces, and d B0 is the discrimination performance on
indicate that this procedure directed the partici-
black faces. This measures the ORB relative to
pants’ attention toward the lower facial features. It
overall performance. Figure 2 shows the mean
was not possible, however, to increase the ORB
ORB scores before and after the training regimes.
using training on white face diagnostic information.
The ORB is present for all participants, except
This may be due to the ORB already being an
Downloaded by [Tufts University] at 11:22 31 October 2014

those who had completed feature-critical training.

These data were entered into a mixed factorial
ANOVA, with type of training as a four-level
between-subjects variable and before and after
being a two-level within-subjects variable. The
interaction between level of training and condition
was significant, F(3, 124) ¼ 3.303, MSE ¼ 0.201,
p , .05. This was indicated by the ORB being sig-
nificantly reduced after training for those receiving
feature-critical training (mean difference ¼ 0.36,
p , .05). After feature-critical training, partici-
pants’ ORB was not significantly different from
zero, suggesting equivalent performance on black
and white faces. No other comparisons were sig-
nificant. The effect size for the removal of the
ORB was large, r ¼ .9 (Cohen’s d ¼ 4.2).
Incidentally, using a simpler method for calcula-
ting the ORB (d 0 w – d B0 ) the results are identical.
Figure 2. ORB scores before and after training. Error bars
represent standard error.
established mechanism within face processing.
In the Introduction, it was surmised that if
black and white faces differ in physiognomy, they
would be stored along different dimensions in
Valentine’s (1991) face-space. It may be possible
that using this type of training regime activates
different dimensions or regions of the face-space
(e.g., Hills, Lewis, & Honey, 2005). Levin’s
(2000) race-as-feature hypothesis suggests that
own-race faces are processed in an individuating
manner. He went on to say that race is used by par-
ticipants to discriminate between other-race faces.
The present data provide evidence to suggest what
dimensions and what individuating information is
required to accurately recognize black faces.
The present data are compatible with Valentine’s
models of face-space, by suggesting particular
dimensions relevant to black faces. They are also
compatible with perceptual learning mechanisms
suggested by such researchers as McLaren and
Mackintosh (2000, 2002) and Hall (1980). In per-
ceptual learning terminology, the unique elements
that distinguish between black faces can be stated
quite firmly as the chin, mouth, nose, and cheeks.
Simply training participants to focus on these
elements reduces the ORB. These features are not
the unique elements when distinguishing white
faces. Of course, when discriminating between
black and white faces, the unique elements may
well be a global construct such as race as Levin
(2000) postulates. However, this study certainly
demonstrates that simply stating that race is a

1000 THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (6)


feature does not accurately portray the perceptual
experience. What exactly it is about the elements
of an other-race face that makes race the feature is
a question unanswered by Levin.
An important aspect of this study is the use of
perceptual learning mechanisms, borrowed from
the animal learning literature, in face recognition.
Using such methods, researchers can be more
explicit in how they define constructs and dimen-
sions in face recognition, which is vital for research
to progress. Moreover, there is much that is
unknown about how face recognition develops.
Perceptual learning can be applied to develop-
mental aspects of face recognition.
One caveat of the present study must be
Downloaded by [Tufts University] at 11:22 31 October 2014
acknowledged. In this study, only white partici-
pants were tested. The effects observed here
could be replicated across black participants who
show an ORB and therefore a lower ability to
recognize white faces than black faces. The
feature-critical condition in this case would be
equivalent to the noncritical training used on
white participants. The effect described in this
study may prove more difficult to obtain in black
participants, since black participants use a wider
range of features to describe faces (Ellis et al.,
1975), which may explain why black participants
are generally better at face recognition tasks than
are white participants (see Shepherd &
Deregowski, 1981). As such, the perceptual learn-
ing effect described here explains why black par-
ticipants show a lower ORB, since they use all
dimensions to recognize faces, while white partici-
pants only use relevant dimensions.
One final point to make at this stage is the
possibility that this simple training may have an
effect on stereotyping and prejudice of the partici-
pants. Since the participants are showing less of an
ORB, it may be that they are processing other-race
faces more appropriately, which may covertly
reduce the stereotyping subsequently used. Social
psychologists may wish to examine this hypothesis
as another method for reducing stereotyping.
In conclusion, the perceptual learning effect can
be used to explain many of the effects in face rec-
ognition. It should become a useful tool in learn-
ing how humans learn about faces and, more
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (6)
REDUCING THE OWN-RACE BIAS
precisely, what are the unique elements which
help us recognize different faces. Clearly, the
ORB is transient enough to be reduced by refocus-
ing of attention, but it is unlikely that an hour’s
training will be sufficient to reverse a lifetime’s
experience. Nevertheless, it is clear that perceptual
learning will be able to assist in discovering the
dimensions of the face-space.
Original manuscript received 2 August 2005
Accepted revision received 15 February 2006
REFERENCES
Anthony, T., Copper, C., & Mullen, B. (1992). Cross-
racial facial identification: A social cognitive integ-
ration. Personality and Social Psychology Bulletin, 18,
296– 301.
Bothwell, R. K., Brigham, J. C., & Malpass, R. S.
(1989). Cross-racial identification. Personality and
Social Psychology Bulletin, 15, 19 – 25
Chance, J. E., & Goldstein, A. G. (1996). The other-
race effect and eyewitness identification. In
S. L. Sporer, R. S. Malpass, & G. Koehnken
(Eds.), Psychological issues in eyewitness identification
(pp. 153– 176). Mahwah, NJ: Lawrence Erlbaum
Associates, Inc.
Ellis, H. D., Deregowski, J. B., & Shepherd, J. W.
(1975). Descriptions of White and Black faces by
White and Black subjects. International Journal of
Psychology, 10, 119– 123.
Galper, R. E. (1973). “Functional race membership” and
recognition of faces. Perceptual and Motor Skills, 37,
455– 462.
Hall, G. (1980). Exposure learning in animals.
Psychological Bulletin, 88, 535– 550.
Hills, P. J., Lewis, M. B., & Honey, R. C. (2005,
August). Stereotype priming in the recognition of unfa-
miliar faces. Paper presented at the BPS Cognitive
Section Conference, Leeds, UK.
Lavrakas, P. J., Buri, J. R., & Mayzner, M. S. (1976). A
perspective on the recognition of other-race faces.
Perception and Psychophysics, 20, 475– 481.
Leippe, M. R. (1995). The case for expert testimony
about eyewitness memory. Psychology, Public Policy,
and Law, 1, 909–959.
Levin, D. T. (2000). Race as a visual feature: Using
visual search and perceptual discrimination tasks to
understand face categories and the cross-race
1001
 HILLS AND LEWIS
recognition deficit. Journal of Experimental
Psychology: General, 129, 559– 574.
Lewis, M. B. (2004). Face-space-r: Towards a unified
account of face recognition. Visual Cognition, 11,
29 – 70.
Macmillan, N. A., & Creelman, C. D. (1991). Detection
theory: A user’s guide. Cambridge, UK: Cambridge
University Press.
Malpass, R. S., Lavigueur, H., & Weldon, D. E. (1973).
Verbal and visual training in face recognition.
Perception and Psychophysics, 14, 285– 292.
McConahay, J. B. (1986). Modern racism, ambivalence,
and the Modern Racism Scale. In S. L. Gaertner &
J. F. Dovidio (Eds.), Prejudice, discrimination, and
racism (pp. 91 – 126). San Diego, CA: Academic
Press.
Downloaded by [Tufts University] at 11:22 31 October 2014
McLaren, I. P. L., & Mackintosh, N. J. (2000). An
elemental model of associative learning: I. Latent
inhibition and perceptual learning. Animal Learning
and Behavior, 28, 211– 246.
McLaren, I. P. L., & Mackintosh, N. J. (2002).
Associative learning and elemental representation: II.
1002 THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (6)
Generalization and discrimination. Animal Learning
and Behavior, 30, 177–200.
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.
Shepherd, J. W., & Deregowski, J. B. (1981). Races and
faces—a comparison of the responses of Africans
and Whites to faces of the same and different
races. British Journal of Social Psychology, 20,
125– 133.
Slone, A. E., Brigham, J. C., & Meissner, C. A. (2000).
Social and cognitive factors affecting the own-race
bias in Whites. Basic and Applied Social Psychology,
22, 71 – 84.
Sporer, S. L. (2001). Recognizing faces of other ethnic
groups—an integration of theories. Psychology,
Public Policy, and Law, 7, 36 – 97.
Valentine, T. (1991). A unified account of the effects of
distinctiveness, inversion and race in face recog-
nition. Quarterly Journal of Experimental Psychology,
43A, 161– 204.