Elsevier

Cognition

Volume 93, Issue 1, August 2004, Pages B1-B9
Cognition

Brief article
A holistic account of the own-race effect in face recognition: evidence from a cross-cultural study

https://doi.org/10.1016/j.cognition.2003.09.011Get rights and content

Abstract

A robust finding in the cross-cultural research is that people's memories for faces of their own race are superior to their memories for other-race faces. However, the mechanisms underlying the own-race effect have not been well defined. In this study, a holistic explanation was examined in which Caucasian and Asian participants were asked to recognize features of Caucasian and Asian faces presented in isolation and in the whole face. The main finding was that Caucasian participants recognized own-race faces more holistically than Asian faces whereas Asian participants demonstrated holistic recognition for both own-race and other-race faces. The differences in holistic recognition between Caucasian and Asian participants mirrored differences in their relative experience with own-race and other-race faces. These results suggest that the own-race effect may arise from the holistic recognition of faces from a highly familiar racial group.

Introduction

In the cross-cultural literature, it is well established that people are better at recognizing faces from their own race relative to faces from other races (for reviews see Bothwell et al., 1989, Brigham and Malpass, 1985, Chiroro and Valentine, 1995). The own-race effect1 is assumed to reflect differences in racial experience such that people have more exposure and practice recognizing faces from their own race relative to faces of other races. Although there is little disagreement that racial experience is linked to improved recognition, an enduring question in the literature revolves around how race-specific experience affects the processes of face recognition (Levin, 2000, Ng and Lindsay, 1994).

It has been suggested that through experience, people acquire the ability to differentiate faces based on the distinctiveness of their features and their configuration (i.e. the spatial relations between the features) (for a full discussion of featural and configural face processes, see Maurer, Le Grand, & Mondloch, 2002). According to the configural account of face recognition, sensitivity to configural information is particularly vulnerable to orientation effects and is disrupted when a face is turned upside down. Consistent with this view, many studies (Carey and Diamond, 1977, Diamond and Carey, 1986, Scapinello and Yarmey, 1970, Yarmey, 1971) have shown that inversion disproportionately impairs the recognition of faces relative to the recognition of other objects (e.g. automobiles, flowers, airplanes) – the so-called face inversion effect (Yin, 1969). Are people more sensitive to the configural information in own-race faces relative to other-race faces? As a test of the configural effects in own-race recognition, Chinese and Caucasian participants were asked to recognize upright and inverted Chinese and Caucasian faces (Rhodes, Tan, Brake, & Taylor, 1989). The main finding of this study was that participants demonstrated a larger inversion effect for their own-race faces relative to other-race faces suggesting that configural processes can be tuned to the recognition of faces from a specific, familiar race.

One limitation of the inversion paradigm is that configural processes are not directly examined, but only inferred by the degree to which inversion disrupts face recognition performance relative to the recognition of other objects (e.g. houses, airplanes) (Valentine, 1988). Therefore, other paradigms in which the spatial relations of features are specifically manipulated provide a more precise test of configural processes. In the Young, Hellawell, and Hay (1987) composite paradigm, for example, the top face half of a well-known person is joined with the bottom half of another well-known person. The participants’ task is to identify the person shown in the top face while ignoring the face shown in the bottom half. Young et al. found that participants were slow to isolate the identity of a person in the top half of the composite face due to the configural interference produced by the face shown in the bottom half. Critically, configural interference was substantially reduced by misaligning the top and bottom halves of the composite face or by turning it upside down.

In contrast to the composite paradigm that measures the effects of global configuration on face recognition, the parts/wholes task measures the interdependence between featural and configural information in the face representation. In this paradigm, participants learned to name a series of upright, inverted or scrambled faces (e.g. Joe) or houses (e.g. Joe's house) (Tanaka and Farah, 1993, Tanaka and Sengco, 1997). After the naming phase, participants’ memory for the face parts (e.g. Joe's nose) or house parts (e.g. Joe's door) was tested when shown in the whole face (or house) and in isolation. The main result was that parts from upright intact faces were better recognized when presented in the whole face than when presented in isolation (Tanaka and Farah, 1993, Tanaka and Sengco, 1997). Moreover, changes in spatial configuration (e.g. increasing inter-eye distance) impaired recognition of the affected features (e.g. eyes) and those features whose spatial location remained unchanged (i.e. nose, mouth) (Tanaka & Sengco, 1997). In contrast, identification performance for parts from inverted faces, scrambled faces and houses was the same whether tested in the whole face (or object) or in isolation. Based on this evidence, Tanaka, Farah, and colleagues argued that in normal face processing, the encoding of a facial feature is combined with its spatial relations to other features in what they referred to as a holistic representation (Tanaka and Farah, 1993, Tanaka and Sengco, 1997).2

As an account of the own-race effect, it is plausible that people recognize faces from their own race more holistically than other-race faces. In the current experiment, the holistic hypothesis was examined by asking Caucasian and Asian participants to recognize face parts from Caucasian and Asian faces in isolation and in the whole face. According to the holistic account, it is predicted that the part-whole advantage should be greater for own-race faces than other-race faces.

Section snippets

Participants

A total of 21 Caucasian undergraduates from the University of Ulm, Germany and 21 Asian undergraduates from the University of Victoria participated. The Caucasian subject group was comprised of 21 (12 female/9 male) right-handed native German participants with an average age of 28.2 years (range 20–44 years). These participants were recruited from a rural area in Southern Germany with a predominantly Caucasian population. The administrative center of this area, the city of Ulm, where subject

Racial experience

According to the participants’ estimates of their own-race and other-race experiences (as shown in Table 1), the Caucasian participants reported extensive exposure to other Caucasians (M=5.00) and relatively little contact with Asians (M=1.41). The Asian participants, on the other hand, reported having slightly more contact with Caucasians (M=3.64) than Asians (M=3.45). The analysis of variance (ANOVA) test with Race of the Participant (Asian, Caucasian) as a between-groups factor and Racial

Discussion

To summarize our results, it was found that Caucasian participants demonstrated holistic processing for the recognition of Caucasian faces and featural processing for the recognition of unfamiliar Asian faces. Asian participants demonstrated holistic recognition for both Asian and Caucasian faces. Given that participants in this study had extensive exposure to members of their own race, these findings indicate that experience of own-race faces promotes holistic processing. Consistent with the

Acknowledgements

We would like to thank Jacques Mehler and two anonymous reviewers for their helpful comments and Jasmine Phua for developing the stimuli used in this study. This research was supported by grants from the National Science Foundation (NSF #9729030 and #0078745) and James S. McDonnell Foundation to J.W.T.

References (23)

  • D.T Levin

    Race as a visual feature: using visual search and perceptual discrimination tasks to understand face categories and cross-race recognition deficit

    Journal of Experimental Psychology: General

    (2000)
  • Cited by (0)

    View full text