- Email: [email protected]
Influence of cross-ethnic social experience on face recognition accuracy and the visual perceptual strategies involved
International Journal of Intercultural Relations 65 (2018) 42–50
Contents lists available at ScienceDirect
International Journal of Intercultural Relations journal homepage: www.elsevier.com/locate/ijintrel
Influence of cross-ethnic social experience on face recognition accuracy and the visual perceptual strategies involved
T
⁎
Alberto Megiasa, , Iga Rzeszewskaa, Luis Aguadob, Andrés Catenaa a b
Learning, Emotion and Decision Group, Mind, Brain, and Behavior Research Center, University of Granada, Granada, Spain Universidad Complutense, Madrid, Spain
A R T IC LE I N F O
ABS TRA CT
Keywords: Intercultural factors Social cognition Face recognition Cross-race effect Eye movements Gypsy ethnicity
The cross-ethnic effect (in the literature, usually termed the cross-race effect) is defined as the greater difficulty in recognizing faces of other ethnicities compared with faces of one’s own. The aims of the present research were: 1) to test the hypothesis that the cross-ethnic effect is due to lack of contact with the other ethnicity. 2) to study possible differences in the perceptual mechanisms employed in face recognition as a function of the contact degree between ethnicities, which may be the basis of the cross-ethnic effect. We compared two ethnic groups with a high degree of contact, but different identities and cultural values: Andalusian Gypsies and Andalusian Caucasians. Both groups had to recognize a set of East Asian, Caucasian, and Gypsy faces while eye movements were monitored. In accordance with the contact hypothesis, our results revealed no differences between Gypsies and Caucasians observers in face recognition success. However, East Asian faces were more poorly recognized than Gypsies and Caucasian faces by both observer groups. With respect to the perceptual strategies, despite achieving similar face recognition performance, Caucasian and Gypsy observers employed different visual exploration strategies. Gypsies focused their attention on the eyes, while Caucasians fixated more on the nose than Gypsies. Our results support the contact hypothesis as an explanation for the cross-ethnic effect, and show how cultural factors imply differences in perceptual strategies even between close ethnic groups.
Introduction People are better at identifying faces of their own ethnicity than those of other ethnicities. This phenomenon is known as the cross-race effect and has been widely studied and replicated in recent decades (MacLin & Malpass, 2003; Meissner & Brigham, 2001; Sporer, 2001). The cross-race effect has usually been conceptualized using the term race; however, we prefer to use the term ethnic, given that race in humans refers to a biological construct created from prevailing social perceptions without scientific evidence from biology, genetics, or physical anthropology (Anderson & Nickerson, 2005; Sternberg, Grigorenko, & Kidd, 2005; Witzig, 1996). The present research aims to study how the degree of contact with other ethnicities modulates the cross-race effect (henceforth crossethnic effect) in face recognition and the visual perceptual mechanisms involved in this process. Several social and cognitive hypotheses have been proposed to account for the cross-ethnic effect, with social attitudes, physiognomic variability between ethnicities, and inter-ethnic contact being the most prominent explanations (Meissner & Brigham, 2001; Ng & Lindsay, 1994). Although there is no total consensus across studies, hypotheses based on inter-ethnic contact have received greater support (Goldinger, He, & Papesh, 2009). The contact hypothesis predicts that the recognition accuracy of other ⁎
Corresponding author at: Centro de Investigación Mente, Cerebro y Comportamiento, Universidad de Granada, Campus de Cartuja s/n, 18071, Granada, Spain. E-mail address: [email protected] (A. Megias).
https://doi.org/10.1016/j.ijintrel.2018.04.008 Received 19 July 2017; Received in revised form 18 April 2018; Accepted 30 April 2018 Available online 06 May 2018 0147-1767/ © 2018 Elsevier Ltd. All rights reserved.
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
ethnicity members is directly related to the quantity and quality of contact with people of that ethnic group (Goldstein & Chance, 1985; Vrij & Winkel, 1989). This approach is empirically supported by the findings of Meissner and Brigham’s (2001) meta-analysis, where authors across 29 independent studies found that inter-ethnic contact plays a significant mediating role in the cross-ethnic effect. For example, studies in ecological contexts have shown that people living in integrated neighbourhoods are better able to discriminate other ethnic faces than those living in segregated neighbourhoods (Feinman & Entwisle, 1976). The effect of inter-ethnic contact level on face recognition could be explained by perceptual learning mechanisms (Goldinger et al., 2009). Gibson (1969) defined perceptual learning as “an increase in the ability to extract information from the environment, as a result of practice and experience with the stimulation coming from it”. According to this theory, and given the existence of physiognomic differences between ethnic groups (Kelly et al., 2005), an increase in the degree of exposure to members of other ethnicities should improve the ability to discriminate between the characteristic facial features of those ethnicities. Thus, a higher degree of contact with the other ethnicity would teach people which perceptual strategies they should use to properly process the faces of that ethnic group. Previous literature has shown that perceptual mechanisms and visual attention patterns employed during face recognition differ as a function of the ethnicity of the observer (Blais, Jack, Scheepers, Fiset, & Caldara, 2008; Hills & Pake, 2013). Visual attention in face processing tends to be restricted to internal facial features, with a systematic visual scan pattern over the eyes, nose, and mouth (e.g., Althoff & Cohen, 1999; Groner, Walder, & Groner, 1984). It has been estimated that approximately 90% of fixation time is spent on these face regions (Henderson, Falk, Minut, Dyer, & Mahadevan, 2001). However, ethnicity, cultural background, and social experience lead to a different weighting of each one of these regions. For example, Blais et al. (2008) showed that Eastern cultures (e.g. East Asians) use more holistic perceptual strategies, with more fixations directed toward the nose (the central region of the face), whilst people of Western Caucasian cultures focus more on the eye region, performing scan paths that involve more analytic strategies (see Miyamoto, Nisbett, & Masuda, 2006, for a study of holistic and analytic perceptual differences between cultures). Similar results have been obtained by Caldara, Zhou, and Miellet (2010), Goldinger et al. (2009) and Hills and Pake (2013), showing that observers from different ethnicities looked at different facial features when they tried to recognize the same faces. Given these differences in the visual scan path and the particular physiognomy of each ethnicity, deficits in cross-ethnic recognition may be due to observers not attending to the most diagnostic facial features of other ethnicities (Hills & Pake, 2013). For instance, an individual may have learnt to use a specific perceptual strategy to recognise faces because this strategy fits well with the physiognomic features of their ethnicity and it is part of their cultural background and social experience. However, this strategy will be problematic for recognising faces from other unfamiliar ethnicities with different physiognomic features, which would lead to poorer performance, and could thus explain the cross-ethnic effect. On the other hand, according to the contact hypothesis, a greater degree of contact with other ethnicities would help adapt the visual exploration strategies to the physiognomy of each ethnicity and thus improve facial recognition accuracy. The aim of this research was to advance our understanding of the perceptual mechanisms underlying the contact hypothesis and their implications for the cross-ethnic effect. We focused on two ethnic groups: Andalusian Caucasians and Andalusian Gypsies. These groups have a high degree of contact, but they differ considerably in terms of their identity and cultural values. Caucasians and Gypsies (Romanies) in Andalusia have shared the same geographical area since the XV century. Approximately 300,000 Gypsies live in this region of Spain, and in Granada – the city where this study was conducted – the Gypsy community is particularly well integrated. There are two historic heritage neighborhoods (Albaicín and Sacromonte) with a large Gypsy population where Andalusian Gypsies and Caucasians co-exist and live in harmony with tourists from many different countries. However, in spite of this co-existence, in many aspects the Gypsy identity remains significantly different from other Andalusians. Gypsies have assimilated a set of cultural factors, traditions, affiliation, and common history that has defined them as a distinct ethnic group for many centuries. To achieve our objective, an Andalusian Gypsy group and an Andalusian Caucasian group performed a standard face recognition task adapted to the study of the cross-ethnic effect (Ng & Lindsay, 1994; Vizioli, Foreman, Rousselet, & Caldara, 2010). In this task, Andalusian Gypsies and Caucasian participants had to view a set of faces from an unfamiliar ethnicity (East Asian faces) and from their own ethnicities (Caucasian and Gypsy faces) under instructions to memorize them for a later recognition test. While the participants were viewing the faces, eye movements were continuously monitored by Eyelink II eye tracking system. Eye movements are a good index of visual attention allocation and play an important functional role in face processing (Henderson, Williams, & Falk, 2005). For example, longer fixation times on specific facial features help to encode, retain, and better retrieve these features, ensuring greater recognition accuracy (Goldinger et al., 2009). In accord with previous studies in the literature (Althoff & Cohen, 1999; Blais et al., 2008), we evaluated the number and duration of fixations on the three face regions that are critical for face processing (eyes, nose, and mouth). We had two main objectives: 1) to test the hypothesis that the cross-ethnic effect is due to the lack of contact with the other ethnicity. Faces from unfamiliar ethnic groups (in our case, East Asians) should be recognized more poorly than faces from their own ethnicity (cross-ethnic effect). On the other hand, the cross-ethnic effect should not occur if the observer has sufficient experience with members of the other ethnicities. Given that Andalusian Gypsies and Caucasians live in close proximity (high or moderate degree of contact), we hypothesized that there will be no differences between Gypsy and Caucasian observers in recognizing faces (Hypothesis 1a). However, both should be particularly poor at recognizing the faces of East Asians, an unfamiliar ethnicity for both observer groups (Hypothesis 1b). Therefore, we expect to find a main effect of the type of face to be recognized (hereinafter called Face Culture). 2) As a second objective, we explore the possible differences in the perceptual mechanisms employed by Gypsies and Caucasians in face recognition of familiar and unfamiliar ethnicities. Differences in the visual exploration strategies as a function of the degree of contact with the other ethnicity (contact hypothesis) could underlie the cross-ethnic effect. First, we expect to find distinct visual strategies in the recognition of Caucasian, Gypsy, and East Asian faces because of the physiognomic differences between them (Hypothesis 2a). Second, due to the degree of contact between Gypsies and Caucasians, both ethnicities should have 43
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
learnt to use similar visual exploration strategies adapted to the physiognomic features of each ethnicity in the recognition of Caucasian and Gypsy faces (Hypothesis 2b). Finally, we propose two possible hypotheses related to the perceptual mechanisms implied in the recognition of East Asian faces. In this case, the observers do not know which visual strategy is better suited to the East Asian physiognomic features, given their low degree of contact with this ethnicity. As a result, they could use the ‘default’ strategy used to recognize faces of their own ethnicity, which we expect to be different between Caucasians and Gypsies given their previous social experience and cultural background (Hypothesis 2c). However, it could also be the case that Gypsy and Caucasian observers easily notice that East Asian faces look different from faces of known ethnicities and they then decide to be guided by the most salient physiognomic features of these faces. In this situation, Gypsy and Caucasian observers could employ quite similar visual strategies in recognizing East Asians (Hypothesis 2d). However, due to the lack of contact and experience with this ethnicity, the strategies used would still not be the most adequate for the recognition of these faces (this requires prior perceptual learning mechanisms), giving rise to the cross-ethnic effect. Taken together, we expected to find a 3-way interaction among the observer group (hereinafter called Participant’s Culture), Face Culture and ROI if Hypothesis 2c is supported, and a 2-way interaction between Face Culture and ROI if Hypothesis 2d is supported. Method Participants Twenty Andalusian Gypsies (17 males, 5 females) with ages ranging from 17 to 54 years and twenty Andalusian Caucasians (9 males, 11 females) with ages ranging from 23–52 years participated in the study. All participants had normal or corrected-to-normal vision. Before starting the experiment, participants signed an informed consent form and were treated in accordance with the Helsinki Declaration. All of them were paid 10 € for their collaboration. Stimuli and apparatus The experiment employed sixty front-view photographs of different human faces. Twenty faces belonging to each of the groups (Caucasian, Gypsies, and East Asians) were used as stimuli, with each group containing ten male and ten female faces. The Caucasian and East Asian face stimuli were obtained from the Productive Aging Lab Face Database (Minear & Park, 2004). All of them had previously been categorized as neutral facial expressions in the database. The Gypsy faces were selected according to similar criteria
Fig. 1. Examples of face stimuli (2 genders × 3 cultures). Regions of interest for eye movement analysis (eyes, nose, and mouth) are indicated by the red boxes on the top-left face. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). 44
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
to those applied to Caucasian and East Asian stimuli (sample stimuli are shown in Fig. 1). Eight volunteer participants assessed the emotional facial expression of Gypsy faces, and we included only those faces for which there was an agreement on the neutral expression of the face. All faces were cropped to remove external facial features such as ears, neck, hair, or jewelry, whilst male faces were clean-shaven. All stimuli were displayed in grayscale where lightness and contrast were balanced by using auto-level in PhotoScape. The image size was 350 × 500 pixels, occupying 14.5 × 10.5 cm on the screen. Photos were presented on a white background on a HD Dell 21” monitor, with a refresh rate of 1000 Hz. The experiment was carried out using Experiment Builder 1.9 (SR Research). The viewing distance from the monitor was fixed at 70 cm using a chin rest. Eye movements of the left eye were recorded by Eyelink II head-mounted eye tracking system (SR Research, Mississauga, Canada). The spatial accuracy was superior to 0.5° and the sampling rate was 500 Hz. Calibration and validation of the eye movements were performed twice before beginning the learning and recognition phase using a nine-point fixation procedure. Fixations around blinks and fixations less than 100 ms in duration were not considered in the analysis. Procedure The experimental procedure consisted of two phases: the learning and recognition phase. In the learning phase, participants were explicitly instructed to memorize the faces displayed on the screen. Thirty face stimuli (five female and five male faces per culture) were presented in random order for 4000 ms each. A fixation point was displayed in the middle of the screen for 750 ms. before each face onset. This learning phase was followed by a pause to re-calibrate the eye tracker system, after which participants performed the recognition phase. This second phase consisted of a forced-choice old-new recognition task, where the thirty faces displayed in the learning phase (old faces) and thirty new faces were presented. Presentation of each face was terminated by the participant's response or until a maximum time of 10 s had elapsed. The participant had to indicate whether the face was old (pressing the “left” mouse button) or new (“right” mouse button). There was no feedback on responses and participants were not informed of the ratio of new/ old faces. The order of presentation of the stimuli was random. The whole experiment lasted approximately 15–20 min. Statistical analysis The average proportion of correct responses and average reaction times for correct responses were submitted to separate analyses of variance. A 2 × 3 mixed ANOVA was used for both analyses with Participant’s Culture (Gypsy, Caucasian) as the between-subject factor and Face Culture (Gypsy, Caucasian, East Asian) as the within-subject factor. An a priori statistical power analysis utilizing the G*Power program (Faul, Erdfelder, Lang, & Buchner, 2007) confirmed the adequate statistical power for these analyses. Total sample sizes exceeding 32 participants would have power higher than 0.95, given a p-value of 0.05, a medium effect size of 0.3, and a correlation among repeated measures of 0.5. With respect to eye movement analysis, following the standard procedure in the face recognition literature (Althoff & Cohen, 1999; Blais et al., 2008), we divided the face stimuli into three non-overlapping regions corresponding to the principal features: eyes, nose, and mouth (see Fig. 1). With these 3 regions of interest (ROIs), we calculated the proportion of fixations and proportion of fixation time on each ROI compared with the total number of fixations and total time per trial, respectively. The learning and recognition phases were analyzed separately. In both phases, the average proportion of fixations and average proportion of fixation time were submitted to two separate 2 × 3 × 3 mixed ANOVAs with Participant’s Culture (Gypsy, Caucasian) as the between-subject factor and Face Culture (Gypsy, Caucasian, and East Asian) and ROI (Eyes, Nose, and Mouth) as within-subject factors. The a priori statistical power analysis also confirmed an adequate statistical power for the eye movement analysis. Total sample sizes exceeding 16 participants have power higher than 0.95. In addition, we decided to use Pearson’s correlations to analyze the relationship between the eye movement variables (number of fixations and fixation time on each region in the learning and recognition phases) and the level of accuracy on the face recognition task. Given the physiognomic differences between ethnicities, we carried out separate correlation analyses for each type of face (Gypsy, Caucasian, and East Asian). Results Two Gypsy and three Caucasian participants were discarded due to problems with the eye tracking calibration or poor recording quality. Thus, the final sample for the analysis was composed of 18 Gypsies and 17 Caucasians. Preliminary analyses Prior to the face recognition task, each participant had to answer a question about the degree of contact with the other cultures included in the task, which allows us to assess the actual degree of contact between cultures in our sample. Participants evaluated the degree of contact with Gypsies, Caucasians, and East Asians on a scale from 0 (none at all) to 4 (very high). Both groups (Gypsies and Caucasians) exhibited high/medium contact scores with Caucasians and Gypsies (Gypsy participants: Mcaucasians = 3.85, Mgypsies = 4; Caucasian participants: Mcaucasians = 4, Mgypsies = 2.05), and very low scores with East Asians (Caucasian participants: Masians = 0.95, Gypsy participants: Masians = 0.6). In order to examine these scores in more detail, we conducted a mixed ANOVA on the degree of contact with Participant’s Culture (Gypsy, Caucasian) as a between-subject factor and Other Group’s Culture (Gypsy, Caucasian, East Asian) as a within-subject factor. The ANOVA revealed significant main effects of Participant’s Culture (F (2, 76) = 32.02, 45
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Fig. 2. Average proportion of correct responses for each face category. Vertical bars represent the standard error of the mean.
p < 0.001, η2p = 0.46) and Other Group’s Culture (F (2, 76) = 535.90, p < 0.001, η2p = 0.93), along with an interaction between both variables (F (2, 76) = 82.62, p < 0.001, η2p = 0.68). Further exploration of this interaction revealed, as expected, a significantly lower degree of contact with East Asians compared with Caucasians and Gypsies for both participant groups (all p < 0.05). In addition, there were no significant differences between Caucasian and Gypsy participants in terms of the degree of contact with Gypsies and East Asians (p > 0.05), but significant differences were found in the degree of contact with Caucasians (p < 0.05). While both Gypsy and Caucasian participants showed a high and similar degree of contact with Caucasians, the Caucasian participants showed a lower degree of contact with Gypsies than the Gypsy participants. It is therefore important to note that, according to perceptual learning theories (Gibson, 1969; Goldinger et al., 2009), support for the proposed hypotheses related to intergroup contact could be particularly evident for the Gypsy group, which have a low degree of contact with Asians, and a high and similar degree of contact with Gypsy and Caucasian cultures. In any case, a better recognition of Gypsy and Caucasian faces compared with East Asian faces was also expected in the Caucasian group, given the differences in the degree of contact with these cultures. Face recognition performance Firstly, Kolmogorov–Smirnov and Levene’s tests verified normality of the distribution and homogeneity of the variances of proportion of correct responses and reaction times for correct responses (all p’s > 0.05). The average proportion of correct responses for each face category are presented in Fig. 2. The 2 × 3 ANOVA conducted on proportion of correct responses revealed a main effect of Face Culture (F (2, 66) = 29.49, p < 0.001, η2p = 0.47). Both the main effect of Participant’s Culture and the interaction were non-significant (F’s < 1). Post hoc analysis of Face Culture showed that Gypsy (M = 0.86) and Caucasian faces (M = 0.78) were better recognized than East Asian faces (M = 0.72). Moreover, Gypsy faces were better recognized than Caucasian faces (p’s < 0.05).
Fig. 3. Average reaction times for correct responses for each face category. Vertical bars represent the standard error of the mean. 46
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Fig. 4. Mean proportion of number of fixations in each region of interest (eyes, nose, and mouth) for Caucasian and Gypsy participants. Vertical bars represent the standard error of the mean.
With respect to the reaction times, the averages for each face category are shown in Fig. 3. The 2 × 3 ANOVA revealed a main effect of Face Culture (F (2, 66) = 7.99, p = 0.001, η2p = 0.19). Participant’s Culture and the interaction were both non-significant (F’s < 1). Post hoc comparisons showed that recognition was slower for both East Asian (M = 1904 ms) and Caucasian faces (M = 1777 ms) compared with Gypsy faces (M = 1669 ms) (p’s < 0.05). As expected, there were no differences between Gypsy and Caucasian observers in recognizing faces (Hypothesis 1a), but both observer groups were poorest at recognizing the faces of East Asians (Hypothesis 1b). Eye movement analysis For the learning task, the 2 × 3 × 3 ANOVAs conducted on the number of fixations and fixation time revealed a main effect of ROI (number, F(2, 66) = 19.26, p < 0.001, η2p = 0.37; time, F(2, 66) = 18.75, p < 0.001, η2p = 0.36) along with significant interactions between Participant’s Culture × ROI (number of fix.: F(2, 66) = 4.28, p = 0.017, η2p = 0.12; fixation time: F(2, 66) = 4.26, p = 0.018, η2p = 0.11) and ROI × Face Culture (number of fix.: F(2, 66) = 6.45, p < 0.001, η2p = 0.16; fixation time: F(2, 66) = 7.03, p < 0.001, η2p = 0.17). No other significant effects or interactions were found. Post hoc analysis of the Participant’s Culture × ROI interaction indicated that Gypsies made a higher number of fixations and showed a longer fixation time on the eyes (Mnumber = 0.27, Mtime = 0.25) than on the nose (Mnumber = 0.16, Mtime = 0.13) and mouth regions (Mnumber = 0.13, Mtime = 0.11) (all p’s = 0.05). Caucasian participants showed a higher number of fixations and a longer fixation time on the eyes (Mnumber = 0.23, Mtime = 0.19) and nose (Mnumber = 0.24, Mtime = 0.19) than on the mouth region (Mnumber = 0.09, Mtime = 0.08) (all p’s = 0.05). Finally, Caucasians showed more fixations on the nose than Gypsies (p < 0.05). The results of the ROI × Participant’s Culture interaction in terms of fixation number are presented in Fig. 4. With respect to the ROI × Face Culture interaction, the eye region received fewer fixations and a lower fixation time for Gypsy faces (Mnumber = 0.22, Mtime = 0.19) than East Asian (Mnumber = 0.26, Mtime = 0.23) and Caucasian faces (Mnumber = 0.27, Mtime = 0.24) (all p’s = 0.05). The nose received fewer fixations and a lower fixation time in the case of East Asian faces (Mnumber = 0.17, Mtime = 0.13) compared with Gypsy (Mnumber = 0.21, Mtime = 0.20) and Caucasian faces (Mnumber = 0.2, Mtime = 0.20) (all p’s = 0.05). The mouth received fewer fixations and a lower fixation time for Caucasian faces (Mnumber = 0.09, Mtime = 0.09) compared with the case of Gypsy faces (Mnumber = 0.13, Mtime = 0.12). As expected, different visual strategies in the recognition of Caucasian, Gypsy, and East Asian faces were found (Hypothesis 2a). However, contrary to Hypothesis 2b, there were visual differences between Gypsy and Caucasian observers in the recognition of both Gypsy and Caucasian faces. Finally, visual differences between observer groups were also found in Gypsy faces, thus supporting Hypothesis 2c as opposed to Hypothesis 2d, but a more detailed discussion is necessary to understand the nature of these results since Hypothesis 2b was rejected (see Discussion section). For the recognition task, ANOVAs conducted on the number of fixations and fixation time revealed main effects of ROI (number of fix.: F(2, 66) = 9.11, p < 0.001, η2p = 0.22; fixation time: F(2, 66) = 6.8, p < 0.002, η2p = 0.17) and Face Culture (number of fix.: F (2, 66) = 5.87, p = 0.005, η2p = 0.15; fixation time: F(2, 66) = 5.45, p = 0.006, η2p = 0.14), along with an interaction between Face Culture and ROI (number of fix.: F(2, 66) = 3.26, p = 0.012; η2p = 0.09; fixation time: F(2, 66) = 5.44, p < 0.001, η2p = 0.14). There are no other significant effects or interactions. Post hoc analysis of the Face Culture × ROI interaction showed that the nose region had a lower number of fixations and a lower fixation time in East Asian faces (Mnumber = 0.25, Mtime = 0.16) compared with Caucasian (Mnumber = 0.29, Mtime = 0.21) and Gypsy faces (Mnumber = 0.29, Mtime = 0.21) (all p’s = 0.05). The mouth region had fewer fixations and a lower fixation time in East Asian (Mnumber = 0.14, Mtime = 0.10) and Caucasian faces (Mnumber = 0.13, Mtime = 0.09) than in Gypsy faces (Mnumber = 0.17, Mtime = 0.13). Similar to the results of the learning task, participants used different visual strategies in the recognition of Caucasian, Gypsy, and East Asian faces (Hypothesis 2a). No differences were found with respect to the Participant’s Culture. 47
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Finally, we analyzed whether the number of fixations and fixation time on each region were related to the success on the face recognition task. There was a significant correlation between accuracy and the number of fixations on the nose in East Asian faces (r = 0.35; p = 0.04). Thus, the fewer the fixations on the nose, the lower the accuracy in face recognition. None of the remaining correlations reached statistical significance. Discussion The present study aimed to examine the influence of the degree of contact with other ethnicities on the cross-ethnic effect and to explore how changes in the visual exploration strategies of the faces are related to this phenomenon. The cross-ethnic effect is described as the poorer recognition of faces of a ethnic group other than our own (Meissner & Brigham, 2001; Sporer, 2001). As an explanation for this effect, the contact hypothesis assumes that visual expertise plays an important role in face recognition and that as a person gains more experience with an ethnicity he/she will be more accurate in recognizing faces of that ethnicity (Caldara et al., 2010; Goldinger et al., 2009). Therefore, the cross-ethnic effect could be due to the fact that the lack of contact does not allow for adjusting the perceptual strategies to the particular physiognomy of the other ethnicity. In order to test the contact hypothesis and the associated perceptual mechanisms, we compared the performance and visual exploration strategies of two ethnic groups that live in close proximity (Andalusian Gypsies and Andalusian Caucasians) in recognizing faces of unfamiliar (Asian faces) and familiar ethnicities (Caucasian and Gypsy faces). In accord with Hypothesis 1a and 1b, there were no differences between Gypsy and Caucasian participants in the recognition of Gypsy and Caucasian faces, and East Asian faces were the most poorly recognized (less accurate and slower responses). Thus, the cross-ethnic effect was only found with faces of cultures with which the participants had no contact. Even Caucasians, who had a moderate degree of contact with the Gypsy ethnicity (see Results section: Preliminary analyses), recognized Gypsy faces at a similar level to Caucasian faces. Based on the intergroup contact hypotheses and perceptual learning mechanisms (Gibson, 1969; Goldinger et al., 2009; Goldstein & Chance, 1985), both Gypsy and Caucasian participants would be expected to show better performance in recognizing Gypsy and Caucasians faces compared with Asians faces, but this better performance should be particularly evident in the Gypsy group, given its higher degree of contact with Caucasians. This fact raises the possibility that a moderate degree of contact with the other ethnicity could be sufficient to achieve adequate recognition. However, it must be taken into account that, in the current study, the similar performance for moderate and high contact could be explained in terms of the low level of difficulty of the task (with an 86% success rate in Gypsy faces), which may overshadow the differences between Gypsies and Caucasians. In any case, faces of cultures with high and moderate contact were better recognized than those with low contact, which suggests that our results are in accordance with the contact hypothesis as a possible explanation for the cross-ethnic facial recognition effect. With respect to the eye tracking data, the results supported Hypothesis 2a, showing differences in the perceptual strategies employed to recognize the faces of each ethnicity, independently of the observer’s culture. The mouth received more fixations and fixation time in the Gypsy faces compared with the remaining cultures in the learning phase. The eyes received fewer fixations and less fixation time for the Gypsy faces compared with the other faces in the learning phase, whilst the nose received fewer fixations and less fixation time for the East Asian faces than for the other faces in both the learning and recognition phases. Regarding the culture of the observer, Caucasians and Gypsies employed different visual exploration strategies in the learning phase, regardless of the type of face to be recognized. Gypsy observer’s fixations were more focused on the eye region than the rest of the regions, while Caucasians fixated more on the nose region than Gypsies, showing a similar proportion of fixations on the eyes and nose. These between-group differences appear to support Hypothesis 2c (as opposed to Hypothesis 2d), but run counter to Hypothesis 2b, since both ethnicities employed different visual strategies in the recognition of both familiar and unfamiliar ethnic groups. However, a deeper analysis is necessary in order to better understand these findings. We therefore discuss this issue in more detail below. Changes in the eye fixation pattern depending on the face culture may be explained in terms of physiognomic differences. There are marked facial physical differences across ethnicities that are key factors in the perceptual discrimination of faces (Kelly et al., 2005). Thus, it is not surprising to think that differences in physiognomic facial features lead to different visual exploration strategies. More importantly for our research, although face culture was associated with changes in the visual exploration pattern, Gypsy and Caucasian participants maintained different perceptual strategies independently of the type of face observed during the learning phase. Differences in the facial features scanned as a function of the observer’s ethnicity are in line with the findings of the studies of Blais et al. (2008), Caldara et al. (2010) and Hills and Pake (2013). However, these results do not fit well with the hypothesis that the perceptual strategies depend on the degree of contact with other ethnic groups (contact hypothesis), since different exploration patterns between Gypsies and Caucasians were not only observed with faces from unfamiliar ethnicities, but also with faces from ethnicities with which they have a high level of contact. In our study, the perceptual mechanisms that are characteristic of each ethnicity seem to have great robustness. The visual exploration strategies were adapted to the physiognomic features of the face to recognize, but certain characteristics associated with the observer’s ethnicity remained stable. Previous literature has pointed to the possibility that cultural force could explain the use of different visual exploration strategies in perceptual tasks (Chua, Boland, & Nisbett, 2005; Nisbett & Miyamoto, 2005). For example, Blais et al. (2008) have provided evidence that Western Caucasians use analytic perceptual strategies, whereas East Asians use holistic strategies for processing faces. In our study, the cultural background of the Gypsies might be guiding the eye movement patterns in processing faces. Gypsies are described as a culture strongly influenced by tradition and spirituality, according to the Schwartz Value Survey (Rzeszewska & Catena, 2012). Not without reason, the eyes have a special status in Gypsy culture, appearing frequently in their folklore, oral traditions, stories, and songs. Thus, the predominance of gazing at eyes in the Gypsy group could be the result of the cultural forces that have shaped the eye movement strategies developed by this ethnic group for hundreds of years. 48
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Our results reveal how participants from distinct cultures achieve similar face recognition performance despite using different scan paths. These findings appear to run counter to the idea that the use of inadequate perceptual mechanisms could be responsible for the cross-ethnic effect. However, along with the perceptual differences between the observer groups, we have also observed changes that are common to both ethnicities in terms of the visual strategies used to recognize the faces of each ethnicity. In the recognition of familiar ethnicities, the previous social experience with the other group would lead to the acquisition of certain changes in the visual exploration patterns, adapting them to the physiognomy of the other ethnic group. These changes could be sufficient for adequate recognition. Thus, different ethnicities could start from different perceptual mechanisms associated with their cultural baggage, and adapt these to the physiognomy of familiar ethnicities, without having to completely modify the characteristic pattern of their own ethnicity. In the case of recognizing faces of unfamiliar ethnicities, such as East Asians, the visual patterns would also be guided by the most salient physiognomy features, but the lack of previous experience with this ethnicity would lead to the use of less successful undeveloped patterns. For example, as was mentioned in the introduction, East Asians use holistic strategies to organize their visual environment, with the nose (the center of the face) being the optimal region to integrate information holistically during face recognition (Blais et al., 2008; Miyamoto et al., 2006). In this vein, Blais et al. (2008) have shown that the nose is the most effective and diagnostic facial feature that is attended to by East Asian observers in order to recognize faces from their own ethnicity. According to the correlational analysis conducted between accuracy and eye movements in our study, a smaller number of fixations on the nose in East Asian faces were associated with lower accuracy on the face recognition task. This finding could be compatible with the fact that East Asian faces were those most poorly recognized since the participants did not correctly apply their visual strategies to the faces of this ethnicity. Another possible explanation for achieving similar performance through the use of different visual strategies comes from a Blais et al.’s (2008) proposal. It is possible that, in spite of the differences in the perceptual strategies employed, Gypsies and Caucasians construct similar mental representations of the faces of those cultures with which they have a high degree of contact, which would lead to good recognition of these faces. Of course, the mental representations of the faces are certainly developed from eye movements and visual attention patterns, but such patterns do not completely constitute direct evidence of these mental representations (Blais et al., 2008). Thus, experience with the other ethnicities could improve face recognition performance by modifying the mental representations of faces and the processes of encoding and retrieval associated with them, without the need to make considerable changes in the visual exploration strategies. In any case, future research should aim to clarify some of these issues. Among the limitations of our study, it is important to note that the absence of differences in face recognition accuracy between Caucasians and Gypsies could also be explained by the smaller physiognomic changes between them compared with other ethnicities such as East Asians. Although studies in the literature have presented evidence of the cross-ethnic effect in numerous different ethnicities, no previous studies have focused on this effect in Gypsies. Our results demonstrated that Caucasians and Gypsies behave differently at the perceptual level, which supports the idea of there being cultural particularities of each ethnic group during the face recognition process. However, more research is needed in order to study the strength of the cross-ethnic effect between Gypsies and Caucasians without cross contact. In addition, Caucasian participants were able to recognize Gypsy faces as well as Gypsy participants, in spite of having only a moderate degree of contact. As previously mentioned, in the current research, the lack of differences between a moderate and high degree of contact could be explained by the low difficulty of our recognition task. Nevertheless, this result suggests that an interesting future line of research could be to explore the degree of contact necessary to achieve improvements in recognition. Finally, the contact hypothesis predicts that the recognition of faces from other cultures is associated with both the quantity and quality of the contact. In our case, we asked that the participants interpret degree of contact as “quantity”. But, it is possible that the quality of the contact could also have had an impact on our results. We therefore believe that the results of the present study are relevant and shed new light on the cross-ethnic effect and the distinctiveness of the Gypsy ethnicity, but further research should corroborate these findings. Moreover, it would be of great interest to apply this line of investigation to real situations that occur in our everyday lives. From a practical point of view, the study of how the cross-ethnic effect could have an impact on the accuracy of face recognition may have important implications in contexts such as eyewitness identification, or screening for individuals at borders, airports, or government buildings. Conclusion The results reported here provide support for the contact hypothesis as an explanation of the cross-ethnic effect. Face recognition performance across ethnicities was similar when there was sufficient contact with members of the other ethnic group. However, ethnicities with less proximity (in our case East Asians) were the most poorly recognized. The cross-ethnic effect only appeared when the observer did not have sufficient experience with members of the other ethnic group. Although there were no between-group differences in the recognition of Gypsy and Caucasian faces, importantly, differences appeared in the visual scan strategies employed. In particular, Gypsies focused their attention primarily on the eyes, while Caucasians distributed their attention equally between the eyes and nose. Thus, we demonstrated how cultural factors could underlie visual perceptual processes in spite of the intense social experience with other ethnicities. Finally, it is worth emphasizing that given the relatively few studies focused on face processing in the Gypsy culture, our work may serve as a good starting point for further research. Acknowledgments This work was funded by Spanish MICINNPSI2012-39292 grant to A. Catena. Alberto Megías is supported by a Juan de la Cierva postdoctoral fellowship from Spanish MINECO (FJCI-2015-25600). 49
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
References Althoff, R. R., & Cohen, N. J. (1999). Eye-movement-based memory effect: A reprocessing effect in face perception. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25(4), 997. Anderson, N. B., & Nickerson, K. J. (2005). Genes, race, and psychology in the genome era: An introduction. American Psychologist, 60(1), 5. Blais, C., Jack, R. E., Scheepers, C., Fiset, D., & Caldara, R. (2008). Culture shapes how we look at faces. PLoS One, 3(8), e3022. Caldara, R., Zhou, X., & Miellet, S. (2010). Putting culture under the ‘spotlight' reveals universal information use for face recognition. PLoS One, 5(3), e9708. Chua, H. F., Boland, J. E., & Nisbett, R. E. (2005). Cultural variation in eye movements during scene perception. Proceedings of the National Academy of Sciences of the United States of America, 102(35), 12629–12633. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. Feinman, S., & Entwisle, D. R. (1976). Children’s ability to recognize other children’s faces. Child Development, 506–510. Gibson, E. J. (1969). Principles of perceptual learning and development. New York: Appleton-Century-Crofts. Goldstein, A. G., & Chance, J. (1985). Effects of training on Japanese face recognition: Reduction of the other-race effect. Bulletin of the Psychonomic Society, 23, 211–214. Goldinger, S. D., He, Y., & Papesh, M. H. (2009). Deficits in cross-race face learning: Insights from eye movements and pupillometry. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35(5), 1105. Groner, R., Walder, F., & Groner, M. (1984). Looking at faces: Local and global aspects of scanpaths. Advances in Psychology, 22, 523–533. Hills, P. J., & Pake, J. M. (2013). Eye-tracking the own-race bias in face recognition: Revealing the perceptual and socio-cognitive mechanisms. Cognition, 129(3), 586–597. Henderson, J. M., Falk, R. J., Minut, S., Dyer, F. C., & Mahadevan, S. (2001). Gaze control for face learning and recognition by humans and machines. In T. Shipley, & P. Kellman (Eds.). From fragments to objects: Segmentation processes in vision (pp. 463–482). Amsterdam: Elsevier. Henderson, J. M., Williams, C. C., & Falk, R. J. (2005). Eye movements are functional during face learning. Memory & Cognition, 33(1), 98–106. Kelly, D. J., Quinn, P. C., Slater, A. M., Lee, K., Gibson, A., Smith, M., ... Pascalis, O. (2005). Three‐month‐olds, but not newborns, prefer own-race faces. Developmental Science, 8(6), F31–F36. MacLin, O. H., & Malpass, R. S. (2003). The ambiguous-race face illusion. Perception, 32(2), 249–252. 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, & Law, 7, 3–35. Minear, M., & Park, D. C. (2004). A lifespan database of adult facial stimuli. Behavior Research Methods Instruments and Computers, 36, 630–633. Miyamoto, Y., Nisbett, R. E., & Masuda, T. (2006). Culture and the physical environment: Holistic versus analytic perceptual affordances. Psychological Science, 17(2), 113–119. Ng, W. J., & Lindsay, R. C. (1994). Cross-race facial recognition failure of the contact hypothesis. Journal of Cross-Cultural Psychology, 25(2), 217–232. Nisbett, R. E., & Miyamoto, Y. (2005). The influence of culture: Holistic versus analytic perception. Trends in Cognitive Science, 9, 467–473. Rzeszewska, I., & Catena, A. (2012). The eye movement strategies in face perception: A cross-cultural study(Master’s thesis). Spain: Universidad de Granada. Sporer, S. L. (2001). Recognizing faces of other ethnic groups: An integration of theories. Psychology, Public Policy, and Law, 7(1), 36. Sternberg, R. J., Grigorenko, E. L., & Kidd, K. K. (2005). Intelligence, race, and genetics. American Psychologist, 60(1), 46. Vizioli, L., Foreman, K., Rousselet, A. G., & Caldara, R. (2010). Inverting faces elicits sensitivity to race on the N170 component: A cross-cultural study. Journal of Vision, 10(1) 15.1–23. Vrij, A., & Winkel, F. W. (1989). Recognition of own and other-race faces: The effects of quality versus quantity of cross-cultural interaction. Journal of Police and Criminal Psychology, 5(2), 24–28. Witzig, R. (1996). The medicalization of race: Scientific legitimization of a flawed social construct. Annals of Internal Medicine, 125(8), 675–679.
50
Contents lists available at ScienceDirect
International Journal of Intercultural Relations journal homepage: www.elsevier.com/locate/ijintrel
Influence of cross-ethnic social experience on face recognition accuracy and the visual perceptual strategies involved
T
⁎
Alberto Megiasa, , Iga Rzeszewskaa, Luis Aguadob, Andrés Catenaa a b
Learning, Emotion and Decision Group, Mind, Brain, and Behavior Research Center, University of Granada, Granada, Spain Universidad Complutense, Madrid, Spain
A R T IC LE I N F O
ABS TRA CT
Keywords: Intercultural factors Social cognition Face recognition Cross-race effect Eye movements Gypsy ethnicity
The cross-ethnic effect (in the literature, usually termed the cross-race effect) is defined as the greater difficulty in recognizing faces of other ethnicities compared with faces of one’s own. The aims of the present research were: 1) to test the hypothesis that the cross-ethnic effect is due to lack of contact with the other ethnicity. 2) to study possible differences in the perceptual mechanisms employed in face recognition as a function of the contact degree between ethnicities, which may be the basis of the cross-ethnic effect. We compared two ethnic groups with a high degree of contact, but different identities and cultural values: Andalusian Gypsies and Andalusian Caucasians. Both groups had to recognize a set of East Asian, Caucasian, and Gypsy faces while eye movements were monitored. In accordance with the contact hypothesis, our results revealed no differences between Gypsies and Caucasians observers in face recognition success. However, East Asian faces were more poorly recognized than Gypsies and Caucasian faces by both observer groups. With respect to the perceptual strategies, despite achieving similar face recognition performance, Caucasian and Gypsy observers employed different visual exploration strategies. Gypsies focused their attention on the eyes, while Caucasians fixated more on the nose than Gypsies. Our results support the contact hypothesis as an explanation for the cross-ethnic effect, and show how cultural factors imply differences in perceptual strategies even between close ethnic groups.
Introduction People are better at identifying faces of their own ethnicity than those of other ethnicities. This phenomenon is known as the cross-race effect and has been widely studied and replicated in recent decades (MacLin & Malpass, 2003; Meissner & Brigham, 2001; Sporer, 2001). The cross-race effect has usually been conceptualized using the term race; however, we prefer to use the term ethnic, given that race in humans refers to a biological construct created from prevailing social perceptions without scientific evidence from biology, genetics, or physical anthropology (Anderson & Nickerson, 2005; Sternberg, Grigorenko, & Kidd, 2005; Witzig, 1996). The present research aims to study how the degree of contact with other ethnicities modulates the cross-race effect (henceforth crossethnic effect) in face recognition and the visual perceptual mechanisms involved in this process. Several social and cognitive hypotheses have been proposed to account for the cross-ethnic effect, with social attitudes, physiognomic variability between ethnicities, and inter-ethnic contact being the most prominent explanations (Meissner & Brigham, 2001; Ng & Lindsay, 1994). Although there is no total consensus across studies, hypotheses based on inter-ethnic contact have received greater support (Goldinger, He, & Papesh, 2009). The contact hypothesis predicts that the recognition accuracy of other ⁎
Corresponding author at: Centro de Investigación Mente, Cerebro y Comportamiento, Universidad de Granada, Campus de Cartuja s/n, 18071, Granada, Spain. E-mail address: [email protected] (A. Megias).
https://doi.org/10.1016/j.ijintrel.2018.04.008 Received 19 July 2017; Received in revised form 18 April 2018; Accepted 30 April 2018 Available online 06 May 2018 0147-1767/ © 2018 Elsevier Ltd. All rights reserved.
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
ethnicity members is directly related to the quantity and quality of contact with people of that ethnic group (Goldstein & Chance, 1985; Vrij & Winkel, 1989). This approach is empirically supported by the findings of Meissner and Brigham’s (2001) meta-analysis, where authors across 29 independent studies found that inter-ethnic contact plays a significant mediating role in the cross-ethnic effect. For example, studies in ecological contexts have shown that people living in integrated neighbourhoods are better able to discriminate other ethnic faces than those living in segregated neighbourhoods (Feinman & Entwisle, 1976). The effect of inter-ethnic contact level on face recognition could be explained by perceptual learning mechanisms (Goldinger et al., 2009). Gibson (1969) defined perceptual learning as “an increase in the ability to extract information from the environment, as a result of practice and experience with the stimulation coming from it”. According to this theory, and given the existence of physiognomic differences between ethnic groups (Kelly et al., 2005), an increase in the degree of exposure to members of other ethnicities should improve the ability to discriminate between the characteristic facial features of those ethnicities. Thus, a higher degree of contact with the other ethnicity would teach people which perceptual strategies they should use to properly process the faces of that ethnic group. Previous literature has shown that perceptual mechanisms and visual attention patterns employed during face recognition differ as a function of the ethnicity of the observer (Blais, Jack, Scheepers, Fiset, & Caldara, 2008; Hills & Pake, 2013). Visual attention in face processing tends to be restricted to internal facial features, with a systematic visual scan pattern over the eyes, nose, and mouth (e.g., Althoff & Cohen, 1999; Groner, Walder, & Groner, 1984). It has been estimated that approximately 90% of fixation time is spent on these face regions (Henderson, Falk, Minut, Dyer, & Mahadevan, 2001). However, ethnicity, cultural background, and social experience lead to a different weighting of each one of these regions. For example, Blais et al. (2008) showed that Eastern cultures (e.g. East Asians) use more holistic perceptual strategies, with more fixations directed toward the nose (the central region of the face), whilst people of Western Caucasian cultures focus more on the eye region, performing scan paths that involve more analytic strategies (see Miyamoto, Nisbett, & Masuda, 2006, for a study of holistic and analytic perceptual differences between cultures). Similar results have been obtained by Caldara, Zhou, and Miellet (2010), Goldinger et al. (2009) and Hills and Pake (2013), showing that observers from different ethnicities looked at different facial features when they tried to recognize the same faces. Given these differences in the visual scan path and the particular physiognomy of each ethnicity, deficits in cross-ethnic recognition may be due to observers not attending to the most diagnostic facial features of other ethnicities (Hills & Pake, 2013). For instance, an individual may have learnt to use a specific perceptual strategy to recognise faces because this strategy fits well with the physiognomic features of their ethnicity and it is part of their cultural background and social experience. However, this strategy will be problematic for recognising faces from other unfamiliar ethnicities with different physiognomic features, which would lead to poorer performance, and could thus explain the cross-ethnic effect. On the other hand, according to the contact hypothesis, a greater degree of contact with other ethnicities would help adapt the visual exploration strategies to the physiognomy of each ethnicity and thus improve facial recognition accuracy. The aim of this research was to advance our understanding of the perceptual mechanisms underlying the contact hypothesis and their implications for the cross-ethnic effect. We focused on two ethnic groups: Andalusian Caucasians and Andalusian Gypsies. These groups have a high degree of contact, but they differ considerably in terms of their identity and cultural values. Caucasians and Gypsies (Romanies) in Andalusia have shared the same geographical area since the XV century. Approximately 300,000 Gypsies live in this region of Spain, and in Granada – the city where this study was conducted – the Gypsy community is particularly well integrated. There are two historic heritage neighborhoods (Albaicín and Sacromonte) with a large Gypsy population where Andalusian Gypsies and Caucasians co-exist and live in harmony with tourists from many different countries. However, in spite of this co-existence, in many aspects the Gypsy identity remains significantly different from other Andalusians. Gypsies have assimilated a set of cultural factors, traditions, affiliation, and common history that has defined them as a distinct ethnic group for many centuries. To achieve our objective, an Andalusian Gypsy group and an Andalusian Caucasian group performed a standard face recognition task adapted to the study of the cross-ethnic effect (Ng & Lindsay, 1994; Vizioli, Foreman, Rousselet, & Caldara, 2010). In this task, Andalusian Gypsies and Caucasian participants had to view a set of faces from an unfamiliar ethnicity (East Asian faces) and from their own ethnicities (Caucasian and Gypsy faces) under instructions to memorize them for a later recognition test. While the participants were viewing the faces, eye movements were continuously monitored by Eyelink II eye tracking system. Eye movements are a good index of visual attention allocation and play an important functional role in face processing (Henderson, Williams, & Falk, 2005). For example, longer fixation times on specific facial features help to encode, retain, and better retrieve these features, ensuring greater recognition accuracy (Goldinger et al., 2009). In accord with previous studies in the literature (Althoff & Cohen, 1999; Blais et al., 2008), we evaluated the number and duration of fixations on the three face regions that are critical for face processing (eyes, nose, and mouth). We had two main objectives: 1) to test the hypothesis that the cross-ethnic effect is due to the lack of contact with the other ethnicity. Faces from unfamiliar ethnic groups (in our case, East Asians) should be recognized more poorly than faces from their own ethnicity (cross-ethnic effect). On the other hand, the cross-ethnic effect should not occur if the observer has sufficient experience with members of the other ethnicities. Given that Andalusian Gypsies and Caucasians live in close proximity (high or moderate degree of contact), we hypothesized that there will be no differences between Gypsy and Caucasian observers in recognizing faces (Hypothesis 1a). However, both should be particularly poor at recognizing the faces of East Asians, an unfamiliar ethnicity for both observer groups (Hypothesis 1b). Therefore, we expect to find a main effect of the type of face to be recognized (hereinafter called Face Culture). 2) As a second objective, we explore the possible differences in the perceptual mechanisms employed by Gypsies and Caucasians in face recognition of familiar and unfamiliar ethnicities. Differences in the visual exploration strategies as a function of the degree of contact with the other ethnicity (contact hypothesis) could underlie the cross-ethnic effect. First, we expect to find distinct visual strategies in the recognition of Caucasian, Gypsy, and East Asian faces because of the physiognomic differences between them (Hypothesis 2a). Second, due to the degree of contact between Gypsies and Caucasians, both ethnicities should have 43
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
learnt to use similar visual exploration strategies adapted to the physiognomic features of each ethnicity in the recognition of Caucasian and Gypsy faces (Hypothesis 2b). Finally, we propose two possible hypotheses related to the perceptual mechanisms implied in the recognition of East Asian faces. In this case, the observers do not know which visual strategy is better suited to the East Asian physiognomic features, given their low degree of contact with this ethnicity. As a result, they could use the ‘default’ strategy used to recognize faces of their own ethnicity, which we expect to be different between Caucasians and Gypsies given their previous social experience and cultural background (Hypothesis 2c). However, it could also be the case that Gypsy and Caucasian observers easily notice that East Asian faces look different from faces of known ethnicities and they then decide to be guided by the most salient physiognomic features of these faces. In this situation, Gypsy and Caucasian observers could employ quite similar visual strategies in recognizing East Asians (Hypothesis 2d). However, due to the lack of contact and experience with this ethnicity, the strategies used would still not be the most adequate for the recognition of these faces (this requires prior perceptual learning mechanisms), giving rise to the cross-ethnic effect. Taken together, we expected to find a 3-way interaction among the observer group (hereinafter called Participant’s Culture), Face Culture and ROI if Hypothesis 2c is supported, and a 2-way interaction between Face Culture and ROI if Hypothesis 2d is supported. Method Participants Twenty Andalusian Gypsies (17 males, 5 females) with ages ranging from 17 to 54 years and twenty Andalusian Caucasians (9 males, 11 females) with ages ranging from 23–52 years participated in the study. All participants had normal or corrected-to-normal vision. Before starting the experiment, participants signed an informed consent form and were treated in accordance with the Helsinki Declaration. All of them were paid 10 € for their collaboration. Stimuli and apparatus The experiment employed sixty front-view photographs of different human faces. Twenty faces belonging to each of the groups (Caucasian, Gypsies, and East Asians) were used as stimuli, with each group containing ten male and ten female faces. The Caucasian and East Asian face stimuli were obtained from the Productive Aging Lab Face Database (Minear & Park, 2004). All of them had previously been categorized as neutral facial expressions in the database. The Gypsy faces were selected according to similar criteria
Fig. 1. Examples of face stimuli (2 genders × 3 cultures). Regions of interest for eye movement analysis (eyes, nose, and mouth) are indicated by the red boxes on the top-left face. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). 44
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
to those applied to Caucasian and East Asian stimuli (sample stimuli are shown in Fig. 1). Eight volunteer participants assessed the emotional facial expression of Gypsy faces, and we included only those faces for which there was an agreement on the neutral expression of the face. All faces were cropped to remove external facial features such as ears, neck, hair, or jewelry, whilst male faces were clean-shaven. All stimuli were displayed in grayscale where lightness and contrast were balanced by using auto-level in PhotoScape. The image size was 350 × 500 pixels, occupying 14.5 × 10.5 cm on the screen. Photos were presented on a white background on a HD Dell 21” monitor, with a refresh rate of 1000 Hz. The experiment was carried out using Experiment Builder 1.9 (SR Research). The viewing distance from the monitor was fixed at 70 cm using a chin rest. Eye movements of the left eye were recorded by Eyelink II head-mounted eye tracking system (SR Research, Mississauga, Canada). The spatial accuracy was superior to 0.5° and the sampling rate was 500 Hz. Calibration and validation of the eye movements were performed twice before beginning the learning and recognition phase using a nine-point fixation procedure. Fixations around blinks and fixations less than 100 ms in duration were not considered in the analysis. Procedure The experimental procedure consisted of two phases: the learning and recognition phase. In the learning phase, participants were explicitly instructed to memorize the faces displayed on the screen. Thirty face stimuli (five female and five male faces per culture) were presented in random order for 4000 ms each. A fixation point was displayed in the middle of the screen for 750 ms. before each face onset. This learning phase was followed by a pause to re-calibrate the eye tracker system, after which participants performed the recognition phase. This second phase consisted of a forced-choice old-new recognition task, where the thirty faces displayed in the learning phase (old faces) and thirty new faces were presented. Presentation of each face was terminated by the participant's response or until a maximum time of 10 s had elapsed. The participant had to indicate whether the face was old (pressing the “left” mouse button) or new (“right” mouse button). There was no feedback on responses and participants were not informed of the ratio of new/ old faces. The order of presentation of the stimuli was random. The whole experiment lasted approximately 15–20 min. Statistical analysis The average proportion of correct responses and average reaction times for correct responses were submitted to separate analyses of variance. A 2 × 3 mixed ANOVA was used for both analyses with Participant’s Culture (Gypsy, Caucasian) as the between-subject factor and Face Culture (Gypsy, Caucasian, East Asian) as the within-subject factor. An a priori statistical power analysis utilizing the G*Power program (Faul, Erdfelder, Lang, & Buchner, 2007) confirmed the adequate statistical power for these analyses. Total sample sizes exceeding 32 participants would have power higher than 0.95, given a p-value of 0.05, a medium effect size of 0.3, and a correlation among repeated measures of 0.5. With respect to eye movement analysis, following the standard procedure in the face recognition literature (Althoff & Cohen, 1999; Blais et al., 2008), we divided the face stimuli into three non-overlapping regions corresponding to the principal features: eyes, nose, and mouth (see Fig. 1). With these 3 regions of interest (ROIs), we calculated the proportion of fixations and proportion of fixation time on each ROI compared with the total number of fixations and total time per trial, respectively. The learning and recognition phases were analyzed separately. In both phases, the average proportion of fixations and average proportion of fixation time were submitted to two separate 2 × 3 × 3 mixed ANOVAs with Participant’s Culture (Gypsy, Caucasian) as the between-subject factor and Face Culture (Gypsy, Caucasian, and East Asian) and ROI (Eyes, Nose, and Mouth) as within-subject factors. The a priori statistical power analysis also confirmed an adequate statistical power for the eye movement analysis. Total sample sizes exceeding 16 participants have power higher than 0.95. In addition, we decided to use Pearson’s correlations to analyze the relationship between the eye movement variables (number of fixations and fixation time on each region in the learning and recognition phases) and the level of accuracy on the face recognition task. Given the physiognomic differences between ethnicities, we carried out separate correlation analyses for each type of face (Gypsy, Caucasian, and East Asian). Results Two Gypsy and three Caucasian participants were discarded due to problems with the eye tracking calibration or poor recording quality. Thus, the final sample for the analysis was composed of 18 Gypsies and 17 Caucasians. Preliminary analyses Prior to the face recognition task, each participant had to answer a question about the degree of contact with the other cultures included in the task, which allows us to assess the actual degree of contact between cultures in our sample. Participants evaluated the degree of contact with Gypsies, Caucasians, and East Asians on a scale from 0 (none at all) to 4 (very high). Both groups (Gypsies and Caucasians) exhibited high/medium contact scores with Caucasians and Gypsies (Gypsy participants: Mcaucasians = 3.85, Mgypsies = 4; Caucasian participants: Mcaucasians = 4, Mgypsies = 2.05), and very low scores with East Asians (Caucasian participants: Masians = 0.95, Gypsy participants: Masians = 0.6). In order to examine these scores in more detail, we conducted a mixed ANOVA on the degree of contact with Participant’s Culture (Gypsy, Caucasian) as a between-subject factor and Other Group’s Culture (Gypsy, Caucasian, East Asian) as a within-subject factor. The ANOVA revealed significant main effects of Participant’s Culture (F (2, 76) = 32.02, 45
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Fig. 2. Average proportion of correct responses for each face category. Vertical bars represent the standard error of the mean.
p < 0.001, η2p = 0.46) and Other Group’s Culture (F (2, 76) = 535.90, p < 0.001, η2p = 0.93), along with an interaction between both variables (F (2, 76) = 82.62, p < 0.001, η2p = 0.68). Further exploration of this interaction revealed, as expected, a significantly lower degree of contact with East Asians compared with Caucasians and Gypsies for both participant groups (all p < 0.05). In addition, there were no significant differences between Caucasian and Gypsy participants in terms of the degree of contact with Gypsies and East Asians (p > 0.05), but significant differences were found in the degree of contact with Caucasians (p < 0.05). While both Gypsy and Caucasian participants showed a high and similar degree of contact with Caucasians, the Caucasian participants showed a lower degree of contact with Gypsies than the Gypsy participants. It is therefore important to note that, according to perceptual learning theories (Gibson, 1969; Goldinger et al., 2009), support for the proposed hypotheses related to intergroup contact could be particularly evident for the Gypsy group, which have a low degree of contact with Asians, and a high and similar degree of contact with Gypsy and Caucasian cultures. In any case, a better recognition of Gypsy and Caucasian faces compared with East Asian faces was also expected in the Caucasian group, given the differences in the degree of contact with these cultures. Face recognition performance Firstly, Kolmogorov–Smirnov and Levene’s tests verified normality of the distribution and homogeneity of the variances of proportion of correct responses and reaction times for correct responses (all p’s > 0.05). The average proportion of correct responses for each face category are presented in Fig. 2. The 2 × 3 ANOVA conducted on proportion of correct responses revealed a main effect of Face Culture (F (2, 66) = 29.49, p < 0.001, η2p = 0.47). Both the main effect of Participant’s Culture and the interaction were non-significant (F’s < 1). Post hoc analysis of Face Culture showed that Gypsy (M = 0.86) and Caucasian faces (M = 0.78) were better recognized than East Asian faces (M = 0.72). Moreover, Gypsy faces were better recognized than Caucasian faces (p’s < 0.05).
Fig. 3. Average reaction times for correct responses for each face category. Vertical bars represent the standard error of the mean. 46
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Fig. 4. Mean proportion of number of fixations in each region of interest (eyes, nose, and mouth) for Caucasian and Gypsy participants. Vertical bars represent the standard error of the mean.
With respect to the reaction times, the averages for each face category are shown in Fig. 3. The 2 × 3 ANOVA revealed a main effect of Face Culture (F (2, 66) = 7.99, p = 0.001, η2p = 0.19). Participant’s Culture and the interaction were both non-significant (F’s < 1). Post hoc comparisons showed that recognition was slower for both East Asian (M = 1904 ms) and Caucasian faces (M = 1777 ms) compared with Gypsy faces (M = 1669 ms) (p’s < 0.05). As expected, there were no differences between Gypsy and Caucasian observers in recognizing faces (Hypothesis 1a), but both observer groups were poorest at recognizing the faces of East Asians (Hypothesis 1b). Eye movement analysis For the learning task, the 2 × 3 × 3 ANOVAs conducted on the number of fixations and fixation time revealed a main effect of ROI (number, F(2, 66) = 19.26, p < 0.001, η2p = 0.37; time, F(2, 66) = 18.75, p < 0.001, η2p = 0.36) along with significant interactions between Participant’s Culture × ROI (number of fix.: F(2, 66) = 4.28, p = 0.017, η2p = 0.12; fixation time: F(2, 66) = 4.26, p = 0.018, η2p = 0.11) and ROI × Face Culture (number of fix.: F(2, 66) = 6.45, p < 0.001, η2p = 0.16; fixation time: F(2, 66) = 7.03, p < 0.001, η2p = 0.17). No other significant effects or interactions were found. Post hoc analysis of the Participant’s Culture × ROI interaction indicated that Gypsies made a higher number of fixations and showed a longer fixation time on the eyes (Mnumber = 0.27, Mtime = 0.25) than on the nose (Mnumber = 0.16, Mtime = 0.13) and mouth regions (Mnumber = 0.13, Mtime = 0.11) (all p’s = 0.05). Caucasian participants showed a higher number of fixations and a longer fixation time on the eyes (Mnumber = 0.23, Mtime = 0.19) and nose (Mnumber = 0.24, Mtime = 0.19) than on the mouth region (Mnumber = 0.09, Mtime = 0.08) (all p’s = 0.05). Finally, Caucasians showed more fixations on the nose than Gypsies (p < 0.05). The results of the ROI × Participant’s Culture interaction in terms of fixation number are presented in Fig. 4. With respect to the ROI × Face Culture interaction, the eye region received fewer fixations and a lower fixation time for Gypsy faces (Mnumber = 0.22, Mtime = 0.19) than East Asian (Mnumber = 0.26, Mtime = 0.23) and Caucasian faces (Mnumber = 0.27, Mtime = 0.24) (all p’s = 0.05). The nose received fewer fixations and a lower fixation time in the case of East Asian faces (Mnumber = 0.17, Mtime = 0.13) compared with Gypsy (Mnumber = 0.21, Mtime = 0.20) and Caucasian faces (Mnumber = 0.2, Mtime = 0.20) (all p’s = 0.05). The mouth received fewer fixations and a lower fixation time for Caucasian faces (Mnumber = 0.09, Mtime = 0.09) compared with the case of Gypsy faces (Mnumber = 0.13, Mtime = 0.12). As expected, different visual strategies in the recognition of Caucasian, Gypsy, and East Asian faces were found (Hypothesis 2a). However, contrary to Hypothesis 2b, there were visual differences between Gypsy and Caucasian observers in the recognition of both Gypsy and Caucasian faces. Finally, visual differences between observer groups were also found in Gypsy faces, thus supporting Hypothesis 2c as opposed to Hypothesis 2d, but a more detailed discussion is necessary to understand the nature of these results since Hypothesis 2b was rejected (see Discussion section). For the recognition task, ANOVAs conducted on the number of fixations and fixation time revealed main effects of ROI (number of fix.: F(2, 66) = 9.11, p < 0.001, η2p = 0.22; fixation time: F(2, 66) = 6.8, p < 0.002, η2p = 0.17) and Face Culture (number of fix.: F (2, 66) = 5.87, p = 0.005, η2p = 0.15; fixation time: F(2, 66) = 5.45, p = 0.006, η2p = 0.14), along with an interaction between Face Culture and ROI (number of fix.: F(2, 66) = 3.26, p = 0.012; η2p = 0.09; fixation time: F(2, 66) = 5.44, p < 0.001, η2p = 0.14). There are no other significant effects or interactions. Post hoc analysis of the Face Culture × ROI interaction showed that the nose region had a lower number of fixations and a lower fixation time in East Asian faces (Mnumber = 0.25, Mtime = 0.16) compared with Caucasian (Mnumber = 0.29, Mtime = 0.21) and Gypsy faces (Mnumber = 0.29, Mtime = 0.21) (all p’s = 0.05). The mouth region had fewer fixations and a lower fixation time in East Asian (Mnumber = 0.14, Mtime = 0.10) and Caucasian faces (Mnumber = 0.13, Mtime = 0.09) than in Gypsy faces (Mnumber = 0.17, Mtime = 0.13). Similar to the results of the learning task, participants used different visual strategies in the recognition of Caucasian, Gypsy, and East Asian faces (Hypothesis 2a). No differences were found with respect to the Participant’s Culture. 47
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Finally, we analyzed whether the number of fixations and fixation time on each region were related to the success on the face recognition task. There was a significant correlation between accuracy and the number of fixations on the nose in East Asian faces (r = 0.35; p = 0.04). Thus, the fewer the fixations on the nose, the lower the accuracy in face recognition. None of the remaining correlations reached statistical significance. Discussion The present study aimed to examine the influence of the degree of contact with other ethnicities on the cross-ethnic effect and to explore how changes in the visual exploration strategies of the faces are related to this phenomenon. The cross-ethnic effect is described as the poorer recognition of faces of a ethnic group other than our own (Meissner & Brigham, 2001; Sporer, 2001). As an explanation for this effect, the contact hypothesis assumes that visual expertise plays an important role in face recognition and that as a person gains more experience with an ethnicity he/she will be more accurate in recognizing faces of that ethnicity (Caldara et al., 2010; Goldinger et al., 2009). Therefore, the cross-ethnic effect could be due to the fact that the lack of contact does not allow for adjusting the perceptual strategies to the particular physiognomy of the other ethnicity. In order to test the contact hypothesis and the associated perceptual mechanisms, we compared the performance and visual exploration strategies of two ethnic groups that live in close proximity (Andalusian Gypsies and Andalusian Caucasians) in recognizing faces of unfamiliar (Asian faces) and familiar ethnicities (Caucasian and Gypsy faces). In accord with Hypothesis 1a and 1b, there were no differences between Gypsy and Caucasian participants in the recognition of Gypsy and Caucasian faces, and East Asian faces were the most poorly recognized (less accurate and slower responses). Thus, the cross-ethnic effect was only found with faces of cultures with which the participants had no contact. Even Caucasians, who had a moderate degree of contact with the Gypsy ethnicity (see Results section: Preliminary analyses), recognized Gypsy faces at a similar level to Caucasian faces. Based on the intergroup contact hypotheses and perceptual learning mechanisms (Gibson, 1969; Goldinger et al., 2009; Goldstein & Chance, 1985), both Gypsy and Caucasian participants would be expected to show better performance in recognizing Gypsy and Caucasians faces compared with Asians faces, but this better performance should be particularly evident in the Gypsy group, given its higher degree of contact with Caucasians. This fact raises the possibility that a moderate degree of contact with the other ethnicity could be sufficient to achieve adequate recognition. However, it must be taken into account that, in the current study, the similar performance for moderate and high contact could be explained in terms of the low level of difficulty of the task (with an 86% success rate in Gypsy faces), which may overshadow the differences between Gypsies and Caucasians. In any case, faces of cultures with high and moderate contact were better recognized than those with low contact, which suggests that our results are in accordance with the contact hypothesis as a possible explanation for the cross-ethnic facial recognition effect. With respect to the eye tracking data, the results supported Hypothesis 2a, showing differences in the perceptual strategies employed to recognize the faces of each ethnicity, independently of the observer’s culture. The mouth received more fixations and fixation time in the Gypsy faces compared with the remaining cultures in the learning phase. The eyes received fewer fixations and less fixation time for the Gypsy faces compared with the other faces in the learning phase, whilst the nose received fewer fixations and less fixation time for the East Asian faces than for the other faces in both the learning and recognition phases. Regarding the culture of the observer, Caucasians and Gypsies employed different visual exploration strategies in the learning phase, regardless of the type of face to be recognized. Gypsy observer’s fixations were more focused on the eye region than the rest of the regions, while Caucasians fixated more on the nose region than Gypsies, showing a similar proportion of fixations on the eyes and nose. These between-group differences appear to support Hypothesis 2c (as opposed to Hypothesis 2d), but run counter to Hypothesis 2b, since both ethnicities employed different visual strategies in the recognition of both familiar and unfamiliar ethnic groups. However, a deeper analysis is necessary in order to better understand these findings. We therefore discuss this issue in more detail below. Changes in the eye fixation pattern depending on the face culture may be explained in terms of physiognomic differences. There are marked facial physical differences across ethnicities that are key factors in the perceptual discrimination of faces (Kelly et al., 2005). Thus, it is not surprising to think that differences in physiognomic facial features lead to different visual exploration strategies. More importantly for our research, although face culture was associated with changes in the visual exploration pattern, Gypsy and Caucasian participants maintained different perceptual strategies independently of the type of face observed during the learning phase. Differences in the facial features scanned as a function of the observer’s ethnicity are in line with the findings of the studies of Blais et al. (2008), Caldara et al. (2010) and Hills and Pake (2013). However, these results do not fit well with the hypothesis that the perceptual strategies depend on the degree of contact with other ethnic groups (contact hypothesis), since different exploration patterns between Gypsies and Caucasians were not only observed with faces from unfamiliar ethnicities, but also with faces from ethnicities with which they have a high level of contact. In our study, the perceptual mechanisms that are characteristic of each ethnicity seem to have great robustness. The visual exploration strategies were adapted to the physiognomic features of the face to recognize, but certain characteristics associated with the observer’s ethnicity remained stable. Previous literature has pointed to the possibility that cultural force could explain the use of different visual exploration strategies in perceptual tasks (Chua, Boland, & Nisbett, 2005; Nisbett & Miyamoto, 2005). For example, Blais et al. (2008) have provided evidence that Western Caucasians use analytic perceptual strategies, whereas East Asians use holistic strategies for processing faces. In our study, the cultural background of the Gypsies might be guiding the eye movement patterns in processing faces. Gypsies are described as a culture strongly influenced by tradition and spirituality, according to the Schwartz Value Survey (Rzeszewska & Catena, 2012). Not without reason, the eyes have a special status in Gypsy culture, appearing frequently in their folklore, oral traditions, stories, and songs. Thus, the predominance of gazing at eyes in the Gypsy group could be the result of the cultural forces that have shaped the eye movement strategies developed by this ethnic group for hundreds of years. 48
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
Our results reveal how participants from distinct cultures achieve similar face recognition performance despite using different scan paths. These findings appear to run counter to the idea that the use of inadequate perceptual mechanisms could be responsible for the cross-ethnic effect. However, along with the perceptual differences between the observer groups, we have also observed changes that are common to both ethnicities in terms of the visual strategies used to recognize the faces of each ethnicity. In the recognition of familiar ethnicities, the previous social experience with the other group would lead to the acquisition of certain changes in the visual exploration patterns, adapting them to the physiognomy of the other ethnic group. These changes could be sufficient for adequate recognition. Thus, different ethnicities could start from different perceptual mechanisms associated with their cultural baggage, and adapt these to the physiognomy of familiar ethnicities, without having to completely modify the characteristic pattern of their own ethnicity. In the case of recognizing faces of unfamiliar ethnicities, such as East Asians, the visual patterns would also be guided by the most salient physiognomy features, but the lack of previous experience with this ethnicity would lead to the use of less successful undeveloped patterns. For example, as was mentioned in the introduction, East Asians use holistic strategies to organize their visual environment, with the nose (the center of the face) being the optimal region to integrate information holistically during face recognition (Blais et al., 2008; Miyamoto et al., 2006). In this vein, Blais et al. (2008) have shown that the nose is the most effective and diagnostic facial feature that is attended to by East Asian observers in order to recognize faces from their own ethnicity. According to the correlational analysis conducted between accuracy and eye movements in our study, a smaller number of fixations on the nose in East Asian faces were associated with lower accuracy on the face recognition task. This finding could be compatible with the fact that East Asian faces were those most poorly recognized since the participants did not correctly apply their visual strategies to the faces of this ethnicity. Another possible explanation for achieving similar performance through the use of different visual strategies comes from a Blais et al.’s (2008) proposal. It is possible that, in spite of the differences in the perceptual strategies employed, Gypsies and Caucasians construct similar mental representations of the faces of those cultures with which they have a high degree of contact, which would lead to good recognition of these faces. Of course, the mental representations of the faces are certainly developed from eye movements and visual attention patterns, but such patterns do not completely constitute direct evidence of these mental representations (Blais et al., 2008). Thus, experience with the other ethnicities could improve face recognition performance by modifying the mental representations of faces and the processes of encoding and retrieval associated with them, without the need to make considerable changes in the visual exploration strategies. In any case, future research should aim to clarify some of these issues. Among the limitations of our study, it is important to note that the absence of differences in face recognition accuracy between Caucasians and Gypsies could also be explained by the smaller physiognomic changes between them compared with other ethnicities such as East Asians. Although studies in the literature have presented evidence of the cross-ethnic effect in numerous different ethnicities, no previous studies have focused on this effect in Gypsies. Our results demonstrated that Caucasians and Gypsies behave differently at the perceptual level, which supports the idea of there being cultural particularities of each ethnic group during the face recognition process. However, more research is needed in order to study the strength of the cross-ethnic effect between Gypsies and Caucasians without cross contact. In addition, Caucasian participants were able to recognize Gypsy faces as well as Gypsy participants, in spite of having only a moderate degree of contact. As previously mentioned, in the current research, the lack of differences between a moderate and high degree of contact could be explained by the low difficulty of our recognition task. Nevertheless, this result suggests that an interesting future line of research could be to explore the degree of contact necessary to achieve improvements in recognition. Finally, the contact hypothesis predicts that the recognition of faces from other cultures is associated with both the quantity and quality of the contact. In our case, we asked that the participants interpret degree of contact as “quantity”. But, it is possible that the quality of the contact could also have had an impact on our results. We therefore believe that the results of the present study are relevant and shed new light on the cross-ethnic effect and the distinctiveness of the Gypsy ethnicity, but further research should corroborate these findings. Moreover, it would be of great interest to apply this line of investigation to real situations that occur in our everyday lives. From a practical point of view, the study of how the cross-ethnic effect could have an impact on the accuracy of face recognition may have important implications in contexts such as eyewitness identification, or screening for individuals at borders, airports, or government buildings. Conclusion The results reported here provide support for the contact hypothesis as an explanation of the cross-ethnic effect. Face recognition performance across ethnicities was similar when there was sufficient contact with members of the other ethnic group. However, ethnicities with less proximity (in our case East Asians) were the most poorly recognized. The cross-ethnic effect only appeared when the observer did not have sufficient experience with members of the other ethnic group. Although there were no between-group differences in the recognition of Gypsy and Caucasian faces, importantly, differences appeared in the visual scan strategies employed. In particular, Gypsies focused their attention primarily on the eyes, while Caucasians distributed their attention equally between the eyes and nose. Thus, we demonstrated how cultural factors could underlie visual perceptual processes in spite of the intense social experience with other ethnicities. Finally, it is worth emphasizing that given the relatively few studies focused on face processing in the Gypsy culture, our work may serve as a good starting point for further research. Acknowledgments This work was funded by Spanish MICINNPSI2012-39292 grant to A. Catena. Alberto Megías is supported by a Juan de la Cierva postdoctoral fellowship from Spanish MINECO (FJCI-2015-25600). 49
International Journal of Intercultural Relations 65 (2018) 42–50
A. Megias et al.
References Althoff, R. R., & Cohen, N. J. (1999). Eye-movement-based memory effect: A reprocessing effect in face perception. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25(4), 997. Anderson, N. B., & Nickerson, K. J. (2005). Genes, race, and psychology in the genome era: An introduction. American Psychologist, 60(1), 5. Blais, C., Jack, R. E., Scheepers, C., Fiset, D., & Caldara, R. (2008). Culture shapes how we look at faces. PLoS One, 3(8), e3022. Caldara, R., Zhou, X., & Miellet, S. (2010). Putting culture under the ‘spotlight' reveals universal information use for face recognition. PLoS One, 5(3), e9708. Chua, H. F., Boland, J. E., & Nisbett, R. E. (2005). Cultural variation in eye movements during scene perception. Proceedings of the National Academy of Sciences of the United States of America, 102(35), 12629–12633. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. Feinman, S., & Entwisle, D. R. (1976). Children’s ability to recognize other children’s faces. Child Development, 506–510. Gibson, E. J. (1969). Principles of perceptual learning and development. New York: Appleton-Century-Crofts. Goldstein, A. G., & Chance, J. (1985). Effects of training on Japanese face recognition: Reduction of the other-race effect. Bulletin of the Psychonomic Society, 23, 211–214. Goldinger, S. D., He, Y., & Papesh, M. H. (2009). Deficits in cross-race face learning: Insights from eye movements and pupillometry. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35(5), 1105. Groner, R., Walder, F., & Groner, M. (1984). Looking at faces: Local and global aspects of scanpaths. Advances in Psychology, 22, 523–533. Hills, P. J., & Pake, J. M. (2013). Eye-tracking the own-race bias in face recognition: Revealing the perceptual and socio-cognitive mechanisms. Cognition, 129(3), 586–597. Henderson, J. M., Falk, R. J., Minut, S., Dyer, F. C., & Mahadevan, S. (2001). Gaze control for face learning and recognition by humans and machines. In T. Shipley, & P. Kellman (Eds.). From fragments to objects: Segmentation processes in vision (pp. 463–482). Amsterdam: Elsevier. Henderson, J. M., Williams, C. C., & Falk, R. J. (2005). Eye movements are functional during face learning. Memory & Cognition, 33(1), 98–106. Kelly, D. J., Quinn, P. C., Slater, A. M., Lee, K., Gibson, A., Smith, M., ... Pascalis, O. (2005). Three‐month‐olds, but not newborns, prefer own-race faces. Developmental Science, 8(6), F31–F36. MacLin, O. H., & Malpass, R. S. (2003). The ambiguous-race face illusion. Perception, 32(2), 249–252. 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, & Law, 7, 3–35. Minear, M., & Park, D. C. (2004). A lifespan database of adult facial stimuli. Behavior Research Methods Instruments and Computers, 36, 630–633. Miyamoto, Y., Nisbett, R. E., & Masuda, T. (2006). Culture and the physical environment: Holistic versus analytic perceptual affordances. Psychological Science, 17(2), 113–119. Ng, W. J., & Lindsay, R. C. (1994). Cross-race facial recognition failure of the contact hypothesis. Journal of Cross-Cultural Psychology, 25(2), 217–232. Nisbett, R. E., & Miyamoto, Y. (2005). The influence of culture: Holistic versus analytic perception. Trends in Cognitive Science, 9, 467–473. Rzeszewska, I., & Catena, A. (2012). The eye movement strategies in face perception: A cross-cultural study(Master’s thesis). Spain: Universidad de Granada. Sporer, S. L. (2001). Recognizing faces of other ethnic groups: An integration of theories. Psychology, Public Policy, and Law, 7(1), 36. Sternberg, R. J., Grigorenko, E. L., & Kidd, K. K. (2005). Intelligence, race, and genetics. American Psychologist, 60(1), 46. Vizioli, L., Foreman, K., Rousselet, A. G., & Caldara, R. (2010). Inverting faces elicits sensitivity to race on the N170 component: A cross-cultural study. Journal of Vision, 10(1) 15.1–23. Vrij, A., & Winkel, F. W. (1989). Recognition of own and other-race faces: The effects of quality versus quantity of cross-cultural interaction. Journal of Police and Criminal Psychology, 5(2), 24–28. Witzig, R. (1996). The medicalization of race: Scientific legitimization of a flawed social construct. Annals of Internal Medicine, 125(8), 675–679.
50