PERCEPTION OF SIX BASIC EMOTIONAL FACIAL EXPRESSIONS BY THE CHINESE

PERCEPTION OF SIX BASIC EMOTIONAL FACIAL EXPRESSIONS BY THE CHINESE KAI WANG First Hospital of Anhui Medical University State Key Laboratory of Brain ...
Author: Holly Ball
0 downloads 0 Views 72KB Size
PERCEPTION OF SIX BASIC EMOTIONAL FACIAL EXPRESSIONS BY THE CHINESE KAI WANG First Hospital of Anhui Medical University State Key Laboratory of Brain and Cognitive Science RUMJAHN HOOSAIN TATIA M. C. LEE University of Hong Kong YU MENG JIA FU First Hospital of Anhui Medical University RENMIN YANG Anhui College of Traditional Chinese Medicine

Thirty morphed emotional faces were developed along a happiness-surprise-fear-sadness-disgust-angerhappiness continuum, and 71 Chinese participants were asked to label each of them in a forced-choice task. The frequency curves for correctly labeling each emotion across the morphed photographs showed six discrete peaks. The labeling results are consistent with categorical perception of basic emotions, although morphed photographs were used, and with the hypothesis of universal facial expressions of emotions, although disgust is not included in the Chinese idiom of seven emotions. Keywords: emotion; facial expression; recognition; fear; disgust; surprise; happiness; anger; sadness; Chinese

In the investigation of facial expressions of emotions, many people have focused on six basic emotions—happiness, surprise, fear, sadness, disgust, and anger—although others have proposed similar numbers of overlapping emotions (cf. Ekman, 1972). Basic emotions have been shown to be displayed or recognized consistently by different literate groups in crosscultural studies (e.g., Izard, 1971) and by preliterate people (e.g., Ekman, Sorenson, & Friesen, 1969), although there are indications that cultural groups can differ in the extent to which they accurately perceive the emotions (cf. Matsumoto, 1992). For the Chinese, L. Wang and Markham (1999) created a set of facial expressions showing the six basic emotions. They reported that among the initial 400 photos posed by 17 people to show the six emotions, they could find fewer photos showing fear and disgust, which were rated as such by 70% or more of judges. When another group of Chinese raters was asked to use a 7-point scale to rate the degree each photograph showed a designated emotion, photos of disgust and fear received the lowest scores. Earlier, Markham and Wang

AUTHORS’ NOTE: This work was supported by the National Natural Science Foundation of China (30370479), National Basic Research Program of China (2005CB522800) Distinguished Youth Science Grants (04043071), and Natural Science Grants of Anhui, China (01043602, 2004kj192zd). Correspondence concerning this article should be addressed to Kai Wang, Department of Neurology, The First Hospital of Anhui Medical University, Hefei, Anhui Province, P.R. China; e-mail: [email protected]. JOURNAL OF CROSS-CULTURAL PSYCHOLOGY, Vol. 37 No. 6, November 2006 623-629 DOI: 10.1177/0022022106290481 © 2006 Sage Publications

623

624

JOURNAL OF CROSS-CULTURAL PSYCHOLOGY

(1996) compared Chinese children in Beijing and Australian children in Sydney in situation discrimination and situation inference or labeling tasks. Both the Chinese and Australians showed poor performance for disgust, but then the Chinese did better than the Australians overall. Disgust has also been found to be difficult to recognize in many other studies (e.g., Gosselin, Kirousac, & Dore, 1995; Pitcairn, 1989). Morphed photographs of facial expressions of emotions (cf. Calder, Young, Perrett, Etcoff, & Rowland, 1996) provide a means to present continuously changing facial features, although not necessarily all found in natural situations, to see if there is an indication of categorical perception of the anchoring basic emotional expressions. The present study aims to create a similar Chinese emotional face continuum, running through happiness, surprise, sadness, fear, disgust, anger, and happiness, and to see whether similar patterns of recognition are found. An indication of categorical perception would be when there is an abrupt change in judgment at category boundaries or when expressions within categories are judged to be more similar than those across categories even though they are respectively spaced along the same continuum (e.g., Calder et al., 1996; Etcoff & Magee, 1992). However, there are challenges to such an interpretation (cf. Massaro, 1998).

METHOD MATERIALS

To obtain digitized black-and-white photographs of prototype facial expressions of the six basic emotions (happiness, surprise, fear, sadness, disgust, and anger), a Chinese actor was asked to envisage emotional conditions for each basic emotion label and pose eight times for each of the emotions. A total of 98 Chinese with a variety of backgrounds were asked to choose 1 of the 6 emotion names to label each of these 48 (6 × 8) photographs. The photograph with the highest number of votes for each label was chosen as the prototype for that emotion. The percentages of votes received by the eventually chosen prototype photos were 100.0% (98 out of 98 persons), 93.98%, 89.8%, 92.9%, 88.8%, and 91.8%, respectively, for the Chinese prototypes of happiness, surprise, fear, sadness, disgust, and anger. The six prototype expressions were ordered by placing each adjacent to the one with which it was most likely to be confused, in the order of happiness, surprise, fear, sadness, disgust, and anger (Calder et al., 1996). Each pair of adjacent expressions was then blended into five morphed images in the proportion of 90:10 (i.e., 90% happiness and 10% surprise for the happiness-surprise part of the continuum), 70:30, 50:50, 30:70, and 10:90, using the software HiJaak Morph V 1.0 for Windows and following the procedure of Calder et al. (1996). The result was a continuum of 30 photographs of morphed images (cf. K. Wang et al., 2002), numbered 1 to 30, spanning the six basic emotions. However, because of limitations, the prototype photos were not Facial Action Coding System coded (cf. Ekman & Friesen, 1978; K. Wang et al., 2002). Other materials were prepared to test the perception of facial information other than emotional expression, including gender, age, gaze direction, and the recognition of both familiar and unfamiliar faces. One purpose for measuring recognition of different types of information other than emotional expression, and the other tests, was to see if recognition of various types of information conveyed by the face involves independent modules of functioning (cf. Bruce & Young, 1986). For the perception of gender and age, five photographs each of the faces of Chinese boys, girls, young men, young women, old men, and old women were prepared. For the perception of gaze direction, there were 15 photographs,

Wang et al. / EMOTIONAL FACIAL EXPRESSIONS

625

3 each with the person depicted looking toward 20° left, 20° right, 20° up, 20° down, or straight ahead. For the recognition of familiar faces, a set of 30 photographs of famous Chinese personalities was used, and for the unfamiliar face test, participants had to identify a photographed person from among six people photographed in different lighting and from different directions. Other tests administered included the Wechsler Adult Intelligence Scale-Revised Chinese Version (WAIS-RC; Gong, 1992), Wechsler Memory Scale-Chinese Version (Gong, 1989), Word Fluency Test (naming animals with four legs, within 1 minute), Benton Test of Judgment of Line Orientation, and Visual Form Discrimination (Benton, Hamsher, Varney, & Spreen, 1983). Tests standardized with Chinese norms were used when possible. PARTICIPANTS

The participants were 71 Chinese (40 male and 31 female) aged from 11 to 72 years (M = 37.32, SD = 17.27), and educated from 4 to 15 years (M = 8.59, SD = 2.65) living in Hefei, Anhui Province of China. Of the participants, 64 were right-handed, 5 were ambidextrous, and 2 were left-handed.

PROCEDURE

Participants were tested individually. First, they were required to produce and describe more than three appropriate examples or situations for each emotion to make sure that they understood the six emotion labels. As an example for disgust, the most common situations mentioned by participants were finding vomit in one’s drink and accidentally stepping on excrement. Then, the morphed emotion expressions were shown one by one in random order, together with six emotion labels, in front of the participants. These six emotion labels were written in Chinese as xi (happiness), jing (surprise), pa (fear), bei (sadness), yan (disgust), and nu (anger). The participants were asked to choose one of the labels for each of the photographs. There was no time limitation for each choice. They could have breaks at any time when they felt tired, to guarantee full attention. No specific feedback was given in the procedure. The entire set of 30 photographs was presented six times, the first time as practice, which was not counted. The responses in the following five blocks of 30 trials were used to plot the frequencies of identification of the emotions. The following procedure was used to compute the correct score (CS) for each basic emotion (cf. Calder et al., 1996; Young et al., 1997). We selected 24 of the 30 morphed photographs, each showing a dominant emotional expression. For example, the happy emotion was represented by photographs of 90% happiness and 10% surprise (No. 1; cf. K. Wang et al., 2002), 70% happiness and 30% surprise (No. 2), 30% anger and 70% happiness (No. 29), and 10% anger and 90% happiness (No. 30). The photograph showing 50% happiness and 50% surprise and the one showing 50% happiness and 50% anger were excluded. A response was considered correct if the label identifying the dominant emotion was provided to characterize each photograph. Thus, in each block of 30 morphs, responses to 4 photographs provided CS for each emotion, with a maximum CS of 20 for each emotion across the 5 blocks. When a participant failed to label the dominant emotion in a photograph, the answer given instead could be one of the adjacent emotions (“adjacent errors”) on either side in the continuum of morphed facial expressions, such as mistaking fear for disgust or sadness. On the other hand, the label provided could be one of an emotion further away, referred to as remote errors (REs; cf. Calder et al., 1996; Young et al., 1997).

Frequency of identification as labeled emotions

626

JOURNAL OF CROSS-CULTURAL PSYCHOLOGY

100% Happiness 80% Surprise 60%

Fear

40%

Sadness

20%

Disgust

Anger 0% 1

3

5

7

9

11

13

15

17

19

21

23

25

27

29

Happiness Surprise Fear Sadness Disgust Anger Happiness Morphs numbered from 1 to 30 and the corresponding emotional region

Figure 1: Frequency of Identification of Morphed Emotional Expressions by 71 Chinese NOTE: The numbers 1 to 30 refer to the morphed photographs (see K. Wang et al., 2002). The percentage of times a morph is identified as expressing happiness, surprise, and so on is shown by a curve for each emotion.

RESULTS The neuropsychological data for the 71 participants showed normal performance for IQ according to WAIS-RC (M = 105), memory, visual perception, and perception and recognition of nonemotion information indicated by the face. The frequency curves for identification of each basic emotion showed discrete peaks (see Figure 1), indicating distinct identification of each emotion of the respective morphs. The CS at the peak was 100.0%, 97.0%, 81.13%, 98.87%, 81.89% and 97.75%, respectively, for happiness, surprise, fear, sadness, disgust, and anger (for morphed photos 1, 5, 10, 15, 21, and 25). The data showed a relatively abrupt shift from one emotion to another around the midpoint of each emotion pair (cf. Calder et al., 1996; Young et al., 1997), indicating a greater sensitivity to change across emotion boundaries than the same amount of change within emotions. Furthermore, there was little tendency for other emotions to intrude at the midpoint of each emotion pair. The morphs at the midpoint were identified as either one of the two anchoring emotions at higher rates than as any other emotions. As an example, Morph 3, which was at the midpoint of happiness-surprise part of the continuum, was identified as happiness 64.23% and as surprise 33.80% of the time but as other emotions only 0.56% of the time. Table 1 shows the CS and the RE results for all six categories of emotional expression recognition. The results showed that all the six emotions were correctly and distinctively identified. The highest CS was for happiness, followed by surprise. The lowest CSs were for disgust and fear. Even for disgust, more than 75% of the morphs were recognized correctly by participants (higher than chance by Wilcoxon matched-paired, signed-ranks test, p < .001). Most errors belonged to the adjacent emotion categories. For example, the morphs in the surprise region were usually identified mistakenly as happiness or fear. This is consistent with the confusability data for the original prototypes. Participants made very few REs, with a mean RE score across the six emotions of .193 out of 20 responses. However, the RE for disgust was highest at .549 out of 20.

Wang et al. / EMOTIONAL FACIAL EXPRESSIONS

627

TABLE 1

The Means and Standard Deviations of Correct Scores and Remote Errors for Each of Six Emotions (Out of 20)

Correct scores M SD Remote errors M SD

Happiness

Surprise

Fear

Sadness

Disgust

Anger

19.86 0.35

18.58 1.18

15.14 2.07

18.14 1.37

15.24 2.15

18.20 1.29

0.028 0.167

0.056 0.232

0.183 0.390

0.113 0.318

0.549 0.650

0.225 0.421

The pairwise correlations of number of CSs for the respective six emotions show that although performance for each of the emotions showed significant correlations with at least some of the other emotions, performance for fear was the only one significantly correlated with the performance for every one of the other emotions, with an r of .20, .32, .32, .66, and .31, respectively, with happiness, surprise, sadness, disgust, and anger. Table 2 shows that perception of emotional expression was not correlated with age, even though there was a wide range of age of the participants. Perception of fear (r = .45) and disgust (r = .54), but not the others, were correlated with education level. For visuospatial abilities, performance in Benton’s Test of Line Orientation and Visual Form Discrimination just managed to reach significance in correlation with the perception of some emotional expressions. There were no significant correlations between recognition of unfamiliar faces and recognition of emotional expressions. There were also no significant correlations between recognition of emotions and recognition of other face conveyed information, including that of age, gender, gaze direction, and famous faces.

DISCUSSION In the present study, the technique of forced-choice labeling of morphed facial expressions with basic emotion names produced results similar to those found in the West using the same procedure (cf. Calder et al., 1996; Young et al., 1997) and to those of L. Wang and Markham (1999) with Chinese participants judging individually posed photos. We found relatively poor performance for fear and disgust, and disgust showed the largest RE, indicating that the confusion was more severe. Even so, we still obtained statistically significant results for all the six basic emotions, considered as indicating universality of facial expressions of the basic emotions (cf. Ekman, 1994). In the case of Chinese, this result is obtained even though disgust is not included as a basic emotion in idiomatic usage of the seven emotions and six motivations. Also, Russell and Yik (1996) reviewed studies of ancient Chinese texts reflecting dominant philosophies including Confucianism, Daoism, and Buddhism and found various sets of noted emotions, ranging from five to seven. Their English translations of all the emotion labels, totaling 12 names, did not include disgust. The present study showed that morphs well within the continuum for a emotional region, such as Numbers 4, 5, 6, and 7, were labeled very similarly, but midpoint morphs of adjacent categories, such as Number 3, was not perceived as indeterminate. The current data of discrete identification peaks for each basic emotion and confusional identification for midpoint of adjacent emotional regions may not be consistent with the dimensional perception

628

JOURNAL OF CROSS-CULTURAL PSYCHOLOGY

TABLE 2

Bivariate Correlations Between Correct Scores for Different Emotions and Age, Education, and Measure of Other Cognitive Abilities Happiness

Surprise

Fear

Sadness

Disgust

Anger

–.028 .153 –.046 –.193 –.064 –.074 .038 –.118 .028 –.119 –.039 –.031

–.129 .109 .048 –.67 –.124 –.222 –.093 .029 .215 .101 .108 –.109

–.014 .448** .155 –.018 .100 .058 .087 –.104 .227 .125 –.156 .072

–.072 .075 .063 –.313** .189 .264* .290* –.001 –.007 .070 –.064 –.150

–.039 .538** .151 –.252* .064 .151 .260* –.033 .072 .029 .009 .132

.005 .266 –.028 .153 .129 .134 .287* –.005 –.040 .012 –.001 .082

Age Education IQ Memory quotient Word fluency Line orientation Visual form discrimination Unfamiliar face recognition Perception of gender Perception of age Perception of gaze direction Famous face identity NOTE: The table presents r values. *p < .05. **p < .001.

of emotional expression but consistent with categorical perception. However, the use of data showing distinct peaks of identification and others showing greater confusion within category boundaries than across boundaries (e.g., Etcoff & Magee, 1992) as evidence for categorical perception has been questioned. Criticisms of the categorical model include that the labels provided could bias possible responses, that the proportion of judgments along the continuum is treated as a linear measure, and that the shape of identification function by itself does not conclusively show categorical perception (cf. Massaro, 1998). It is also pointed out that the procedure does not explicitly test the categorical model against continuous models. The finding that recognition of both fear and disgust is correlated with education suggests that it is not something specifically related to disgust or its linguistic or cultural characteristics but with the common characteristic of fear and disgust being more difficult to recognize. The significant correlations with scores for Benton Test of Visual Form Discrimination suggest that perception of emotional expressions, as a visual recognition task, depends on certain more basic visuospatial functions. The lack of correlations between recognition of emotions and recognition of other kinds of information conveyed by the face is consistent with the functional face processing model of Bruce and Young (1986), in which emotion processing is distinguished from processing of familiarity and other information conveyed by the face.

REFERENCES Benton, A. L., Hamsher, K. S., Varney, N. R., & Spreen, O. (1983). Contributions to neuropsychological assessment: A clinical manual. New York: Oxford University Press. Bruce, V., & Young, A. (1986). Understanding face recognition. British Journal of Psychology, 77, 305-327. Calder, A. J., Young, A. W., Perrett, D. I., Etcoff, N. L., & Rowland, D. (1996). Categorical perception of morphed facial expressions. Visual Cognition, 3, 81-117. Ekman, P. (1972). Emotion in the human face. New York: Pergamon.

Wang et al. / EMOTIONAL FACIAL EXPRESSIONS

629

Ekman, P. (1994). Strong evidence for universals in facial expression: A reply to Russell’s mistaken critique. Psychological Bulletin, 115, 268-287. Ekman, P., Sorenson, E. R., & Friesen, W. V. (1969). Pan-cultural elements in facial displays of emotions. Science, 164(3875), 86-88. Etcoff, N. L., & Magee, J. J. (1992). Categorical perception of facial expression. Cognition, 44, 227-240. Gong, Y. X. (1989). Wechsler Memory Scale—Revised in China. Changsha, China: Mapping Press. Gong, Y. X. (1992). Wechsler Adult Intelligence Scale—Revised in China version. Changsha, China: Hunan Medical College. Gosselin, P., Kirousac, G., & Dore, F. (1995). Components and recognition of facial expression in the communication of emotion by actors. Journal of Personality and Social Psychology, 68, 83-96. Izard, C. E. (1971). The face of emotion. New York: Appleton-Century-Crofts. Markham, R., & Wang, L. (1996). Recognition of emotion by Chinese and Australian children. Journal of CrossCultural Psychology, 27, 616-643. Massaro, D. W. (1998). Perceiving talking faces: From speech perception to a behavioral principle. Cambridge, MA: MIT Press. Matsumoto, D. (1992). American-Japanese cultural differences in the recognition of universal facial expressions. Journal of Cross-Cultural Psychology, 23, 72-84. Pitcairn, T. (1989). Origins and processing of facial expressions. In A. Young & H. Ellis (Eds.), Handbook of research on face processing (pp. 71-76). Amsterdam: Elsevier. Russell, J. A., & Yik, M. S. M. (1996). Emotion among the Chinese. In M. H. Bond (Ed.), The handbook of Chinese psychology (pp. 166-188). Hong Kong: Oxford University Press. Wang, K., Hoosain, R., Li, X., Zhou, J., Wang, C., Fu, X., et al. (2002). Impaired recognition of fear in Chinese man with bilateral cingulate and unilateral amygdala damage. Cognitive Neuropsychology, 19, 641-652. Wang, L., & Markham, R. (1999). The development of a series of photographs of Chinese facial expressions of emotion. Journal of Cross-Cultural Psychology, 30, 397-410. Young, A. W., Rowland, D., Calder, A. J., Etcoff, N. L., Seth, A., & Perrett, D. I. (1997). Facial expression megamix: Test of dimensional and category account of emotion recognition. Cognition, 63, 271-313.

Kai Wang, PhD, MD, professor, is the director of the Neurology Department, Medical Psychology Department and Cognitive Neuropsychology Laboratory at Anhui Medical University. His recent research interests include the neural basis for emotion, social cognition, attention and memory, and cognitive recovery after brain damage. Rumjahn Hoosain is an honorary professor in the Psychology Department of the University of Hong Kong. His recent publications include the edited volumes with Farideh Salili titled Language in Multicultural Education and Religion in Multicultural Education. Tatia M. C. Lee, PhD, is the director of the Neuropsychology Laboratory at the University of Hong Kong. Her research interests are deception, emotion regulation, brain recovery, and neural plasticity. Yu Meng, MD, is the acting director of neurology at the Anhui Medical University. His research interests include emotion and memory deficit after stroke. Jia Fu, MD, is the acting director of neurology at the Anhui Medical University. Her research interests include cognitive deficit after stroke and brain recovery after brain damage. Renmin Yang, MD, is the chair professor and director of the Neurology Institute of Anhui College of Traditional Chinese Medicine. He is also the head of the Anhui Branch of Neurology Section of Chinese Medicine Association. His major research interests are Wilson’s disease and other motor disorders. Recently, he is interested in basal ganglia involved in cognition, particularly in emotion or memory.

Suggest Documents