PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES Universals and Cultural Differences in the JudgmentsofFacial Expressions of Emotion Paul Ekman and Wallace V. Friesen University of California, San Francisco Maureen O'Sullivan University of San Francisco Irene Diacoyanni-Tarlatzis Aristotlian University, Thessaloniki, Greece Rainer Krause University des Saarlandes, Saarbriicken, Federal Republic of Germany Tom Pitcairn University of Edinburgh, Scotland Klaus Scherer University of Geneva, Switzerland Anthony Chan Chinese University of Hong Kong Karl Heider University of South Carolina William Ayhan LeCompte University of Maryland Pio E. Ricci-Bitti University of Bologna, Italy Masatoshi Tomita Waseda University, Tokyo,Japan Athanase Tzavaras Aristotelian University, Thessaloniki, Greece We present here new evidence of cross-cultural agreement in the judgment of facial expression, Subjects in 10 cultures performed a more complex judgment task than has been used in previous cross-cultural studies. Instead of limiting the subjects to selecting only one emotion term for each expression, this task allowed them to indicate that multiple emotions were evident and the intensity of each emotion. Agreement was very high across cultures about which emotion was the most in- tense. The 10 cultures also agreed about the second most intense emotion signaled by an expression and about the relative intensity among expressions of the same emotion. However, cultural differ- ences were found in judgments of the absolute level of emotional intensity. In the last 10 years, opinion has shifted about whether facial expressions of emotion are universal. The earlier viewthat what a facial expression signifies is completely different from culture to culture (Birdwhistell, 1970 LaBarre, 1947 Leach, 1972)is no longer accepted within psychology,although it is still main- tained by some anthropologists (Howell, 1985). Those who have become persuaded by the evidence of universal facial expres- sions of emotion can cite consistent findings across three quite different types of research. Those who remain skeptical, how- ever, can cite flaws in each. Our study was designed to remedy some of these flaws. We will consider the strengths and weak- nesses in each type of research on the universality of facial ex- pressions of emotion. In one type of investigation (Ekman & Friesen, 1971), mem- Paul Ekman's work is supported by Research Scientist Award MH 06092 from the National Institute of Mental Health. Correspondence concerning this article should be addressed to Paul Ekman, Human Interaction Laboratory, University of California School of Medicine, 401 Parnassus Avenue, San Francisco, California 94143. bers of one culture were asked to show how their face would look if they were the person in each of a number of different emotional contexts (e.g., "you feel sad because your childdied," "you are angry and about to fight"). Universality was demon- strated when observers in another culture did far better than chance in identifying which emotional contexts the expressions were intended to portray. This finding had unusual import be- cause the persons displaying the expressions were members of a visually isolated New Guinea culture (the South Fore). The ability of Americans to understand these NewGuinean expres- sions could not be attributed to prior contact between these groups or to both having learned their expressions from mass media models. Three problems limit these findings. First, there has been only one such study. It has not been repeated in another preliter- ate, visually isolated culture, nor for that matter in a literate, non-Western or Western culture. Second, not all sixemotions portrayed were accurately recognized. Anger, disgust, happi- ness, and sadness were distinguished from each other and from fear and surprise, but the American observers could not distin- guish the New Guineans portrayals of fear and surprise. Third, Journal of Personality and Social Psychology, 1987, Vol. 5 3, No. 4,712-717 Copyright 1987by the American Psychological Association, Inc. 0022-3514/87/$00.75 712

UNIVERSALS 713 the facial expressions were posed, and Mead (1975) argued that establishing that posed expressions are universal need not im- ply that spontaneous facial expressions of emotion are univer- sal. The next type of research design answered this criticism. Facial expressions shown by Japanese and by Americans while they watched stress-inducing films (bodily mutilation) and neutralfilms(nature scenes) were measured. When the sub- jects in each culture watched the films alone, unaware of a hid- den camera, virtually the same facial responses were emitted regardless of culture (Ekman, 1972). However, when a scientist was present when they watched the films, the Japanese more than the Americans masked negative expressions with smiles (Friescn, 1972). In addition to examining spontaneous facial expressions, this study was the first to show how cultural differ- ences in the management of facial expressions (what Ekman & Friesen, 1969, had termed display rules) can mask universal facial expressions. Two problems limit these findings. First, again it is but a sin- gle study no one has yet attempted to replicate it. Also, the mutilation films elicited only a few emotions (disgust and fear), not allowing determination of whether the full range of sponta- neous emotional expressions is universal. The next type of re- search met these two criticisms. Photographs of facial expressions were shown to observers who were asked to judge the emotion displayed. Very high agreement was found across 12 literate cultures in the specific emotions attributed to facial expressions. The strength of this evidence is its many replications. Unlike the first two kinds of research, this type of study has been repeated in many cultures, by different researchers (Ekman, Sorenson, & Friesen, 1969 Izard, 1971), and with different photographs of facial expres- sion. Four questions can be raised about the value of such judg- ment studies in which the same set effaces is shown to observers in different cultures. First, the observers were shown posed rather than spontaneous expressions. This criticism is at least partially met by the fact that universality was also found in one other judgment study (Ekman, 1972) in which the observers saw spontaneous facial behavior. The expressions of the Japan- ese and American subjects in the study described earlier, in which subjects had watched body mutilation and neutral films, were shown to Japanese and American observers. These observ- ers were asked to judge whether each person's expressions oc- curred in reaction to a stressful or a neutral film. Thejudgments made by the Japanese and American observers were highly cor- related and did not differ as a function of whether they were interpreting the expressions of their own or the other culture. Second, all the cultures had some contact either with each other or with media presentations of facial expressions, and therefore their similar judgments could be the consequence of having learned a common set of facial expressions. This criti- cism is met by judgment studies in two different, visually iso- lated, preliterate NewGuinean cultures, the South Fore and the Dani. The New Guineans discriminated most but not all of the emotions distinguished by the literate-culture observers. In both New Guinean cultures, happiness, sadness, disgust, and surprise were discriminated from each other and from anger and fear. In the South Fore, fear was not distinguished from surprise (Ekman & Friesen, 1971), but this discrimination was made by the other New Guinean culture, the Dani (Heider & Rosch, reported in Ekman, 1973). The Dani did not discrimi- nate anger from disgust, but the South Fore did. The next two criticisms raise questions due to limitations in the judgment task that the observers used to register their im- pressions. The third one is that the judgment tasks might have concealed cultural differences in the perception of secondary blended emotions. Many students of emotion have noted that facial expressions may contain more than one message (Ekman & Friesen, 1969 Izard, 1971 Plutchik, 1962 Tomkins, 1963). The two emotions in a blend may be of similar strength, or one emotion may be primary, much more salient than the other sec- ondary emotion. In prior cross-cultural studies, the investiga- tors presumed that the expressions they showed displayed a sin- gle emotion rather than a blend and therefore did not provide those who observed the expressions the opportunity to choose more than one emotion for each expression. Without such data, however, it is not possible to ascertain whether an expression conveys a single emotion or a blend, and if there is blend, whether cultures agree in their judgment of the secondary emo- tion. Prior evidence of cross-cultural agreement in the judg- ment of expressions might be limited just to the primary mes- sage, not to the secondary blended emotions. Fourth, despite agreement about which emotion is depicted, there might be differences in the strength of the perceived emo- tion. Only one cross-cultural judgment study (Ekman, 1972) obtained intensity judgments, and no differences were found. Further investigation is warranted as only five cultureswere ex- amined. To summarize, there has been no cross-cultural study of whether cultures differ in the perception of secondary blended emotions. To do so requires that the observers be allowed to indicate that an expression shows multiple emotions. Our study was designed to fill this gap and also to replicate the finding that the intensity ascribed to an emotional expression is also universal. Observers were asked to judge the emotions shown in each photograph twice, once restricted to a single choice for each expression and once allowed to register up to seven emo- tions and the relative strength of each. Hypothesis 1: There will be agreement across cultures about which emotion is shown in each expression when observers are limited to a single choice. This hypothesis simply predicts that earlierfindings with a single-choice judgment task will be replicated. Hypothesis 2: There will be agreement across cultures about which emotion is the strongest one shown in each expression when ob- servers are allowed to register the presence of up to seven emotions. This hypothesis predicts that allowing multiple-emotion judg- ments will not eliminate cross-cultural agreement. Hypothesis 3: There will be agreement across cultures about which emotion is perceived as the second strongest emotion in each ex- pression. This prediction is more tenuous, for whether or not there will be universality about the secondary emotion is not implied by the prior evidence. We make this prediction extrapolating from Ekman and Friesen's (1975) finding on Americans that the muscu- lar display in the expression predicted the secondary emotion that was attributed to the expression. Hypothesis 4: There will be agreement across cultures in the judg- ment of the strength of an emotional expression. Whereas Ekman and Friesen (1969) described how cultural differences in display rules could lead to differences in the judgment of emotional inten- sity, Hypothesis 4 is based on their finding (Ekman, 1972) of cross- cultural agreement in intensity judgments.