THE ACCURACY OF INTENSITY RATINGS OF EMOTIONS FROM FACIAL EXPRESSIONS

The results of a study on the accuracy of intensity ratings of emotion from facial expressions are reported. The so far research into the field has shown that spontaneous facial expressions of basic emotions are a reliable source of information about the category of emotion. The question is raised of whether this can be true for the intensity of emotion as well and whether the accuracy of intensity ratings is dependent on the observer’s sex and vocational orientation. A total of 228 observers of both sexes and of various vocational orientations rated the emotional intensity of presented facial expressions on a scale-range from 0 to 8. The results have supported the hypothesis that spontaneous facial expressions of basic emotions do provide sufficient information about emotional intensity. The hypothesis on the interdependence between the accuracy of intensity ratings of emotion and the observer’s sex and vocational orientation has not been confirmed. However, the accuracy of intensity rating has been proved to vary with the category of the emotion presented.


Aleksandra Kostić 1
Faculty of Philosophy, University of Niš The results of a study on the accuracy of intensity ratings of emotion from facial expressions are reported.The so far research into the field has shown that spontaneous facial expressions of basic emotions are a reliable source of information about the category of emotion.The question is raised of whether this can be true for the intensity of emotion as well and whether the accuracy of intensity ratings is dependent on the observer's sex and vocational orientation.A total of 228 observers of both sexes and of various vocational orientations rated the emotional intensity of presented facial expressions on a scale-range from 0 to 8. The results have supported the hypothesis that spontaneous facial expressions of basic emotions do provide sufficient information about emotional intensity.The hypothesis on the interdependence between the accuracy of intensity ratings of emotion and the observer's sex and vocational orientation has not been confirmed.However, the accuracy of intensity rating has been proved to vary with the category of the emotion presented.

INTRODUCTION
An insight into the relevant literature in the field reveals that the examination of the accuracy of intensity ratings of basic emotions on the basis of spontaneous facial expressions has been mostly ignored.Rather than spontaneous, it is posed facial expressions that have been the focus of the so far research studies.In addition, the attention has been primarily paid to the examination of the factors that affect the perception of the category of emotion, rather than its intensity.This, naturally, does not mean that research on the accuracy of intensity rating is less important.Several studies have been carried out in which the intensity of emotion has been investigated in addition to its category.These studies, cross-cultural in character, found that different cultures highly agree with one another in identifying emotions displayed by the face (Ekman & Friesen, 1969;Ekman, Friesen & Ellsworth, 1972;Izard, 1971).Ekman et al. (1987) also revealed a cross-cultural agreement in the estimation of relative intensity of two different expressions of one and the same emotion.Despite this agreement, however, they also found a cross-cultural disagreement about the estimation of absolute intensity of some of the expressions.The discrepancy was explained by differences in cultural norms that, among other things, affect the modes of interpretation of emotional intensity.
Can the high agreement in the so far studies about recognition of the category of basic emotions, thus, also apply to their intensity ratings on the basis of facial expressions?Is there an agreement about the estimation of absolute intensity of expressions within a single culture?Are the findings obtained through posed stimulus-materials applicable to spontaneous expressions of emotions as well?Is it possible for these findings to be generalized?
The key issue addressed in the paper, therefore, is whether the face can provide reliable information about the intensity of emotion and whether observers can distinguish with utmost accuracy between low, moderate and high intensity of spontaneous facial expressions of six basic emotions (happiness, sadness, fear, anger, surprise, disgust).If so, the question is raised of whether the accuracy of intensity ratings of emotion from a facial expression may be dependent on the type of the emotion expressed and the observer's sex and vocational orientation.The so far research has shown that there exist gender differences in the skills used to decode non-verbal signs -women have been shown to be more superior decoders than men (Zuckerman, Lipets, Hall Koivumaki, Rosenthal, 1975;Hall, 1978;Rosenthal, De Paulo, 1979).When it comes to the relationship between the success in decoding non-verbal signs and the profession of the observer, no empirical data can be found.We find it most important that the role of the observer's profession should be put under investigation, as there is an assumption that the subjects from professions based on communication with people are more sensitive to non-verbal signs sent by others, compared to those professions that are based on working with things.The expectation, therefore, exists that more sensitive subjects are more interested in the information sent by others and are, thereby, capable of providing more accurate ratings of the intensity of emotional expressions.This would mean that students of medicine, humanities, and social sciences would be more successful raters than would students of technical sciences.

DEFINITION OF TERMS
The intensity of emotion is commonly defined as a quantitative attribute, the intensity of a specific behavior -in this case, emotional behavior (English and English, 1972).Dragan Krstić (1991, pg 230) points out that the intensity may be referred to as "a degree of manifestation along a dimension of behavior".In his study, the intensity of emotion is directly associated with the intensity of facial muscle movements.This means that the intensity of muscle movement represents the index of the intensity of emotional experience, implying a full correspondence between the intensity of the emotion that is experienced and the intensity of its expression.Such a correspondence occurs in all situations in which emotions are experienced and expressed spontaneously.
Spontaneous facial expressions of emotion stand for the behavior of the face that occurs when a person displays their emotion involuntarily, with no prior planning or intention.
The accuracy of intensity ratings of emotion is the obtaining of correct information about the intensity of a spontaneously experienced and expressed emotion.The criterion for accuracy of intensity ratings of the six observed basic emotions is based on the analysis of facial behavior components that are relevant for emotional intensity communication, as well as on stimulus-subjects' verbal report on the intensity of the emotion they have experienced, and on the degree of agreement among the observers.

METHOD
Two major procedures have been used: that of encoding and that of decoding facial signs of emotions.

Subjects-decoders
A total of 228 third and fourth year students of medicine, humanities, social and technical sciences at the University of Niš (Serbia) were included.They were randomized and divided into six groups, each comprising 38 males and females from each subject area.

Facial stimuli
The stimulus-material comprised 28 video recordings of spontaneous expressions of six basic emotions and neutral facial expressions (four recordings per each basic emotion -anger, fear, surprise, disgust, happiness, sadness -and four recordings of neutral expression).Each specific category of emotion and each neutral facial expression were illustrated by two recordings of a male face and two of a female face (to guard against a possible influence of facial morphology).Some of the spontaneous facial expressions were obtained under experimental conditions, by a sort of manipulation.Male and female subjects of the age similar to that of the subjects-decoders were presented stimulations likely to elicit spontaneous emotional reactions (a wailing sound of the siren, whistle, scenes from horror movies and melodramas, unexpected touch, unexpected appearance of some people).Their facial behavior was recorded with a hidden camera.The other source of facial stimuli included recordings of basic emotions' facial expressions taken in everyday situations (sports competitions, funerals, weddings).
All selected expressions were encoded by independent encoders (2) using the Facial Action Coding System (FACS) according to Ekman and Friesen (1978). 2 The encoders examined each expression in detail, by decomposing it to specific action units (AU) and identifying which of the units or their combination best describes the observable change in the face.The scores for each facial expression were presented on a list of AUs that generate the emotion in question.The duration and intensity of each facial movement, as well as each bilateral asymmetry, were specified and the reliability obtained was .91.The FACS encoding technique confirmed that the type and intensity of facial muscle movement correlated with those muscles that were of relevance to six "universal" emotions: happiness, sadness, disgust, anger, fear, and surprise (Ekman & Friesen, 1975). 3In our study, facial stimuli were "pure" expressions of happiness, sadness, fear, anger, surprise, and disgust, the intensity of which being precisely specified (FACS).

Research procedure
The recorded emotional expressions of males and females (28) were twice presented to students-decoders.In either observation the grouped subjects were shown all 28 recordings, one at a time and always in the same order.Each expression was viewed for 10 seconds.The interval between each recording presentation was also 10 seconds.During the first viewing, the subjects were asked to choose from a list of 14 emotion terms (surprise, happiness, sadness, envy, fear, confusion, shame, spite, interest, disgust, neutral, contempt, love, anger) one emotion word that best illustrates the emotion presented.Upon recognition of all expressions, the subjects were shown the facial stimuli again and asked to rate the intensity of each expression on a scale from 0 to 8 (0 -absence of emotion; 1,2,3low intensity; 4, 5, 6 -moderate intensity; and 7, 8 -high intensity).

RESULTS AND DISCUSSION
The relationship between the independent variables of "sex" and "vocational orientation" of the observer and the dependent variable of "the accuracy of intensity ratings of emotional facial expression" were tested on a cohort of 228 subjects.The interactions between the factors of "sex", "vocational orientation" and "type of emotion" were examined.The data obtained were processed by using the procedure of variance analysis.Interaction tests were also used.
1.In order to determine the impact of the observer's sex upon the accuracy of intensity ratings of emotion, a mono-factor analysis of the variance was done (bivalent design).The results have shown that differences in the accuracy of intensity ratings of facial expressions of emotions by observers of different sex are not statistically significant (t=-0.608;Sig.0.544).
2. The impact of the observer's vocational orientation upon the accuracy of intensity ratings of facial expressions of emotion was also examined.The Omnibus test was employed through the F-test.The testing of significance (F=1.849;df=2; Sig.0.160) has shown that the accuracy of intensity ratings of facial expressions of emotion is not dependent on the observer's vocational orientation, namely, that no significant differences in the accuracy of rating have been found between the observers studying medicine, humanities, social or technical sciences.
3. Differences in the accuracy of intensity ratings of different emotions' facial expressions were examined irrespective of the observer's sex and vocational orientation.A mono-factor analysis of the variance was done (mono-factor multivalent design -seven levels of the independent variable).The obtained results have shown that differences exist in the accuracy of rating the intensity of expressions of different emotions and that these differences are statistically significant (F=37.092;df=6; Sig.0.000).Irrespective of their sex and vocational orientation, the observers most successfully rated the so-called "null" intensity (lack of emotion), less successfully the intensity of facial expressions of fear, surprise, and disgust, whereas sadness intensity was rated least successfully.
4. The interaction of the factors of "sex" and "type of emotion" was tested against the accuracy of intensity ratings.The results of the tests of significance in this dual-factor multivalent design (2x7) have shown that males and females do not significantly differ in their accuracy of intensity ratings of different emotions' facial expressions.This is indicated by the t-values calculated for each emotion category (sadness: t=-1.468,Sig.0.144; fear: t=-0.082,Sig.0.935; happiness: t=-0.527,Sig.0.599; disgust: t=-1.041,Sig.0.299; anger: t=-1.375,Sig.0.170; surprise: t=-0.295,Sig.0.768).The only significant difference occurs in the ratings of "null" intensity from the category of "emotional neutrality", with females being more observant (t=-2.710,Sig.0.007).The factors of "sex" and "type of emotion" do not significantly correlate with each other (F=0.887,df=6, Sig.0.504), as their impacts on the dependent variable ("the accuracy of intensity ratings of emotional expressions") are independent of each other.The impact of "sex" on the accuracy of intensity ratings of emotional expression is the same at all levels of the "type of emotion" factor, that is, the influence is the same irrespective of the type.
5. The interaction of the factors of "sex" and "vocational orientation" was examined with respect to the accuracy of intensity ratings of emotion.The obtained results have shown that no statistically significant difference exists between "sex" and "vocational orientation" with respect to the accuracy of intensity ratings of emotional expression (F=0.408,Sig.0.666).This means that the impact of "sex" is the same at various levels of "vocational orientation", that is, the effect of one factor is independent of the other.The results have also shown that there exists no significance of simple effects of "sex" (F=0.371,Sig.0.543) and "vocational orientation" (F=1.834,Sig.0.162) when these are taken separately.
6.We tried to answer the question of whether differences exist in the accuracy of intensity ratings of different emotions' expressions among the observers of different sex and vocational orientation.The testing of the factors of "sex", "type of emotion", and "vocational orientation" have shown that their interaction is not statistically significant, that is, the aforementioned factors have a mutually independent effect upon the accuracy of intensity ratings of emotional expression.Statistical indicators are given in the table below.On the basis of the results presented, the following conclusions can be drawn: 1.The accuracy of intensity ratings of emotion varies in accordance with the type of emotion.
2. The interaction between the factors of "different types of emotion" and "sex" is not statistically significant.The accuracy of intensity ratings of facial expressions of different emotions is independent of the effect of the "sex" variable.
3. The interaction between the factors of "type of emotion" and "vocational orientation" is not statistically significant either.
4. Triple-factor interaction (type of emotion, sex, and vocational orientation) is not statistically significant.These factors exert a mutually independent effect on the dependent variable (the accuracy of intensity ratings of emotional expression).
The results of the study suggest that the accurate rating of intensity is dependent on the intensity of a spontaneously expressed emotion.Accordingly, more successful ratings were achieved for "high" intensity (52.56% observers) but less successful for "low" intensity (19.30%).The accuracy of intensity ratings of emotional expressions showed an upward tendency when the observers were presented strong negative emotions (fear, anger).This is quite understandable for accurate ratings of high intensity of negative emotions have a functional value for our social interactions, affecting the development of adequate reactions.On the other hand, the failure to rate "low" intensity was also expected in the light of the fact that the expressions of so displayed an emotion are not that obvious and are difficult to perceive and recognize in 10 seconds' intervals of viewing.When presented an emotion with a "null" intensity", however, the observers showed the greatest success (80.00%).
The results, therefore, suggest that in the events of "high" intensity and of no intensity at all (neutral face), the observers do the rating quite easily.When they were not quite sure about the intensity category, however, the observers preferred the category of "moderately" high intensity, assuming that the choice of "mid-high" intensity would be a less erroneous rating.It is a well-known fact that in a majority of situations of this type some participants tend to provide "mid" answers.A significant number of accurate ratings of this intensity category have been obtained in our study as well, namely, for the expressions showing disgust, anger, surprise, sadness, and fear.
Given all above, a conclusion may be drawn that the observer who is asked to estimate the relative intensity of an emotion, by rating it as low, moderate or high, will do so with relative success, with no prior training.The agreement between the results of this study and the results obtained by cross-cultural studies on intensity ratings is supportive of such reasoning (Ekman et al., 1987).
The hypothesis of the interrelatedness between the observer's sex and the accuracy of intensity ratings of basic emotions has been rejected.Given the findings of this study, women cannot be said to be superior raters of emotional intensity.
The results have not supported the hypothesis of a significant effect of vocational orientation upon the accuracy of intensity ratings of emotion.
In some social situations, the accurate rating of intensity is more significant than is the accurate rating of emotion category.The obtained results show that accurate intensity rating is relevant to all observers, irrespective of their sex and vocational orientation, as it makes the everyday life in the society more successful and allows for an easier establishment of new contacts, adaptation to new situation and performance of various roles in a diversity of complex social situations.The underestimation of the intensity of negative emotions, as well as the overestimation of the intensity of positive ones, can lead to misunderstandings and make social communication more difficult.
The finding that the most successful ratings were obtained for "high" intensity of "anger" and "moderate" intensity of "surprise" and "disgust" (with the exception of "null" intensity) cannot be considered as merely accidental.Accurate ratings of the intensity of these three emotions allow people to take up the appropriate role within the context of a situation, including, among other things, an act of protecting themselves or others.If an attempt is made to explain this finding in terms of an easier and more accurate rating of "high" intensity, then the question can be raised of why this did not happen in the case of "happiness" and "sadness".The "high" intensity of "happiness" was recognized with more accuracy than was the same intensity of "sadness" (a sort of perceptual defence), but not as successfully as was the high intensity of "fear" or "anger".The explanation may lie in the fact that "happiness" and "sadness" are not the emotions the intensity of which exerts the effect of a direct threat or harm to the environment.Accordingly though, the intensity of "happiness" and the intensity of "sadness" were rated with the least success.Accurate intensity ratings of emotions, thus, might be said to have an instrumental value.It may seem as if we are not interested and "awake" to accurately identify the intensity of all emotions.Surely, the majority of "unpleasant" emotions (with the exception of "sadness") occupy the first place on a scale of accurate intensity ratings, as these are emotions that draw bad consequences if rated with imprecision.There follow the majority of "pleasant" emotions, the inaccurate intensity rating of which neither threatens nor does harm to the observer.