Multilevel analysis of facial expressions of emotion and script: self-report (arousal and valence) and psychophysiological correlates

Background: The paper explored emotion comprehension in children with regard to facial expression of emotion. The effect of valence and arousal evaluation, of context and of psychophysiological measures was monitored. Indeed subjective evaluation of valence (positive vs. negative) and arousal (high vs. low), and contextual (facial expression vs. facial expression and script) variables were supposed to modulate the psychophysiological responses. Methods: Self-report measures (in terms of correct recognition, arousal and valence attribution) and psychophysiological correlates (facial electromyography, EMG, skin conductance response, SCR, and heart rate, HR) were observed when children (N = 26; mean age = 8.75 y; range 6-11 y) looked at six facial expressions of emotions (happiness, anger, fear, sadness, surprise, and disgust) and six emotional scripts (contextualized facial expressions). The competencies about the recognition, the evaluation on valence and arousal was tested in concomitance with psychophysiological variations. Specifically, we tested for the congruence of these multiple measures. Results: Log-linear analysis and repeated measure ANOVAs showed different representations across the subjects, as a function of emotion. Specifically, children’ recognition and attribution were well developed for some emotions (such as anger, fear, surprise and happiness), whereas some other emotions (mainly disgust and sadness) were less clearly represented. SCR, HR and EMG measures were modulated by the evaluation based on valence and arousal, with increased psychophysiological values mainly in response to anger, fear and happiness. Conclusions: As shown by multiple regression analysis, a significant consonance was found between self-report measures and psychophysiological behavior, mainly for emotions rated as more arousing and negative in valence. The multilevel measures were discussed at light of dimensional attribution model.