Network analyses of emotion components: an exploratory application to the component process model of emotion

Abstract Emotion is an episode involving changes in multiple components, specifically subjective feelings, physiological arousal, expressivity, and action tendencies, all these driven by appraisal processes. However, very few attempts have been made to comprehensively model emotion episodes from this full componential perspective, given the statistical and methodological complexity involved. Recently, network analyses have been proposed in the field of emotion and cognition as an innovative theoretical and statistical framework able to integrate several properties of emotions. We therefore addressed the call for more multi-componential evidence by modeling the network of a comprehensive list of emotion components drawn from the Component Process Model of Emotion. Five-hundred students were confronted with mildly ambiguous scenarios from everyday life, and reported on their situational appraisals and emotion responses. Network analyses were applied to the emotion components related to a positive and a negative scenario to explore 1) how the components organize themselves into networks and dimensions; 2) which components are the most central within networks and dimensions; and 3) the patterns of components relation between and within dimensions. A three-dimensional solution emerged in both scenarios. Additionally, some appraisals and responses appeared to be differentially relevant and related to each other in both scenarios, highlighting the importance of context in shaping the strength of emotion component relations. Overall, we enriched the field of affective science by exploring the connections between emotion components in three novel ways: by using network analyses, by integrating them into a multi-componential framework, and by providing context to our emotion components. Our results can also potentially inform applied research, where understanding the interconnections and the centrality of components could aid the personalization of interventions.