affective dynamics
Definition
Affective dynamics refers to the processes by which emotions continually change, fluctuate, synchronize, and influence one another over time, rather than operating as stable traits that simply switch on or off. Applied to political communication, affective dynamics addresses how the facial expressions of emotions in political speeches form patterns of covariation tied to the positive or negative sentiment of speech content. In one study using dynamic Exploratory Graph Analysis on facial expression recognition data from 220 videos of global political leaders, a two-dimensional network structure emerged, with positive and negative emotions showing distinct patterns of co-occurrence and rate of change. Populist leaders displayed a less connected network in which anger was expressed more autonomously, while happiness became more contingent on the expression of other positive emotions.
Sources: Tomašević & Major (2024)
Related Terms
- network analysis (1 shared article)
- emotions (1 shared article)
- exploratory graph analysis (1 shared article)
- populism (1 shared article)
- computational social science (1 shared article)
Applications
Affective Dynamics and Populist Political Communication
Populist leaders show a distinctive affective dynamic structure compared to less populist counterparts: anger operates with greater autonomy, showing weaker connections to other emotions, while happiness becomes more dependent on co-occurring positive emotional expressions. This pattern suggests that the use of emotion characteristic of populist communication manifests not only in which emotions are expressed but in how those emotions are structurally related to one another across the course of a speech.
Sources: Tomašević & Major (2024)
Affective Dynamics and Facial Expression Recognition
Facial expression recognition data provide a time-series record of emotion intensity that can be used to map affective dynamics in naturalistic settings. In the analysis of political leaders' videos, transformer-based zero-shot machine learning was applied to extract intensity scores for six emotions across 220 recordings, yielding the multivariate time series on which dynamic network models of emotional covariation were estimated.
Sources: Tomašević & Major (2024)
