Taken at Face Value: How Emotion Expression Does and Does Not Affect Protest Dynamics

Understanding the role of emotions in protest is a growing field ofresearch, but existing research does not address the role of emotionsonce protests start. By applying computer vision models to the expressed emotions of the 37,558 faces in 7,824 geolocated protest images across twelve protest waves in tencountries, this article makes five contributions to the study ofemotions and protest. Most importantly, it measures emotions withinprotest waves, not before them. It also investigates emotions' temporaleffects, multiple emotions simultaneously, connects them directly toactual protests, and does so across multiple countries. The resultssuggest that anger, disgust, fear, happiness, sadness, and surpriseoccur simultaneously throughout a protest, though happiness peaks on thefirst day. Emotions sometimes correlate with protest size in unexpecteddirections, and the coefficient signs differ by country. The mostconsistent finding is that models without lagged terms outperform thosewith lags, suggesting emotions and protests covary more than the formercauses changes in the latter.