Neural responses to facial expressions: A story of arousal and valence?

Abstract

Facial expressions of emotion provide important information about our social partner’s internal state. Historically facial expressions have been conceptualized as discrete categories (i.e., happy, angry), as well as along a continuum of arousal (activated to calm) and valence (positive to negative). Both behavioural and event-related potential (ERP) research attribute emotion category effects (i.e., happy vs angry) to arousal and valence, however, no study has directly linked these constructs. The aim of the present study was to understand if early ERP components involved in visually processing facial expressions are sensitive to these continuous factors (i.e., arousal and valence) at the stimulus level. ERPs were recorded while 80 participants viewed faces expressing fear, anger, happiness, and no emotion, and performed a gender discrimination task. Following the ERP task, participants then viewed each face again and rated them on arousal and valence using 1-9 Likert scales. Individual ratings from each face were linked back to ERP data, trial by trial. ERPs were analyzed in a data-driven way (all time points, all electrodes) using mass univariate statistics. Paired contrast between expressions were analyzed using three different hierarchical models: without (original) and with valence or arousal ratings. Results from models with ratings highly overlapped with the original model, although were more restricted. The N170 component was the most impacted by arousal and valence ratings, where most emotion contrasts revealed significant valence or arousal interactions at this peak. On the P2 emotion differences were driven by more negative amplitudes for angry than happy or neutral faces, but this was unrelated to any rating. On the Early posterior negativity (EPN) happy faces elicited more negative amplitudes than fearful and neutral faces, a difference related to both arousal and valence ratings. However, no other contrast was significant on the EPN. Overall, our analyses show that ERP emotion effects are related to the participants’ perceived arousal and valence of the stimuli, although these relationships highly depend on the specific contrast analyzed. We conclude that perceived arousal and valence of individual stimuli is a critical source of ERP variability, likely contributing to inconsistent findings in the field. Future research must consider the relationship between stimuli ratings and ERP amplitudes to fully unpack how and when our brain extracts emotional information from faces.