Dynamic emotion intensity estimation from physiological signals facilitating interpretation via appraisal theory

Area of Science:

• Cognitive Science
• Affective Computing
• Computational Neuroscience

Background:

• Emotion recognition research often overlooks the dynamic nature of emotional processes, limiting interpretability within appraisal theories like Scherer's Component Process Model (CPM).
• Existing methods typically focus on discrete emotion classification rather than capturing the continuous dynamics of emotional experiences.

Purpose of the Study:

• To estimate emotion intensity from physiological features using dynamical models.
• To investigate the relationship between physiological dynamics and perceived emotion intensity within the CPM framework.
• To develop computational models that capture emotion dynamics for enhanced interpretability.

Main Methods:

• Utilized nonlinear autoregressive exogeneous (NARX) models to analyze physiological signals associated with the neurophysiological component of the CPM.
• Induced emotions of varying intensities and measured physiological signals while participants reported subjective feelings in real-time.
• Trained intrasubject and intersubject intensity models using a genetic algorithm on extracted physiological features.

Main Results:

• Dynamical models, specifically NARX, outperformed traditional sliding-window linear regression in estimating emotion intensity.
• Interpreted NARX model parameters revealed consistent heart rate parameters in intersubject models.
• These findings suggest a significant temporal contribution in physiological dynamics that aligns with CPM-predicted changes.

Conclusions:

• Dynamical modeling offers a robust approach to understanding the link between physiological dynamics and perceived emotion intensity.
• NARX models provide interpretable parameters aligned with appraisal theory, enhancing our understanding of emotion processes.
• This research bridges computational modeling and appraisal theory, offering insights into the neurophysiological underpinnings of emotion.