Emotion Modeling

This theme focuses on the computational modeling of emotional processes guided by appraisal theories, specifically leveraging neural networks to simulate the elicitation, differentiation, and dynamic patterning of emotions. It addresses the challenge that traditional discrete and dimensional emotion models provide limited computational tractability, and explores the potential of componential models, such as the Component Process Model (CPM), to underpin neural architectures that capture the recursive appraisal mechanisms linking cognition and emotion.

This research strand investigates the incorporation of affective states—detected through physiological, behavioral, or contextual cues—into agent models operating in intelligent environments such as smart homes and learning systems. The goal is to improve agent adaptivity, personalize user interactions, and optimize decision-making by leveraging emotion-aware input. This includes both the development of affective models for user behavior prediction and the design of systems that respond appropriately to detected emotional states, thus illuminating the interplay between emotion, cognition, and control in pervasive computing.

This theme explores computational models designed for recognizing, categorizing, and simulating emotions based on multimodal inputs including facial expressions, speech, EEG signals, and behavioral data. It investigates statistical, probabilistic, and machine learning methods to capture the temporal dynamics of emotions and the challenges of modality integration. The research includes efforts to refine corpora, improve annotation paradigms, and develop real-time classifiers that advance the interpretability and robustness of emotion recognition systems relevant for human-computer interaction and affective computing applications.