Assaying Affective States During Video Gameplay Using Persistence Landscape-Based Topology

Affective gaming is a growing research area that leverages human affect to enhance player experiences. Video games can elicit various emotional responses, including aggression, sadness, and addiction. Physiological signals provide a reliable modality for understanding affective states, offering promising applications in affective computing. However, conventional approaches often fail to capture the dynamic emotional shifts during video gameplay. EEG signals, reflecting neural activity in brain regions, deliver precise insights into psychological states and are widely used in emotion recognition. Nonliear analysis and advanced feature extraction techniques have been instrumental in classification tasks; however, fails to capture the geometric structure of EEG signals linked to affective states. Recently, topological data analysis (TDA) has emerged as a powerful tool to explore time-delay space properties, leveraging persistent homology for computing topological signatures. This study proposes a persistence landscape-based topological nonlinear dynamics analysis of EEG signals. Experimental validation on a publicly available GAMEEMO dataset, demonstrates the proposed frame-work's efficacy, achieving 97.68% average accuracy in binary classification task (positive and negative emotion) and 94.31% in multi-class classification (LALV, LAHV, HALV, and HAHV), outperforming existing methods. These results highlight TDA as a potential tool for biomedical signal analysis and its applications in adaptive brain-computer interfaces.