Abstract

Emotion recognition is an expanding field with applications in psychology and neuroscience. Pupillometry has proven to be an effective technique for assessing emotional responses due to its high temporal resolution and its ability to reflect autonomic nervous system activity. This study explores how pupil characteristics contribute to identifying emotional states in response to visual stimuli.First, a clustering analysis based on affective dimensions (Approach, Commitment, and Identity) was performed. Then, a segmentation based exclusively on pupillary characteristics was applied, allowing for the identification of significant differences in parameters such as dilation amplitude, contraction latency, and pupil variability. For more robust statistical results, we use generalized linear mixed models (GLMMs).The results confirm that pupillometry is an effective tool for distinguishing emotional states, reinforcing its usefulness as a non-invasive method for emotional assessment. Pupillometry alone shows a significant relationship to the emotional dimensions of Commitment and Attention. Additionally, the importance of controlling external factors such as lighting and eye movement is highlighted to improve analysis accuracy. © 2025 IEEE.