Proceedings of the Annual Meeting of the Cognitive Science Society

Blink synchronization is closely associated with cognitive state and comprehension during communication. This study developed a real-time model to estimate listener comprehension using blink synchronization. Data were collected from 22 participants who watched a 3-minute explanatory video. An estimation model was trained on the full dataset, with blink synchronization as a key feature. The model discriminated between comprehension levels, with relatively balanced performance (accuracy = 0.727, F1-score = 0.750, and ROC-AUC = 0.726). Real-time comprehension probabilities were then calculated using a 60-second sliding window updated every second. These estimates generally aligned with overall comprehension labels and suggest group-level stability, with limited late-stage deviations among some low-comprehension participants. Overall, the findings highlight the potential feasibility of blink synchronization as a dynamic indicator of listener understanding and contribute to the development of tools for real-time communication support.