محسن صفائي

The eye, as one of the primary channels for receiving information, plays a crucial role in attention an‎d visual perception through its movements. Among these movements, microsaccades—small, involuntary shifts—serve as important indicators of cognitive an‎d emotional processes. This study aimed to examine the influence of emotional content on microsaccade characteristics during visual scene perception. Eye movement data were recorded from 16 participants (40 trials) using the Eyelink 1000 Plus system an‎d analyzed with the Engbert–Kliegl algorithm. Images with positive, negative, an‎d neutral emotional valence were presented, followed by classification of neutral images into three categories: “human,” “animal,” an‎d “object.” The results showed that prior emotional content significantly affected microsaccade rates, an‎d these effects persisted during the classification of neutral images. In addition, changes in pupil diameter were correlated with the emotional intensity of the stimuli. Furthermore, machine learning algorithms (Ran‎dom Forest an‎d XGBoost) successfully predicted image classification based on microsaccade features, achieving an accuracy of 89.60%. These findings highlight that microsaccades, beyond their role in attentional mechanisms, can also serve as objective markers of emotional processing. They hold potential applications in the development of cognitive–affective models an‎d intelligent systems.

ميكروساكاد , طبقه‌بندي تصاوير , Eyelink1000plus , جنگل تصادفي , XGBOOST

Microsaccade , Eyelink 1000 Plus , Image categorization , Random Forest , XGBoost

Mohsen Safaei

Mohammadreza Daliri