Sex difference of saccade patterns in emotional facial expression recognition

In this work, we conduct two experiments: the first aims at emotional (sadness, happiness, disgust, fear, anger, shame) and neutral facial expressions recognition and the second investigates how the emotional audio (cry, laugh) affects the recognition for emotional (sadness, happy) and neutral facial expressions. The eye movements data in both experiments are recorded by an SMI eye tracker with 120 Hz sampling frequency. The hidden Markov model (HMM) is then applied to extract the fixation patterns in data. For each emotional block, two HMMs are learned, which are related, respectively, to the emotional and neutral faces. The HMM models are trained using 70% saccade data and tested using the remaining 30% data. The first experiment indicates that the saccade patterns not only relate to the task, but also relate to the emotional stimulations and contexts. In particular, the males and females show significant difference in that the females are more easily affected by sadness facial expressions (negative emotion). The second experiment maintains that the emotional audio has greater impact on the females than on the males while the subjects recognizing the neutral facial expressions. This implies that men and women also display significant difference in audio modulated facial expressions cognition. The findings in this study may be still inconclusive, possibly the results of methodological differences.