Implementation of eye tracking technique for human behavior quantification

Abstract

In this modern era of technology, numerous eye-mind relationship studies have dominated discussion on the ability of computer to interpret human personality traits. This phenomenon shows that eye tracking technique is becoming an important area of study. However, the study of eye movement system on validating the interpretation of actual human behaviour is still deficient. Thus, this study designs an experiment to detect human behaviour of Big Five personality traits - extraversion, agreeableness, neuroticism, openness and conscientiousness - through eye tracking system. It investigates the correlation between eye movement and human behaviour, and evaluates and validates the observational response and human personality traits. To test the hypothesis that there is a significant relation between eye movement and human behaviour, a personality test of Big Five model and experiment were conducted with 30 engineering undergraduates from one premier public university in Malaysia. The respondents were randomly picked, and the personality test was distributed to the respondents before starting the experiment. The fixation duration stimulated by film clips of different arousal contents and graphic health warning labels on cigarette packs were explored using Tobii TX300 eye tracker device. Each subject was analysed by studying their eye movement using five types of emotional video stimuli including joy, amusement, neutral, fear and sad, as well as six graphic health warning labels that are currently used by government of Malaysia. The results of eye gazing of emotion-related clips and warning labels stimuli were compared with Big Five personality test to study the relationship of eye movements and human behaviour. The results were analysed using statistical analysis of variance (ANOVA) which indicated there is no significant relation between fixation duration and human behaviour. As the actual results of positive emotion scored the longest fixation at 2570ms while negative emotion scored 2380ms, were unexpectedly different from the expected outcome based on previous studies. The data also indicated that the validation of eye-tracking technique from the emotion clips and warning labels did not have the consistent response from each subject personality, related towards their behaviour. These results suggest that their eye movement did not likely portray the participant behaviour. Hence, the concepts of validating system from eye tracker device should be further analysed in future studies for better understanding of human – computer interaction.