A novel approach to exploring infant gaze patterns with AI-manipulated videos

Abstract

Abstract Eye tracking is a widely used tool to study infant development, but creating diverse stimuli while maintaining high control over confounding variables can be challenging. In this proof-of-concept study, we examined an innovative way to generate ecologically valid stimuli using AI technology, in order to create videos that can be used in culturally diverse settings. Using the eye-mouth-index (EMI), a commonly used paradigm in infant eye tracking, we examined the consistency of eye tracking measures across original videos and two types of AI-manipulated videos in a sample of 46 infants aged 12–14 months. We found a very strong correlation of the EMI across original and AI videos (r = 0.873–0.874), and there were no statistically significant differences between mean EMI in the original and AI conditions. Additionally, we created culturally diverse videos to measure gaze following, and found that children followed the gaze of the people in the AI-manipulated videos in an expected manner. In conclusion, AI technology provides promising tools to create ecologically valid and culturally diverse stimuli, that can be used to conduct studies in a wide range of settings and to examine the generalizability of earlier findings in the field of developmental psychology.