AI Optimizes Facial Emotion Perception Studies in Autism Research

A new research paper introduces an AI-driven approach to enhance studies on facial emotion perception in individuals with autism. The method involves training artificial neural networks to predict how autistic and neurotypical participants judge facial expressions. These models then identify specific images that highlight perceptual differences between the groups. The AI framework was used to select novel faces predicted to maximize group separation, which indeed produced larger behavioral differences in an independent cohort compared to random images. Furthermore, the models, combined with a generative adversarial network, transformed diagnostic images to reduce predicted group separation, validating the approach. This establishes a novel, model-guided way to create and modify stimuli for studying population-specific perceptual variations.