Skip to main content

Achievement

3D simulation of infant social learning

Research Achievements

3D simulation of infant social learning

IGERT trainee Joshua Lewis collaborated with IGERT faculty Gedeon Deák, IGERT faculty Jochen Triesch (now at the Frankfurt Institute for Advanced Studies) and Hector Jasso to develop a realistic three dimensional simulation of social infant learning environments. Combined with a reward system based on computer vision techniques and a reinforcement learning model from machine learning, a simulated infant learned to look more frequently in the direction of a simulated caregiver's gaze (a phenomenon known as gaze following). Behavioral data from videotaped interactions between (real) infants and caregivers were used to set simulation parameters, representing a novel end-to-end approach where behavioral data collection, analysis and synthesis, and computational modeling are part of a coherent workflow in one developmental lab. The results from this work will be presented at the 2010 Annual Meeting of the Cognitive Science Society.

SEE MORE: