Brain Organoids Achieve Goal-Directed Cart-Pole Learning

Researchers at the University of California, Santa Cruz reported in Cell Reports that lab-grown brain organoids were coached via electrical feedback and a reinforcement-learning algorithm to solve the cart-pole control task. The team increased organoid cart-pole win rates from 4.5% to 46%, demonstrating goal-directed, trial-and-error learning in minimal cortical tissue. This finding suggests organoids can serve as in vitro models to study learning mechanisms and neuromodulation.