Check Alex and Andrew’s reinforcement learning paper with Oxford. It’s now published in Vision Research.