Reward-dependent learning in neuronal networks for planning and decision making.