Efficient Reinforcement Learning by Discovering Neural Pathways