Some Experiments on Reinforcement Learning - The Case of the “Shortest Path Problem”