Investigating Navigation Strategies in the Morris Water Maze through Deep Reinforcement Learning