Navigation is a complex process that involves several interacting brain regions and has been the subject of intense research for several decades. In general, spatial navigation involves strategies using one of two reference frames: egocentric and allocentric. However, it remains unclear why a particular strategy is chosen over another, and how the neural spatial representations should be related to the chosen strategy. Here, we use a deep reinforcement learning model to investigate whether a navigation strategy could arise spontaneously during spatial learning without imposing a bias onto the model. We then examine the spatial representations that emerge in the network to support different navigational strategies. To this end, we study two ethologically valid tasks, which we refer to as guidance and aiming, respectively. In guidance, the agent navigates to a goal location fixed in allocentric space from different start locations in an environment with stable external landmarks. In aiming, the goal is marked by a visible cue, which is shifted to a different position in each trial. We find that when both strategies are available to the agent, the solutions it develops for guidance and aiming are heavily biased towards the allocentric or the egocentric strategy, respectively, as one would expect. Nevertheless, the agent is able to learn both tasks using either strategy in principle, although learning efficiency is higher for the preferred strategy. Furthermore, we find that place-cell-like allocentric representations emerge preferentially in guidance when using an allocentric strategy, whereas egocentric vector representations emerge when using an egocentric strategy in aiming. We thus find that alongside strategy, the nature of the task plays a pivotal role in the type of spatial representations that emerge.