Investigating the Interplay of Prioritized Replay and Generalization