Search papers, labs, and topics across Lattice.
This study investigates the interplay between episodic exploration bonuses and neural memory architectures in partially observable reinforcement learning environments. By conducting a controlled analysis across three distinct environments, the authors reveal that an identical exploration bonus can lead to three different interaction patterns depending on how memory content is acquired, highlighting that exploration and memory are complementary processes. The findings emphasize that the effectiveness of exploration bonuses is contingent on the reward structure, with implications for optimizing policy convergence in reinforcement learning tasks.
Exploration bonuses can either amplify or neutralize memory architecture differences, depending on how memory content is acquired and supervised.
In partially observable reinforcement learning, agents face a dual bottleneck: they must explore to encounter rewarding states and retain that experience in memory to optimize their policies. Exploration bonuses and memory architectures are traditionally evaluated in isolation, leaving their interaction unmeasured, and standard notions of sparse reward conflate temporal signal density with what the reward actually supervises. We present a controlled study crossing episodic exploration bonuses with diverse neural memory architectures across three environments that vary how the content of memory is acquired. An identical bonus signal yields three distinct interaction patterns: it amplifies architectural capacity differences where memory content must be actively discovered and retained unsupervised; equalizes architectures to a shared ceiling where the content, once sought out, is a single reward-supervised cue; and is null where the observation stream is purely scheduled. Controlled reward manipulations verify that these patterns track reward structure rather than density: a dense reward neutralizes a bonus only if it directly supervises the required latent memory, and a small avoidable penalty on exploratory actions (leaving the optimum unchanged) induces policy convergence to suboptimal stationary states, which either bonus resolves. We then formalize reward sparsity with observation-anchored reward machines, separating structural sparsity (an automaton reproduces the return without the task-required history) from potential sparsity (the one-step reward misprices local exploratory actions); the resulting vocabulary organizes the three regimes by the retention burden each task exposes. Together, these results show exploration and memory are complements, not substitutes: a bonus induces exposure, and only memory converts exposure into return.
