A neurally plausible model learns successor representations in partially observable environments

TL;DR

This paper introduces a biologically plausible model that learns successor representations in noisy, partially observable environments, enabling efficient reinforcement learning when direct policy learning is challenging.

Contribution

The paper proposes a novel distributional successor features model that incorporates neural-like uncertainty representation for learning in noisy, partially observable settings.

Findings

Supports reinforcement learning in noisy environments

Enables rapid value computation and adaptation

Handles uncertainty in sensory observations

Abstract

Animals need to devise strategies to maximize returns while interacting with their environment based on incoming noisy sensory observations. Task-relevant states, such as the agent's location within an environment or the presence of a predator, are often not directly observable but must be inferred using available sensory information. Successor representations (SR) have been proposed as a middle-ground between model-based and model-free reinforcement learning strategies, allowing for fast value computation and rapid adaptation to changes in the reward function or goal locations. Indeed, recent studies suggest that features of neural responses are consistent with the SR framework. However, it is not clear how such representations might be learned and computed in partially observed, noisy environments. Here, we introduce a neurally plausible model using distributional successor features, which builds on the distributed distributional code for the representation and computation of uncertainty, and which allows for efficient value function computation in partially observed environments via the successor representation. We show that distributional successor features can support reinforcement learning in noisy environments in which direct learning of successful policies is infeasible.