Modeling the effects of environmental and perceptual uncertainty using deterministic reinforcement learning dynamics with partial observability

TL;DR

This paper introduces a deterministic reinforcement learning model for agents with partial observability, revealing unexpected benefits and complex dynamics such as limit cycles and critical slowing down, applicable across biological, social, and machine learning contexts.

Contribution

It provides a parsimonious, systematic framework for modeling partially observable agents, enabling analysis of emergent behaviors and complex dynamics with broad applicability.

Findings

Partially observant agents can learn better outcomes faster and more stably.

Partial observability can help agents overcome social dilemmas.

Emergence of catastrophic limit cycles and critical slowing down in learning dynamics.

Abstract

Assessing the systemic effects of uncertainty that arises from agents' partial observation of the true states of the world is critical for understanding a wide range of scenarios. Yet, previous modeling work on agent learning and decision-making either lacks a systematic way to describe this source of uncertainty or puts the focus on obtaining optimal policies using complex models of the world that would impose an unrealistically high cognitive demand on real agents. In this work we aim to efficiently describe the emergent behavior of biologically plausible and parsimonious learning agents faced with partially observable worlds. Therefore we derive and present deterministic reinforcement learning dynamics where the agents observe the true state of the environment only partially. We showcase the broad applicability of our dynamics across different classes of partially observable agent-environment systems. We find that partial observability creates unintuitive benefits in a number of specific contexts, pointing the way to further research on a general understanding of such effects. For instance, partially observant agents can learn better outcomes faster, in a more stable way and even overcome social dilemmas. Furthermore, our method allows the application of dynamical systems theory to partially observable multiagent leaning. In this regard we find the emergence of catastrophic limit cycles, a critical slowing down of the learning processes between reward regimes and the separation of the learning dynamics into fast and slow directions, all caused by partial observability. Therefore, the presented dynamics have the potential to become a formal, yet practical, lightweight and robust tool for researchers in biology, social science and machine learning to systematically investigate the effects of interacting partially observant agents.