Statistical Properties of Decision-Making Governed by Reinforcement Learning with Status Quo Bias

Abstract

Within the paradigm of human intermittent control over unstable systems human behavior admits the interpretation as a sequence of point-like moments when the operator makes decision on activating or deactivating the control. These decision-making events are assumed to be governed by the information about the state of system under control which the operator accumulates continuously. In the present work we propose the concept of reinforcement learning with decision inertia (the status quo bias) that opens a gate to applying the formalism of reinforcement learning to describing human intermittent control. The basic feature of such reinforcement learning is that human behavior in a sequence of selecting available options exhibits quasicontinuous dynamics. Numerical simulation based on a fairly simple model demonstrates that the proposed formalism does possess the required properties of quasicontinuous behavior.