Exploration, novelty, surprise, and free energy minimization

TLDR

The free energy principle has been argued to preclude novelty and complexity by positing that biological systems minimize long‑term surprise to maintain homeostasis. The paper reviews recent developments of the free energy principle that provide a normative, Bayesian‑inference‑based perspective on classical economic decision‑making. The authors cast decision‑making as a variational problem in which agents infer policies that minimize surprise by reducing the relative entropy between expected and desired outcomes, thereby balancing exploitation of high‑utility goal‑states with exploration of alternative states, leading to behavior governed by both entropy and expected utility. Recent formulations demonstrate that minimizing surprise naturally generates exploration and novelty bonuses, increases the value of current states by visiting new states, and resolves the apparent conflict between surprise minimization and novelty seeking.

Abstract

This paper reviews recent developments under the free energy principle that introduce a normative perspective on classical economic (utilitarian) decision-making based on (active) Bayesian inference. It has been suggested that the free energy principle precludes novelty and complexity, because it assumes that biological systems-like ourselves-try to minimize the long-term average of surprise to maintain their homeostasis. However, recent formulations show that minimizing surprise leads naturally to concepts such as exploration and novelty bonuses. In this approach, agents infer a policy that minimizes surprise by minimizing the difference (or relative entropy) between likely and desired outcomes, which involves both pursuing the goal-state that has the highest expected utility (often termed "exploitation") and visiting a number of different goal-states ("exploration"). Crucially, the opportunity to visit new states increases the value of the current state. Casting decision-making problems within a variational framework, therefore, predicts that our behavior is governed by both the entropy and expected utility of future states. This dissolves any dialectic between minimizing surprise and exploration or novelty seeking.