Bayesian Mechanics for Stationary Processes

TL;DR

This paper introduces a Bayesian mechanics framework for adaptive systems using Markov blankets, linking inference, active control, and steady-state dynamics to explain adaptive behavior.

Contribution

It develops a novel Bayesian mechanics approach for modeling adaptive systems with Markov blankets, connecting inference, control, and steady-state dynamics.

Findings

States internal to the blanket encode information about external states.

Internal states infer external states via variational inference.

Active states perform active inference and stochastic control.

Abstract

This paper develops a Bayesian mechanics for adaptive systems. Firstly, we model the interface between a system and its environment with a Markov blanket. This affords conditions under which states internal to the blanket encode information about external states. Second, we introduce dynamics and represent adaptive systems as Markov blankets at steady-state. This allows us to identify a wide class of systems whose internal states appear to infer external states, consistent with variational inference in Bayesian statistics and theoretical neuroscience. Finally, we partition the blanket into sensory and active states. It follows that active states can be seen as performing active inference and well-known forms of stochastic control (such as PID control), which are prominent formulations of adaptive behaviour in theoretical biology and engineering.