Original title: Bayesian mechanics of self-organising systems
Authors: Takuya Isomura
This study explores Bayesian mechanics, framing dynamical systems as Bayesian inference processes. It identifies that generic dynamical systems’ Hamiltonian can function as a generative model, equating their Helmholtz energy with variational free energy. The self-organization in these systems aims to minimize Helmholtz energy, aligning the system’s Hamiltonian with the environment’s—a process leading to generalized synchrony emergence. Essentially, these systems perform Bayesian inference of their interacting environment. The research demonstrates these concepts across various systems like coupled oscillators, simulated and living neural networks, and quantum computers. These insights offer foundational understanding of self-organizing systems and their interactions with the environment, shedding light on potential mechanisms underlying the emergence of intelligence.
Original article: https://arxiv.org/abs/2311.10216