Generative models as parsimonious descriptions of sensorimotor loops

TL;DR

This paper proposes that generative models in the brain are better understood as efficient descriptions of sensorimotor interactions relevant for behavior, rather than exact representations of the environment, challenging traditional views.

Contribution

It introduces a perspective that generative models serve as parsimonious, task-relevant descriptions of sensorimotor loops instead of accurate environmental models.

Findings

Generative models can effectively describe sensorimotor relationships.

This approach shifts focus from veridical perception to functional behavior.

It offers a new framework for understanding brain function beyond traditional predictive coding.

Abstract

The Bayesian brain hypothesis, predictive processing and variational free energy minimisation are typically used to describe perceptual processes based on accurate generative models of the world. However, generative models need not be veridical representations of the environment. We suggest that they can (and should) be used to describe sensorimotor relationships relevant for behaviour rather than precise accounts of the world.