Nathaniel Virgo, Martin Biehl, Manuel Baltieri, and Matteo Capucci revisit the Conant–Ashby good regulator theorem from the standpoint of embodied agents. They develop a formulation in which an observer can interpret a successful regulator as updating beliefs about its environment. The notion of a ‘model’ is observer-dependent rather than necessarily a literal internal representation; the possible model may also be trivial.
The paper addresses the familiar tension between models as internal mechanisms and models attributed through an observer’s interpretation. It does not establish that every effective regulator contains a detailed explicit representation of the world.
Original paper: ‘A “Good Regulator Theorem” for Embodied Agents’ (arXiv:2508.06326v2, 21 August 2025; CC BY 4.0). Authors: Nathaniel Virgo, Martin Biehl, Manuel Baltieri, and Matteo Capucci. For an earlier, different critique, see ‘Fixing The Good Regulator Theorem’ – John Wentworth (2021), already covered on SysCoI.