Morphogenesis : Basal Cognition :: Self-Organisation : Maximum Entropy

I gave a talk to the Levin lab about my recent work on gauge symmetries, inference, and self-organisation.

Abstract:

We give an account of inference in the context of self-organisation, with particular focus on structures like Turing patterns and developing cell lines—objects which seem too simple to be cognitive, but which self-organise nonetheless. We introduce the idea of a morphological constraint, also called an ontological potential. Two main consequences are discussed. Under the principle of constrained maximum entropy, and the closely related free energy principle, we can argue that anything which organises into some sort of individuated structure performs inference over (i) that pattern, and (ii) the perturbations from the environment dissipating that pattern. This allows connections to basal cognition, in the sense that these objects exhibit a sort of elemental inference which closely resembles the kind of representation-based cognition that humans do. Our second point relates self-organisation to a gauge force arising from constraints on what states can be occupied by a system, introducing the idea of a morphological gauge field. Open questions about the generation of target morphologies and Markov blankets are also raised.

Slides: Levin Lab May 2022.pdf

As a precursor to this talk, I gave an account of some technical aspects of this work to the Levin lab and collaborators. That talk was hosted by the VERSES Research Lab.

Slides: Levin Lab March 2022.pdf