Consciousness as a Functor
arXiv Preprint Archive August 25, 2025 Sridhar Mahadevan
Consciousness can be modeled as a mathematical functor that transfers contents between unconscious and conscious memory, offering a formal version of the Global Workspace Theory. The framework represents unconscious processes as a topos category of coalgebras and defines an internal language of thought called MUMBLE. Information flow from conscious short-term working memory to long-term unconscious memory is described using Universal Reinforcement Learning, while the reverse transmission from unconscious long-term memory into resource-constrained short-term memory is modeled with a network economic approach.