Work

How one brain substrate generates many ideas. A Geometro-Diffusion Theory model

Year: 2026
Type: preprint
Abstract: A brain has one anatomical network and one set of physiological laws, yet it holds a very large number of different ideas. We propose a mechanism for this variety within the Geometro-Diffusion Theory (GDT) of consciousness, which associates experience with the geometry generated by propagation across the brain. In the model, every input induces, at the time at which the geometry is read, an effective network: the regions whose activity is, or has been, above a threshold. Each effective network carries its own part of the brain's geometry and corresponds to one idea, which a reader, the diffusion-based machine, reads and associates with what is felt. The dynamics derives from the dynamic mean-field model of the cortex. On the human connectome, tens of thousands of inputs produce tens of thousands of different effective networks, the same network is reached from different inputs, and geometric richness varies widely. Along the coupling axis fitted to anaesthesia and psychedelic data, the networks become poorer at loss of responsiveness and richer under psilocybin. The picture and the name of an item produce different networks, which the reader associates after learning both. Having learned the propagation from short repeated episodes, the reader reads the geometry of every effective network immediately. With two images present, both networks are generated and a serial reader reads them in turn, weighted by richness and familiarity, as in binocular rivalry. Finally, the predicted networks overlap the cortical parcels that respond to electrical stimulation in 780 patients, well above chance.
Language: English
Open Access status: green
OpenAlex ID: W7236592383
Cites: 52
Cited by:
Related to: 10