Why Current AI Architectures are Not Conscious: Neural Networks as Spinfoam Networks in a Theory of Quantum Gravity
IPI Letters December 31, 2025 Trevor Nestor
The authors propose a theoretical framework called Neural Spinfoam Networks (NSNs), which combines loop quantum gravity and the Orchestrated Objective Reduction (Orch-OR) theory of consciousness to address limitations of classical deep neural networks, such as energetic inefficiency and lack of integrated binding. They recast neural layers as spin-networks and learning updates as spinfoam transitions, using gravitational collapse at phase transitions and Majorana-fermion braiding to achieve one-shot credit assignment for the NP-hard perceptual binding problem. They argue that this model offers a more plausible mechanism for backpropagation and weight transport, and they discuss criticisms of Orch-OR while citing recent demonstrations of microtubule superradiance and time-crystalline oscillations as supporting evidence.