Self-Organizing Cognitive Reasoning Architecture with Triple-network Executive System
A computational architecture inspired by the Triple Network Model of human cognition decomposes artificial intelligence into interacting modules, including a global workspace, an affective-motivational token system with dopaminergic dynamics, and a meta-cognitive self-monitoring layer. Formal stability analysis shows subsystem dynamics are bounded: inhibitory coupling converges to a bistable attractor with spectral radius under 0.29, the token economy recovers from total depletion within 21 ticks, and meta-cognitive drift rate is bounded at 0.049 per tick. The authors propose seven experimental protocols to test whether this design improves reasoning stability, epistemic robustness, and long-horizon planning compared to monolithic inference systems. They do not claim machine consciousness.