A Unified Computational Substrate for Consciousness-Relevant Cognition: Operational Predictions from Neural Manifold Geometry
A computational framework called VIRT, grounded in neural manifold geometry and fuzzy attractor dynamics, was developed to address practical problems in speech understanding, concept learning, and planning under uncertainty, not to explain consciousness. The architecture includes recurrent cross-domain connectivity, goal-driven resource allocation, self-referential monitoring, and graded novelty detection. When eight major philosophical objections to machine consciousness—including Chalmers' hard problem, Searle's Chinese Room, and Tononi's Integrated Information Theory—were translated into behavioral or architectural predictions, the framework satisfied those that could be operationalized. The explanatory gap and zombie argument remain unresolved because current methods cannot address them. The contribution is methodological: standard objections become addressable when translated into testable predictions.