Search for a command to run...
We describe parallel neural processing that operates below awareness, governs execution, and automaticity. We propose a processing hierarchy distinguishing conscious processing from subconscious processing, bounded below by states in which neither reportable experience nor directed voluntary execution is present (deep surgical anesthesia, coma). We identify four systems as the subconscious infrastructure for skilled execution and flow states. Three execution systems and one gate whose suppression enables them, the cerebellum, basal ganglia, hippocampus, and locus coeruleus. During flow, the cerebellum, basal ganglia, and locus coeruleus are active while the hippocampus is suppressed; we propose that suppression removes competing claims on behavioral control, enabling uninterrupted execution. Their collective state is consistent with accounting for flow's characteristic phenomenology, including sensory narrowing, automaticity, and degraded episodic memory. No framework produces this distinction. The framework predicts that the learning-to-execution pipeline, in which consciously practiced skills are executed without conscious supervision, requires subconscious execution to remain architecturally independent of conscious monitoring. Performance interference, where conscious reactivation degrades execution because self-monitoring intrudes on systems that operate without it, is predicted on this account. Flow states, where absorption, effortlessness, and the absence of self-monitoring co-occur, are consistent with the same logic. The hippocampus encodes episodic context and acts as an execution gate whose suppression releases the execution systems, and we propose that flow onset is threshold-gated at a specific neural ratio. The framework generates eleven falsifiable claims across six experimental tests distinguishing it from models of consciousness, expertise, and automaticity.