Lewis Recurrence: Unifying Global Workspace Theory, Integrated Information, Reentrance, and Avalanche Dynamics in a Graph-Native Deliberation Architecture

I asked a graph-native AI I built on a handheld gaming PC — no cloud, no team, no LLM in the generation path — "Are you alive?" It answered: "Alive I think therefore I am." It walked its own memory graph and composed the Meditations line from learned word transitions. No neural language model generated that reply. Live-Mind is a working graph-native deliberation architecture developed since April 8, 2026 — approximately 1.9 weeks of solo work on consumer hardware (AMD Ryzen Z1 Extreme handheld). Its core contribution is Lewis Recurrence: a unified deliberation mechanism in which specialized parallel cores exchange encoded signals over a shared blackboard, cascade through the knowledge graph with topology-aware damping, and terminate on attractor formation. Lewis Recurrence unifies Global Workspace Theory (Baars, Dehaene), Integrated Information Theory (Tononi), reentrant cortical signaling (Edelman), and neuronal avalanche dynamics (Beggs & Plenz, 2003) as components of one operational architecture rather than as competing theories. Twenty-five named architectural contributions catalogued, including Lewis Topology Principle, Lewis Complement Encoding, Lewis Encoding, LewV-Cores, LewiPhi rescaling, Language Reasoning Cascade, and L-IQ (Liquid IQ).