SANRE: A Quantifiable Closed-Loop Framework for Consciousness as Active Negentropy Maintenance and Rule Embedding

We propose SANRE (Stability--Active Negentropy--Rule Embedding), a unified and quantifiable framework for modeling consciousness as a dynamical, substrate-independent closed loop. Consciousness is defined as a self-evidencing process in which a system actively influences its future sensory distribution, minimizes variational free energy under energy constraints, and produces negentropy to sustain its structural identity. The framework rests on four axioms and quantifies consciousness along three axes---Stability (S), Rule Modeling (R), and Negentropy Production (N)---via computable information-theoretic and dynamical metrics. The unified consciousness level is given by the hybrid function \[ C = (S × R × N)1/3 × (1 - λ · max(0, 1 - min(S,R,N))) \]with bottleneck penalty. We provide explicit algorithms for each axis, derive falsifiable predictions, and map current LLMs to Level~3 while offering a concrete path to Level~4 autonomous AGI. SANRE bridges the Free Energy Principle, Integrated Information Theory, and the 2025 Beautiful Loop Theory, while addressing qualia as the system's low-dimensional epistemic control interface.