Acupuncture Points as Local Optima of Shannon Entropy: An Information-Theoretic Hypothesis for Acupuncture Analgesia

Abstract This paper proposes a hypothesis grounded in the Free Energy Principle (FEP)—a framework typically formulated in terms of variational free energy (an upper bound on surprise). We argue that a more direct and empirically tractable quantity for understanding acupuncture is Shannon entropy, the measure of uncertainty in a probability distribution. We hypothesize that acupuncture points function as local optima of Shannon entropy in the brain's generative model of the body. That is, these points are body locations where stimulation can bidirectionally modulate neural entropy—increasing it when pathologically low (e.g., chronic pain, rigidity) or decreasing it when pathologically high (e.g., acute pain, anxiety)—thereby restoring a healthy optimal range. This “optima” hypothesis subsumes the FEP’s prediction error minimization, explains both entropy‑increasing and entropy‑decreasing effects of acupuncture, and generates testable predictions that distinguish it from purely descriptive accounts. The hypothesis unifies existing evidence, parsimoniously explains point specificity without requiring meridian anatomy, and opens computational bridges to placebo effects and interoceptive inference. Not limited to pain relief, the hypothesis can also be generalized to other conditions (e.g., inflammation, nausea) where acupuncture modulates subjective experience.