Thermodynamically Consistent Two-Phase Decision Dynamics in Open Dissipative Systems: Theory, Numerical Validation, and Bifurcation Analysis

We present a complete theoretical and computational framework for modeling decision-making as anon-equilibrium phase transition in open dissipative systems. The architecture exhibits a sequentialtwo-phase structure: (1) stochastic exploration via Langevin dynamics and (2) dissipative attractorselection. We introduce a locally-computable criterion for phase transition based on accumulatedFisher information, bypassing the need for global probability density estimation. Through extensivenumerical simulations (N = 5×104trials) of a bistable decision landscape, we validate the existenceof an optimal noise intensity minimizing mean first-passage time (stochastic resonance). We furtheranalyze the thermodynamic cost of computation via stochastic entropy production and identifya Hopf-like bifurcation in the transition dynamics. The framework predicts a trade-off betweenenergetic efficiency and decision speed, with implications for both biological and artificial inferencesystems. This project is now under legal monitoring. All commercial implementations must contact the author.