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.
Valeriy Kotenko (Tue,) studied this question.