Skip to Main Content
We propose to model learning of optimal decision rules using a mathematically rigorous descriptive model given by a modified set of Cohen-Grossberg neural network equations with reaction-diffusion processes in an environment of uncertainty. Our theory, which we call Neurohydrodynamics, naturally arises within the framework of neural networks while utilizing the foundations of Decision Field Theory (DFT) for describing the cognitive processes of the mammalian brain in the decision-making processes. Human cognition and intelligence requires more than an algorithmic description by a formal set of rules for its operation; it must possess what Alan Turing called an uncomputable human “intuition” (oracle) as a guide for decision-making processes. We draw an analogy with an idea from Quantum Hydrodynamics, namely, that a “pilot wave” guides quantum mechanical particles along a deterministic path by stochastic “forces” naturally arising from Schrodinger's wave equation. This type of equation was also investigated by Turing, and the reaction-diffusion processes of real neurons have been shown to aid in pattern formation while exhibiting self-organization. Because searching over all courses of action is costly, in both resources and time, we seek to include a mechanism that shortcuts the decision-making processes as described by DFT. Some empirical research has determined that diffusion does occur in the cognitive processing of real human brains. We propose a model for high-level decision processes by combining diffusion with other mechanisms (e.g., adaptive resonance and neuromodulation) for the interactions between different brain regions. For dynamic decision-making tasks, the diffusion processes within certain parts of the frontal lobes and basal ganglia are assumed to interact in hierarchical networks that integrate emotion and cognition incorporating both heuristic and deliberative decision rules.