Resource Bounded Bayesian Minds: Complexity, Constraints, and Predictive Processing

This study asks a simple but profound question: How does the mind solve problems without drowning in complexity? Drawing on computational theory, neuroscience, and philosophy, we argue that cognition is built around three tractable primitives: collapsing high‑dimensional uncertainty (MAP), passing messages locally (SUM), and modulating probes according to cost (Control). We first construct synthetic families of tasks to show how these primitives behave in principle, then design experiments with EEG to test whether the same signatures appear in human brains. The results suggest that cognition is not a brute‑force machine but a system tuned to efficiency: frontal signals collapse variance, parietal networks synchronize like message‑passing graphs, and somatosensory responses scale with cost. Philosophically, the work stresses that information is not abstract but physical, behaving like energy—conserved, transformed, and optimized. Cognition, on this view, is the physics of information, a lawful system designed to navigate complexity without succumbing to intractability. The broader implication is a unified science of tractability: bridging computation, neuroscience, and philosophy to show that the limits of thought are principled, measurable, and adaptive. This framework not only clarifies how minds work but also sets an agenda for understanding intelligence as a natural law of efficiency.