Hierarchy as Constraint on Revision Randomness: A Conceptual Account of Why Layered Adaptive Systems Learn When Flat Ones Merely React

This paper proposes that hierarchy in adaptive systems functions to direct error-driven revision by means of accumulated structural knowledge — a claim formalized as the Hierarchical Constraint Principle (HCP). The paper makes four operationally distinct contributions. First, it provides a three-criterion definition of functional hierarchy — temporal separation of layers, top-down constraint on revision priors, and class-level content — with explicit satisfaction thresholds that enable principled adjudication of ambiguous cases, including a formal analysis of why standard end-to-end trained deep networks satisfy Criterion 1 and 3 but are borderline non-satisfying on Criterion 2 at threshold, with that weakness predicting a specific and independently testable generalization limitation. Second, it operationalizes revision entropy (H(E)) with a pre-registered taxonomy of four structurally distinct revision categories derived from task architecture prior to data collection, blocking measurement circularity; it further provides a Monte Carlo power analysis demonstrating that the predicted entropy difference between prefrontal-lesion and control groups (Cohen’s d = 0.65–0.80) is detectable at β = .80 with n = 28–34 per group, rendering the experimental protocol feasible with standard clinical samples. Third, it argues that the HCP’s constraint-on-revision relationship is temporally prior to the constraint-on-prediction relationship central to predictive processing (PP), making the two accounts complementary rather than competitive: PP explains how accurate predictions direct action; the HCP explains how the revision process through which accurate predictions are acquired must itself be structurally directed. Fourth, it derives a failure prediction not made by hierarchical reinforcement learning, perceptual control theory, Simon’s near-decomposability, or Dennett’s Tower: higher-layer disruption should increase trial-to-trial revision variance rather than merely decreasing revision frequency or accuracy. The experimental protocol leads with a prefrontal lesion design as the primary test and treats a cognitive-load manipulation as convergent replication, with both studies powered by the Monte Carlo analysis.