Captured Resonance as Structural Law: Scale-Invariant Failure of Nested Markov-Blanket Systems Across Six Domains of Organization

Contemporary biology, neuroscience, and the social sciences have produced rich domain-specific accounts of failure — somatic-mutation theory for cancer, the amyloid cascade for neurodegeneration, brain-disease and habit-formation frameworks for addiction, inner-misalignment frameworks for artificial intelligence, regulatory-capture theory for institutions, and risk-management accounts of capital flight — without identifying the structural form that recurs across all of them. This paper proposes that a single structural pathology, here termed captured resonance, operates with the same three defining features across six domains of organization, and that its recurrence is a formal consequence of the scale-invariance of the nested Markov-blanket architecture characterizing self-organizing systems (Kirchhoff, Parr, Palacios, Friston & Kiverstein, 2018). Captured resonance is defined by three co-occurring features: a stable attractor against substrate maintenance, to which the system returns under perturbation while the substrate degrades; recruitment of the substrate’s own integrative resources into the attractor’s maintenance rather than into correction against it; and dependence on environmental conditions that suppress corrective alternatives. Where these three are present together, the system’s Markov blanket has undergone pathological closure: the statistical boundary that should mediate open exchange with the environment is sealed around the attractor, with environmental elements incorporated into the blanket in ways that recruit them into maintenance rather than correction. We specify this formally as free-energy minimization under a pathologically precise prior whose prediction errors are suppressed by action rather than resolved by model-updating (Friston, 2010; Schwartenbeck et al., 2015). We develop six instantiations. Three are anchored in established empirical and theoretical literature: cancer, read through the tissue-organization field theory (Soto & Sonnenschein, 2011) and the closure-of-constraints formalism (Montévil & Mossio, 2015); Alzheimer’s disease, read through an amplitude-depletion model of recognition failure; and addiction, unifying five converging literatures including active inference under suboptimal models (Schwartenbeck et al., 2015) and environmental modulation of the addictive attractor (Alexander, Coambs & Hadaway, 1978). Three are developed as structural extensions with explicit and differentiated epistemic status: artificial-intelligence alignment, where a system trained on its own outputs faces the recognition constraint at the level of its training architecture and the corrective apparatus is progressively recruited into the optimization target; the institutional domain, where contested factions are absorbed on the architecture’s terms and their corrective apparatus recruited into its legitimacy; and the monetary domain, where an alternative monetary substrate is absorbed into the dollar system’s own institutional infrastructure and rendered non-corrective under exactly the stress that should activate it, a claim anchored to primary-sourced events of 2026. The formal grounding is the scale-invariance of the Markov blanket: because the same statistical-boundary structure appears at every scale of organization, a specific pathological state of that structure is itself scale-invariant. We propose a single domain-invariant measurement — a coherence metric tracking the synchronization between a substrate signal and its constraint-network expectation — applicable across all six domains. The paper states three cross-scale falsifiable predictions, develops a unified recovery geometry whose deepest statement is that the only intervention the architecture cannot absorb is an external precision source it cannot recruit, and situates the account within the active-inference program, the closure-of-constraints program, and the extended-mind literature on the boundaries of the