Synkyria and the Free Energy Principle: Boundary-Centred Accountable Viability vs Inference-Centred Self-Organisation

This record presents Version 1.0 of “Synkyria and the Free Energy Principle: Boundary-Centred Accountable Viability vs Inference-Centred Self-Organisation,” part of the Comparative Diagnostics under Finite Capacity (CDFC) series of the Synkyria Project. The paper compares Synkyria’s boundary-centred viability grammar with the inference-centred Free Energy Principle and active-inference frameworks. Its aim is not to merge the two frameworks or claim that one subsumes the other, but to isolate genuine overlap, clarify structural differences, and state the conditions under which an active-inference agent may be viewed as operating within a Synkyrian accountability grammar. The central distinction is that FEP/active inference may function as a mechanism layer for uncertainty, generative inference, and policy selection, while Synkyria provides a grammar layer of admissibility, refusal, holding, re-description, and boundary-legible witness under finite horizons. The paper argues that inference-only optimisation does not, by itself, guarantee accountable boundary preservation unless explicit REFUSE/HOLD outcomes and witness-legible boundary contact are made available. Version 1.0 includes a conservative no-go result, an embedding schema for active inference inside Synkyrian accountability grammar, and a minimal POMDP-style appendix demonstrating how boundary-blind exported records can fail to support auditability. It builds on Fractal Onto-Praxis, Admissibility–Execution–Witness coupling, Field–System Coupling, the Scale-Invariant Viability Transition Theorem, and the preceding CDFC-01 living-systems comparison.