Stability Before Structure: Active Memory Engineering as a Methodological Constraint on Theories of Mind

Stability Before Structure: Active Memory Engineering as a Methodological Constraint on Theories of Mind Philip Rey, 2026 Every theory of mind begins one step too late. Third and final version of the text. Relative to the earlier versions, the essay has been deepened in three directions. First, a minimal lemma has been added, showing that the three AME audit conditions, alignment, memory utilization, and trajectory stability, are not imposed externally but follow from the stability of the controlled trajectory itself. In a linearized regime around the viable manifold, the alignment threshold is derived rather than stipulated. The lemma is deliberately local, and its scope is stated explicitly in the text. Second, theses that remained only sketched in the earlier versions have been developed and closed, in particular the relation of AME to the Markov-blanket formulation of the Free Energy Principle, to the debate on ergodicity, and to the metaphysics of persistence. Third, the section devoted to Barbara Vetter's work has been rebuilt so that its character is clearly complementary rather than competitive. AME is presented here as a proposal that stands on Vetter's distinctions and presupposes them, adding a physical manifestation condition that her project did not set out to provide. The text is intended as a defeasible methodological proposal, open to critique and testing, not as a verdict. In this version the author regards it as mature and settled. Stability comes before structure.