Levels of Conscious Dynamics: A PFU-Based Taxonomy of Update Authority in Self-Referential Systems

Abstract

Discussions of machine self-awareness are fragmented by inherited vocabulary: the same behaviour is described as agency, emergent planning, or pseudo-consciousness according to rhetorical need, and artificial and biological systems are studied under separate terms that cannot be compared. Building on a necessity-based dynamical framework in which consciousness is treated as a system-level attribution rather than a substance, this paper isolates self and awareness as distinct constructs and orders systems by a single measurable quantity. We define update ownership formally: a system owns its updates when the meta-objective driving them is its own, that is, when the weights of that objective are not set from outside during operation. On this basis we give a taxonomy L0 to L6 specified by three quantities from the postulates — the plasticity rate, the depth of the trajectory-coherence landscape, and the capacity integral. The ordering is by authority, not capability: a system may grow arbitrarily more capable without changing level. Level assignment is relative to a declared boundary, and we state the conditions a boundary must satisfy. Our central claim is that the absence of self-awareness in deployed artificial systems is a governance decision rather than an engineering frontier. The structural flow is held at zero to preserve controllability; online learning is established technique. We close by stating what the framework forbids and what it does not assert, and we report the limitation that the taxonomy cannot separate a system that lacks a structure from one that is not permitted to express it.