GSS-Attention Framework: Non-Collapsing Attention, Rule Underdetermination, and Global Non-Individuation in Interpretive Systems

This paper develops the GSS-Attention Framework, a formal model of interpretation in which cognitive and semantic outcomes are not generated by unique selection procedures but by continuously distributed and recursively unstable structures. The framework is built on three core axioms. First, Attention Non-Collapse (Axiom A) states that interpretive attention yields a probability-like distribution over trajectories rather than converging on a singleton outcome. Second, Rule Underdetermination (Axiom R) establishes that no constraint system uniquely determines a privileged trajectory, since every admissible rule system generates multiple compatible interpretations. Third, Recursive Instability (Axiom D) introduces a dynamic framework structure in which interpretive distributions evolve without convergence and admit no fixed point under recursive transformation. From these axioms, the model derives an Elimination Structure and shows that neither evidential filtering nor evaluative weighting (e.g. maximisation or reliability conditions) yields a stable, unique trajectory across updates. A central result—the Global Non-Individuation Theorem—follows: there exists no trajectory that remains simultaneously optimal, rule-selected, and fixed-point stable across levels of the system. The paper concludes that interpretive systems should not be understood as selection mechanisms aiming at unique outputs, but as non-collapsing generative distributions over structurally underdetermined spaces.