NRR-Core: Non-Resolution Reasoning as a Computational Framework for Contextual Identity and Ambiguity Preservation

Abstract

Ambiguity loss is a persistent concern in language-processing systems that optimize for a single resolved output. When context is incomplete, competing interpretations can be compressed too early into one response state. We propose Non-Resolution Reasoning (NRR), a computational framework that treats ambiguity retention as a valid reasoning mode rather than a defect to be eliminated. NRR introduces three principles: (1) Non-Identity ($A \neq A$)--the same symbol refers to different entities across contexts; (2) Approximate Identity ($A \approx A$)--entities share partial structural overlap without being identical; and (3) Non-Resolution--conflicting interpretations can coexist without forced convergence. We formalize these principles through Multi-Vector Embeddings, Non-Collapsing Attention, and Contextual Identity Tracking (CIT). Functional verification in a synthetic two-turn disambiguation task shows that NRR-lite maintains high entropy ($H = 0.91$ bits, near-maximum $1.0$) at the ambiguous turn, while a matched single-state baseline collapses early ($H = 0.15$ bits). NRR challenges the assumption that meaning must collapse to be useful: it targets premature collapse, not commitment itself. Alternatives remain available while evidence is incomplete, without treating retention as repeated full branchwise comparison, and commitment occurs at explicit output or action gates. The question is not whether AI should resolve ambiguity, but when, how, and under whose control.