Answers are grounded in this paper and cite their sources.
A capability should not be identified with the implementation that happens to realize it. This paper develops a cognitive ABI: a semantic contract boundary through which cognitive operations are composed, qualified, authorized, realized, and evaluated independently of a particular human, model, program, or system backend. Hybrid Semantic Compiler (HSC) theory supplies the distinction between Fuzzy Semantic Compiler (FSC) operation without established deterministic semantic closure and Deterministic Semantic Compiler (DSC) operation within a declared closure regime. The Cognitive Kernel (CogK) situates work through observation, epistemic position, Exigence, attention, custody, and sufficiency. The Three Hash Rule supplies independent commitments to state, admissibility, and meaning, with optional locality and separate receipt binding. The ABI provides contracts both for work CogK proposes and for operations implementing CogK itself, without assigning CogK execution authority. Typed semantic residue preserves unresolved obligations and connects them to responsibility-bearing work. An evidence-carrying Workflow IR spans cognitive operations and system effects while distinguishing formal proof, deterministic checking, empirical qualification, and attested observation. An author-verified vertical slice combines evidence custody binding, Action IR correspondence checking, deterministic fault evaluation, a bounded CGSOS system test under a one-shot warrant, and residue separation. It performs four DSC cognitive operations and one governed system effect, with zero model calls, network calls, or production changes. Focused tests and 105 bounded qualification trials support cross-plane feasibility. Genuine FSC residue discharge, the integrated CogK work-custody cycle, independent reproduction, and public artifact release remain outstanding. This architecture-and-artifact account makes meaning, competence, authority, effects, unresolved work, and evidence independently inspectable across changing realizations.
No takes yet. Share an insight, caveat, or question.
Adam Ableman (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: