AICOS_OS_Decision_Authority_Science_Yasin_Kalafatoglu_2026.pdf

This work introduces AICOS OS (Decision Authority Operating System), a formal governance architecture designed to enforce human-final decision authority in AI-assisted systems. Contrary to prevailing AI paradigms that conflate intelligence, alignment, and oversight with authority, this paper demonstrates—through neuroscience, cognitive psychology, systems engineering, formal methods, and legal theory—that decision authority is a non-computational, non-emergent property that must be explicitly enforced. The paper establishes Decision Authority Science as a new interdisciplinary field and presents AICOS OS as its reference implementation. AICOS OS encodes non-autonomy as a system invariant, requiring cryptographically verifiable human authorization prior to any admissible decision. The architecture integrates formal verification of decision admissibility, explicit authority representation, irreversibility constraints, cognitive load regulation, and deterministic audit and replay suitable for regulatory and legal scrutiny. Neuroscientific evidence shows that human decision authority is mediated by prefrontal cortical gating mechanisms sensitive to irreversibility and responsibility—mechanisms absent in artificial systems. Cognitive psychology demonstrates that automation bias, authority illusion, and decision fatigue predictably erode procedural human oversight. AICOS OS compensates for these structural limits by enforcing authority as system law rather than policy. This publication constitutes public prior art as of 08 February 2026 and provides a patent-grade, audit-ready foundation for lawful, accountable, and sustainable AI deployment in high-stakes domains.