Istisarelism: Proportionate Objection, Epistemic Autonomy and Reasoned Termination in Human–AI Deliberation

This conceptual paper proposes Istisarelism as a provisional normative and methodological framework for reasoned, proportionate disagreement in human–AI deliberation.

Seven principles govern the framework: fallibility, priority of reasons, within-dialogue revisability, proportionate objection, risk-sensitive caution, effective human autonomy, and reasoned termination.

The proposal is compared with established human–AI deliberation, Socratic AI, critical-thinking interventions, appropriate reliance, and the 2026 AACT and SYNC frameworks.

Its candidate operational mechanism, the Reasoned Objection Threshold (ROT), evaluates evidence, decision relevance, marginal informational contribution, conversational history, and cognitive cost, with separate safeguards for consequential risks.

The paper distinguishes useful cognitive effort from redundant interaction burden. A two-arm feasibility pilot followed by a factorial study is proposed to isolate objection selection and termination.

A prospective non-inferiority criterion for reasoning quality and a reduction in redundant objections jointly define the principal research aim.

Research status: This is a conceptual preprint. No empirical results, independent peer review, or definitive priority claim are asserted. The paper presents a falsifiable research agenda rather than a validated new theory.

Generative AI assisted with drafting and internal critical review. Responsibility for the research claims and publication rests with the human author.