From Explanation to Effective Contestation: An Objection-to-Remedy Evaluation Framework for AI-Supported Decisions

People may possess a legitimate basis for disputing an AI-supported decision without being able to inspect, reproduce, or understand the model that contributed to it. Existing work has already distinguished explanation, recourse, contestability, institutional review, and effective remedy; it has also identified the importance of evidence access, human intervention, and authority. This paper therefore does not claim to discover the need for contestability or the institutional chain linking notice to remedy. Instead, it proposes a falsifiable, challenge-grounded evaluation framework for determining when a decision subject's lack of model expertise is, and is not, a barrier to justified review. The framework treats a contested decision as an episode in which a claim must move through objection formulation, claim-specific evidence acquisition, substantive assessment, authorized disposition, and implementation. It specifies separately observable failures at these transitions, distinguishes a substantiated objection from entitlement to reversal, and proposes an outcome vector that rewards justified corrections and justified non-reversals while penalizing unsupported reversals, unresolved valid objections, inaccessible review, and disproportionate burden. A proposed four-arm comparison contrasts competent conventional review, technical explanation, structured challenge-oriented evidence, and assisted review with defined authority. Cases would include erroneous inputs, omitted overriding evidence, misapplied rules, model-level challenges, justified adverse decisions, and genuinely ambiguous decisions. The proposal is positioned against contestable-AI design, algorithmic recourse, legal access rights, decision provenance, prior reconstruction-burden work, and recent institutional-chain accounts. No implementation, participant study, effect size, empirical superiority, legal certification, or claim of universal scientific novelty is reported. The intended contribution is a preregisterable means of examining the incremental value and failure conditions of contestation support under matched evidence, competence, authority, cost, and consequence constraints.