Preference-Based Semantics for Defeasible Conditional Obligations

This article introduces a dual-layered semantic framework centered on preferences for modeling defeasible conditional obligations, responding to a concern highlighted by Horty. It enhances Hansson-Lewis style preference semantics within dyadic deontic logic by incorporating a nonmonotonic reasoning approach that permits the retraction of previously established obligations in light of new, conflicting information. The framework utilizes two distinct orderings—ideality and normality—to address limitations found in previous methods, with each employing its own ranking system. The nonmonotonic component examines various postulates, such as antecedent strengthening and inclusion. Additionally, a link is drawn to constrained input/output (I/O) logic, a recognized standard for normative reasoning based on an alternative methodology. This paper falls under Computer Science > Logic in Computer Science and is accessible on arXiv.