Arguments, responsibilities and moral dilemmas in abductive default logic

Abstract

We assume that an agent is not responsible for rule-induced extensions of its theory about the world; responsibility requires the presence of a choice. This supports the attractive conclusion that responsibility for rule-based agents can only arise when the agent faces a “dilemma” regarding how to apply the rules. Default logic offers precise formulations of this intuition. However, it turns out that existing definitions force us to recognise too many dilemmas when reasoning about rules. Specifically, not all moral conflicts are moral dilemmas; the crucial element of choice is sometimes missing. To address this, we first present a refined definition for normal default theories, before going on to present a generalisation that applies to abstract argumentation frameworks.