We introduce the Deontic Action Model Logic (DAML), a dynamic modal framework for reasoning about obligations over actions in multi-agent systems. DAML extends the epistemic Action Model Logic by incorporating deontic evaluation mechanisms that assess agents' actions in terms of both the desirability and the likelihood of their outcomes. Obligations arise for those actions that maximize expected deontic value among an agent's available alternatives at a given decision point, yielding a formal account for reasoning about conditional and context-sensitive obligations in settings involving strategic interaction and incomplete information. DAML supports principled action selection in norm-governed multi-agent systems, and is the first such framework to derive these obligations using the action model logic machinery. We provide an axiomatization of the logic and prove soundness and completeness with respect to its semantics. Finally, we demonstrate the expressive power of our framework through applications to the Miners' Puzzle and other multi-agent deontic scenarios.

翻译：我们提出了道义行为模型逻辑（DAML），这是一个用于在多智能体系统中推理动作义务的动态模态框架。DAML通过纳入道义评估机制扩展了认知行为模型逻辑，这些机制根据行为结果的合意性和可能性来评估智能体的动作。义务产生于那些在给定决策点上最大化智能体可用备选方案中预期道义值的行为，从而为涉及策略互动和不完全信息的环境中有关条件性和情境敏感性义务的推理提供了一个形式化基础。DAML支持规范约束的多智能体系统中的原则性行为选择，并且是首个利用行为模型逻辑工具推导这些义务的此类框架。我们提供了该逻辑的公理化，并证明了其相对于语义的可靠性和完备性。最后，通过应用于矿工之谜及其他多智能体道义场景，我们展示了本框架的表达能力。