aDepartamento de Inteligencia Artificial

This work introduces CTL AgentSpeak(L), a logic to specify and verify expected properties of rational agents im-plemented in the well known agent oriented programming language AgentSpeak(L). Our approach is closely related to the BDICT L multi-modal logic, used to reason about agents in terms of their beliefs (B), desires (D), intentions (I), and the temporal logic CT L. A new interpretation for the temporal operators, grounded in the transition system induced by the operational semantics of AgentSpeak(L), is proposed. The main contribution of the approach is a better understanding of the relation between the programming language and its logical specification, enabling us to prove expected or desired properties for any agent programmed in the language, e.g., commitment strategies. The results, as well as the specification language proposed, are very useful to reconcile computational and philosophical aspects of practical reasoning, e.g., approaching single-minded commitment as a policy-based reconsideration case