Artificial Intelligence Architecture and Modelling : Proceedings of the

The title of this paper suggests two distinct aspects of the models that we propose and consider. The rst of these is the modelling of other agents by motivated agents. That is to say that the act of modelling is itself motivated and constrained by the agent doing that modelling. The second aspect is that all such models will also be of motivated agents. It is not su cient merely to know what other agents are like, but also to know why they are like that. This why aspect is what provides the extra information that allows a greater understanding of the interactions between entities in the world, and consequently provides for more resilient agents capable of e ectively dealing with new and unforeseen circumstances in an uncertain world. Previous work has described a formal framework for agency and autonomy in which agents are viewed as objects with goals, and autonomous agents are agents with motivations. This paper considers the nature of cooperation within that framework. We identify distinct kinds of interaction, depending on the nature of the entities involved. In particular, we describe and specify the di erences that arise in these interactions which we characterise as engagements of non-autonomous agents, and cooperation between autonomous agents.