The ability to regulate one’s affective state and to induce an affective state in another person is a key component of social and emotional interaction. In this paper we describe an architecture design aimed at modeling these two closely related processes. We first define the high-level functional components of each process, and then use these to refine the specific knowledge necessary for their implementation. Based on this analysis we propose a cognitive architecture capable of modeling affect regulation in the self, and affect induction in the other. We illustrate the functioning of this architecture by way of an example from a peacekeeping training simulation. The architecture is currently being implemented in the context of this simulation training task.