It is therefore natural to use logic-based methods for reasoning about the agent's behavior, and for designing an agent that can reason about its environment and about itself. Logics of actions and change are intended to serve this purpose, and form an important research area within Knowledge Representation.
I discuss the present shift of paradigm in the research on logics of actions and change, where the older ad-hoc methodology which was based on intuition and test examples, is replaced by a systematic methodology which is based on a precise definition of intended models with respect to certain ontological and epistemological assumptions. Given a definition of intended models, one proceeds to finding and proving the range of applicability for the various nonmonotonic inference methods that have been proposed and are being proposed.
I present a framework for defining intended models which in itself attempts to capture characteristic aspects of situated agents. Thus, while the concrete situated agents are the intended users of the logics, abstracted situated agents form the theoretical basis for the study of those same logics.