Pierre Bonzon

The First International and Interdisciplinary Conference on Modelling and Using Context, Rio de Janeiro, January 1997, gave rise to the present book, which contains a selection of the papers presented there, thoroughly refereed and revised. The treatment of contexts as bona fide objects of logical formalisation has gained wide acceptance, following the seminal impetus given by McCarthy in his Turing Award address. The field of natural language offers a particularly rich variety of examples and challenges to researchers concerned with the formal modelling of context, and several chapters in the volume deal with contextualisation in the setting of natural language. Others adopt a purely formal-logical viewpoint, seeking to develop general models of even wider applicability. The 12 chapters are organised in three groups: formalisation of contextual information in natural language understanding and generation, the application of context in mechanised reasoning domains, and novel non-classical logics for contextual application.

We consider Baars’ "Global Workspace" theory of consciousness and discuss its possible representation within a model of intelligent agents. We first review a particular agent implementation that is given by an abstract machine, and then identify the extensions that are required in order to accommodate the main aspects of consciousness. According to Baars’ theory, this amounts to unconscious process coalitions that result in the creation of contexts. These extensions can be formulated within a reified virtual machine encompassing a representation of the original machine as well as an additional introspective component. This computational framework is illustrating throughout using a simple working example.

We consider the problem of executing conscious behavior i.e., of driving an agent’s actions and of allowing it, at the same time, to run concurrent processes reflecting on these actions. Toward this end, we express a single agent’s plans as reflexive dialogs in a multi-agent system defined by a virtual machine. We extend this machine’s planning language by introducing two specific operators for reflexive dialogs i.e., conscious and caught for monitoring beliefs and actions, respectively. The possibility to use the same language both to drive a machine and to establish a reflexive communication within the machine itself stands as a key feature of our model.