COGNITION IN SYSTEMS

This paper proposes a four-level architecture for the cognitive appara tus of future autonomous systems, addresses the handling of nested tim e scales, tries a first step toward a technical approximation of consc iousness as a linear abstract control channel within a massively paral lel architecture, and describes a number of interplays between an intu itive (subsymbolic) and a symbolic level of information processing. Fu rthermore, concrete models, namely a task model, model of the environm ent, partner model, and eigenmodel of a system, are distinguished. Som e comments on test beds are included, as are hints about the project A MOS at FAW that addresses some of the topics discussed within a platfo rm-based realization, particularly the interplay between subsymbolic a nd symbolic forms of information processing.