Interaction of simulated actors with the environment

The current paper explores the possibilities of an improvement of the interaction of an ACT-R actor with its environment including other actors. This is done in the framework of a project aiming at a multi-actor simulation environment based on the ACT-R architecture. Two objections against traditional cognitive architectures like Soar and ACT-R, namely the lack of physical grounding and the lack of symbol grounding, are explained. For a possible improvement of this situation, organizational semiotics and simulation of emotion seem to offer promising perspectives. Organizational semiotics offers us concepts for the encoding of the environment in the form of affordance signs, social constructs and social norms. This leads to new declarative chunk types in ACT-R. An emotion subsystem can maintain an emotional state that encourages task performance, learning, and social behaviour. An awareness subsystem enables task switching based on the emotional state and the selection of those social constructs and norms that are applicable to the current situation.