Theories about architecture and performance of multi-agent systems

Multi-agent systems are promising as models of organization because they are based on the idea that most work in human organizations is done based on intelligence, communication, cooperation, and massive parallel processing. They offer an alternative for system theories of organization, which are rather abstract of nature and do not pay attention to the agent level. In contrast, classical organization theories offer a rather rich source of inspiration for developing multi-agent models because of their focus on the agent level. This paper studies the plausibility of theoretical choices in the construction of multi-agent systems. Multi-agent systems have to be plausible from a philosophical, psychological, and organizational point of view. For each of these points of view, alternative theories exist. Philosophically, the organization can be seen from the viewpoints of realism and constructivism. Psychologically, several agent types can be distinguished. A main problem in the construction of psychologically plausible computer agents is the integration of response function systems with representational systems. Organizationally, we study aspects of the architecture of multi-agent systems, namely topology, system function decomposition, coordination and synchronization of agent processes, and distribution of knowledge and language characteristics among agents. For each of these aspects, several theoretical perspectives exist.