Toward Constructing a Modular Model of Distributed Intelligence

Abstract

Multi-agent social systems (MASSes) are systems of autonomous interdependent agents, each pursuing its own goals and interacting with other agents and environment. The dynamics of the MASS cannot be adequately modeled by the methods borrowed from statistical physics because these methods do not reflect the main feature of social systems, viz., their ability to percept, process, and use external information. This important quality of distributed (swarm) intelligence has to be directly taken into account in a correct theoretical description of social systems. However, discussion of distributed intelligence (DI) in the literature is mostly restricted to distributed tasks, information exchange, and aggregated judgment, i.e., to the “sum” or “average” of independent intellectual activities. This approach ignores the empirically well-known phenomenon of “collective insight” in a group, which is a specific manifestation of MASS DI. In this paper, the state of art in modeling social systems and investigating intelligence per se is briefly characterized and a new modular model of intelligence is proposed. This model makes it possible to reproduce the most important result of intellectual activity, viz., the creation of new information, which is not reflected in the contemporary schemes (e.g., neural networks). In the framework of the modular approach, the correspondence between individual intelligence and MASS DI is discussed and prospective directions for future research are outlined. The efficiency of DI is estimated numerically by computer simulation of a simple system of agents with variable kinematic parameters (k_i) that move through a pathway with obstacles. Selection of fast agents with a positive mutation of the parameters provides ca. 20% reduction in the average passing time after 200–300 generations and creates a swarm movement whereby agents follow a leader and cooperatively avoid obstacles.