System Dynamics and Intelligent Agent-Based Simulation: Where is the Synergy?

摘要

Pedagogical research has demonstrated that while traditional teaching methods can be useful for imparting factual information ("cold knowledge"), simulations and video games are more effective in teaching decision-making processes ("warm knowledge"). Advances in video game creation allow the development of multi-agent, artificial society simulators with capabilities for modeling physiology, stress, emotion, and course-of-action decision-making. This new approach enables superior understanding of the complexity in organizations and their relevant business environments. This in turn provides an opportunity for game-play that helps promote better decision-making. System Dynamics also allows managers to make their understanding of business problems explicit and improve upon them. This occurs by modelling structures (e.g., relationships, policies, incentives, etc) that underlie behaviour of systems. While system dynamics acknowledges the critical role of personal and organizational mental models (e.g., motivations, values, norms, biases, etc.) as the foundation or key influencers of structure, it does not explicitly model mental models, nor does it take into account decision makers ‘mood’. In contrast, in Agent-Based Modelling (ABM), organizations are modelled as a system of semiautonomous decision-making parts (purposeful individuals) called agents). Macro-behaviour is not simulated; it emerges from the micro-decisions of individual agents. In this work, each agent individually assesses its situation and makes decisions based upon value hierarchies of goals for action, preferences for artefacts, and standards for behaviour. In ABM, agents have a bounded rationality that is subject to stress, time pressure, and emotive forces. At the simplest level, an agent-based model consists of a system of agents and the relationships between them. Experience with agent-based modelling shows that even a simple agent-based model can exhibit complex behaviour patterns and provide valuable information about the dynamics of the real world system that emulates them. In this paper the two different simulation approaches to learning effectiveness, i.e., the agent-based modelling and systems dynamics are compared conceptually and the potential synergy between them is discussed. As such this paper is theoretical and exploratory in nature. Further studies are needed to provide empirical evidence to the observations and theories put forward in this paper.