The article discusses agent-based simulation as a tool of sociological understanding. Based on an inferential account of understanding, it argues that computer simulations increase our explanatory understanding both by expanding our ability to make what-if inferences about social processes and by making these inferences more reliable. However, our ability to understand simulations limits our ability to understand real world phenomena through them. Thomas inferences about social processes and by making these inferences more reliable. The inferential account also suggests a number of ways in which the use of simulation methodology might give rise to illusory understanding.The structure of the article is as follows. I begin by describing the challenge of explaining macro social phenomena in terms of causal mechanisms. I then introduce the idea of ABS and explain why an increasing number of social scientists find it a promising tool for a mechanistic understanding of social phenomena. The sociological opportunities raised by ABS will be illustrated by Thomas Schelling's famous checkerboard model of ethnic segregation. The latter part of the paper will concentrate on giving a philosophical account of the contribution agent-based simulation models can make to sociological understanding. Then, employing the inferential account of explanatory understanding, I show how computer simulation can be understood as a case of extended cognition. I conclude the article with two applications of the proposed account of understanding. First, I articulate a possible source of illusory understanding in simulation studies: the confusion between understanding a simulation and understanding the target phenomenon using the simulation. Second, I reconstruct some reasons why highly abstract how-possibly explanations -such as Schelling's checkerboard model -are seen as providing explanatory insight.
Social Mechanisms and Agent-Based SimulationOne of the central challenges of sociology is to make sense of macro social facts. Such facts concern for example the distribution of individuals into different classes and status groups, social networks existing between entrepreneurs, the social norms characterizing social groups, and the segregation of members of various groups into different social activities or residential areas.While there is no general sociological agreement about the nature of sociological explanation, it has become increasingly popular to think that social explanation should explicate the causal mechanisms that generate the phenomenon to be explained (Hedström & Ylikoski 2010). As this view is almost unanimously accepted by social scientists who advocate ABS, I will use it as the starting point of my discussion.One of the basic premises of the mechanistic approach is that a proper understanding of collective processes requires paying attention to the entities that mechanisms are made of (the agents, their properties, actions, and relations) rather than treating them as black boxes. To account for macro-level phenomena i...