Abstractions on steroids: A critical realist approach to computer simulations

Abstract

AbstractComputer simulations are increasingly used within various fields of the social sciences, including analytical sociology and computational social science. However, the method has yet to fully find its place within critical realism, generally regarded as reductionist and methodological individualist and thus inadequate to the task of grasping the full and enigmatic complexity of social life. This article contributes to the discussion by suggesting an alternative approach to computer simulations in the social realm. Instead of representing reality, simulations are seen as a way of boosting abstraction by enabling us to isolate and study the effects of emergent mechanisms. The need for such tools is growing more urgent in an age characterized by rapid change and global networks of causal interrelation. Hence there is a need not only for new tools to deal with causal complexity within critical realism, but also for a critical realist perspective to fill the meta‐theoretical vacuum on which the simulation approach is largely based. The approach developed here provides a challenge to the predominant ways in which simulations are utilized today, with increasingly sophisticated models aiming towards realistic representation on the basis of empirical data.