The Physical Modelling of Human Social Systems

Publisher Website
Google Scholar
Abstract
One of the core assumptions in the study of complex systems is that there exist ‘universal’ features analogous to those that characterize the notion of universality in statistical physics. That is to say, sometimes the details do not matter: certain aspects of complex behaviour transcend the particularities of a given system, and are to be anticipated in any system of a multitude of simultaneously interacting components. There can be no tougher test of this idea than that posed by the nature of human social systems. Can there really be any similarities between, say, a collection of inanimate particles in a fluid interacting via simple, mathematically defined forces of attraction and repulsion, and communities of people each of whom is governed by an unfathomable wealth of psychological complexity? The traditional approach to the social sciences has tended to view these psychological factors as irreducible components of human social interactions. But attempts to model society using the methods and tools of statistical physics have now provided ample reason to suppose that, in many situations, the behaviour of large groups of people can be understood on the basis of very simple interaction rules, so that individuals act essentially as automata responding to a few key stimuli in their environment. This is clearly a challenging, perhaps even disturbing, idea. I will review some of the evidence in its favour. I will show, with examples ranging from pedestrian dynamics to social choice theory, economics, demography, and the formation of businesses and alliances, that modelling the behaviour of individuals and social and political institutions according to the viewpoint of statistical physics does seem capable of capturing some of the important features of social systems. These models reveal many of the characteristic elements displayed by other complex systems: collective dynamics that changes via abrupt shifts (phase transitions), metastability, critical phenomena and scale-free statistical variations. I will discuss what this implies for the notions of human free will and determinism.