The statistical physics of decision-making in insect colonies

We apply the stochastic methods of statistical physics to analyse collective-decision making in social insect colonies, allowing us to derive the colony-level behaviour from an individual-level model. This contrasts with the traditional approach where a differential equation model, with or without arbitrary noise terms, is assumed. Social insect colonies vary in size from on the order 100 to 10,000,000 individuals, and such a statistical physics approach allows us explicitly to derive equations for both the average behaviour and the noise in the system, across this entire scale. We develop such a framework by building upon an existing stochastic model of opinion formation to model the decision-making processes in emigrating ant colonies. This new model is both driven by and evaluated against results from experiments with rock ants. This allows us to elucidate rigorously the role played by the individual-level phenomena of direct switching in the colony-level decision-making process, which optimality theory has predicted to be of crucial importance, and which we compare with our experimental results. This illustrates the power of the stochastic methods of statistical physics for understanding social insect colonies as complex systems.