Multiple complex emergent phenomena in a noise-driven statistical mechanical model of social dynamics

We introduce an agent-based model of bilateral conflict and cooperation, unique in its construction and ability to generate three complex nonlinear emergent behaviors. It considers a conceptual index of cooperativity and consists of two partially coupled lattices containing three agent types: individualists, networkers, and reciprocators. The emergent phenomena are as follows. (1) A threshold process arises: if reciprocators out-number individualists by a certain margin, society-wide cooperativity is trapped in its initial condition; otherwise, it proceeds to a quasi-dynamical equilibrium. (2) Some asymmetric parameterizations between the populations produce aperiodic oscillations and a scale-free power spectrum in the time series of system-wide mean cooperativity with frequent small fluctuations and rare large events. This is a dynamical fractal akin to but genetically very different from that generated by Bak-Tang-Wiesenfeld or other cellular automaton models. (3) Omitting reciprocators gives a power-law degree distribution in the complex network formed by inter-agent correlations, with many poorly connected agents and a few super-nodes. This is a scale-free network akin to but genetically very different from that generated by, for example, the preferential attachment model.