What Matters in Social Networks: Defining Factors of Interest for Agent-­‐based Socio-­‐cognitive Simulations

This paper examines dyadic social tie formation, and how cognitive and environmental factors influence the formation of dyadic ties. Using agent models instantiated in ACT-R that interact in a large-scale social simulation, we capture the affect of environmental factors including population size, running time, and map configurations, as well as memory constraints. To explore and test these relationships, we ran simulations using a factorial design and used the simulation data to generate a large corpus of networks. Our analyses suggest five interesting conclusions: first, that the three environmental factors all influence both network density and some aspects of network structures; second, that agent memory strongly and decisively alters the network’s density and structure; third, the growth pattern of these networks approximates a power law distribution; fourth, that the environment structure influences the networks’ generation speed; and finally, certain map configurations tend to have more asymmetric activation patters. These findings are interesting in that they imply that the size of a social network primarily depends on internal cognitive factors rather than environmental factors, providing support for and deepening our understanding of Dunbar’s(1998) number. These findings also suggest that future simulations examining generative social networks should account for and carefully report these factors. Published as: (2015). Building social networks out of cognitive blocks: factors of interest in agent-based socio-cognitive simulations. Computational and Mathematical Organization Theory, 21(2), 115-149.