Socio-cognitive networks: Modeling the effects of space and memory on generative social structures

This paper examines the relationships between environmental and cognitive factors as they influenced the formation of social networks through modeling how these factors affect tie formation between pairs of agents in a simulated world. We modeled worlds consisting of 20, 40, and 60 ACT-R agents and examined the influence of population size, run time, map configuration, and navigation strategies, comparing the density and clustering of the resulting networks. We found that all these factors affect tie formation for agents with perfect memories, with population size having the greatest effect. We also examined the effect of these exogenous factors on the ties' strength in the agents' memories by combining and analyzing egonets. We found that changes to each of these exogenous factors affected the network's average memory activation value of each tie, with population size having a negative effect and run time having a positive effect. Map configuration and navigation strategy both influenced network structure. Further, we found that using the agents' activation values as a threshold for network inclusion was a useful way for identifying core groups and subgroups within the network. These findings provide further insights into the cognitive dimensions underlying networks and their structures, as reflected by Dunbar's (1998) number and Simmelian numbers. These results also show that these factors need to be reported when describing network simulations.