Decentralized Core-periphery Structure in Social Networks Accelerates Cultural Innovation in Agent-based Modeling

Previous investigations into creative and innovation networks have suggested that intellectual, artistic, and technological innovations often occur with the interaction of core and peripheral actors, with both individuals and teams at intermediary positions well-situated to innovate. In this work, we investigate the effect of global coreperiphery network structure on the speed and quality of cultural innovation. Drawing on differing notions of core-periphery structure from [22] and [2], we distinguish decentralized core-periphery, centralized core-periphery, and affinity network structure. We generate networks of these three classes from stochastic block models (SBMs), and use them to run an agent-based model (ABM) of collective cultural innovation, in which agents can only directly interact with their network neighbors. In order to discover the highestscoring innovation, agents must discover and combine the highest innovations from two completely parallel technology trees. We find that decentralized core-periphery networks outperform both centralized core-periphery networks and affinity networks, in terms of mean crossover time for this final innovation. We hypothesize that decentralized core-periphery structure provides a more fruitful environment for collective problem-solving, by allowing for the relative shielding of periphery nodes from the optimal innovations known by the core community at any given time. This prevents the disincentive for parallel explorations that emerges in a highly connected, centralized network. We then build upon the "Two Truths" hypothesis regarding community structure in spectral graph embeddings first articulated in [22]. The Two Truths hypothesis suggests that the adjacency spectral embedding (ASE) captures core-periphery community structure in a graph, while the Laplacian spectral embedding (LSE) captures affinity structure. For a given network generated from an SBM, we use ASE and LSE to resample new networks of similar structure, using either to parametrize a random dot product graph (RDPG) model. We find that, for core-periphery networks, ASE resampling best recreates networks with similar performance on the innovation SBM. Since the Two Truths hypothesis suggests that ASE captures core-periphery structure, this result further supports our hypothesis.