Modelling cultural contagion using Social Impact Theory

In the context of computational social modelling, culture represents the broad set of attributes over which people may influence each other. Existing agent-based models typically have simple dyadic means by which individuals copy cultural traits from others. Building on these models, we instead embed forces from Social Impact Theory as the basis on which simulated agents influence each other. Agents are influenced based on their social status as well as their similarity to, and distance from, multiple connected contacts. These principles are gradually added to a novel agent-based model, allowing us to isolate and observe their effects. We find that these mechanisms cause a shift away from the global polarisation of existing models, toward a diverse state where cultures can overlap and mix. Application of the new model to different network topologies offers insights into the contexts which encourage either cultural convergence or pluralism. We develop methods for the generation of generalised hierarchical structures containing ‘team’ subgraphs, allowing us to investigate the spread of cultural traits within hypothetical organisations. The new model is also tested against existing cultural models on a real-world dataset to determine its ability to approximate actual behavioural contagion. The results support the notion that models incorporating elements of Social Impact Theory, particularly multiple sources of influence, more closely resemble reality. Our findings offer new insights into factors affecting the diffusion of traits and cultural overlap. These insights are potentially relevant to modelling the spread of misinformation, political polarisation, and the design of social and organisational networks with regard to their cultural repercussions