Temporal higher order interactions in social dynamics: From theoretical models to controlled experiments

Objective

This project investigates how group interactions, also known as higher-order interactions (HOIs), shape individual traits/opinions, with a particular focus on online social media environments. While traditional network science has largely focused on pairwise interactions, evidence from sociology, experimental psychology and complex systems theory suggests that group-level dynamics play a critical and qualitatively different role in shaping opinions, beliefs, and behaviors. Understanding these dynamics is essential for explaining phenomena such as polarization, radicalization, and fake-news diffusion, as well as for designing interventions to foster civic engagement and social cohesion. To capture the complexity of group interactions, the project will adopt higher-order network (HON) representations and develop novel methods to infer genuine HOIs from social media data using tools from multivariate information theory. These methods will enable the construction of realistic HONs that reflect actual group dynamics, overcoming limitations of current models that rely on simplified assumptions. The project will then analyze how group-level properties, such as size, internal communication patterns, and trust, affect individual attributes, including political leaning or stance on controversial topics. Building on these findings, a mechanistic model grounded in sociological theory will be developed and validated against empirical data. To ensure causal validity, the project will also include online social experiments with a control group, designed to isolate and test the mechanisms assumed in the model. By integrating theoretical modeling, empirical data analysis, and experimental validation, the project will contribute to a deeper understanding of online social dynamics and inform practical strategies to mitigate polarization and fragmentation in digital societies.