From Online Opinion Polarization to Collective Action: A Sociotechnical Framework for Behavior–Opinion Coevolution
Main article
Abstract
The emergence of social media platforms as dominant arenas of public discourse has fundamentally transformed opinion formation processes and collective behavioral dynamics, raising urgent questions about how individual viewpoints and group actions coevolve within complex sociotechnical ecosystems. Existing research on opinion dynamics typically examines belief evolution in isolation from behavioral outcomes, while classical collective action theories often underestimate the cognitive dimensions underpinning participation decisions. To bridge this gap, this paper proposes a Behavior–Opinion Coevolution (BOC) framework grounded in sociotechnical systems theory. The framework integrates a DeGroot-inspired weighted opinion update mechanism on an opinion-network layer with a three-mechanism behavioral choice model—comprising neighbor imitation, payoff-driven adjustment, and cognitive consistency maintenance—on a parallel interaction-network layer. Through systematic agent-based simulation across four canonical network topologies (Erdős–Rényi, Barabási–Albert, Watts–Strogatz, and regular lattice) under both synchronous and asynchronous update regimes, we characterize the coevolutionary trajectories of group opinion consensus and cooperative behavior diffusion. Results reveal a consistent behavior-leading, opinion-lagging pattern: cooperative behaviors stabilize significantly earlier than corresponding opinion convergence, on average by 60–90 time steps. Scale-free networks demonstrate superior convergence speed and higher equilibrium cooperation rates compared to random and lattice structures, reflecting the amplifying role of hub nodes in opinion–behavior coevolution. Empirical validation using longitudinal survey data on environmental attitudes and pro-environmental behaviors in China confirms the model’s explanatory power, with behavioral intentions preceding attitudinal convergence by approximately 45–60 days in the sampled panel. The framework offers actionable insights for platform governance, suggesting that targeted norm-anchoring interventions at structurally central network positions can accelerate collective action without requiring prior consensus formation.
