Emotional Contagion Dynamics with Dual Recovery Channels on a Real-World Social Network: An Agent-Based Complex Systems Model

Collective emotion is an emergent property of social systems: individual affective states propagate along social ties and, in the aggregate, produce group-level patterns that rise, peak, and stabilize. This paper models emotional contagion as a dynamical process on a real social network, extending the susceptible–influenced–susceptible (SIS) framework with a third, aware state—the susceptible–influenced–aware (SIA) model—and separating two recovery routes that earlier single-channel models conflate: natural recovery, the passive fading of an emotional state, and aware recovery, the active acquisition of emotion-regulation capacity that confers partial protection against re-influence. We implement the agent-based model on the Facebook friendship network (N = 4039, E = 88,234). As a theory-building model rather than a data-calibrated one, it generates, under baseline parameters and across 30 independent simulations, a transient peak of roughly 33% influenced density followed by a stable plateau near 28%, with the aware density lagging behind throughout; these are model outcomes, not empirical estimates of real-world prevalence. Mean-field analysis yields a basic reproduction ratio R₀ = β/(μ + α) ≈ 1.94 governing a transcritical bifurcation between extinction and self-sustaining collective emotion; because the aware channel enters this ratio symmetrically with natural recovery, building regulation capacity is a first-order control on whether collective emotion self-sustains, not merely an aid to recovery. A second analytical result is that the two channels interact non-trivially: the steady-state fraction that acquires regulation capacity is non-monotonic in both the influence rate and the capacity-building rate, peaking at an interior optimum, so the most resilient population is produced by an optimal rather than maximal level of intervention. A social network analysis shows the network is small-world, heavy-tailed, and organized into sixteen strongly modular communities, with a small set of hub-and-bridge nodes mediating cross-community spread. By representing the two recovery channels as independent parameters, the model separates environment-oriented from capacity-oriented interventions and generates testable predictions for the regulation of collective emotion in digital and organizational settings.