Macro-level reinforcement tunes the transition order of reversible social contagion

Abstract

Social contagion is often shaped by reinforcement: individuals become more likely to adopt a new behavior, opinion, or product as exposure accumulates or adoption becomes widely visible. Existing network models mainly capture this effect through local mechanisms, such as threshold responses or higher-order interactions. However, how macro-level reinforcement reshapes reversible spreading remains unclear. Here we study a SIS-like process in which pairwise transmission is reinforced by global prevalence. Combining quasistationary simulations and bifurcation analysis, we show that global feedback can produce first-order transition and hysteresis loop, with distinct activation and collapse thresholds. We further show how network localization promotes local ignition while weakening the global prevalence signal required for abrupt macroscopic activation, thereby raising the reinforcement threshold. Our results reveal how onset--retreat asymmetry emerges from global feedback coupled to network structure, providing a minimal mechanism for abrupt, history-dependent reversible social contagion.