ABSTRACT This paper investigates the evolution of cooperation behaviors within innovation networks through a novel network public goods game framework. We extend the classic model by incorporating dual‐centered game structures and heterogeneous relationship strengths, where enterprises simultaneously participate in both self‐centered and neighbor‐centered cooperative innovation games. Our framework introduces a probability‐weighted participation mechanism where involvement in neighbor‐centered games is determined by both edge weights and a participation proportion parameter. Using numerical simulations (5000 steps) in scale‐free networks, we analyze the evolution of cooperative strategies under varying initial cooperation rates and self‐centered degrees. The results reveal a fundamental trade‐off between stability and dynamism: While high self‐centeredness () ensures rapid stabilization, moderate neighbor‐reliance (specifically ) drives dynamic breakthroughs essential for escaping low‐cooperation traps. Contrary to the expectation of serving as stable anchors, high‐enhancement factor nodes and high‐centrality nodes are found to amplify systemic volatility, particularly under medium initial conditions. Furthermore, the relationship between cooperative payoffs and strategy adoption exhibits a universal inverse non‐linear correlation, highlighting a temptation paradox where the spread of cooperation paradoxically lowers the relative payoff advantage. Additionally, the proposed mechanism demonstrates remarkable resilience, enabling the system to recover from unfavorable initial conditions through either stable rebounding or oscillatory surges. These findings advance our understanding of cooperative dynamics in innovation systems and provide practical guidance for enterprises in managing their dual roles as innovation leaders and partners.
Wang et al. (2026) studied this question.