激励机制下合作行为的演化动力学

Evolutionary dynamics of cooperation driven by incentives

中文关键词  演化博弈论  合作行为  激励机制

英文关键词  evolutionary game theory  cooperation  incentives

中文摘要

      探索复杂系统中控制自私个体合作演化的激励机制, 是国际系统与控制等诸多领域关注的重要科学问题. 研究 者曾探讨了引入惩罚机制对系统中个体合作的促进作用. 但实际中, 惩罚者可能会受到奖励的正激励, 以及被惩罚个体 报复的负激励影响. 本文基于演化博弈论分别研究分散式和集中式惩罚机制下的合作演化, 重点关注对惩罚行为施加不 同的激励作用对系统演化的影响. 研究发现, 当系统中存在对惩罚行为的负激励时, 分散式惩罚机制对合作行为的促进 作用被完全抑制; 然而集中式惩罚机制却依然可以促进合作行为的涌现. 另一方面, 对惩罚行为的正激励可以有效地抵 制负激励的负面影响, 从而使合作涌现. 特别地, 在采用集中式激励的大规模种群中, 合作行为和惩罚行为能够在自然 选择意义下占优. 激励机制的研究对于复杂系统优化调度与配置、群体机器人协同控制等实际问题具有重要的理论意 义和应用价值.

英文摘要

      Cooperation, a conundrum from the evolutionary perspective, has received increasing attention from various realms. Punishment has been found to be effective among all the mechanisms to promote cooperation. However, retaliation undermines the positive role of punishment in the evolution of cooperation. This results in the query on the extensive application of punishment in nature. Moreover, reward for punishment which is ubiquitous has been overlooked in previous studies. To explore the underlying principle of such phenomena, we propose a simple model and study the evolution of cooperation when retaliation and reward are allowed in the context of decentralized and centralized punishments. The results illustrate that cooperation is never facilitated by decentralized punishment while it can dominate defection when punishment is centralized in the presence of retaliation. On the other hand, reward for altruistic punishment significantly improve the situation of cooperation such that cooperation can be advantageous over defection no matter whether the retaliation exists or not. Moreover, we find that cooperation and punishment can emerge in large scale populations with centralized punishment. Our study paves the way for understanding the roles of incentives in the evolution of cooperation. Furthermore, it has important theoretical significance and engineering value for some practical problems, such as optimal scheduling and configuration in complex systems, and coordinated control of swarm robot systems.