Evolution and Ecology Program, International Institute for Applied Systems Analysis, Laxenburg, A-2361, Austria; Program in Quantitative and Computational Biology, Princeton University, Princeton, NJ 08544, USA.
Evolution and Ecology Program, International Institute for Applied Systems Analysis, Laxenburg, A-2361, Austria; Department of Mathematics and Mathematical Statistics, Umeå University, Umeå, 901 87, Sweden.
J Theor Biol. 2018 Apr 14;443:56-65. doi: 10.1016/j.jtbi.2018.01.005. Epub 2018 Jan 11.
Cooperation is ubiquitous in biological and social systems, even though cooperative behavior is often costly and at risk of exploitation by non-cooperators. Several studies have demonstrated that indirect reciprocity, whereby some members of a group observe the behaviors of their peers and use this information to discriminate against previously uncooperative agents in the future, can promote prosocial behavior. Some studies have shown that differential propensities of interacting among and between different types of agents (interaction assortment) can increase the effectiveness of indirect reciprocity. No previous studies have, however, considered differential propensities of observing the behaviors of different types of agents (information assortment). Furthermore, most previous studies have assumed that discriminators possess perfect information about others and incur no costs for gathering and storing this information. Here, we (1) consider both interaction assortment and information assortment, (2) assume discriminators have limited information about others, and (3) introduce a cost for information gathering and storage, in order to understand how the ability of discriminators to stabilize cooperation is affected by these steps toward increased realism. We report the following findings. First, cooperation can persist when agents preferentially interact with agents of other types or when discriminators preferentially observe other discriminators, even when they have limited information. Second, contrary to intuition, increasing the amount of information available to discriminators can exacerbate defection. Third, introducing costs of gathering and storing information makes it more difficult for discriminators to stabilize cooperation. Our study is one of only a few studies to date that show how negative interaction assortment can promote cooperation and broadens the set of circumstances in which it is know that cooperation can be maintained.
合作在生物和社会系统中普遍存在，尽管合作行为通常是昂贵的，并且有被不合作者利用的风险。有几项研究表明，间接互惠，即群体中的一些成员观察他们同伴的行为，并利用这些信息来歧视未来不合作的代理人，可以促进亲社会行为。一些研究表明，不同类型的代理人之间相互作用的倾向性差异（相互作用分类）可以提高间接互惠的有效性。然而，以前没有研究考虑过不同类型的代理人之间观察行为的倾向性差异（信息分类）。此外，大多数以前的研究都假设鉴别器具有关于他人的完美信息，并且不花费任何成本来收集和存储这些信息。在这里，我们（1）同时考虑相互作用分类和信息分类，（2）假设鉴别器对他人的信息有限，（3）引入了信息收集和存储的成本，以便了解鉴别器稳定合作的能力如何受到这些朝着增加现实性的步骤的影响。我们报告了以下发现。首先，当代理人优先与其他类型的代理人相互作用，或者当鉴别器优先观察其他鉴别器时，即使他们的信息有限，合作也可以持续。其次，与直觉相反，增加鉴别器可用的信息量会加剧背叛。第三，引入信息收集和存储的成本使得鉴别器更难以稳定合作。我们的研究是迄今为止为数不多的研究之一，表明了负面相互作用分类如何促进合作，并扩大了已知合作可以维持的情况范围。
