Dissertation Research: Generalized Reciprocity: Understanding the Social Contagion of Altruistic Behavior

匿名的慷慨行为在日常生活中很普遍：人们献血，停下来帮助一个被困的司机，写匿名的在线评论，或者为开源软件贡献代码，通常没有任何直接的奖励或认可。这个谜题的一个重要关键可能是帮助行为具有传染性的可能性——在接受或观察帮助的情况下提供帮助，有可能引发远远超出原始行为的慷慨链。连接两大但基本上互不相关的现有文献——利他行为和社会传染——本论文项目确定了两种改变帮助陌生人可能性的机制：广义互惠（接受帮助的人可能会把帮助传递出去）和第三方模仿（给予帮助的观察者可能会效仿）。假设这些机制会对行为产生不同的影响：多次接受陌生人的帮助会增加帮助陌生人的可能性，而观察陌生人之间的帮助会先增加然后减少这种可能性；然而，对于那些接受过帮助的人来说，后者的减少幅度会小一些。因此，当第三方模仿与广义互惠相互作用时，慷慨的级联预计会更频繁，但影响的人会更少。为了检验宏观影响的有效性和稳健性以及假设机制的经验相关性，提出了两项研究：基于智能体模型的模拟实验和人类受试者的在线实验。更广泛的影响该项目将推进关于帮助行为及其传播和在线行为实验方法的理论知识。此外，慷慨的潜在连锁反应意味着，当我们帮助别人的时候，我们也可能间接地帮助更多下游的人。更好地了解触发这些连锁反应的机制，有可能通过更有效地针对那些过去受益的人，为动员自愿捐款和亲社会行为的努力提供信息。特别是，可以利用广义互惠和第三方模仿来加强大规模在线协作项目和用户生成内容社区的创造性参与和社区形成。

Michael MacyMilena TsvetkovaCornell UniversityAnonymous acts of generosity are widespread in daily life: people donate blood, stop to help a stranded motorist, write anonymous online reviews, or contribute code to open-source software, often without any direct reward or recognition. An important key to this puzzle may be the possibility that helping behavior can be contagious -- helping in response to receiving or observing help has the potential to trigger a chain of generosity that reaches far beyond the original act..Bridging two large but largely disconnected existing literatures -- on altruistic behavior and social contagion -- this dissertation project identifies two mechanisms that alter the probability to help a stranger: generalized reciprocity (a recipient of help is likely to pay it forward) and third party imitation (an observer of help giving is likely to emulate it). These mechanisms are hypothesized to affect behavior differently: while receiving help from strangers multiple times will increase the probability to help a stranger, observing help between strangers will initially increase and then decrease that probability; this latter decrease, however, will be smaller for those who have received help. As a result, when third-party imitation interacts with generalized reciprocity, cascades of generosity are expected to be more frequent but reach fewer people. To test the validity and robustness of the macro-implications and the empirical relevance of the hypothesized mechanisms, two studies are proposed: a simulation experiment with an agent-based model and an online experiment with human subjects.Broader ImpactsThe project will advance theoretical knowledge about helping behavior and its diffusion and methodologies for online behavioral experiments. In addition, the potential ripple effect of generosity means that when we help others, we may also be indirectly helping many more people downstream. Better understanding of the mechanisms that can trigger these cascades has the potential to inform efforts to mobilize voluntary contributions and pro-social behavior by more effectively targeting those who have benefited in the past. In particular, generalized reciprocity and third-party imitation can be harnessed to enhance creative participation and community formation in large-scale online collaborative projects and user-generated-content communities.