Exploration-exploitation tradeoff and opinion diversity in real-world groups

社区在探索问题的新解决方案和利用现有解决方案之间摇摆不定，这取决于手头的任务。因此，关于可能解决方案的意见的多样性会随着时间的推移而消长。这些意见多样性的变化如何影响群体表现是一个悬而未决的问题，现有的研究发现了积极和消极的影响。这个项目将调查这些不一致的发现是否可以通过多样性如何与任务复杂性和群体威胁相互作用来解释。当群体面临复杂的任务时，对可能的解决方案的意见的较高多样性可能是适应性的，而较低的多样性则有助于集体解决简单的任务。此外，拥有更一致观点的群体，因此能够更好地协调，通常在群体间冲突中具有优势。因此，当群体威胁导致意见多样性减少时，它可能对复杂任务的表现产生负面影响，但对简单任务的表现产生中性甚至积极的影响。此外，群体可以根据任务复杂性和所经历的威胁使用策略来自适应地操纵其多样性水平。因此，该项目还将调查这些策略的适应价值，包括维持治安、排斥和诋毁外围群体。总的来说，了解促进或抑制意见多样性的因素在今天的国家安全问题上非常重要。当前的通信技术为与许多不同的人互动提供了前所未有的机会，但它们似乎也促进了具有相似观点的人形成相对孤立的社区。要维持和促进健康的民主对话，就必须从理论上和经验上更好地了解影响现实世界社区意见多样性的因素，以及缓和多样性对寻找良好社会解决办法的价值的因素。为了促进这一领域的进一步研究，该项目还将为认知科学与社会科学、计算语言学和网络科学交叉领域的学生提供独特的培训机会，并包括两个跨学科工作组，讨论理解现代社会多样性的理论和方法进展的有希望的方向。为了研究任务复杂性和群体威胁如何影响意见多样性和由此产生的群体表现，并调查现实世界社区在应对这些因素时用来管理其多样性的机制，该项目将首先整合先前关于探索-开发权衡、威胁下的群体行为和决策、社会学习、利他惩罚、排斥和刻板印象的一些不相关的文献。由此产生的假设将使用三种不同的方法进行测试：计算建模、小组实验和计算语言学分析，这些方法涉及数千名用户在很长一段时间内参与在线新闻网站的讨论。这将允许鲁棒性检查并降低仅依赖一种方法的风险。这些理论和方法的发展将促进对现代人类社会的进一步科学研究。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Communities oscillate between exploring new solutions to problems and exploiting existing ones, depending on the task at hand. As a consequence, the diversity of opinions about possible solutions waxes and wanes over time. How these changes in diversity of opinion affect group performance is an open question, with extant studies finding both positive and negative effects. This project will investigate whether these inconsistent findings can be explained by how diversity interacts with task complexity and group threat. Higher diversity of opinions about possible solutions can be adaptive when groups face complex tasks, while lower diversity can help in collective solving of simple tasks. In addition, groups that have more homogeneous views, and are therefore better coordinated, often have an advantage in intergroup conflict. As a consequence, when group threat leads to decreased diversity of opinions, it might have negative effects on performance in complex tasks, but neutral or even positive effects for simple tasks. Furthermore, groups can be expected to use strategies to adaptively manipulate their level of diversity according to task complexity and experienced threat. This project will therefore also investigate the adaptive value of such strategies, including policing, ostracism, and denigrating outgroups. Overall, understanding factors that promote or suppress the diversity of opinions is immensely important today for national security issues. Current communication technologies offer unprecedented opportunities for interactions with many different people, and yet they also seem to facilitate formation of relatively isolated communities of people with similar opinions. Better theoretical and empirical understanding of factors that influence the diversity of opinions in real-world communities, and that moderate the value of diversity for finding good societal solutions, is necessary for maintaining and promoting a healthy democratic dialogue. To promote further research in this area, the project will also provide unique training opportunities for students at the intersection of cognitive and social sciences, computational linguistics, and network science, and include two interdisciplinary working groups discussing promising directions for theoretical and methodological progress in understanding diversity in modern societies.To study how task complexity and group threat influence opinion diversity and the resulting group performance, and to investigate mechanisms that real-world communities use to manage their diversity in response to these factors, the project will first integrate several previously disconnected literatures on exploration-exploitation trade-offs, group behavior and decision making under threat, social learning, altruistic punishment, ostracism, and stereotyping. The resulting hypotheses will be tested using three different methodologies: computational modeling, group experiments, and computational linguistics analyses of discussions on online news sites involving thousands of users over prolonged time periods. This will allow for robustness checks and mitigation of risks of relying on only one method. These theoretical and methodological developments will foster further scientific research of modern human sociality.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.