Agent-Based Models of In-Group Favoritism and Out-Group Hostility

该项目使用基于主体的模型来深入了解群体内偏爱和群体外敌意的基本方面，包括种族中心主义、种族冲突以及基于肤色、宗教或国籍等因素的歧视。这些问题的重要性是显而易见的。在美国，种族歧视是造成经济不平等的主要原因。在全球一级，种族冲突是地方性的，几乎有100个种族冲突同时发生。以前的模型假设群体的存在和成员是固定的，共享的成员关系会带来偏见。在这个项目中，基于主体的模型能够放弃这些限制性假设，从而允许调查个人如何形成他们的社会身份，内部群体如何变得连贯，以及为什么感知到的相似性往往成为偏爱的基础。基于代理的建模始于对个体（称为代理）、它们如何相互作用以及种群如何随时间变化的特定假设。一个基于代理的模型作为计算机模拟运行来生成人工历史。然后分析人工历史，看看随着时间的推移会发生什么以及原因。在这个项目中，模型强调为了洞察而简单，而不是试图对任何一个案例进行完全准确的描述。该项目建立在每个人都有一些可观察到的、相对稳定的特征的前提下，比如肤色、语言和宗教。然后，其他人可以利用这些特征来确定这个人是被视为“我们”中的一员还是“他们”中的一员。第一个模型旨在调查(1)群体内偏爱可能产生并持续存在的条件；(2)为什么种族群体之间的敌意如此普遍；(3)使歧视成本更高的可能影响是什么；(4)定居模式如何影响对移民的容忍。一个更先进的模型，称为多特征模型，被用来研究(1)为什么某些特征在歧视中比其他特征更被强调；(2)在群体之间划分边界时，代理人倾向于强调哪些社会类别；(3)即使在资源稀缺的世界里，什么样的临时干预可以对宽容、合作和平等产生持久的有益影响。该项目使用了广泛的社会科学学科，特别是政治学和社会学的事实、概念和理论。该项目还借鉴了进化生物学和计算机科学的概念和理论。这些发现为跨学科研究提供了新的机会，为分析群体内/群体外动态提供了新的视角，并为后续的实证检验提供了新的理论基础假设。对社会的广泛影响将是双重的。1. 理论模型的见解将为制定公共政策以预防、抑制或纠正群体内偏爱和群体外敌意所引起的问题提供更健全的基础。虽然理论模型本身永远无法提供有用的指导，但简单的正式模型的见解可以帮助分析师和政策制定者提供一个框架，在这个框架中提出可能富有成效的问题。特别是，研究结果将提出新的思路，思考如何将有限的资源最有效地集中在减少歧视、减轻移民带来的社会紧张局势、通过有针对性的教育增加宽容以及干预种族冲突上。2. 该项目将通过为高中到研究生院的学生提供一个网站，将教学和研究结合起来。这个网站将提供所有需要的资源，让一个几乎没有受过数学训练的学生运行基于主体的模型来探索群体内偏爱和群体外敌意的动态。学生和研究人员也将能够修改Java源代码来进行他们自己设计的新实验，并看到QuickTime电影显示他们的人口如何随着时间和空间而演变。该网站还将包括存档数据、建议学生练习以及未解决问题列表。

This project uses agent-based models to develop a deeper understanding of fundamental aspects of in-group favoritism and out-group hostility, in all their manifestations including ethnocentrism, ethnic conflict, and discrimination based on factors such as skin color, religion or national origin. The importance of these problems is manifest. In the US, racial discrimination is a major cause of economic inequality. At the global level, ethnic conflict is endemic, with almost one hundred ethnic conflicts active at the same time. Previous models have assumed that the existence and membership of groups are fixed, and that shared membership entails partiality. The agent-based models in this project are able to drop these restrictive assumptions, and thereby allow the investigation of how individuals form their social identity, how in-groups become coherent, and why perceived similarity often becomes the basis of favoritism. Agent-based modeling starts with specific assumptions about individuals (called agents), how they interact, and how the population changes over time. An agent-based model is run as a computer simulation to generate artificial histories. The artificial histories are then analyzed to see what happens over time and why. In this project, the models emphasize simplicity for the sake of insight, rather than try for a completely accurate depiction of any one case. The project is built on the premise that every person has some observable and relatively stable characteristics such as skin color, language, and religion. These characteristics can then be used by someone else to determine whether that person will be treated as one of "us" or one of "them". The first model is designed to investigate (1) the conditions under which in-group favoritism is likely to arise and persist, (2) why hostility between ethnic groups is so common, (3) what are the likely effects of making discrimination more costly, and (4) how settlement patterns can affect tolerance for immigrants. A more advanced model, called the Multi-Trait Model, is used to investigate (1) why some characteristics are emphasized more than others in discrimination, (2) what social categories agents tend to emphasize in drawing boundaries between groups, and (3) what temporary interventions can have lasting beneficial effects on tolerance, cooperation, and equality, even in a world with scarce resources. The project uses facts, concepts and theories from a broad range of social science disciplines, especially political science and sociology. The project also draws on concepts and theories from evolutionary biology and computer science. The findings provide new opportunities for interdisciplinary research, new perspectives for analyzing in-group/out-group dynamics, and new theoretically grounded hypotheses for later empirical testing. The broader impacts for society will be twofold. 1. The insights from the theoretical models will provide a sounder basis on which to make public policies to prevent, inhibit or correct the problems caused by in-group favoritism and out-group hostility. While a theoretical model can never by itself provide useful guidance, the insights of a simple formal model can be helpful to both analysts and policy makers in providing a framework in which to ask potentially fruitful questions. In particular, the results will suggest new ways of thinking about how limited resources might be most effectively focused to reduce discrimination, lessen social tensions from immigration, increase tolerance through targeted education, and intervene in ethnic conflicts. 2. The project will integrate teaching and research by providing a web site for students in high school through graduate school. The web site will have all the resources needed for a student with little mathematical training to run agent-based models to explore the dynamics of in-group favoritism and out-group hostility. Students and researchers will also be able to modify the Java source code to conduct new experiments of their own design, and to see QuickTime movies showing how their population evolves over time and space. The web site will also include archived data, suggested exercises for students, and a list of unsolved problems.