Learning-based motivation of intergroup aggression

上个世纪有超过2亿人在种族灭绝、战争和其他形式的团体冲突中丧生，其中1.7亿是平民。更常见的群体间侵略形式，如政治冲突，弥漫在日常生活中，因此对经济的总体影响可能至少同样代价高昂。虽然个人可能会因为个人和群体间冲突而受到伤害他人的动机，但源自群体间环境的伤害动机可能特别危险。这种动机可能会增加攻击性，因为它允许在没有任何个人不满的情况下，将伤害作为道德上必要的正当理由。此外，想要攻击一个群体外成员的愿望可能会推广到他们的整个群体。因此，当进攻性动机在社会群体中展开时，理解它尤其重要。研究人员米娜·奇卡拉(哈佛大学)提出，对群体外的痛苦感到快乐是一种自然反应，它会使人们更容易学会一种原本令人反感的行为：积极地伤害他人。如果观察团队外成员的痛苦与感受快乐是一致的，随着时间的推移，人们可能会学会支持，甚至以有害的方式向团队外目标发泄。这个项目采用了一种新颖的、跨学科的方法，通过整合社会心理学和认知心理学来理解这些问题。该项目还解决了关于集团间侵略的出现和升级的知识方面的一个重大差距，并可提供加强国家安全的见解。在这个项目中测试的模型假设，群体间攻击的能力可能部分是通过基本的学习原则发展起来的。也就是说，基本的强化学习过程将感受到的快乐和群体外的痛苦结合在一起，可能会帮助人们克服对伤害他人的自然厌恶。一系列使用政治、国家和种族认同的实验测试了竞争性的外部群体(相对于内部群体和中立的外部群体)是否更有可能成为攻击的目标。实验背景包括一系列扩展的交互作用，它们测试攻击性是否在交互过程中升级。进一步的实验测试如果学习被打乱，攻击性是否会减少。这种学习中断要么是通过来自组内成员的负面社会反馈发生的，要么是当每一次激进的行动都需要评估相关成本和收益时发生的。这个项目建立在经典和当代学习理论的基础上。因此，它与其他科学研究领域，包括行为神经科学、经济学和生物学，以及从研究其他动物中获得的知识，有几个接触点。该项目的结果有可能为集中和廉价的认知行为干预提供信息，以减轻群体间攻击行为。这些发现可能会对有权制定和执行政策的政治和教育机构感兴趣。

The last century has seen over 200 million people, 170 million of which were civilians, killed in acts of genocide, war, and other forms of group conflict. More mundane forms of intergroup aggression such as political conflicts pervade everyday life, and as a consequence may be at least as costly in total impact on the economy. Although individuals can be motivated to harm others because of personal as well as inter-group conflict, motivation to harm that originates from intergroup contexts may be especially dangerous. Such motivation can increase aggression because it allows harm to be justified as being morally necessary in the absence of any personal grievance. Moreover, the desire to aggress against one out-group member may generalize to their entire group. Thus, the motivation to aggress is especially important to understand as it unfolds in social groups. The investigator Mina Cikara (Harvard University) proposes that feeling pleasure in response to out-group pain is a natural response that makes it easier to learn a behavior which is otherwise repugnant to individuals: actively doing harm to others. If observing the pain of out-group members is consistently linked with feeling pleasure, people may learn over time to support and even act out in harmful ways toward out-group targets. This project takes a novel, interdisciplinary approach to understanding these questions by integrating social and cognitive psychology. This project also addresses a major gap in knowledge regarding the emergence and escalation of intergroup aggression, and can provide insights that enhance national security. The model tested in this project posits that the capacity for intergroup aggression may have developed partly through basic learning principles. That is, basic reinforcement-learning processes that couple feeling pleasure and out-group pain may help people overcome a natural aversion to hurting others. A series of experiments using political, national, and ethnic identities test whether competitive out-groups (relative to in-group and neutral out-groups) are more likely to be targeted with aggression. The experimental contexts include an extended sequence of interactions and they test whether aggression escalates over the course of the interaction. Further experiments test whether aggression is reduced if learning is disrupted. This learning disruption takes place through either negative social feedback from in-group members or when each aggressive action requires evaluating the associated costs and benefits. This project builds on classic and contemporary theories of learning. As such, it makes several points of contact with other areas of scientific inquiry including behavioral neuroscience, economics, and biology, and with the knowledge gained from studying other animals. The results of this project have the potential to inform focused and inexpensive cognitive behavioral interventions to attenuate intergroup aggression. The findings may be of interest to political and educational institutions with the power to make and implement policy.