Deceit and Interactional Synchrony In Different Social Constellations

不同社会格局中的欺骗与互动同步

• 批准号: 1651118
• 负责人: Mark Frank
• 金额: $ 31.85万
• 依托单位: SUNY at Buffalo
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-05-01 至 2022-04-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1651118&HistoricalAwards=false
• 关键词: Deceit; Interactional; Synchrony; Different; Social

Humans are a social species, but the nature of the connection between any two persons varies. Typically, two friends have a stronger relationship than do two strangers. The nature of the interaction also varies depending on the messages conveyed, such as when a truth versus a lie is told. A measure that captures the fundamental social dynamics between people is called interactional synchrony, and it reflects the extent to which the behavior of two or more individuals correlates within a short time window. Past research shows that interactional synchrony strengthens a relationship. However, a cohesive synchrony measure that captures multiple facets of behavior has yet to be tested. This project explores the role of deception in interactional synchrony in three different social constellations: individuals who have built rapport with a stranger, individuals who are friends, and individuals who belong to the same social organization. The synchrony between an interviewer and an interviewee, of varying levels of social relationship, will be captured when the interviewee lies and tells the truth about an action. The research uses computer vision and machine learning techniques to analyze and uncover synchrony in the faces, vocal tones, bodies, postures, and sub-visible physiological responses of interviewers and interviewees. The results will inform the need to consider social ties in research on deceptive behavior and when assessing demeanor during real world interactions, such as interviews with criminal suspects or potential terrorists.This project explores the role of deception in interactional synchrony in three different social constellations: individuals who have built rapport with a stranger, individuals who are friends, and individuals who belong to the same social organization. Interactional synchrony is expected to be higher for interactions that involve high rapport, closer friendship, and shared group membership. The deception scenarios involve both sanctioned and unsanctioned lying. The variation in interactional synchrony as a result of lying compared to truth telling within participants and as a function of truth/lie interactions with rapport, friendship, and shared group membership will be evaluated. The analysis of synchrony using computer vision and machine learning algorithms applied to different channels, ranging from more controllable behaviors, such as posture, to those less controllable, such as electrodermal responses, will enable a global interactional synchrony measure and also aid in the understanding of interactional synchrony as a concept, separating intentional mimicry from synchrony that is more automatic and unconscious. This refinement in the measurement technology of interactional synchrony will be of special value in detecting deceit, thereby contributing to the nation's security priorities.

人类是一个社会性的物种，但任何两个人之间的联系的性质是不同的。通常，两个朋友的关系比两个陌生人的关系更牢固。互动的性质也会因所传达的信息而异，比如说，什么时候说真话，什么时候说假话。一个捕捉人与人之间基本社会动态的指标被称为时间间隔同步，它反映了两个或多个人的行为在短时间内的相关程度。过去的研究表明，跨期同步加强了一种关系。然而，一个内聚同步措施，捕捉行为的多个方面还有待测试。这个项目探讨了欺骗在三个不同的社会群体中的互动同步中的作用：与陌生人建立融洽关系的个体，朋友的个体，以及属于同一社会组织的个体。采访者和被采访者之间的同步性，不同层次的社会关系，将被捕获时，被采访者说谎，并告诉真相的行动。该研究使用计算机视觉和机器学习技术来分析和揭示面试官和受访者的面部，语调，身体，姿势和亚可见生理反应的同步性。研究结果将有助于我们在研究欺骗行为时，以及在评估真实的世界互动中的行为举止时，考虑到社会联系的必要性，例如与犯罪嫌疑人或潜在恐怖分子的访谈。与陌生人建立了融洽关系的人，朋友，和属于同一个社会组织的个体。对于涉及高度融洽、更密切的友谊和共享的群体成员的互动，预期间歇同步性更高。欺骗的场景包括认可和未经认可的谎言。将评估作为说谎与在参与者中讲真话相比的结果以及作为真实/谎言与融洽、友谊和共享组成员关系的交互作用的函数的间期同步的变化。使用应用于不同渠道的计算机视觉和机器学习算法对同步性进行分析，从更可控的行为（如姿势）到那些不太可控的行为（如皮肤电反应），将实现全球间歇性同步性测量，并有助于将间歇性同步性理解为一个概念，将有意模仿与更自动和无意识的同步性分开。这种对时间间隔同步测量技术的改进将在检测欺骗方面具有特殊价值，从而有助于国家的安全优先事项。

项目成果

Enhancing Human Face Recognition with an Interpretable Neural Network

使用可解释的神经网络增强人脸识别

• DOI: 10.1109/iccvw.2019.00064
• 发表时间: 2019
• 期刊: IEEE/CVF International Conference on Computer Vision Workshop (ICCVW
• 作者: Zee, Timothy;Gali, Geeta;Nwogu, Ifeoma
• 通讯作者: Nwogu, Ifeoma