Model-Based Investigation of Aberrant Neural Face Representation in Autism

项目摘要 人脸通常传达着丰富的信息，而处理人脸是大多数社交活动的焦点 互动。当我们看到一个人的脸时，我们可以很容易地识别出他们的独特身份和一般特征 例如种族、性别和年龄。面部处理的格式塔使我们能够对一个 人的情绪或其他方面，如他们的可信度水平。然而，这个简单的感性任务是不同的fi崇拜 对于患有自闭症谱系障碍(ASD)的个人，花有限时间参与的人群 面对面的眼神交流或一般的社交互动。虽然有一大堆关于脸的文献 感知和许多研究都记录了ASD患者的异常面部处理，大多数存在 研究的重点是对面孔和情绪表达的识别，或对特定社会的感知 属性(例如，可信度)。在很大程度上还不清楚大脑是如何代表和评估面孔的。 一般情况下，以及这种机制在ASD中是否/如何不同。人脸处理在ASD中的研究非常多 这很重要，因为它不仅有助于我们了解这种疾病的社会原因，而且还提供了一种 独特的机会研究与正常人脸加工的功能专业化相关的因素。 在这个项目中，我们建议进行一项非常fi的研究来研究神经脸。 在患有自闭症的个体中的表现，并描绘涉及编码面部特征的大脑区域。 重要的是，通过使用同步功能磁共振成像(FMRI)和眼睛跟踪和拍摄 利用深度神经网络(DNN)的最新进展，我们能够提取基于关联的 任何人脸的特征，并合成新的人脸，以验证模型预测。的主要目标 本研究主要做了两个方面的工作：(1)通过构造和建立人脸的一般神经表征 验证神经面孔模型，以及(2)比较ASD患者之间的面孔神经表征 和控制。认知神经科学和计算机科学在这个项目中的合作提供了 这是一个更好地了解自闭症患者如何感知人脸的独特机会，特别是fi 一般情况下，大脑机制涉及到表征面孔。达到这种程度的理解 面部表征的神经计算基础将是我们对面部理解的独特之处 在没有ASD的对照组和有ASD的对照组中进行处理。反过来，这项研究可能会为 这种弥漫性的defi在自闭症中的发展轨迹和潜在的干预目标。

Project Summary Faces often convey a wealth of information and processing the human face is at the focal point of most social interactions. When we see a person's face, we can easily recognize their unique identity and general features such as race, gender, and age. The gestalt of facial processing enables us to make judgments about a person's mood or other aspects such as their level of trustworthiness. Yet, this simple perceptual task is difficult for individuals with autism spectrum disorder (ASD), a population that spends limited amounts of time engaged in face-to-face eye contact or social interactions in general. Although there is a large body of literature on face perception and many studies have documented abnormal face processing in people with ASD, most existing studies focus on the recognition of faces and emotional expressions or on perception of a particular social attribute (e.g., trustworthiness). It remains largely unclear how the brain represents and evaluates faces in general, and whether/how this mechanism differs in ASD. The study of face processing in ASD is very important because it will not only help us understand the social deficits of this disorder but also provide a unique opportunity to study the factors related to the functional specialization of normal face processing. In this project, we propose to conduct one of the very first studies to investigate neural face representation in individuals with ASD and delineate those brain regions involved in coding facial features. Importantly, by using concurrent functional magnetic resonance imaging (fMRI) and eye tracking and taking advantage of recent advances in deep neural networks (DNNs), we are able to extract association-based features from any face and synthesize new faces for validating model predictions. The primary objectives of this research are two-fold: (1) to establish a general neural representation of faces by constructing and validating neural face models, and (2) to compare neural representations of faces between people with ASD and controls. The collaboration between cognitive neuroscience and computer science in this project provides a unique opportunity to better understand how individuals with ASD perceive human faces, specifically what brain mechanisms are involved in representing faces in general. Obtaining this level of understanding of the neural computational underpinnings of face representation will be unique to our understanding of face processing in controls without ASD as well as those with ASD. In turn, this research may provide insights into the developmental trajectory of this pervasive deficit in autism and potential targets for intervention.