The speed of holistic face processing

熟悉的朋友或同事的脸会立刻被认出来，而不需要有意识的努力或预先考虑。尽管人脸识别发生得很快，但人们对准确的时间、潜在的认知操作和调节快速识别的大脑机制知之甚少。我提出的研究将引入一种新颖而强大的技术来测试人脸识别的“如何”和“何时”。我们假设快速人脸识别需要整体处理，其中人脸被识别为整个面部刺激，而不是识别其单个部分。人脸识别文献中一个未经检验的问题是整体人脸处理发生的速度。在我和一名博士后实习生进行的快速整体处理（RHP）的初步测试中，一张脸在屏幕上闪现17到500毫秒，然后戴上白噪声面具。关键的发现是，在最短的17ms时间内，整个面部的眼睛比单独的眼睛更容易被识别（Tanaka & Xu, 2018）。虽然我们的研究结果表明，整体人脸识别发生得非常快，但需要更严格的快速人脸识别心理物理测试来确定RHP的阈值；也就是说，整个面部优势出现的精确曝光时间。在接下来的五年里，我的研究小组将首先通过融合心理物理范式来评估RHP阈值，使用超高速240 Hz显示器，能够以4毫秒的速度呈现。在确定RHP阈值之后，我们将继续研究RHP的概括性和局限性。我们将尝试通过颠倒面部刺激或打乱其内部特征来破坏面部的RHP。我们还将测试RHP是否扩展到对面部情绪和非面部物体（即房屋）的感知。在这些实验的基础上，我们将通过调查RHP在整个生命周期中的影响，以及在现实世界和实验室训练的专家中研究感知经验和学习的影响。在相关工作中，我们将使用脑电图（EEG）中的频率标记方法分离快速整体识别面部身份和情绪的大脑动力学，其中单个面部部位（眼睛，鼻子，嘴巴）将以不同的频率显示。通过关注整体处理的时间动态，我们的研究将在人脸和物体识别方面开辟令人兴奋的，迄今为止尚未探索的研究途径。这些实验将产生重要的概念见解，将加深我们对涉及人脸识别速度的认知和神经基础理论的理解。这项工作将极大地推动这一领域的发展，并为评估健康成年人和患有面部识别能力缺陷的人（如自闭症患者、精神分裂症患者和痴呆症患者）的面部识别带来新的方法。

The face of a familiar friend or colleague is recognized instantaneously without conscious effort or forethought. Although face recognition occurs quickly, little is known about the exact timing and the underlying cognitive operations and brain mechanisms that mediate fast recognition. My proposed research will introduce a novel and powerful technique to test the "how" and "when" of face recognition. We assume that rapid face recognition requires holistic processing in which a face is recognized as a whole face stimulus rather than recognition of its individual parts. An untested question in the face recognition literature is the speed at which holistic face processing occurs. In a preliminary test of rapid holistic processing (RHP) I conducted with a postdoctoral trainee, a face was flashed on the screen for 17 to 500 ms, followed by a white-noise mask. The key finding was that the eyes were better recognized in the whole face than in isolation when presented for the shortest duration of 17ms (Tanaka & Xu, 2018). Although our results suggest that holistic face recognition occurs very quickly, more stringent psychophysical tests of rapid face recognition are warranted to identify the threshold of RHP; that is, the precise exposure duration at which the whole face advantage emerges. Over the next five years, my research group will first assess the RHP threshold across converging psychophysical paradigms with an ultra-high speed 240 Hz display monitor, capable of presentation rates of 4 ms. After determining the RHP threshold, we will move on to examine the generalizability and limits of RHP. We will attempt to disrupt RHP for faces by inverting the face stimulus or scrambling its internal features. We will also test whether RHP extends to the perception of facial emotion and non-face objects (i.e., houses). Building on these experiments, we will examine the effects of perceptual experience and learning by investigating the RHP across the lifespan and in real-world and laboratory-trained experts. In related work, we will isolate the brain dynamics of rapid holistic recognition of face identity and emotion by using a frequency tagging method in electroencephalography (EEG) where individual face parts (eye, nose, mouth) will be shown at different frequency rates. By focusing on the temporal dynamics of holistic processing, our research will open up exciting, and as yet, unexplored avenues of research in face and object recognition. The experiments will yield important conceptual insights that will deepen our understanding of cognitive and neural-based theories involving the speed of face recognition. The work will significantly advance the field and lead to new measures for assessing face recognition in healthy adults and in people who experience deficits in their face recognition abilities, such as individuals with autism, schizophrenia and dementia.