Feature and location interactions in visual object processing

该项目旨在了解特征和位置偏好如何在人类视觉对象识别中相互作用，包括这种相互作用的感知后果及其潜在的神经机制。人类非常擅长识别复杂的视觉物体，尤其是人脸。我们的大脑是如何做到这一点的？视觉识别被认为依赖于下颞叶皮层（IT）的神经处理。IT子区域的神经反应显示出对不同视觉对象类别（例如面孔或地点）的明显偏好，并且传统上被描述为“位置不变”（即与视觉对象出现在视网膜上的位置无关）。然而，一系列的研究结果指出了刺激位置对物体识别的作用，并表明IT中的神经元可以具有高度的空间选择性。在此基础上视觉识别是如何工作的呢？最近来自申请人和猴子的独立研究的发现表明：特征和位置偏好以一种反映共同输入属性的方式匹配。拟议中的项目将在行为和神经成像实验中测试这一想法，以及它与健康和疾病中面部感知的其他方面的关系。例如，先前的研究结果表明，面部的眼睛和嘴巴分别最常出现在上下视野中——甚至在观察者的眼球运动中也是如此。此外，当观察者的眼睛和嘴出现在各自的典型视野位置时，他们确实更善于识别孤立的眼睛和嘴。这可以解释偏爱眼睛或嘴巴的神经群具有匹配的空间偏好。拟议的项目开发了实验范式来测试这一假设，以及空间和特征敏感性之间的相互作用可以解释长期研究的颠倒脸的感知效应和面孔失认症（“脸盲”）的症状。最后，它将测试这种特征-位置交互是否扩展到书面文字识别，这是一种获得视觉专业知识的视觉对象。

The proposed project aims to understand how feature and location preferences interact in visual object recognition in humans, including the perceptual consequences of such interactions and their underlying neural mechanisms. Humans are remarkably good at recognising complex visual objects, especially faces. How do our brains enable this? Visual recognition is thought to rely on neural processing in the inferior temporal cortex (IT). Neural responses of IT subregions show a clear preference for different visual object categories (e.g. faces or places), and are traditionally portrayed as 'location invariant' (that is, independent of where on the retina a visual object appears). However, a range of findings point to a role of stimulus location for object recognition and show that neurons in IT can be highly spatially selective. How can visual recognition work on this basis? Recent findings from the applicant and independent studies in monkeys suggest the following: Feature and location preferences are matched in a way reflecting common input properties. The proposed project will test this idea in behavioural and neuroimaging experiments and how it relates to other aspects of face perception in health and disease. For instance, previous results show that eyes and mouths of faces appear most often in the upper and lower visual field, respectively - even across eye movements of the observer. Furthermore, observers are indeed better at recognising isolated eyes and mouths when they appear at their respective typical visual field position. This could be explained neural populations with a preference for eyes or mouths having a matching spatial preference. The proposed project develops experimental paradigms to test this hypothesis, as well as the ones that such interactions between spatial and feature sensitivity can explain long-studied perceptual effects of turning faces upside down and symptoms of prosopagnosia ('face-blindness'). Finally, it will test whether such feature-location interactions extend to written word recognition, a type of visual object for which visual expertise is acquired.