The role of shape and texture within the neural representation of faces.

拟议中的博士项目将加入一个动态和快速增长的文献主体，该文献涉及人类大脑如何计算和编码视觉刺激。长期以来，人们已经证实，大脑中有一些区域对某些类型的刺激(参照物)更敏感，例如位于枕叶外侧皮质的区域(纹状体外区域)已被证明对身体及其组成部分做出反应，但对物体的反应最小(Downing等人，2001年)。同样，神经成像研究揭示了枕叶和颞叶区域的网络，这构成了对人脸进行视觉分析的核心系统(Haxby，Hoffman&Gobbini，2000；Kanwisher，McDermott&Chun，1997)。这些研究一致地发现，在这个网络中，对面孔的反应比其他视觉刺激更强的区域是：梭状脸区(FFA)、枕面区(OFA)和后上颞沟(PSTS)。这些组成区域被发现处理面孔的不同方面，例如，FFA与身份的加工有关，PSTS与面孔的变化方面的加工有关，如表情或观点(Haxby，Hoffman&Gobbini，2000)。虽然有证据支持这种普遍的分离，但这些区域的神经编码尚不清楚。例如，神经编码是基于身份和表情等高级属性的，还是更基本、更一般的神经编码支持这些人脸区域的表征。许多研究表明，纹理(可以被认为是由于面部特征的形状和位置以及面部内部更广泛的反射模式而导致的反射率的突然变化)在用于识别熟悉的人脸的不变表征中起着关键作用(Burton，Jenkins，Hancock&White，2005；Russell&Sinha，2007)。例如，如果纹理呈现在标准化形状上(Burton，Jenkins，Hancock&White，2005)，或者当形状通过拉伸图像而扭曲时(Hole，George，Eaves&Rasek，2002)，熟悉的人脸识别不会受到实质性的影响。相比之下，缺乏任何表面属性的脸部线条画通常不足以识别(Davies，Ellis，&Shepherd，1978)。综上所述，这些研究表明，纹理是人脸识别的主要线索(Burton，2013)。相比之下，表情的判断通常被认为主要基于关键表情特征的形状和位置，如眉毛、眼睛、鼻子和嘴巴。形状线索在面部表情识别中的重要性的证据来自对比度反转(如在照片底片中)。在对比度反转的图像中，定义特征形状属性的边保持在相同的位置，尽管总体表面属性(纹理)发生了巨大变化。尽管对比否定对面部身份识别的破坏性是众所周知的(Bruce&Young，1998)，但事实证明，基于面部表情的判断在对比反转的图像中仍然是可能的(Bruce&Young，1998；Harris，Young&Andrews，2014；White，2001)。目前的提案将确定形状和纹理在人脸区域中的作用。这样做的目的是为了更好地理解大脑是如何表示关于人脸的信息的。这项研究的主要假设是，面部区域的神经表征不是直接与身份或表情等高级概念联系在一起，而是基于更基本的底层表征，如形状或纹理。这种类型的表示将提供更通用的代码，可以用来以更灵活的方式解码来自人脸的信息。

The proposed PhD project will join a dynamic and rapidly growing body of literature concerning the unraveling of how the human brain computes and codes for visual stimuli. It has been long established that within the brain there are areas that are more sensitive for certain types of stimuli (reference) for example an area located in the lateral occipital cortex (the extrastriate body area) has been shown to respond to bodies and their constituent components but minimally for objects (Downing et al, 2001). Likewise neuroimaging studies have revealed a network of regions in the occipital and temporal lobes, which form a core system for the visual analysis of faces (Haxby, Hoffman & Gobbini, 2000; Kanwisher, McDermott & Chun, 1997). These studies have consistently found regions within this network that show stronger responses to faces than other visual stimuli: the fusiform face area (FFA), the occipital face area (OFA) and the posterior superior temporal sulcus (pSTS). These constituent areas have been found to process different aspects of faces, for example; the FFA is linked to the processing of identity and the pSTS is linked to the processing of changeable aspects of faces such as expression or viewpoint (Haxby, Hoffman & Gobbini, 2000). Although there is evidence in support of this general dissociation, the neural code in these regions is not clear. For example, is the neural code based on high-level properties such as identity and expression or can a more basic and general neural code underpin the representation in these face regions.A number of studies have shown that the texture (which can be thought of as abrupt changes in reflectance due to the shapes and positions of facial features and a more broader pattern of reflectance within the face) plays a critical role in the invariant representation that is used for the recognition of familiar faces (Burton, Jenkins, Hancock & White, 2005; Russell & Sinha, 2007). For example, familiar face recognition is not substantially affected if the texture is presented on a standardized shape (Burton, Jenkins, Hancock & White, 2005) or when the shape is distorted by stretching the image (Hole, George, Eaves & Rasek, 2002). In contrast, line drawings of faces, which lack any surface properties, are not usually sufficient for recognition Davies, Ellis, & Shepherd, 1978). Together, these studies suggest that texture is the dominant cue in face recognition (Burton, 2013).In contrast judgements of expression are often thought to be based primarily on the shapes and positions of critical expressive features such as the eyebrows, eyes, nose and mouth. Evidence for the importance of shape cues in facial expression recognition comes from contrast reversal (as in a photo negative). In a contrast-reversed image, the edges that define feature shape properties remain in the same positions, despite the huge change in overall surface properties (texture). Although contrast negation is well-known to be very disruptive of facial identity recognition (Bruce & Young, 1998) it turns out that judgements based on facial expression are still possible in contrast-reversed images (Bruce & Young, 1998; Harris, Young & Andrews, 2014; White, 2001). The current proposal will determine the role of shape and texture in human face regions. The aim is to better understand how the brain represents information about faces. The main hypothesis of this study is that the neural representation in face regions, rather than being directly linked to high-level concepts such as identity or expression, is based on a more basic underlying representation, such as shape or texture. This type of representation would provide a more general code that could be used to decode information from faces in a more flexible way.