Visual pattern representation in extrastriate cortex

项目摘要/摘要 这项研究的目标是发现决定反应特性的皮质计算 视觉皮质V2和V4区的神经元，V2和V4是初级视觉皮质外的两个最大的视觉区域 灵长类动物。V2从V1接收强大的直接输入，并依赖于V1的功能完整性来实现其视觉 响应性。因此，我们将开发一个两阶段模型，在该模型中，响应由 V1传入的适当组合，每个阶段的设计遵循共同的规范形式。这 模型被设计为尽可能经济地考虑神经元的视觉响应属性， 这使得它可以与从单个神经元记录的数据相匹配。该模型还将能够产生人口 能够预测感知能力的表征。反过来，V4从V2接收其大部分直接输入。 我们将测量V4的响应，目标是构建一个类似的模型来转换来自V2的输入。 V4中的一些神经元对物体和形状的图像做出选择性反应，而其他神经元似乎更多 对本地图像统计数据敏感。 V2模型结构的动机，以及我们对其成功的信心，来自 三项以前工作的融合：(1)我们已经开发、拟合和验证了类似的两阶段 MT区神经元反应的模型，纹外区也接受来自 区域V1；(2)我们有代表V2运行的关键模型元素的初步证据 神经元作为其输入空间的导数，以及(3)我们已经证明了这种结构的模型可以 解释了我们之前的发现，许多V2细胞对自然纹理刺激的反应更强烈 而不是匹配的噪音。我们将通过分析自然图像的局部结构来补充建模 以及人类观察者在该模型旨在捕捉的空间中的心理物理表现。 我们还没有准备好构建V4如何组合来自V2的输入的原则性模型。我们认为V4的S 功能电路将类似于V2，但在我们将为V2构建的模型之前，我们缺少 对V4的S输入的足够精确的说明。但我们从之前对V4和人类的研究中得到了证据 神经成像数据表明，V4的S反应的一个关键成分是它对复杂形式的选择性，所以我们将 沿着自然形式和图像统计之间的关键图像连续体测量V4响应，以及 还将衡量语境信号在建立V4‘S选择性中的作用。 这项工作的结果是对V2和V4的功能有了新的认识，V4起着举足轻重的作用 在大脑皮层中详细阐述视觉信息。

Project Summary / Abstract The goal of this research is to discover the cortical computations that determine the response properties of neurons in visual cortical areas V2 and V4, two of the largest visual areas outside the primary visual cortex in primates. V2 receives a strong direct input from V1, and depends on the functional integrity of V1 for its visual responsiveness. We will therefore develop a two-stage model, in which responses are constructed from a suitable combination of V1 afferents, with the design of each stage following a common canonical form. This model is designed to account for the visual response properties of neurons as economically as possible, allowing it to be fit to data recorded from single neurons. The model will also be able to generate population representations that can predict perceptual capabilities. V4, in turn, receives the bulk of its direct input from V2. We will measure V4 responses with the goal of building a similar model of the way it transforms input from V2. Some neurons in V4 respond selectively to images of objects and shapes while others seem to be more sensitive to local image statistics. The motivation for the structure of the V2 model, and our confidence in its success, comes from the convergence of three pieces of previous work: (1) we have developed, fit, and validated a similar two-stage model for neuronal responses in area MT, an extrastriate area which also receives primary afferent drive from area V1; (2) we have preliminary evidence for a key model element, which is to represent the operation of V2 neurons as derivatives over the space of its inputs, and (3) we have shown that models of this structure can account for our previous discovery that many V2 cells respond more vigorously to naturalistic texture stimuli than to matched noise. We will complement the modeling by analyzing the local structure of natural images and the psychophysical performance of human observers in the space that the model is designed to capture. We are not yet ready to build a principled model of how V4 combines inputs from V2. We believe that V4's functional circuitry will be similar to that of V2, but in advance of the model we will build for V2 we lack a sufficiently precise account of V4's inputs. But we have evidence from previous studies of V4 and from human neuroimaging data that a key component of V4's response is its selectivity for complex forms, so we will make measurements of V4 responses along a key image continuum between natural form and image statistics, and will also measure the role of contextual signals in establishing V4's selectivity. The outcome of this work will be a new understanding of the functions of V2 and V4, which plays a pivotal role in the elaboration of visual information in the cerebral cortex.