Visual pattern representation in extrastriate cortex

描述(申请人提供)：这项研究的目标是发现腹侧视区V2区神经元的反应中包含的皮质计算。由于V2的视觉反应在很大程度上依赖于V1的输入，我们将开发一个两阶段模型，其中反应由V1传入的适当组合构建，每个阶段的设计遵循共同的规范形式。这个模型旨在尽可能经济地解释神经元的视觉反应特性，同时不一定反映神经元电路的细节。这种简单性是经过深思熟虑的，因为它将允许模型适应从单个神经元记录的数据，并在组装成一个群体时预测感知能力。模型结构的动机，以及我们对其成功的信心，来自于先前工作的三个方面的融合：(1)我们已经开发、拟合和验证了类似的MT区神经元反应的两阶段模型，MT区是背流区，也接受来自V1区的初级传入驱动；(2)我们开发了一个用于视觉纹理表示的两阶段模型，该模型使用与V2细胞的大小相匹配的空间整合区域来捕捉自然图像的可感知识别的结构。我们已经证明了合成的具有匹配模型反应的图像对于人类观察者是不可区分的；以及(3)我们已经获得了初步数据，表明大多数V2细胞对合成纹理刺激的反应比对光谱匹配的噪声刺激更强烈，而V1细胞则不是。本研究分为三个部分。首先，我们将收集电生理数据来剖析那些模型生成的特征，这些特征是V2对纹理刺激的更高响应性的基础。同时，我们将使用功能磁共振成像收集提高反应性的证据，这将使我们能够同时比较V1和V2中神经群体的平均测量反应。其次，我们将开发一个在生理上看似合理的纹理模型实例化，并开发方法来适应来自单个神经元的数据。最后，我们将通过同时测量清醒行为猕猴的神经元反应和知觉判断，将我们发现的新的功能反应特性与感知联系起来。为了探索对自然特征的敏感性，我们将把心理测量和单神经元神经测量功能联系起来，并使用选择概率将反应与行为表现联系起来。

DESCRIPTION (provided by applicant): The goal of this research is to discover the cortical computations embodied in the responses of neurons in ventral visual area V2. Since V2 is heavily dependent on input from V1 for its visual responsiveness, we will 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 intended to account for the visual response properties of neurons as economically as possible, while not necessarily reflecting the details of neuronal circuitry. This simplicity is deliberate, as it will allow the mdel to be fit to data recorded from single neurons, and, when assembled into a population, to predict perceptual capabilities. The motivation for the structure of the model, and our confidence in its success, comes from the convergence of three strands of previous work: (1) we have developed, fit, and validated a similar two-stage model for neuronal responses in area MT, a dorsal stream area which also receives primary afferent drive from area V1; (2) we have developed a two-stage model for visual texture representation that captures perceptually recognizable structures of natural images using spatial integration regions matched in size to those of V2 cells. We've shown that images synthesized to have matching model responses are indistinguishable to human observers; and (3) we've obtained preliminary data indicating that most of V2 cells respond more vigorously to synthetic texture stimuli than to spectrally matched noise stimuli, whereas V1 cells do not. The research is divided into three parts. First, we will gather electrophysiological data to dissect those model-generated features that underlie the increased responsiveness of V2 to texture stimuli. We will, in parallel, gather evidence for the increased responsiveness using fMRI, which will allow us to compare simultaneously measured responses averaged over neural populations in V1 and V2. Second, we will develop a physiologically plausible instantiation of the texture model and develop the methods to fit it to data from single neurons. Finally, we will link the novel functional response properties we have discovered to perception by simultaneously measuring neuronal responses and perceptual judgments in awake behaving macaques. To explore sensitivity to naturalistic features, we will relate psychometric and single-neuron neurometric functions, and use choice probability to link responses to behavioral performance.