Cortical Basis of Perceptual Grouping

描述（由申请人提供）：拟议研究的长期目标是了解哺乳动物视觉系统场景分割的神经机制。在这个过程中，视觉系统被认为是通过识别共同特征之间的关系来将场景解析为其组成对象。分割中的一个关键步骤是将定向特征分组为形成对象边界的轮廓。目前的证据表明，纹状皮层作出了根本性的贡献，轮廓整合的过程中，但底层的神经元表示没有完全理解。在拟议的研究中，我们计划监测行为猕猴的纹状皮质中两个或多个位点的神经元组的活动，以测试轮廓整合的神经元相关的三个具体假设。我们将进行两组实验，在这些实验中，我们将建立轮廓显著性、神经元活动和行为表现之间的关系。在第一个实验中，动物将被要求通过在中心固定目标消失后向其位置进行扫视眼球运动来发出它们对嵌入背景中的外围轮廓的检测的信号。在第二个实验中，动物将被提示选择嵌入在背景中的两个相邻轮廓之一，通过在中心注视目标熄灭后对提示轮廓进行扫视眼球运动。在这两个实验中，轮廓的显着性，因此，任务的难度，将控制通过改变的方向组成的轮廓的元素。视觉刺激将由随机取向的Gabor贴片阵列组成，其中一个或两个贴片子集将对齐以形成闭合轮廓。轮廓或背景的元素将被定位在细胞感受野上以激活记录的神经元。这种模式将使我们能够保持激活细胞的局部刺激恒定，同时系统地操纵图形/背景关系，感知显著性和轮廓的行为相关性。我们将分析由此产生的活动，以确定轮廓整合是否与放电率，时间相关性，响应延迟或这些变量的某些组合的特定变化相关。总之，这些实验将使我们能够调查纹状体皮层的轮廓整合的神经元相关性，并推进我们对场景分割过程的理解。

DESCRIPTION (provided by applicant): The long-term objective of the proposed research is to understand the neuronal mechanisms underlying scene segmentation by the mammalian visual system. In this process, the visual system is thought to parse a scene into its component objects by identifying relationships among common features. A critical step in segmentation is the grouping of oriented features into contours that form the boundaries of objects. Current evidence indicates that striate cortex makes a fundamental contribution to this process of contour integration, but the underlying neuronal representations are not fully understood. In the proposed research, we plan to monitor the activity of groups of neurons at two or more sites in striate cortex of behaving macaque monkeys, in order to test three specific hypotheses for the neuronal correlate of contour integration. We will conduct two sets of experiments in which we will establish the relations between contour salience, neuronal activity and behavioral performance. In the first experiment, the animals will be required to signal their detection of a peripheral contour embedded in a background by making a saccadic eye movement to its location after a central fixation target is extinguished. In the second experiment, the animals will be cued to select one of two adjacent contours embedded in a background by making a saccadic eye movement to the cued contour after a central fixation target is extinguished. In both experiments, the contour salience, and hence the task difficulty, will be controlled by varying the orientation of the elements comprising the contour. The visual stimuli will be composed of arrays of randomly oriented Gabor patches, in which one or two subsets of the patches will be aligned to form a closed contour. Elements of the contour(s) or background will be positioned over the cellular receptive fields to activate the recorded neurons. This paradigm will enable us to keep the local stimuli activating the cells constant while systematically manipulating the figure/ground relationships, the perceptual salience and the behavioral relevance of the contours. We will analyze the resulting activities to determine if contour integration is correlated with specific changes in firing rates, temporal correlations, response latencies or some combination of these variables. Together, these experiments will enable us to investigate the neuronal correlate of contour integration in striate cortex, and advance our understanding of scene segmentation processes in general.