Computational and Psychophysical Studies of Visual Perceptual Learning

提案#9817979宁倩我们毫不费力地看到我们周围的事物。 由于这个原因，我们通常很难理解大脑中视觉信息处理的巨大复杂性。 研究人员一直在试图制造能够模仿人类大脑各种功能的机器。事实证明，比起下棋或解决大学物理问题，视觉是一项更难模仿的任务。事实上，虽然已经有一台国际象棋机器，IBM的深蓝，可以挑战国际象棋世界冠军，但我们还远远没有一台视觉机器接近平均三岁儿童的视觉能力。为了找出大脑如何实现视觉的非凡壮举，研究人员一直在研究视觉系统的各个方面，使用各种技术。 这个应用程序中提出的项目围绕着我们视觉系统最重要的功能之一：它通过从经验中学习不断重新校准和改进自己。 这种所谓的视觉感知学习现象在所有年龄段的人类受试者中都有观察到。 理论和实验方法将被应用，以综合大量的实证研究结果到一个连贯和逻辑的框架，与理解的视觉系统中的感知改善的神经机制的最终目标。 这种综合不能仅仅通过数据积累自动实现，需要定量理论来理解和联系不同的实验数据。 （类似地，计算机如何工作不能仅仅通过测量晶体管和计算机内部其他组件的连接和活动来理解;需要操作系统和数据结构等理论概念。 本研究的重点是基于现有的生理学和解剖学数据，构建视觉感知学习的定量理论。 这些项目的结果不仅将促进我们对视觉感知和可塑性的大脑机制的理解，而且还将提供可能具有实际工程应用的重要知识。

PROPOSAL #9817979NING QIANWe see things around us effortlessly. For this reason it is often difficult for us to appreciate the enormous complexity involved in visual information processing in the brain. Researchers have been trying to build machines that can mimic various human brain functions. It turns out that seeing is a much more difficult task to emulate than, for example, playing chess or solving college physics problems. In fact, while there is already a chess machine, IBM's Deep Blue, that can challenge the world chess champion, we are still far from having a vision machine that approaches the visual capabilities of an average three-year old.To find out how the brain achieves the remarkable feat of seeing, researchers have been studying various aspects of the visual system using a variety of techniques. The projects proposed in this application center around one of the most important functions of our visual system: It constantly recalibrates and improves itself by learning from experience. This so-called visual perceptual learning phenomenon has been observed in human subjects of all ages. Both theoretical and experimental methods will be applied in order to synthesize a large body of empirical findings into a coherent and logical framework, with the ultimate goal of understanding the neural mechanisms of perceptual improvement in the visual system. This synthesis cannot be achieved automatically through data accumulation alone, quantitative theories are required for understanding and relating different pieces of experimental data. (Analogously, how a computer works cannot be understood only through measuring connectivity and activity of the transistors and other components inside the computer; theoretical concepts such as operating system and data structure are required.) The focus of the proposed projects is to construct such quantitative theories for visual perceptual learning based on existing physiological and anatomical data. Results from these projects will not only advance our understanding of the brain mechanisms of visual perception and plasticity but will also provide important knowledge that might have practical engineering applications.