Two-Dimensional Models of Distributed Neural Processing in the Saccadic System

伊本98-08204凯勒。这项研究项目的目标是更深入地了解大脑如何控制运动对视觉输入的反应。用来检查视觉环境的眼球运动称为扫视。眼跳运动在二维空间中旋转视线，是人类能够执行的最快速、最准确的运动之一。同时，与多关节肢体控制系统相比，由眼睛及其肌肉组成的轨道上的机械系统相对简单。因此，眼跳系统已经成为探索感觉输入如何转化为运动行为这一普遍问题的模型系统。在目前的项目中，凯勒博士将开发作为眼跳控制基础的感觉运动过程的分布式模型。该模型是分布式的，因为模型中没有单一的动态处理元素(模型中的元素旨在表示大脑中简化的单个神经元)编码将视觉输入转换为扫视所需的控制信号。相反，这样的信号在转化的每个阶段都由大量活跃的元素代表。在模型中，通过使用有偏随机游走算法的优化过程来确定单元层内的各种元素之间以及单元层之间的假设连通性。该算法通常被认为具有生物学相关性。运动系统疾病的一个常见迹象是，用于将手从一个点移动到另一个点或眼睛从一个方向移动到另一个方向的轨迹变得高度弯曲，而不是遵循正常运动中看到的直线方法。将通过在模型中的连接或处理元素中放置模拟损伤来探索弯曲轨迹的潜在基础。该模型将探索眼跳系统中使用的反馈类型，以确保在视觉系统中面对长时间处理延迟时动作的准确性。该模型将有助于统一和解释目前在其他研究中收集的大量复杂的神经生理学数据，并将为进一步的眼跳系统运作的解剖学和生理学研究提供指导。此外，该模型的涌现特性可能揭示出适用于具有大神经元集合的分布式运动控制的一般原理。

IBN 98-08204 KELLER. The goal of this research project is to provide a deeper understanding of how the brain controls movements in response to visual inputs. The eye movements used to examine the visual surround are called saccades. Saccadic eye movements, which rotate the line of sight in two- dimensional space, are among the most rapid and accurate movements that humans are capable of executing. At the same time the mechanical system in the orbits consisting of the eye and its muscles is relatively simple compared to multi-jointed limb control systems. Therefore the saccadic system has served as a model system for exploring the general problem of how sensory input is converted to motor behavior. In the current project Dr. Keller will develop distributed models of the sensorimotor processes underlying the control of saccades. The model is distributed because no single dynamic processing element in the model (elements in the model are meant to represent simplified single neurons in the brain) codes the control signals needed to transform visual inputs to saccades. Rather such signals are represented by large populations of active elements at each stage of the transformation. The hypothesized connectivity between various elements within layers of cells and between layers of cells are determined in the model by an optimization process that uses a biased random walk algorithm. The algorithm is generally considered to have biological relevance. One common sign of disease in motor systems is that the trajectories used to move, for example, the hand from one point to another, or the eye from one direction to another, become highly curved instead of following the straight-line approach seen in normal movements. The underlying basis for curved trajectories will be explored by placing simulated lesions in the connections or processing elements in the model. The type of feedback used in the saccadic system to insure accuracy of the movements in the face of long proce ssing delays in the visual system will be explored with the model. The model will help to unify and explain large amounts of complex neurophysiological data currently being gathered in other studies, and will also provide guidance for further anatomical and physiological research into the operation of the saccadic system. In addition, emergent properties of the model may reveal general principles applicable to distributed motor control with large ensembles of neurons.