CRCNS: Extracting Dynamical Structure Embedded in Motor Preparatory Activity

CRCNS：提取运动准备活动中嵌入的动态结构

• 批准号: 7109167
• 负责人: Krishna V Shenoy
• 金额: $ 33.67万
• 依托单位: STANFORD UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2005
• 资助国家: 美国
• 起止时间: 2005-09-01 至 2010-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7109167
• 关键词: Primates; action potentials; brain electrical activity; brain mapping; computer data analysis; computer program /software; computer system design /evaluation; electrodes; mathematical model; motor cortex; nervous system disorder therapy; neural information processing; prosthesis; psychomotor function

DESCRIPTION (provided by applicant): Spiking activity from neurophysiological experiments often exhibits dynamics beyond that driven by external stimulation, presumably reflecting the extensive recurrence of neural circuitry. Characterizing these dynamics may reveal important features of neural computation, particularly during internally-driven "cognitive" operations. For example, neurons in premotor cortex (PMd) are active during a "planning" period separating movement-target specification and a movement-initiation cue. Recent evidence suggests that PMd neural activity settles to a movement-specific state during this period. Can trial-to-trial variation in behavior be predicted from the dynamics of settling? Current methods to characterize recurrent neural dynamics on a trial-by-trial basis, and thus answer this and related questions, are limited. Standard methods average activity from different trials or different cells, and so cannot express variable dynamics. The proposed research will test the hypothesis that the dynamics underlying PMd plan activity can be described by a low-dimensional hidden non-linear dynamical systems (HNLDS) model, with underlying recurrent structure and stochastic point-process output. Such a model is capable of expressing rich dynamics, but the task of learning the model parameters from spike data is challenging. The proposed research will develop and validate algorithms for parameter estimation, and then characterize the dynamics in PMd data recorded from an electrode array while monkeys perform delayed-reach tasks. Single trial estimates of underlying dynamics can then be used to predict variation in details of reaching motor behavior. The proposed research program will directly inform cortically-controlled neural prosthesis research in our laboratory and elsewhere. Such motor and communication prostheses could dramatically improve the quality of life for patients with upper spinal cord injuries, amputations, ALS and other neuro-degenerative diseases. The proposed research program will increase our understanding of how PMd rapidly prepares movements, and thereby help increase the speed and accuracy of prosthetic systems.

描述(由申请人提供)： 神经生理学实验中的放电活动通常表现出外部刺激以外的动态变化，这可能反映了神经回路的广泛重现。描述这些动力学特征可能会揭示神经计算的重要特征，特别是在内部驱动的“认知”操作期间。例如，运动前皮质(PMD)中的神经元在分离运动目标指定和运动启动线索的“计划”阶段是活跃的。最近的证据表明，PMD的神经活动在这一时期稳定在一种运动特有的状态。是否可以从和解的动态来预测行为的审判之间的差异？目前在逐个试验的基础上描述递归神经动力学，从而回答这个问题和相关问题的方法是有限的。标准方法平均不同试验或不同细胞的活性，因此不能表达可变的动力学。这项研究将检验这样一个假设，即PMD计划活动背后的动力学可以用一个具有潜在递归结构和随机点过程输出的低维隐藏非线性动力系统(HNLDS)模型来描述。这样的模型能够表达丰富的动力学，但从尖峰数据中学习模型参数的任务是具有挑战性的。这项拟议的研究将开发和验证参数估计算法，然后表征从电极阵列记录的PMD数据在猴子执行延迟到达任务时的动力学。然后，对潜在动力的单次试验估计可以用来预测达到运动行为的细节的变化。 拟议的研究计划将直接为我们实验室和其他地方的皮质控制神经假体研究提供信息。这种运动和通讯假体可以显著提高上段脊髓损伤、截肢、肌萎缩侧索硬化症和其他神经退行性疾病患者的生活质量。拟议的研究计划将增加我们对PMD如何快速准备动作的理解，从而有助于提高假肢系统的速度和准确性。