Control of Human Arm Movement

从写签名到骑自行车，人类执行各种各样的熟练运动任务，然而，了解人类如何计划和控制即使是简单的运动仍然是神经科学的一个主要挑战。大量的关节和肌肉为中枢神经系统（CNS）提供了许多自由度的运动和控制；运动可以通过无限数量的关节旋转和肌肉激活模式来实现。令人惊讶的是，研究揭示了个体内部和个体之间一致的、刻板的运动和肌肉激活模式，但是中枢神经系统如何控制这些自由度的问题仍然是运动控制和感觉运动神经科学中一个未解决的关键问题。在本提案中，我们描述了旨在直接测试关于CNS如何控制运动的假设的研究，在当前主要的运动控制和运动学习的计算框架内，即最优反馈控制（OFC）。我们的方法结合了人类经验工作，使用定制设计的高性能机器人设备来测量和干扰手臂运动，并使用手臂神经肌肉系统的生理现实计算模型进行计算机模拟，以产生特定的预测。我们使用机器人在视觉引导的到达运动中向手臂传递精确的运动依赖负载，并使用虚拟显示系统通过操纵看不见的手和代表手位置的光标之间的关系来传递视觉运动扰动。使用力场和视觉运动扰动的组合来检验关于假定的运动学习成本函数的假设。我们还将通过使用过期气体分析来测量运动学习过程中VO2的变化，对代谢能量被中枢神经系统最小化的假设进行实证检验。这项研究将扩大我们对自主运动是如何计划和控制的，以及大脑是如何实现运动学习的知识。这项工作的结果也有可能推动应用工程和技术领域的发展，如神经假肢、人机界面和拟人机器人。

Humans perform a wide variety of skilled motor tasks, from writing an autograph to riding a bicycle, yet, understanding how humans plan and control even simple movements is still a major chal- lenge in neuroscience. A large number of joints and muscles provides the central nervous system (CNS) with many degrees of freedom in movement and control; movements can be achieved by an infinite number of joint rotations and muscle activation patterns. Surprisingly, research has revealed consistent, stereotypical patterns of movement and muscle activation, both within and across individuals, but the question of how the CNS controls these many degrees of freedom is still a key unresolved problem in motor control and sensory-motor neuroscience. In this proposal we describe studies aimed at directly testing hypotheses about how the CNS controls movement, within the current major computational framework of motor control and motor learning, namely optimal feedback control (OFC). Our approach combines human empirical work using custom-designed, high performance robotic devices to both measure and perturb arm movements, with computer simulations using physiologically realistic computational models of the arm neuromuscular system to generate specific predictions. We use the robots to deliver precise motion-dependent loads to the arm during visually-guided reaching movements, and a virtual display system to deliver visuo-motor perturbations, by manipulating the relationship between the unseen hand and a cursor representing hand position. A combination of force-fields and visuo-motor perturbations are used to test hypothesis about the putative cost function for motor learning. We will also empirically test the hypothesis that metabolic energy is minimized by the CNS, by using expired gas analysis to measure VO2 changes over the course of motor learning. This research will expand our knowledge of how voluntary movements are planned and controlled, and how motor learning is achieved by the brain. The results of this work also has the potential to advance fields in applied engineering and technology such as neural prosthetics, human-machine interfaces and anthropomorphic robotics.