Sensory and motor plasticity in learning

我们大脑中的运动控制系统非常擅长快速适应环境和身体的持续变化。然而，这种感觉运动学习背后的机制在很大程度上仍不清楚。这项提议的目的是1)更好地理解用于灵活学习和适应动作的信息来源，2)探索大脑中作为感觉运动学习基础的过程，以及3)研究这一过程如何扩展到与工具和其他物体的互动。*能够适应我们环境和身体的变化取决于我们估计肢体，特别是我们的手的位置的能力。这被称为状态估计，是所有电机控制模型中的关键元素。状态估计将感觉反馈与大脑前向模型使用的与传出运动命令(传出副本)相关的信息结合起来，以预测传出运动命令的感觉后果。然而，大多数运动控制模型认为，适应完全是预测更新的结果，而不是感知更新的结果。我的实验室的工作表明，在学习过程中，对手的本体感觉估计确实会发生变化，这在一定程度上解释了运动性能的变化。为了更好地了解本体感觉和预测如何以及在多大程度上对运动学习做出贡献，我们开发了一种方法来分离它们的不同贡献。在拟议研究的第一个目标中，我们将测试每个人对已知的影响运动学习的多种因素的敏感性，以及测量他们在整个学习过程中的单独时间进程并建立模型。在第二个目标中，我们将用脑电(EEG)来补充这些行为和建模研究，以识别反映正在进行的感觉和预测错误处理的大脑活动，以及在整个运动适应过程中状态估计和运动规划的结果变化。运动学习不仅包括形成和更新我们效应器的前向模型，还包括建立与我们交互的对象和工具的运动相关表示。然而，尽管人类运动的多样性和复杂性涉及到预测和熟练操纵物体的能力，但这方面的学习在很大程度上被忽视了。因此，为了我们的第三个目标，我们使用3D虚拟现实(VR)技术来操纵与操作熟悉的工具和物体相关的正常物理来表征工具技能的获得。*拟议的研究将扩大我们对人类感觉运动学习的理解，包括潜在的大脑活动和熟练的工具使用。这反过来可以带来更好的康复，更好的人机界面，以及更好的运动技能训练，适用于所有水平的表现。

The motor-control systems in our brain are remarkably good at rapidly adapting to continuous changes in our environment and changes in our body. Yet, the mechanisms that underlie such sensorimotor learning remain largely unknown. The aims of this proposal are to 1) better understand the sources of information used to flexibly learn and adapt movements, 2) to explore processes in the brain that underlie sensorimotor learning, and 3) to investigate how this extends to interacting with tools and other objects.*******Being able to adapt to changes in our environment and body relies on our ability to estimate the location of our limbs, in particular, our hand. This is called state estimation and is a key element in all models of motor control. A state estimate combines sensory feedback with information related to outgoing motor commands (efferent copies) used by forward models in the brain to predict the sensory consequences of the outgoing motor command. Yet, most models of motor control assume that adaptation exclusively results from updates in prediction, but not updates in perception. Work from my lab shows that proprioceptive estimates of the hand do change during learning and partly account for changes in motor performance. To better understand how and to what extent both proprioception and prediction contribute to motor learning, we have developed a method for isolating their distinct contributions. In the first aim of the proposed research, we will test the sensitivity of each to multiple factors known to impact motor learning, as well as measure and model their separate time courses throughout learning. In the second aim, we will complement these behavioural and modeling studies with electroencephalography (EEG) to identify brain activity reflecting ongoing processing of sensory and prediction errors and the resulting changes in state estimation and motor planning throughout motor adaptation. Motor learning does not only involve forming and updating forward models of our effectors, but also building motor-related representations of objects and tools we interact with. Yet, this aspect of learning has been largely overlooked even though the diversity and complexity of human movements involves the ability to predict and skillfully manipulate objects. Thus, for our third aim we use 3D virtual reality (VR) technology to manipulate the normal physics associated with operating familiar tools and objects to characterize tool-skill acquisition.****The proposed research will expand our understanding of human sensorimotor learning, including the underlying brain activity and skilled tool-use. This in turn could lead to improved rehabilitation, better human-machine interfaces, and better motor-skills training for all levels of performance.******