The feedback response to error as a teaching signal for motor learning

对错误的反馈响应作为运动学习的教学信号

• 批准号: 9232909
• 负责人: Scott Tyler Albert
• 金额: $ 4.4万
• 依托单位: JOHNS HOPKINS UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-12-01 至 2018-11-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9232909
• 关键词: Affect; Behavior; Brain; Cerebellar Ataxia; Cerebellum; Clinic; Clinical; Complex; Data; Disease; Dissociation; Educational process of instructing; Electromyography; Environment; Feedback; Goals; Investigation; Learning; Methods; Modeling; Motor; Motor output; Movement; Muscle; Nervous System Trauma; Nervous system structure; Pattern; Process; Psychophysics; Recording of previous events; Rehabilitation therapy; Sensory; Shapes; Signal Transduction; Space Models; System; Techniques; Testing; Therapeutic; Time; Training; Update; Work; abstracting; base; behavioral study; design; experience; experimental study; improved; insight; motor control; motor learning; motor recovery; motor rehabilitation; neural stimulation; public health relevance; rehabilitation paradigm; rehabilitation strategy; relating to nervous system; response; teacher

 DESCRIPTION (provided by applicant): When we perform a motor task, unexpected perturbations produce errors in the intended sensory consequences of our movements. Behavioral studies have suggested that the nervous system updates its motor commands in order to minimize these `sensory predictions errors' on a trial-to-trial basis, a phenomenon known generally as motor learning. By abstracting psychophysical results into scalar variables, state-space models have proven useful in explaining the relationship between error and learning. However, a fundamental limitation of these models in the interpretation of motor behavior, is that movements involve complex activation of muscles, patterns that are outside the scope of current scalar models that have been used to represent learning and error. In order to investigate the mechanism of trial-to-trial motor learning we must move beyond the state-space framework and consider how these temporal patterns of muscle activation are altered after experiencing a perturbation. Here, we propose that the motor commands that the nervous system produces in response to perturbations, i.e., the feedback responses that correct for the prediction error during a movement, also serve as a teacher for the motor learning process. The goal of our proposal is to use electromyography (EMG) to systematically evaluate the relationship between feedback responses and motor learning. We will test our proposed hypothesis in three ways. First, we will systematically alter the shape of the feedback response (through manipulation of perturbations), and test whether this alteration results in consistent modulation of trial-to-trial motor learning. Second, we will modulate the gain of the feedback response (through manipulation of task parameters), and test whether this gain modulation results in corresponding changes in trial-by-trial motor learning. Third, we will increase or decrease the amount that the brain learns from error (through manipulation of error history), and test whether this modulation of learning coincides with modulation of the feedback response. Finally, we will stimulate the cerebellum, a region that is thought to be critical for motor learnig, and determine to what extent this neural substrate affects the relationship between feedback responses and trial-to-trial learning. From a clinical standpoint, understanding the relationship between feedback and learning has direct implications in the design of motor rehabilitation paradigms for neurotrauma or disease. The results of this work may suggest that rehabilitation strategies should allow the nervous system to make predictions and correct for its errors, in order to more effectively promote motor learning. In addition, the learning-feedback correlation and its relation to error-sensitivity may offer insight into potential rehabilitation techniques tht could enhance the rate of learning outside the clinic. Finally, our investigation of cerebellar involvement in learning and feedback may improve our understanding of cerebellar ataxias. We may also identify methods to alter the learning-feedback response relationship using neural stimulation, which could enhance therapeutic benefits of rehabilitation paradigms.

 描述（由申请人提供）：当我们执行运动任务时，意外的扰动会在我们运动的预期感觉结果中产生错误。行为研究表明，神经系统更新其运动指令，以便在一次又一次的试验基础上最大限度地减少这些“感觉预测错误”，这种现象通常被称为运动学习。通过将心理物理结果抽象为标量变量，状态空间模型已被证明在解释错误和学习之间的关系方面是有用的。然而，这些模型在解释运动行为方面的一个基本限制是，运动涉及肌肉的复杂激活，这些模式超出了用于表示学习和错误的当前标量模型的范围。为了研究试验到试验运动学习的机制，我们必须超越状态空间框架，并考虑这些肌肉激活的时间模式是如何在经历扰动后改变的。 在这里，我们提出神经系统响应扰动产生的运动命令，即，在运动期间校正预测误差的反馈响应也用作运动学习过程的教师。我们的建议的目标是使用肌电图（EMG）系统地评估反馈反应和运动学习之间的关系。 我们将从三个方面来检验我们提出的假设。首先，我们将系统地改变反馈反应的形状（通过操纵扰动），并测试这种改变是否会导致试验到试验运动学习的一致调制。其次，我们将调节反馈反应的增益（通过操纵任务参数），并测试这种增益调节是否会导致逐个试验的运动学习发生相应的变化。第三，我们将增加或减少大脑从错误中学习的量（通过操纵错误历史），并测试这种学习的调制是否与反馈反应的调制一致。最后，我们将刺激小脑，一个被认为对运动学习至关重要的区域，并确定这个神经基质在多大程度上影响反馈反应和试验到试验学习之间的关系。 从临床的角度来看，了解反馈和学习之间的关系，在神经创伤或疾病的运动康复范例的设计有直接的影响。这项工作的结果可能表明，康复策略应该允许神经系统做出预测并纠正其错误，以便更有效地促进运动学习。此外，学习反馈相关性及其与错误敏感性的关系可能会提供潜在的康复技术，可以提高学习率的诊所外。最后，我们对小脑参与学习和反馈的研究可能会提高我们对小脑共济失调的理解。我们还可以确定使用神经刺激来改变学习反馈反应关系的方法，这可以增强康复范例的治疗益处。