Trial-by-Trial Human Generalization of Sense into Action

描述（由申请人提供）：尽管运动学习是基础和临床神经科学的中心，但我们对神经系统在将感觉转化为适应性控制的过程中所起的定量作用知之甚少。这里提出的实验将挑战正常的人类受试者，包括不同强度、持续时间、应用频率、偏差和运动内部和运动之间的空间复杂性的扰动。这些扰动的可变性将产生一个模板，我们将用它来识别每次试验适应中的细微之处。这些询问，再加上新的状态空间分析，将使我们能够彻底理解将特定的感官体验转化为预测控制中的即时、增量适应，从而大大增强我们对人类运动适应的定量理解。经验使我们能够建立运动环境的内部动态模型。这种内部动态适应的研究人员假设了学习的两个组成部分：从以前的动作中提取错误信号，并应用这种错误来指定或概括整个动作空间的学习。迄今为止，人类不断尝试的适应表明，适应随着感觉错误不断扩大，并在整个运动空间中广泛推广。然而，PI的初步结果发现了学习的两个组成部分令人惊讶的灵活性：感觉反馈可以诱导与运动误差显著不成比例的适应，环境可以诱导跨运动空间的泛化缩小。在这里，我们建议确定必要的感官体验，以诱导人们在将单一运动感觉转化为增量适应的基本计算中新建立的变化。这些结果将阐明神经系统如何进行实时信号处理以改善运动表现。由此产生的模型将有助于神经科学和生物建模界更好地将行为与其潜在的生理基础联系起来。这些见解也将用于研究正常运动控制的全部功能以及控制如何在疾病状态下失效。我们的目标是制定科学的基础，康复如何最好地帮助患者推广临床训练之外的运动功能，以改善他们的日常生活。

DESCRIPTION (provided by applicant): Despite the centrality of motor learning to basic and clinical neuroscience, we know very little about the quantitative role neural systems play in the transformation of senses into adapted control. The experiments presented here will challenge normal human subjects with perturbations of varying strengths, durations, frequency of application, biases, and spatial complexities both within and across movements. The variability of these perturbations will generate a template with which we will identify subtleties in trial-by-trial adaptation. These interrogations, coupled with novel state space analyses, will enable a thorough understanding of the transformation of specific sensory experiences into immediate, incremental adaptations in predictive control, greatly enhancing our quantitative understanding of human motor adaptation. Experience enables us to build internal dynamic models of our movement environment. Investigators of this internal dynamic adaptation have hypothesized two components of learning: the abstraction of an error signal from previous movements and the application of this error to either specify or generalize learning across movement space. Human trial-by-trial adaptation has, to date, suggested that adaptation constantly scales with sensed error and generalizes broadly across movement space. However, preliminary results from the PI have discovered surprising flexibility in both components of learning: sensory feedback can induce adaptation strikingly disproportional to movement error, and environments can induce narrowing of generalization across movement space. Here we propose to identify the necessary sensory experiences to induce these newly established changes in the fundamental computations people execute to transform single movement sense into incremental adaptation. These results will illuminate how the nervous system performs real-timed signal processing to improve motor performance. The resultant models will help the neuroscience and biological modeling community to better connect behavior to its underlying physiological basis. These insights will also be of use to investigate the full repertoire of normal motor control and how control fails in disease states. We aim to formulate the scientific basis of how rehabilitation can optimally help patients generalize beyond clinical training to improve motor function in their daily lives.