Learning and plasticity in the human brain

这项研究的主要目标是：一)确定学习和经验如何影响大脑的结构和功能；二)确定学习和可塑性如何被调节。 目前正在进行的研究主要有两个领域： 1)截肢后大脑皮层手部表征有何变化？ 我们一生中的感觉运动经历被认为塑造了身体的神经表征。当成人大脑失去一个关键的输入来源时，例如，在截肢后，它会发生什么？我们纵向研究了手部选择性截肢前后大脑皮层手部表征的稳定性。两名患者在截肢前和截肢后两个不同的时间点：3个月和6个月进行了两次脑功能成像。此外，我们在相同的时间范围内对15名年龄匹配的健全对照组参与者进行了扫描(总共60次扫描)。使用数字地形图、表征相似性分析和随时间的解码，我们发现截肢前的手和截肢后的幻觉(缺失)手的指间表征结构非常一致。总体而言，这项工作为截肢前和截肢后保留假手(缺失)的再现提供了第一个纵向证据。这些证据对于思考脑机接口(BCI)和假肢设备的发展至关重要。 2)调节学习--假肢训练对手部表征的影响(NCT00001360) 在之前的工作中，我们发现专家工具与手的表征变得更加不同(即更少体现)。截肢者的假肢也有类似的结果。然而，假肢设备的发展强调以体现为目标，设备的设计和控制变得更加仿生(模仿身体)。在此背景下，值得注意的是，在上肢截肢者群体中，假手使用率很低，甚至完全排斥装置的情况很常见。如果对设备的体验不能让它更像是真实的身体部位，那么设计外观和动作都与人体相同的假手的追求可能是误导的，实际上可能会阻碍使用。在这项研究中，我们调查了不同的训练方案，特别是直觉(仿生)和任意控制机制，如何影响假肢(仿生)和真实手的表现。我们训练身体健全的参与者使用人造手模拟器，类似于假肢，可以绑在前臂上，通过前臂的肌电(EMG)读数进行控制。参与者接受了仿生控制系统(将机械手动作与真实手部动作相匹配)或任意控制系统(将机械手动作与任意真实手部动作相匹配)的培训。对于这两个训练组，训练改善了仿生肢体控制，减少了认知依赖，并增加了对仿生手的体现。仿生使用者在训练的早期有更直观和更快的控制。在后来的训练中，任意用户的仿生表现都与之相匹配。此外，任意用户对一种新的控制策略表现出更多的概括性。总而言之，我们的发现表明，仿生和任意控制策略提供了不同的好处。最优策略可能不是严格的仿生，而是根据用户、可用的培训机会和用户要求，在仿生到任意范围内的一种灵活策略。 除了行为学数据外，我们还收集了训练前后的功能磁共振测量数据。在正在进行的分析中，我们正在研究真实手和仿生手的视觉和运动表征如何在训练过程中发生变化。 建立成人大脑皮质可塑性的性质、程度和后果，为神经系统损伤或功能障碍后康复脑治疗的可能性提供重要的见解。

The primary goals of this research are i) to establish how learning and experience impact the structure and function of the brain, and ii) to determine how learning and plasticity can be modulated. There are two main areas of ongoing research: 1) How does the cortical hand representation change following amputation? Sensorimotor experiences throughout our lifespan are thought to shape the neural representation of the body. What happens to the adult brain when it loses a key source of input, for example, following the amputation of an arm? We longitudinally investigated the stability of the cortical hand representation, before and after elective hand amputation. Two patients underwent functional brain imaging twice pre-amputation and at two separate time-points following amputation: 3 months and 6 months. Additionally, we scanned 15 age-matched able-bodied control participants across the same timescale (60 scans in total). Using mapping of digit topography, representational similarity analysis and decoding over time, we show a remarkably consistent inter-digit representational structure of the pre-amputation hand and the post-amputation phantom (missing) hand. Overall, this work provides the first pre- and post-amputation longitudinal evidence for preserved representation of the phantom (missing) hand following amputation. Such evidence is critical for thinking about the development of brain computer interfaces (BCI) and prosthetic devices. 2) Modulating learning - Impact of prosthesis training on hand representations (NCT00001360) In prior work, we found that expert tools become more differentiated from the representation of the hand (i.e. less embodied). A similar result has been found for prosthetic limbs in amputees. However, the development of prosthetic devices emphasizes embodiment as the goal with the design and the control of devices becoming more biomimetic (mimicking the body). In this context it is notable that there are low rates of prosthetic hand usage and even complete device rejection are common in upper-limb amputee populations. If experience with a device does not lead to representing it more like a real body part, then maybe the quest to design artificial hands that look and act in the same way as the human body is misguided and may actually hinder usage. In this study, we investigated how different training regimens, and in particular intuitive (biomimetic) versus arbitrary control mechanisms, impact the representation of prosthetic (bionic) and real hands. We trained able-bodied participants to use an artificial hand simulator, similar to a prosthesis, that can be strapped to the forearm and controlled via electromyographic (EMG) readings from the forearm. Participants were trained with either a biomimetic control system (matching robotic hand movements to real hand movements) or an arbitrary control system (matching robotic hand movements to arbitrary real hand movements). For both trained groups, training improved bionic limb control, reduced cognitive reliance, and increased embodiment over the bionic hand. Biomimetic users had more intuitive and faster control early in training. Arbitrary users matched biomimetic performance later in training. Further, arbitrary users showed increased generalization to a novel control strategy. Collectively, our findings suggest that biomimetic and arbitrary control strategies provide different benefits. The optimal strategy is likely not strictly biomimetic, but rather a flexible strategy within the biomimetic to arbitrary spectrum, depending on the user, available training opportunities and user requirements. In addition to the behavioral data, we also collected functional MRI measures before and after training. In ongoing analyses we are investigating how visual and motor representations of the real and bionic hand change over the course of training. Establishing the nature, degree and consequences of plasticity in the adult cortex provides important insights into the potential for rehabilitative brain therapies following injury or dysfunction in the nervous system.