Integrating perception and action: the multi-channel model of visuo-motor control

当一个熟练的板球运动员伸出手去接球时，多个大脑系统同时预测球和手臂的位置。基于这些视觉信息，运动系统必须快速纠正正在进行的运动，以确保成功抓住球的方式驱动手。类似的视觉和运动通路在所有运动中都有应用，运动员需要尽可能快速准确地对由视觉反馈信号发出的世界微小变化做出反应。事实上，这些过程是日常生活中大多数熟练动作的基础。视觉感知的变化既可以指自己的身体，也可以指外部世界的物体（比如一个球）。通过将这些视觉变化分配给自身或外部物品，视觉运动控制系统可以以正确的方式快速响应，以确保熟练的动作。在这个项目中，我们将研究构成视觉到运动动作回路的神经通路。具体来说，我们将研究三个主要问题：第一个问题是这些不同的信号通路，无论是我们的手的位置还是目标的位置（例如球），是如何相互作用的。目前的科学观点是，大脑只是计算目标和手的位置之间的差异，并利用这种差异来纠正运动。然而，我们的初步实验表明，这是不正确的，而是表明，这两种途径导致部分独立的反应。使用机器人设备，我们将仔细测量这两个反馈路径的相互作用和独立性。第二个问题研究的是大脑的哪些部分负责处理这两种反馈途径。我们将使用机器人设备研究主动伸手运动，同时使用功能性磁共振成像测量大脑活动。每个区域的个体活动模式将揭示目标和手的信息是如何在不同的大脑区域表示的，以及这些区域是如何相互作用的。第三个问题是大脑如何将视觉信号分配给一个人自己的动作，或其他与动作相关的物体。例如，大量的运动使用球拍或球拍，作为受试者自己手的延伸。因此，大脑必须为这些物体分配代理，将它们标记为自我，以便对这些物体的视觉变化做出正确的反应，这可能与对外部物体（如球）的变化做出反应不同。我们将研究这种情况发生的过程，并试图将其与注意机制区分开来。这个项目调查了基础视觉到运动动作路径，这些路径是熟练动作的基础。了解这些通路以及大脑利用它们进行快速动作的方式，将有助于改进高性能运动的训练制度。在许多运动中，最高水平的表现需要对微小的、几乎无法察觉的视觉信息做出准确、快速的反应。这项研究还对直接参与动作学习的途径产生了基本的理解。因此，它提供了关于学习和再训练技能和动作的机制的重要信息。这与脑损伤（如中风）后的康复特别相关。目前正在开发广泛的技术，使用机器人设备进行脑损伤后的再训练，并提供视觉反馈。通过详细了解这些视觉信息是如何以及在哪里被处理的，可以开发出最佳的中风康复训练设计，其中考虑到各种反馈回路中的个体缺陷。

When a skilled cricket player reaches out to catch a ball, multiple brain systems simultaneously predict the position of the ball and the arm. Based on this visual information, the motor system must rapidly correct the ongoing movement, driving the hand in a manner that ensures successful grasp of the ball. Similar visual and motor pathways are used in all sports, where an athlete is required to respond as quickly and accurately as possible to the small changes in the world signalled by visual feedback. Indeed, these processes are fundamental to most skilled movements in everyday life. The visually sensed changes may refer either to one's own body, or to objects in the external world (such as a ball). By assigning these visual changes to either itself or to an external item, the visuo-motor control system can respond quickly in the correct manner to ensure skilled action. In this project, we will investigate the neural pathways that constitute this vision to motor action loop. Specifically, we will study three major questions: The first question is how these different pathways, signalling either our hand location or the target location (e.g. ball), interact with one another. The current scientific view is that the brain simply calculates the difference between the target and hand locations and uses this difference to correct the movement. However, our preliminary experiments demonstrate that this is not true, but rather suggest that the two pathways lead to partially independent responses. Using a robotic device, we will carefully measure the interactions and independence of these two feedback pathways.The second question investigates which parts of the brain are dedicated to the processing of the two feedback pathways. We will investigate active reaching movements using a robotic device while measuring brain activity using functional magnetic resonance imaging. The individual activity pattern in each region will reveal how target and hand information are represented in different brain regions, and how these regions interact. The third question is how the brain assigns visual signals to one's own movements, or other action-relevant objects. For example, a huge number of sports utilize bats, or rackets that act as an extension of the subjects own hand. The brain must therefore assign agency to these objects, marking them as self, in order to respond correctly to visual changes in these objects, which may be a different action than responding to changes in external objects such as the ball. We will investigate the process by which this occurs and attempt to distinguish it from attention mechanisms. This project investigates the basic vision to motor action pathways that underlie skilled movements. Understanding these pathways and the manner in which the brain utilizes them for fast action will lead to improvements of training regimes for high-performance sports. In many sports, the highest level of performance requires the ability to respond accurately and with exceptional speed to small, barely detectible visual information. The research also produces an essential understanding of the pathways directly involved in learning of action. As such, it provides important information on the mechanisms used in learning and retraining skills and movements. This has particular relevance for rehabilitation after brain injury, such as stroke. Extensive techniques are being developed which use robotic devices for retraining after brain injury, where feedback is also provided visually. By understanding in detail how and where this visual information is processed, optimal training designs for stroke rehabilitation can be developed, which take into account individual deficits in the various feedback loops.