Credit assignment in movement planning and movement execution

我们如何计划和执行动作是神经科学的基本问题之一。这两个过程中的不准确通常会导致运动错误，从而提供对潜在大脑过程的洞察。例如，当计划到达时，头部倾斜到一个肩膀，观察到的头部滚动相关的到达误差与大脑低估的头部角度有关，所需的头部角度是反旋转视网膜上倾斜的视觉信息所需的。类似地，人们可以将手臂重量改变(例如，肘部扛着一个满满的购物袋)下到达终点的系统性偏差解释为大脑错误估计改变的手臂动力学的结果。问题是，这些解释依赖于这样的假设，即计划和控制在功能上是分开的，是可以分开的。考虑到大脑规划和控制区域的巨大重叠，这一假设是否合理值得怀疑；但这一假设永远无法明确测试，因为该领域的一个关键缺失部分是一个无缝集成运动规划和动作执行以消除运动错误来源的模型。模型可以将大脑中的功能机制与行为联系起来。如果没有一个集成的模型，我们就有可能错误地将错误归因于错误的潜在机制，从而减缓理解大脑的进展。因此，至关重要的是要解决这一问题，并根据行动规划和执行相结合的模式审查行动错误的潜在原因，以确认或纠正目前的解释。 要将运动误差分配给具体的过程，需要对潜在原因及其相互作用进行综合描述。为了实现这一点，我们需要一种新的模型，将已知的运动规划特性与现代运动控制理论相结合。规划过程包括将感觉信号转换到不同的坐标系和整合互补的信息源。最优反馈控制最好地捕捉动作执行，依赖于对身体状态的最优估计、身体动力学的内部模型和噪声感觉反馈的使用。我们的实验室在计划和控制方面都有专业知识；因此，我们唯一有资格开发计划和设计新的行为、成像和刺激实验，以消除替代误差源的歧义。这个多学科的项目将提供一个理想的培训环境；总部将领导所有目标。

How we plan and execute movements is one of the fundamental questions in neuroscience. Inaccuracies in either process generally result in movement errors that provide insight into underlying brain processes. E.g. when planning reaches with the head tilted to one shoulder, observed head-roll-dependent reach errors correlate with brain under-estimated head angles required to counter-rotate tilted visual information on the retina. Similarly, one can interpret systematic biases in reach endpoints under altered arm weight (e.g. while carrying a full grocery bag in the elbow) as a consequence of the brain mis-estimating the altered arm dynamics. The problem is, these interpretations rely on the assumption that planning and control are functionally separate and separable. Given the large overlap of brain areas for planning and control, it is questionable whether this assumption is justified; but this could never be explicitly tested because a key missing piece in the field is a model that seamlessly integrates motor planning and movement execution to disambiguate the origin of movement errors. Models can link functional mechanisms in the brain to behaviour. Without an integrated model, we run the risk of erroneously crediting errors to the wrong underlying mechanism, slowing progress in understanding the brain. It is thus crucial to address this problem and examine the potential cause of movement errors in the light of a combined movement planning and execution model to either confirm or correct current interpretations. Assigning movement errors to specific processes requires an integrated characterization of the potential causes and their interactions. To achieve this, we need a new model combining known movement planning features with modern motor control theories. Planning processes include converting sensory signals to different coordinate frames and integrating complementary sources of information. Optimal feedback control best captures movement execution, relying on the optimal estimation of body state, an internal model of the body dynamics and the use of noisy sensory feedback. Our lab has expertise in both planning and control; we thus uniquely qualified to develop a planning n design new behavioural, imaging and stimulation experiments that disambiguate alternative error sources. This multi-disciplinary project will provide an ideal training environment; HQP will lead all aims.