An investigation into the neurobehavioral interactions between sensory- and reward-prediction errors during motor learning

项目总结 当我们遇到预测错误，预期和观察之间的差异时， 大脑使用错误传递的信息来指导适应性学习，从而提高未来的准确性。这个 拟议的研究侧重于同时可用预测误差通知的机制 我们运动行为的改变。以前的工作将预测误差分为两类，根据 导致它们的条件和它们诱导的学习。奖励预测误差(RPE)在以下情况下出现 表现好于预期或低于预期，并导致影响行动选择的学习。感官- 另一方面，预测误差(SPE)发生在运动的预期之间存在不匹配时 观察感觉反馈，并驱动对特定动作的适应。先前的工作已将RPE联系在一起 对于基底节和SPE与小脑的关系，导致了这两种错误对 控制不同学习过程的不同网络。事实上，SPE(但不是RPE)会导致隐含适应 意识水平以下，而RPE(但不是SPE)会导致外显学习，这表明 不同的神经系统使用不同的误差信号来支持不同的学习过程。然而， 最近的研究表明，联合SPE和RPE对内隐和外显学习产生的影响不同于 或者是孤立的，这表明基于SPE的学习网络和基于RPE的学习网络相互作用。这些影响可能 由SPE和RPE信号之间的直接相互作用调节，其中一个错误的呈现影响 神经教学信号，通常驱动学习对另一个作出反应。或者，交互可以是 间接的，在基于SPE和RPE的系统中并行学习收敛到一个共同的神经目标上 影响行为输出。为了理解这些交互作用背后的机制，我将提出不同的 视觉运动到达适应(VMR)的预测误差及其独立效应和联合效应的测量 关于行为和fMRI的大胆信号。神经心理学调查将补充这些研究和 深入了解疾病期间错误及其支持系统之间的功能交互作用。 这些研究集中在错误和处理系统之间相互作用的神经基础上 它们将为支持适应性运动学习的事件和信号提供有价值的见解。我的发现 还可以根据有关能力的新知识为康复方案的设计提供洞察力 不同的错误信号在受疾病影响或幸免的系统中驱动学习。

PROJECT SUMMARY When we experience a prediction error, a difference between what was expected and what was observed, the brain uses information conveyed by the error to guide adaptive learning that improves future accuracy. The proposed research focuses on the mechanisms by which concurrently available prediction errors inform changes in our motor behaviors. Prior work divided prediction errors into two categories, according to the conditions that cause them and the learning that they induce. Reward-prediction errors (RPE) arise when task performance is better or worse than expected, and cause learning that affects action selection. Sensory- prediction errors (SPE), on the other hand, occur when there is a mismatch between a movement’s expected and observed sensory feedback, and drive adaptation of that particular movement. Prior work has linked RPEs with the basal ganglia and SPEs with the cerebellum, leading to the idea that the two kinds of error operate on distinct networks to control distinct learning processes. Indeed, SPE (but not RPE) causes implicit adaptation beneath the level of conscious awareness, while RPE (but not SPE) causes explicit learning, suggesting that different neural systems use the different error signals to support separate learning processes. However, recent work has revealed that joint SPE and RPE produce different effects on implicit and explicit learning than either in isolation, suggesting that SPE-based and RPE-based learning networks interact. These effects may be mediated by a direct interaction between SPE and RPE signals, with presentation of one error affecting the neural teaching signal that normally drives learning in response to the other. Alternately, the interaction may be indirect, with parallel learning in SPE- and RPE-based systems converging on a common neural target to influence behavioral output. To understand the mechanism underlying these interactions, I will present different prediction errors during visuomotor reach adaptation (VMR) and measure their independent and joint effects on behavior and fMRI BOLD signals. Neuropsychological investigations will complement these studies and provide insight into the functional interactions between errors and their supporting systems during disease. These studies, by focusing on the neural basis for interactions between errors and the systems that process them, will provide valuable insight into the events and signals that support adaptive motor learning. My findings may also provide insight for the design of rehabilitation protocols based on new knowledge about the capacity for different error signals to drive learning in systems affected or spared by disease.