Decision-making and assignment policies in sensorimotor learning

感觉运动学习中的决策和分配策略

• 批准号: RGPIN-2018-05589
• 负责人: Carter, Michael
• 金额: $ 2.11万
• 依托单位: McMaster University
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2020
• 资助国家: 加拿大
• 起止时间: 2020-01-01 至 2021-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=713234
• 关键词: Decision; making; assignment; policies; sensorimotor

Learning from the consequences of our actions is the cornerstone of sensorimotor learning and human cognition. Thus, accurately specifying the source of errors (i.e., assignment) and initiating, from one movement to the next, the necessary adjustments (i.e., decision-making) is critical for adaptive, goal-oriented behaviour. Yet, the mechanisms and strategies that allow us to respond to errorswhich are essential to understanding the decision-making processes that underpin sensorimotor learningare poorly understood. The broad aim of the proposed research is to better understand how we incorporate information about motor errors into decisions that shape sensorimotor learning. While it is recognized that the best way to learn new motor skills is through extensive training, we know not all forms of practice are equally effective. There is converging evidence that self-controlled feedback schedules are more effective for motor learning compared to experimentally-imposed schedules. However, the mechanisms underlying this learning benefit are not well understood. Project 1 will test between two competing explanations for this benefit to better understand the cognitive and neural mechanisms of self-controlled learning advantages. Real-world action tasks involve learning a number of task variations, such as different hockey shots (e.g., slapshot, backhand, wristshot). While a coach can tell us how to practice these variation, the reality is that most of our training time is spent alone. Project 2 will investigate the factors that affect our training choices during multiple task learning. Unlike previous research, a key aim of this project is to determine if these choices are optimal when the amount of training between skills can vary. Motor interactions between multiple agents is common in everyday life. In recent years, the prevalence of assistive robots has increased in a variety of rehabilitative and training settings (e.g., surgery) where the human and robot collaborate to achieve a common goal. This redundancy in coordination poses an appreciable computational challenge during learning. That is, the human must assign the error to a source even though the true source (i.e., self, robot, combination) is ambiguous. Project 3 will investigate the effectiveness and optimization of human-robot interactions for sensorimotor learning. The proposed research will contribute to our understanding of how the brain controls and learns real-world action tasks. Specifically, we will gain greater insight regarding the complex interplay between motor errors and decision-making. Having a solid understanding of these mechanisms may be able to promote and facilitate learning in workplace, surgical, and sport settings.

从我们行为的后果中学习是感觉运动学习和人类认知的基石。因此，准确地确定错误的来源(即分配)，并启动从一个动作到下一个动作的必要调整(即决策)，对于适应和以目标为导向的行为至关重要。然而，让我们对错误做出反应的机制和策略是理解支持感觉运动学习的决策过程所必需的，人们对此知之甚少。这项拟议研究的广泛目标是更好地理解我们如何将有关运动错误的信息纳入塑造感觉运动学习的决策中。 虽然人们认识到学习新运动技能的最好方法是通过广泛的训练，但我们知道并不是所有形式的练习都同样有效。有一致的证据表明，与实验强加的时间表相比，自我控制的反馈时间表对运动学习更有效。然而，这种学习益处背后的机制还没有被很好地理解。项目1将测试这一益处的两种相互竞争的解释，以更好地理解自我控制学习优势的认知和神经机制。 现实世界的动作任务包括学习许多任务变体，例如不同的曲棍球击球(例如，拍击、反手击球、腕击)。虽然教练可以告诉我们如何练习这些变化，但现实是，我们的大部分训练时间都是独自度过的。项目2将调查在多任务学习过程中影响我们培训选择的因素。与以前的研究不同，这个项目的一个关键目标是确定当不同技能之间的培训量可能不同时，这些选择是否是最佳的。 多个智能体之间的运动相互作用在日常生活中很常见。近年来，在人类和机器人合作以实现共同目标的各种康复和训练环境(例如外科手术)中，辅助机器人的普及程度有所增加。这种协调上的冗余在学习过程中构成了一个明显的计算挑战。也就是说，人类必须将误差分配给一个源，即使真正的源(即，自我、机器人、组合)是不明确的。项目3将调查人-机器人交互对感觉运动学习的有效性和优化。 这项拟议的研究将有助于我们理解大脑是如何控制和学习现实世界中的动作任务的。具体地说，我们将更深入地了解运动错误和决策之间的复杂相互作用。对这些机制有一个扎实的理解也许能够促进和促进在工作场所、外科手术和运动环境中的学习。