Brain-wide neural population dynamics during decision making and learning

项目总结/摘要 决策需要大脑中的神经元群体共同处理感官证据并选择 适当的行为反应。神经种群动力学（NPD），描述了一个神经元的反应是如何发生的。 随着时间的推移，单个神经元的群体展开，可以提供对这些决策过程的深刻见解。 关于个体神经元在选择决策任务中如何反应，我们已经了解得很多。然而， 关于神经元群体如何动态地执行决策计算，以及如何在一个 大脑区域的结构涉及许多决策任务。NPD中跨多个任务的共享功能可以 指示作为给定神经电路的多功能性的基础的统一的神经计算机制。 该建议旨在揭示人类和非人类中这些决策相关NPD的细节和结构 灵长类动物的背侧前运动皮层，这一区域与决策功能和决策功能障碍密切相关。 将特别注意审查如何在多个决策任务中组织国家方案设计 以及在学习新任务时这些NPD是如何出现的。在K99阶段，将使用新的分析工具， 开发用于从同时记录的神经群体活动中提取NPD，重要的是， 在NPD和它们在决策相关的神经计算中的作用之间提供可解释的联系。目标 为了确定与决策相关的计算的统一原则，大规模分析将整合现有的 以及新收集的涉及人类执行的多个决策任务的神经生理数据集 和非人类灵长类动物。在R 00阶段，拟议的研究将侧重于了解如何 NPD是在学习做出新的决策时出现的。该提案假定，容易，速度， 学习的效率和效果都取决于关键神经回路能够在多大程度上利用预先存在的神经回路。 计算的机制。这些概念将在协作的人类和非人类灵长类动物neu中进行测试， 生理学实验，由人工神经网络的询问提供指导， 类似的决策相关的学习任务。完成后，拟议的研究将提供新的基础 关于i）运动前皮层在许多决策过程中的多功能和适应性作用的知识 任务和ii）支持这种灵活性的神经计算的统一原则。更好地理解决策- 相关回路及其神经机制可以最终阐明许多精神疾病的基础。 损害决策的疾病，这最终可能会改善这些疾病的诊断和治疗 使人衰弱的状况。

Project Summary / Abstract Decision-making requires populations of neurons in the brain to collectively process sensory evidence and select appropriate behavioral responses. Neural population dynamics (NPDs), which describe how the responses of a population of individual neurons unfold over time, can provide an insightful view into these decision processes. Much is known about how individual neurons respond in select decision making tasks. However, little is known about how populations of neurons dynamically perform decision computations, and how resulting NPDs within a brain area are structured across many decision-making tasks. Shared features in NPDs across many tasks could indicate unifying neural mechanisms of computation that underlie the multi-functionality of a given neural circuit. This proposal aims to uncover the details and structure of these decision-related NPDs in human and nonhuman primate dorsal premotor cortex, an area tightly linked to both the function and dysfunction of decision making. Particular attention will be devoted to examining how NPDs are organized across multiple decision-making tasks and how those NPDs emerge during learning of new tasks. During the K99 phase, novel analytical tools will be developed for extracting NPDs from simultaneously recorded neural population activity and, importantly, for providing interpretable links between NPDs and their role in decision-related neural computation. With the goal of identifying unifying principles of decision-related computation, large-scale analyses will then integrate existing and newly collected neurophysiological datasets involving multiple decision-making tasks performed by humans and nonhuman primates. During the R00 phase, the proposed research will pivot toward understanding how NPDs emerge during learning to make new types of decisions. The proposal postulates that the ease, speed, and efficacy of learning all hinge on the extent by which critical neural circuits can leverage pre-existing neural mechanisms of computation. These concepts will be tested in collaborative human and nonhuman primate neu- rophysiological experiments, with guidance provided by interrogations of artificial neural networks posed with similar decision-related learning tasks. Upon completion, the proposed research will provide new fundamental knowledge concerning i) the multi-functional and adaptive role of premotor cortex across many decision-making tasks and ii) unifying principles of neural computation that support this flexibility. Better understanding decision- related circuits and their neural mechanisms could ultimately elucidate the basis for the numerous psychiatric disorders that impair decision making, which could eventually lead to improved diagnosis and treatment of these debilitating conditions.