Multi-regional neural circuit dynamics underlying short-term memory

摘要 短时记忆（STM）是一种核心认知功能，对于推理、决策和灵活性至关重要。 行为虽然它已经被认为是一个关键的功能感兴趣的几十年来，关键的神经 对STM的基底了解很少。记录实验发现，分布式持续活动与 然而，目前还不清楚关键的回路节点是什么，STM是否是 由单个分布式电路支持或同时涉及许多不同的并行表示，以及 在不同的大脑区域，与短时记忆相关的活动之间是否存在因果关系。中的这一空白 知识源于缺乏对产生短时记忆的神经动力学的整体观点。全面 在一个单一的短时记忆行为中检查多个大脑区域的研究严重缺乏。此外，委员会认为， 单靠记录无法评估多个大脑区域的信号如何相互关联， 行为瞬态扰动是研究递归神经网络的一种有效方法。的响应 瞬态扰动之后的神经动力学与网络结构有关。的目的 一项建议是在多个大脑区域使用区域特异性和时间精确的扰动来探测关键的 STM的电路节点。这将证明我们的方法识别多区域电路的潜力 驾驶行为，并产生关于其网络结构的假设。我们把注意力集中在额叶皮层 区域（ALM），这是我们开发的STM行为的关键。小鼠进行感觉辨别，然后 一个延迟时期，在此期间，他们保持对即将到来的运动反应的STM。ALM神经元表现出持续的 与即将到来的运动有因果关系的活动。STM神经动力学通过ALM 投射到脑干来触发行为反应。这些属性使ALM成为理想的切入点 读出由不同电路节点的相互作用引起的动态，并将动态与 行为我们将首先获得一个全面的数据集，在那里我们瞬时扰动所有候选大脑 STM区域（使用记录和解剖学识别），同时监测它们对STM的影响 使用最新的硅探针技术（Aim1）进行种群记录，然后我们将 对扰动动力学进行理论分析，得出与特定动力学相关的模型类 行为，并作为底层电路结构（Aim2）的假设。如果成功，这种方法将 在一个单一的行为中识别关键的解剖学基底和相关的STM神经动力学。模型 其结果将指导对多地区STM进行真正有针对性的全面调查 电路.

Abstract Short-term memory (STM) is a core cognitive function, critical for reasoning, decision-making, and flexible behavior. Though it has been recognized as a key function of interest for many decades, the critical neural substrate of STM is little understood. Recording experiments have found distributed persistent activity related to STM across multiple brain areas, yet it is unclear what the critical circuit nodes are, whether STM is supported by a single distributed circuit or involves many distinct parallel representations at the same time, and what if any are the causal relations between STM related activity in different brain areas. This gap in knowledge stems from lacks of a holistic view of the neural dynamics giving rise to STM. Comprehensive studies that examine multiple brain regions within a single STM behavior are severely lacking. Moreover, recordings alone cannot assess how signals across multiple brain regions are related to each other and to behavior. Transient perturbation is a powerful approach to probe recurrent neural networks. The response of neural dynamics following a transient perturbation are related to the network structure. The objective of this proposal is to use region-specific and temporally-precise perturbations in multiple brain regions to probe critical circuit nodes for STM. This will demonstrate the potential of our approach for identifying multi-regional circuits driving behavior and generate hypotheses regarding their network structure. We focus on a frontal cortical region (ALM) that is critical for a STM behavior we developed. Mice make a sensory discrimination, followed by a delay epoch in which they maintain a STM of their upcoming motor response. ALM neurons exhibit persistent activity that are causally related to the upcoming movement. STM neural dynamics are funneled through ALM projections to the brainstem to trigger a behavioral response. These properties make ALM an ideal entry point to read out the dynamics arising from the interactions of different circuit nodes and relate the dynamics to behavior. We will first obtain a comprehensive dataset where we transiently perturb all of the candidate brain regions for STM (identified using recordings and anatomy) while monitoring their consequences on STM dynamics in ALM with population recordings using the latest silicon probe technology (Aim1). We will then perform theoretical analysis of the perturbed dynamics leading to model classes that relate specific dynamics to behavior and serve as hypotheses for the underlying circuit structure (Aim2). If successful, this approach will identify the critical anatomical substrate and the relevant STM neural dynamics in a single behavior. The model outcome will subsequently guide a truly targeted yet comprehensive investigation of the multi-regional STM circuit.