CRCNS: Neural computations underlying sequence memory consolidation in sleep

可以说，存储和检索顺序相关信息的能力是智能 行为。它使我们能够预测感官情况的结果，通过产生 一系列的运动动作，在精神上探索不同导航或运动的可能结果 选择，并最终通过灵活链接生成的复杂言语序列进行交流 更简单的元素序列是在童年时期学到的。睡眠从被学习的人中提取不变的特征 信息，导致显性知识和洞察力的产生。尽管取得了显著的进步， 包括PI和该项目的共同PI所做的工作，关于睡眠在记忆中的作用，许多关键问题仍然存在 和学习。在这里，我们建议通过计算的发展来解决这些问题 通过在小鼠身上的体内实验来探索和验证的模型。我们将探索海马体 (HC)和新皮质(NC)机制是如何获得序列以及随后如何获得的 在慢波睡眠(SWS)期间通过离线重放进行整合，最大限度地减少 重叠和/或反向序列之间的干扰以及NC如何链接序列片段 在一起。我们结合了计算机建模(巴泽诺夫)的尖峰神经网络，模拟清醒和 SWS脑动力学，包括NC慢振荡和HC尖锐波纹(SWR)，具有高密度 在包括序列在内的受控行为环境中的小鼠的神经系综记录(McNaughton) 学习和随后的化学遗传诱导的SWS，这使得观察如何学习成为可能 NC中的序列表示在长时间的SWS中自发地进化。私家侦探已经 在过去的几年里就这个主题进行了合作和讨论，从而产生了具体的假设 可以在真正的大脑中进行探索。项目成果将使我们更好地了解 是从经验中提取出来的，涉及什么大脑回路，以及大脑动力学是如何由 建立丰富的世界内部模型，包括预测当前 情况和自己在这种情况下的行动。 相关性(请参阅说明)： 存储和检索顺序相关信息的能力是智能行为和 大脑执行功能。这种能力的缺陷是由大脑回路中断引起的，在 抑郁症、精神分裂症和创伤后应激障碍。更好地理解这些机制和脑动力学 顺序信息的获取、合并和检索的基础将导致干预 改善健康受试者和精神疾病患者的认知表现、记忆和学习。

The ability to store and retrieve sequentially related information is arguably the foundation of intelligent behavior. It allows us to predict the outcomes of sensory situations, to achieve goals by generating sequences of motor actions, to 'mentally' explore the possible outcomes of different navigational or motor choices, and ultimately to communicate through complex verbal sequences generated by flexibly chaining simpler elemental sequences learned in childhood. Sleep extracts invariant features from the learned information, leading to the generation of explicit knowledge and insight. Despite remarkable progress, including work by PI and co-PI of this project, many critical questions remain about role of sleep in memory and learning. Here we propose to address these questions through the development of computational models that are probed and validated through in vivo experiments in mice. We will explore the hippocampal (HC) and neocortical (NC) mechanisms underlying how sequences are acquired and subsequently consolidated through off-line replay during Slow Wave Sleep (SWS) in a manner that minimizes interference between overlapping and/or reversed sequences and how NC may chain sequence fragments together. We combine computer modelling (Bazhenov) of spiking neural networks that mimic awake and SWS brain dynamics, including NC slow oscillations and HC Sharp Wave Ripples (SWR), with high density neural ensemble recordings (McNaughton) in mice, in a controlled behavioral setting including sequence learning and subsequent, chemogenetically induced SWS, which makes it possible to observe how learned sequence representations in NC evolve spontaneously over prolonged periods of SWS. The PIs have been collaborating on and discussing this topic for the past several years, resulting in specific hypotheses that can be explored in real brains. The project outcome will provide a better understanding of how knowledge is extracted from experience, what brain circuits are involved and how brain dynamics are shaped by the development of a rich internal model of the world, including the ability to predict the outcomes of current situations and one's own actions in that context. RELEVANCE (See instructions): The ability to store and retrieve sequentially related information is the foundation of intelligent behavior and brain executive function. Deficits in this ability, resulting from disruption of brain circuits, are seen in depression, schizophrenia and PTSD. Better understanding of the mechanisms and brain dynamics underlying the acquisition, consolidation and retrieval of sequential information will lead to interventions to improve cognitive performance, memory and learning in healthy subjects and patients with mental illness.