Decoding sleeping brain activity: Integrating experiments and computational neuroscience to elucidate the cognitive benefit of sleep

大脑的各种功能，如感知、学习、记忆和决策，都是从神经计算中产生的。利用电生理学和计算方法，我们一直特别关注“学习和记忆”，因为它提供了其他高级认知功能的基础。记忆大致可分为外显记忆和内隐记忆。外显记忆是一种你可以有意识地回忆和口头描述的记忆(例如，你今天早上吃的早餐)。内隐记忆是一种你无法有意识地回忆或口头描述的记忆(例如，骑自行车所需的运动技能)。关于这些记忆的两个重要问题是它们在大脑中如何表现，以及它们如何/何时在神经回路中巩固。一种被广泛支持的理论是，一组在解剖上/功能上相连的神经元，称为“细胞集合”，代表记忆。也有人提出，记忆的巩固部分发生在睡眠期间。睡眠的这种记忆功能得到了实验的支持，实验表明，行为诱导的神经活动在睡眠期间重新激活(记忆重新激活)，而记忆在睡眠期间通过电刺激进行修改。虽然我们有这些记忆和睡眠的一般框架，但术语‘细胞组装’缺乏严格的定义，我们仍然不明白不同类型的记忆是如何由细胞组装来代表的。我们也不知道细胞组件在睡眠期间是如何被修改的。此外，通过睡眠期间的电刺激来改变记忆的实验受到了最近负面结果的挑战。因此，在这个方案中，我们将进行电生理实验，以澄清记忆和电刺激的混合实验结果，开发系统的细胞组装检测方法来了解外显和内隐记忆是如何被表征的，并开发一个计算模型来了解睡眠如何改变神经回路从而改变记忆。这一提议的结果将突出外显记忆和内隐记忆之间的异同，提供更好的存储容量估计，澄清睡眠期间电刺激的影响，并提供更好的机制洞察睡眠期间突触可塑性如何改变神经活动。这项工作还将有助于改进大脑启发的技术。例如，深度学习可以通过实现睡眠的记忆功能来变得更灵活，获得更好的分类能力。

Various brain functions such as perception, learning and memory and decision making emerge from neural computations. Using electrophysiological and computational approaches, we have been specifically focusing on `learning and memory' because it provides the basis of other high-order cognitive functions. Memory is broadly categorized into explicit and implicit memories. Explicit memory is a type of memory that you can recall consciously and verbally describe (for example, the breakfast you had this morning). Implicit memory is a type of memory that you cannot recall consciously nor verbally describe (for example, the motor skill required to ride a bike). Two important questions regarding these memories are how they are represented in the brain and how/when they are consolidated in the neural circuits. A widely supported theory is that a group of anatomically/functionally connected neurons, called a `cell assembly', represents memory. It has been also proposed that consolidation of memory occurs in part during sleep. This memory function of sleep has been supported by experiments showing that behaviourally induced neural activity reactivates during sleep (memory reactivation) and that memory is modified by electrical stimulation during sleep. Although we have these general frameworks for memory and sleep, the term `cell assembly' lacks the stringent definition and we still do not understand how different types of memory are represented by cell assemblies. We do not know how cell assemblies are modified during sleep either. Furthermore, experiments in which memory was modified by electrical stimulation during sleep were challenged by recent negative results. Therefore, in this proposal, we will perform electrophysiological experiments to clarify the mixed experimental results on memory and electrical stimulation, develop a systematic cell assembly detection method to understand how explicit and implicit memories are represented, and develop a computational model to understand how sleep modifies neural circuits and hence memories. The outcomes of this proposal will highlight similarities and differences between explicit and implicit memory, provide better estimates of storage capacity, clarify the effect of electrical stimulation during sleep, and provide better mechanistic insights into how synaptic plasticity during sleep modifies neural activity. The work will also help improve brain-inspired technologies. For instance, deep learning may become more flexible and obtain better categorization ability by implementing memory function of sleep.