Decoding memory reprocessing during sleep

在编码之后，新的记忆必须进行转换才能长期存在。在这种巩固过程中，支持新记忆的神经底物从海马区转移到新皮质区域，这一过程被称为系统记忆巩固。睡眠被认为在这种转变中扮演着重要的角色。通过重新激活与学习相关的神经元活动，信息可能会被转移到神经系统，这些系统比最初编码记忆的网络更不可塑性，但因此提供了更稳定的记忆存储。直到最近，人们还很难研究支持人类睡眠期间记忆巩固的隐蔽过程。随着机器学习方法被引入认知神经科学，检测大脑活动中反映学习相关信息处理的隐藏模式已经成为可能。在提议的项目中，我将更详细地研究作为睡眠对记忆有益影响的关键机制的重新激活的特性。我将与我的东道主普林斯顿计算记忆实验室的肯尼斯·诺曼教授一起开发机器学习方法，以解码睡眠中的记忆重新处理。诺曼教授是多变量模式分析方面的专家。一个主要目标将是确定可用作模式分类分析特征的有前景的数据转换。这应该使个体和记录会话之间的滋扰差异保持最小，同时最大化与记忆处理内容相关的信息。然后，这些方法将被应用于研究自发记忆再激活和使用外部再激活线索的定向记忆再激活的性质。特别是，我们将(1)研究有助于巩固记忆的振荡活动，(2)检查睡眠大脑的不同区域在晚上什么时候处理与学习有关的信息，以及(3)描述睡眠大脑中的不同区域在处理与学习相关的信息时如何相互作用。我们的结果有助于阐明在人脑中导致稳定记忆存储的隐蔽过程。通过确定哪些振荡事件参与了记忆材料的再处理，以及这种活动如何跨不同的大脑区域相互作用，我们可以更好地理解睡眠是如何促进系统记忆巩固的。在未来，这可能有助于有效地针对这些机制，促进长期记忆的形成。

After encoding, new memories have to be transformed to last for the long-term. During the course of this consolidation, the neural substrate supporting new memories shifts from hippocampal to neocortical regions, a process that has been termed systems memory consolidation. Sleep has been proposed to play an important role in this transition. By a reactivation of learning-related neuronal activity, which emanates from the hippocampus and spreads across the cortex, information may be transferred to neural systems that are less plastic than the network that initially encoded the memory, but consequently offer more stable memory storage. Until recently, it has been difficult to study the covert processes that support memory consolidation during sleep in humans. With the introduction of machine learning methods to cognitive neuroscience, it has become possible to detect the hidden patterns in brain activity that reflect processing of learning-related information. In the proposed project, I will examine in more detail the properties of reactivation as a key mechanism underlying the beneficial effects of sleep on memory. Together with my host Prof. Kenneth Norman of the Princeton Computational Memory Lab, who is expert in multivariate pattern analysis, I will develop machine learning approaches to decode memory reprocessing during sleep. A main objective will be to identify promising data transformations that can be used as features for pattern classification analysis. These should keep nuisance variance between individuals and recording sessions minimal, while maximizing the information related to the content of memory processing. These methods will then be applied to study the properties of both spontaneous memory reactivation and targeted memory reactivation using external reactivation cues. In particular, we will (1) investigate oscillatory activity that contributes to memory consolidation, (2) examine at what times of the night different regions of the sleeping brain process learning-related information, and (3) describe how different regions in the sleeping brain interact when processing learning-related information. Our results can help to clarify the covert processes the lead to stable memory storage in the human brain. By identifying which oscillatory events are involved in reprocessing mnemonic material and how this activity interacts across different brain regions, we can get a better understanding of how sleep contributes to systems memory consolidation. In the future, this may help to effectively target these mechanisms to facilitate long-term memory formation.