Temporal Patterns in Sleep Mechanisms of Learning

睡眠学习机制的时间模式

• 批准号: 7086997
• 负责人: DANIEL MARGOLIASH
• 金额: $ 24.93万
• 依托单位: UNIVERSITY OF CHICAGO
• 依托单位国家: 美国
• 财政年份: 2002
• 资助国家: 美国
• 起止时间: 2002-09-20 至 2007-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7086997
• 关键词: behavior prediction; behavioral /social science research tag; brain electrical activity; computational neuroscience; electrophysiology; ethology; juvenile animal; learning; mathematical model; memory; method development; model design /development; neural information processing; neurophysiology; sleep; songbirds; statistics /biometry

DESCRIPTION (provided by applicant): Sleep is a universal behavior. One functional hypothesis proposes that during sleep perceptual or procedural activity experienced during the prior day is recapitulated, with some memories consolidated and others erased. This hypothesis is supported by observations of individual neurons in the zebra finch forebrain nucleus RA that exhibit neuronal replay, similar patterns of bursting activity during singing and later during undisturbed sleep and in response to song playback. Directly testing the hypothesis, however, is hampered by the lack of procedures to analyze temporal patterns of neuronal activity, especially during sleep in the absence of a time-reference. The proposed research will develop essential statistical modeling and neurophysiological experiments to directly examine functional models of sleep. In the first experiment, the hypothesis of sleep replay in RA will be rigorously statistically tested. Patten filtering methods will be developed to assess matching not only for individual bursts but also for trains of bursts. The statistical significance of replay patterns will be described and a more complete description of the replay phenomenon at the single cell level will be derived. In the second experiment, the hypothesis that sleep replay occurs in nuclei afferent to RA will be tested. The activities of different projection classes of neurons in the afferent nucleus HVc will be recorded. Information-based and metric-based alignment techniques for temporal alignment will be developed to accommodate the lower temporal resolution of these neurons, as compared to RA neurons. The internal noise, and context-dependent temporal precision will be considered to maximize spike alignment. In the third experiment, the specific hypothesis that afferent input from IMAN modifies RA activity on a nocturnal cycle will be tested. A new statistical model, adaptive pattern filtering, will be developed to assess when during sleep are burst patterns first recognized that are destined to be expressed in singing behavior the following day. The memory consolidation hypothesis will be tested by assessing the prediction that after IMAN lesions there should be fewer or no changes in RA burst patterns following periods of sleep. The fourth experiment will record directly from RA in juvenile birds learning to sing. In this case, behavior (singing) and neurophysiological activity during the day as well as during sleep is likely to exhibit high variability which will require statistical modeling. Changes in behavior will be assessed with entropy and other statistical measures. The hypothesis that RA burst patterns change during sleep in a direction that predicts singing behavior the following day will be directly tested.

描述（由申请人提供）：睡眠是一种普遍的行为。一种功能假说认为，在睡眠过程中，前一天经历的知觉或程序性活动被重现，一些记忆得到巩固，另一些记忆被抹去。这一假设得到了对斑胸草雀前脑核RA中单个神经元的观察的支持，这些神经元表现出神经元重放，在唱歌期间和随后在不受干扰的睡眠期间以及对歌曲重放的反应中出现类似的爆发活动模式。然而，由于缺乏分析神经元活动的时间模式的程序，特别是在没有时间参考的睡眠期间，直接测试这一假设受到了阻碍。拟议的研究将发展必要的统计建模和神经生理学实验，以直接检查睡眠的功能模型。在第一个实验中，我们将对睡眠重放假说进行严格的统计检验。模式滤波方法将被开发出来，不仅用于评估单个突发的匹配，也用于评估一系列突发的匹配。将描述重放模式的统计意义，并推导出单细胞水平上重放现象的更完整描述。在第二个实验中，睡眠重放发生在RA传入核的假设将被验证。记录传入核HVc中不同投射类型神经元的活动。与RA神经元相比，将开发基于信息和基于度量的时间定位技术，以适应这些神经元的较低时间分辨率。内部噪声和上下文相关的时间精度将考虑最大化尖峰对准。在第三个实验中，将测试来自IMAN的传入输入改变夜间周期的类风湿性关节炎活动的具体假设。一种新的统计模型，即自适应模式过滤，将被开发出来，以评估睡眠中什么时候会首次识别出爆发模式，这些模式注定会在第二天的歌唱行为中表达出来。记忆巩固假说将通过评估预测来验证，即在IMAN损伤后，睡眠后RA爆发模式的变化应该更少或没有变化。第四个实验将直接记录幼年鸟类学习歌唱的RA。在这种情况下，行为（唱歌）和白天以及睡眠期间的神经生理活动可能表现出高度的可变性，这将需要统计建模。行为的变化将用熵和其他统计方法来评估。睡眠期间RA爆发模式的变化方向可以预测第二天的歌唱行为，这一假设将直接得到验证。