认知与发展心理学研究室知识库

研究者们利用正电子成像(Positron Emission Tomography, PET), 功能磁共振成像(functional Magnetic Resonance Imaging, fMRI), 脑磁图(Magnetoencephalography，MEG)对静息态下大脑活动的研究揭示了大脑内在的功能组织形式，确证了静息态网络(Resting-State Networks)的存在；利用多种实验范式，研究者们发现这些静息态网络与大脑功能之间存在关联，还可以帮助揭示精神疾病的机制。近来的研究指出，静息态网络中的一个与人格、反省等重要功能有关的成员：默认网络（Default Mode Network, DMN）与另一个与情景记忆、刺激新旧性的判断有关的顶叶记忆网络(Parietal Memory Network, PMN)存在解剖位置上的易混淆性。这种易混淆性再加上任务范式或静息态研究的实验条件限制，使得我们难以预测这两个静息态网络在复杂的真实生活情景中是否还具有功能的区别。另一方面，由于静息态网络的重要性质和意义，在一些采用任务范式或静息态的研究中它们已经隐含地被认为可以编码大脑的活动：不同认知活动的差异或具有不同认知特质的人群的区别反映在静息态网络的变化上。但还缺少证据说明静息态网络可以编码大脑在真实生活中的活动。本课题以自然观影(Natural Viewing)为真实生活情景的一种模拟，通过研究DMN 和PMN 在两个自然观影实验条件下的功能分离来了解这两个静息态网络是否在真实生活情景中仍然具有功能分离，并进一步探索可否利用静息态网络对自然观影条件下的大脑活动进行编码。本课题通过网络的代表性时间序列、网络间功能连接、网络内部及外部的功能一致性来反映功能分离，发现在观看完整视频时，相对于观看画面顺序被打乱的乱序视频，DMN 的代表性时间序列的方差和偏度更低，而PMN 呈现出相反的变化方式；将位于初级视觉皮层的初级视觉网络(medial Visual network, mV)作为参照加入分析后，在观看视频时PMN 总有比DMN 更强的与mV的功能连接；以种子点功能连接方法计算的网络内部、外部功能一致性分析发现DMN、PMN在两个自然观影条件下都呈现出内部强、外部弱的功能一致性模式，表明这两个网络具有相对独立性。这些结果表明静息态网络DMN、PMN 在自然观影条件下不仅具有功能分离，还反映了大脑的功能组织形式。以此为启发我们将静息态网络作为编码大脑活动的基础编码了两种自然观影状态下的大脑活动，将编码后的大脑活动与自然观影条件通过分类器建立关联，最后以这分类器的分类准确率的统计显著性来反映这种编码的有效性。结果发现PMN、DMN、mV 的代表性时间序列的方差、偏度以及网络间功能连接形成的分类器可以有效的实现编码。这些结果表明，静息态网络DMN、PMN 在自然观影条件下存在功能分离，应当作为不同的两个网络进行研究；这些静息态网络也反映了大脑在自然观影时的功能组织形式，可以对大脑活动进行编码。

In recent years, Positron Emission Tomography (PET), functional Magnetic Resonance Imaging (fMRI), Magnetoencephalography (MEG) have helped researchers to reveal the functional organization of brain during resting state and revealed the existence of resting-state networks (RSNs). By applying experiment paradigms researchers also revealed the relationships between RSNs and brain functions. Some recent studies propose that an RSN called Default Mode Network (DMN), which plays important roles in trait and introspection, is spatially adjacent to another RSN called Parietal Memory Network (PMN), which is related to the episodic memory and the oldness of the stimuli. Such spatial overlap and the limitations in the conventional experiment setting: the highly controlled experiment paradigms or simply resting, constrain our understanding about whether these two RSNs still functionally segregate in the complex real live. On the other hand, in some conventional studies RSNs have implicitly become the basis in encoding brain functions: the difference in cognitive processes or the diversity among populations with different traits are measured and revealed by the alternations of RSNs. However, it is unclear whether RSNs could encode brain functions in real live, which contains complex and temporal continuous information. The current study introduced participants to freely viewing movies in the scanner to simulate the real-life situation and to investigate whether DMN and PMN were functionally segregated during the Natural Viewing (NV) conditions. Such functional segregation is crucial for encoding brain functions with RSNs. The functional segregation was reflected by the representative time series, functional network connectivity and intra/inter network functional homogeneity. The results showed that, compared with while watching the scrambled version, during watching the original normal movie the representative time series of DMN showed lower variance and skewness, while PMN was modulated in an inverse way. We introduced a medial visual network (mV) located at the primary visual cortex as a reference network and compare the functional network connectivity between DMN, PMN and mV, the results showed that PMN was higher connected with mV compare with DMN, in both movie conditions. The seed-based functional connectivity analysis revealed higher intra-network functional connectivity homogeneity compared with inter-network homogeneity, which reflected the relative functional independence between DMN and PMN. Those results evidence the functional segregation between DMN and PMN during NV, also indicate that these RSNs still reflect the functional organization in reallife setting. Inspired by this, we encoded the fMRI data during watching movies with the RSNs and tested the encoding paradigm by assessing the performance of a classifier relating the encoded data and the movie conditions. The results showed that the classifier could predict the movie condition by the variance, skewness and the functional network connectivity derived from the representative time series. The results indicate that the resting-state network DMN and PMN are functionally segregated during watching movies, and they reflect the functional organization of brain during the real-live setting. Thus, the brain activity could be encoded by the RSNs.