社会与工程心理学研究室知识库

重性抑郁障碍(Major Depressive Disorder, MDD)是目前最为流行的精神疾病之一，给社会和患者本人都带来巨大的负担和痛苦。静息态功能磁共振成像(Resting state functional Magnetic Resonance Imaging, R-fMRI)由于其具有的实验设计简单易行、对人体无任何伤害、便于大数据累积等独特优势，近年来被广泛应用于MDD生物标记物的探寻。然而目前R-fMR】领域研究结果较差的可重复性使得这方面的探究进展缓慢。这一方面可能是由于样本量、实验设计、统计方法等方法学因素，另一方面也可能是个体在静息状态下自发思维的复杂性导致的。本研究首先使用真实的R-fMRI数据验证了目前R-fMRI领域研究的可重复性现状。随后针对一种常见的与MDD密切相关的自发思维形式一一反当思维背后的认知神经机制进行了探究，希望从两个方面回应较差的可重复性制约MDD生物标记物搜寻的问题，从新的方向探索1VIDD脑机制。

尽管己有大量理论分析指出，目前R-fMRI研究的结果的可重复性不高，但是目前尚无实证研究利用真实的R-fMR】数据对该领域的可重复性水平进行验证，也缺乏针对可重复性的影响因素的实证研究。因此，研究一使用国际公开的R-f1VIR】共享数据集验证了主流R-fMRI指标的可重复性水平，并探究了多重比较矫正的选择、实验设计和样本量等因素对R-fMRI研究结果可重复性的影响。结果表明，目前R-fMRI研究结果的可重复性不容乐观，但是采用合适的多重比较矫正方法、使用被试内实验设计和采集合适样本量的数据可以有效地提升结果的可重复性。

既往有关反当思维脑机制的研究提示，反当思维与三个大尺度脑网络:默认网络(Default Mode Network, DMN、执行控制网络(Central Executive Network,CEN)以及凸显网络(Salience Network, SN)的相关脑区有关。然而这三个脑网络之间的交互作用以及默认网络内部三个子系统与反当思维之间的关系尚不明确。目前传统的反当思维研究范式存在局限性，难以回答这些问题。因此研究二采用较为新颖的“反当思维状态”范式，针对目前该范式存在的缺陷进行了改进，并结合研究一的结果设计了重复测量的被试内设计的“反当思维状态”任务，以较大样本量(N=41)的健康成人作为被试在三台不同的扫描仪上探究了可重复的反当思维脑网络机制。研究二的结果表明，反当思维状态下，个体默认网络内部核心子系统和内侧颗叶子系统之间的功能连接显著上升，但核心子系统和背内侧前额叶子系统之间的功能连接显著下降，这一结果可以在所有三台扫描仪上得到重复。研究二的结果进一步凸显了默认网络在反当思维脑机制中扮演的关键作用，并提示默认网络不同子系统之间的交互在反当思维状态下存在着独特的模式。

反当思维在现象学上表现为自我强化、循环反复、往往持续时间很久的心理状态，因此很有可能在脑活动的动态性特征上存在独有的特征。研究三使用“滑动窗口”C Sliding window)分析技术探讨了反当思维状态下各个功能磁共振指标随时间的动态变化特征。结果表明，在反当思维状态下，镜像同伦功能连接(Voxel-Mirrored Homotopic Connectivity, VMHC)指标在内侧前额叶以及颗顶联合区的脑活动动态性显著上升。此外，我们还发现全局信号相关(Global SignalCorrelation, GSCorr)指标在外侧颗叶脑区的动态性变化程度显著上升。这些结果可以在不同的扫描仪上得到重复。研究三的结果验证了研究二中发现的默认网络，特别是默认网络核心子系统在反当思维中扮演的关键作用。并进一步提示了这些脑区在反当思维过程中表现出的偏侧化倾向。外侧颗叶脑区的相关发现则提示了反当思维可能也与语义性长时记忆过程有关。

既往研究还发现，反当思维在不同的个体之间存在着相关，然而对这种反当思维状态下脑活动的被试间相关还没有实证研究进行过探究。此外，这种被试间相关算法还可以排除一部分被试间和被试内变异对结果的影响，具有提升结果可重复性的潜力。因此研究四对反当思维状态下脑活动的被试间相关进行了探究。结果表明，反当思维状态下腹外侧前额叶以及外侧颗叶脑区的被试间相关显著上升，并且这一结果可以在不同的扫描仪上得到重复。研究四的结果提示了不同的个体在进行反当思维时同样地有有关语义性长时记忆过程的参与，暗示反当思维过程中，与客观标准之间的比对是一种跨被试的重要的认知神经过程。

总之，本研究的结果表明，经过精心设计并分析得当的fNIRI研究能够得到可重复的结果;默认网络在反当思维状态的脑机制中扮演着核心的作用，但不同的子系统之间的交互表现为不同的模式;反当思维可能与两个不同的脑环路有关:由内侧前额叶一后侧扣带回一内侧颗叶组成的与个体化自传体记忆有关的，具有较强情绪性的环路，和由腹外侧前额叶一外侧颗叶构成的与语义性长时记忆有关的偏认知的环路。本研究首先指出了提升目前R-fMRI研究可重复性的可行方案，另外，本研究从脑网络、动态性以及被试间相关三个角度发现的反当思维可重复的脑机制具有作为MDD新的生物标记物的巨大潜力。

Major depressive disorder (MDD) is one of the most prevalent psychiatric disorders worldwide, leading to heavy burden and pain to both society and patients. Resting state functional magnetic resonance imaging (R-fMRI) is being increasingly adopted in searching biomarkers of MDD due to its easy acquisition of data, safety and amenability to aggregating data across different sites. However, concerns regarding its reproducibility have been raised recently. This "reproducibility crisis" may be caused by a) the prevalence of small sample size, the small effect size of experimental design and the abuse of liberal multiple comparison strategies or b) the complexity of the subjects' spontaneous thoughts during the resting state. Thus, we first verified the reproducibility of typical R-fMRI research with real R-fMRI data, then systematically investigated the reproducible neural underpinnings of rumination, a common and important kind of spontaneous thought that is closely related to MDD.

Despite the large amount of literature criticizing the reproducibility of R-fMRI research theoretically, no empirical study has demonstrated this issue with real data until now. Thus, study 1 intended to verify the current status of typical R-fMRI research with openly shared R-fMRI datasets and to investigate the impact of multiple comparison strategies, sample size and experimental design on the reproducibility. Results of study 1 revealed a relatively low reproducibility of present R-fMRI research. Luckily, we also found that with certain kind of multiple comparison correction, relatively large sample size and within-subject design, the reproducibility of R-fMRI study can be improved substantially.

Previous research indicated a connection between rumination and brain regions belong to three large-scale brain networks: the default mode network (DMN), the central executive network (CEN) and the salience network (SN). Nevertheless, the association between interactions among these three networks and the three sub-systems of DMN and the neural mechanisms of rumination is largely unknown. Furthermore, the traditional paradigm of rumination research is not suitable for addressing this issue. Hence, in study 2, we recruited a relatively large sample of young healthy adults (N=41) and conducted a modified new "rumination state" task to unravel the brain network neural mechanisms behind rumination. All participants finished the "rumination state" task on all 3 distinct scanners and results are reported only if they can be reproduced by different scanners. Study 2 revealed that during rumination sate, functional connectivity between two DMN sub-systems: the core sub-system and the Medial Temporal Lobe (MTL) subsystem was significantly enhanced while functional connectivity between core sub-system and DMPFC sub-system is reduced. Importantly, these results can be repeated by all 3 distinct scanners. These findings further emphasize the critical role DMN played in the neural mechanism behind rumination and give a more nuanced description of the network underpinnings of rumination.

Rumination is deemed as a self-strengthened, cycled and long-lasted mental state. These phenomenological characteristics may lead to unique dynamic brain activity modes. To investigate these possible dynamic features, study 3 used sliding window method to analyze the dynamic properties of main-stream fMRI metrics in rumination state. We found the variation of voxel-mirrored homotopic connectivity (VMHC) is raised in medial prefrontal cortex (MPFC) and temporal parietal conjunction (TPJ) in rumination state. Moreover, an elevated variation of global signal correlation (GSCorr) was revealed in the lateral temporal cortex (LTC). These results can be reproduced through different scanners. Study 3 verified the findings from study 1 emphasizing the key role of DMN, especially the core sub-system of DMN, in rumination's neural underpinnings and further indicated the lateralization of these brain regions in rumination state. Findings regarding LTC may imply an involvement of semantic long-term memory m rumination.

Previous literature also elaborated a correlation among different individual's rumination pattern and different time points of one person, but no empirical research has been conducted to investigate the brain mechanisms behind this phenomenon. Besides, the algorithm of this inter-subject correlation can partly rule out the impact of between-subject variation on the reproducibility of fMRI research and may help improve the reproducibility. Thus, study 4 investigated the inter-subject correlation in rumination state and found that the inter-subject correlation is significantly elevated in the inferior frontal gyrus (IFG) and LTC during rumination state. This finding can be reproduced through distinct scanners. Stu街4 indicated that different individual similarly engages in the semantic long-term memory in rumination state and implied that the comparison between current status and objective criterion may be an inter-subject mental process underlying rumination.

In summary, the present study demonstrated that results from a well-designed and decently-analyzed fMRI research can be reproducible; DMN plays a critical role in the brain mechanisms underlying rumination while the interactions between different sub-systems revealed a distinct pattern. Rumination's neural underpinnings may consist of 2 circuits that contain different brain regions and underlie different mental process: an emotional MPFC-PCC-MTL circuits that is associated with individualized autobiographical memory and a cognitive IFG-MTL circuit that is related to the semantic long-term memory. The present study firstly pointed out a possible way to improve reproducibility of R-fMRI research. The revealed reproducible neural mechanisms of rumination can also be served as a potential biomarker for MDD.