It has been 17 years since Shulman et al. (1997) first noted that a constellation of areas in the human cerebral cortex consistently reduced its activity while performing various novel, non-self-referential, goal-directed tasks (Shulman et al. 1997) when these tasks were compared with a control state of quiet repose (i.e., a resting state of eyes closed or visual fixation). That these localized reductions in activity were occurring at all was surprising, and their consistency across a wide variety of tasks made it all the more remarkable. The immediate challenge was to prove that these activity decreases were not due to activations in the resting state caused by experimentally uncontrolled cognition.
自从 Shulman 等人（1997）首次指出人类大脑皮层中的一组区域在执行各种新颖的、非自我相关的、目标导向的任务时，其活动始终减少（Shulman et al. 1997），已经过去了 17 年。这些任务与安静休息状态（即闭眼或视觉固定的休息状态）进行比较时，这些局部活动的减少令人惊讶，而它们在各种任务中的一致性更是引人注目。眼下的挑战是证明这些活动减少并非由于实验中未控制的认知引起的休息状态激活。

In 2001 we used positron emission tomography (PET) measurements of regional blood flow and oxygen consumption to show, by established metabolic criteria for activation, that areas consistently exhibiting activity reductions during task performance were not activated in the resting state. Our article was titled, “A Default Mode of Brain Function” (Raichle et al. 2001). We concluded that the brain areas observed to decrease their activity during attention-demanding, goal-directed tasks were not activated in the resting state but, rather, were indicative of a heretofore-unrecognized organization within the brain's intrinsic or ongoing activity. Parenthetically, it had not occurred to us that others would anoint the constellation of areas exhibiting this unique behavior as the brain's default mode network. The name obviously caught on.
在 2001 年，我们使用正电子发射断层扫描（PET）测量区域血流和氧气消耗，通过已建立的代谢激活标准，显示在任务执行过程中持续表现出活动减少的区域在静息状态下并未被激活。我们的文章标题为《大脑功能的默认模式》（Raichle 等，2001）。我们得出的结论是，在需要注意和目标导向的任务中观察到活动减少的大脑区域在静息状态下并未被激活，而是表明大脑内在或持续活动中存在一种之前未被认识的组织。顺便提一下，我们没有想到其他人会将表现出这种独特行为的区域星座称为大脑的默认模式网络。这个名称显然得到了广泛认可。

Research on the brain's default mode network and the brain's intrinsic activity more generally has moved in many directions producing a literature that has become quite extensive. In preparing this review, we examined this literature to look for general trends, which we summarize in Figure 1 for the interested reader. Rather than attempting a detailed analysis of this entire body of work, I focus on topics within it that have been of particular interest to me and that provide a sense of the future of this work.
关于大脑默认模式网络及大脑内在活动的研究已经朝多个方向发展，形成了相当庞大的文献。在准备这篇综述时，我们审视了这些文献，以寻找一般趋势，我们在图 1 中为感兴趣的读者进行了总结。与其对整个研究领域进行详细分析，我更关注其中一些特别引起我兴趣的话题，并提供对这一领域未来发展的感知。

Figure 1 图 1  The evolving literature on the default mode network (DMN). Since the publication of “A Default Mode of Brain Function” (Raichle et al. 2001), nearly 3,000 papers have been published on this topic. To identify the broad topics of inquiry in the DMN field, we conducted a search for papers citing Raichle et al. (2001). The search returned 2,988 papers with digital object identifiers in the Information Sciences Institute (ISI) database. ISI was then used to return the reference list for each of the 2,988 papers. These data were used to construct a 2,988 × 2,988 binary adjacency matrix, in which each element (i, j) indicated whether paper i cited paper j. Once constructed, the adjacency matrix was imported into Gephi, the open-source network visualization tool. To simplify the network graph, all nodes with degrees (number of citations from other papers in the network) less than or equal to 20 were removed. The number of network modules was then determined using a fast, modularity-optimization method included within Gephi. Each node was assigned to a specific module, and each module was assigned a unique color for visualization purposes. A topic was assigned to each module by identifying the focus of the highest cited papers within each module. Finally, the size of each node within the graph was weighted according to the node degree. This algorithm represents a data-driven approach to identify clusters of papers with common citation patterns. Each cluster represents a broad topic of investigation in the DMN field. The purple cluster contains task-driven papers typical of cognitive neuroscience. Papers in the blue cluster are broadly focused on the relationship between disease states and the DMN. Papers in the red cluster concentrate on functional connectivity processes. The green cluster is composed of papers on self-referential processing and mind wandering. Articles in the yellow cluster are related to neurophysiology and cell biology. At the center, in yellow, is Raichle et al. (2001).
关于默认模式网络（DMN）不断发展的文献。自从“脑功能的默认模式”（Raichle 等，2001 年）发表以来，已经有近 3000 篇论文在这一主题上发表。为了识别 DMN 领域的广泛研究主题，我们对引用 Raichle 等（2001 年）的论文进行了检索。检索结果返回了 2988 篇在信息科学研究所（ISI）数据库中具有数字对象标识符的论文。随后，使用 ISI 返回了这 2988 篇论文的参考文献列表。这些数据用于构建一个 2988 × 2988 的二元邻接矩阵，其中每个元素（i，j）表示论文是否引用了论文。一旦构建完成，邻接矩阵被导入到 Gephi 这一开源网络可视化工具中。为了简化网络图，所有度数（来自网络中其他论文的引用数量）小于或等于 20 的节点被移除。然后，使用 Gephi 中包含的快速模块优化方法确定网络模块的数量。每个节点被分配到一个特定的模块，每个模块被分配一个独特的颜色以便于可视化。 每个模块都分配了一个主题，通过识别每个模块内被引用次数最高的论文的焦点来实现。最后，图中每个节点的大小根据节点度数进行了加权。该算法代表了一种数据驱动的方法，用于识别具有共同引用模式的论文集群。每个集群代表了 DMN 领域内的一个广泛研究主题。紫色集群包含了典型的认知神经科学的任务驱动论文。蓝色集群的论文广泛关注疾病状态与 DMN 之间的关系。红色集群的论文集中于功能连接过程。绿色集群由关于自我参照处理和思维游离的论文组成。黄色集群的文章与神经生理学和细胞生物学相关。在中心位置，黄色部分是 Raichle 等人（2001）。

Figure 1 

Click to view 点击查看

Download as PowerPoint  下载为 PowerPoint

 

This article begins with a review of our initial work because of its centrality in establishing the legitimacy of the brain's default mode network.
本文首先回顾了我们的初步研究，因为它在确立大脑默认模式网络的合法性方面具有重要意义。

By the early 1980s, PET began to receive serious attention as a potential functional neuroimaging device in human subjects (Raichle 2009). The study of human cognition with neuroimaging was aided greatly by the involvement of cognitive psychologists in the 1980s. Their experimental strategies for dissecting human behaviors fit well with the emerging capabilities of functional brain imaging (Posner & Raichle 1994) in which one measures the time required to complete specific mental operations isolated by the careful selection of task and control states. This approach, in various forms, has dominated the cognitive neuroscience agenda ever since, with remarkably productive results (e.g., see Price 2012).
到 1980 年代初，正电子发射断层扫描（PET）开始受到严肃关注，作为一种潜在的人体功能性神经成像设备（Raichle 2009）。在 1980 年代，认知心理学家的参与极大地促进了神经成像对人类认知的研究。他们用于剖析人类行为的实验策略与功能性脑成像的新兴能力相得益彰（Posner & Raichle 1994），该技术通过精心选择任务和控制状态来测量完成特定心理操作所需的时间。这种方法以各种形式主导了认知神经科学的议程，并取得了显著的成果（例如，见 Price 2012）。

For the better part of the decade following the introduction of subtractive methodology to neuroimaging, the vast majority of changes reported in the literature were activity increases (or activations, to use the jargon of the cognitive neuroscience field). Activity increases but not decreases are expected in subtractions of a control condition from a task condition as long as the assumption of pure insertion is not violated. To illustrate, using an example based on mental chronometry, say that one's control task requires a key press to a simple stimulus, such as the appearance of a point of light in the visual field, whereas the task state requires a decision about the color of the light prior to the key press. Assuming pure insertion, the response latency difference between conditions is interpretable as the time needed to perform the color discrimination. However, the time needed to press a key might be affected by the nature of the decision process itself, violating the assumption of pure insertion. More generally, the brain state underlying any action could be altered by introducing an additional process.
在引入减法方法用于神经成像的十年间，大多数文献中报告的变化都是活动增加（或称为激活，使用认知神经科学领域的术语）。在从任务条件中减去控制条件时，只要不违反纯插入的假设，通常会预期到活动增加而非减少。举个例子，假设一个人的控制任务要求对简单刺激（例如视觉场中出现的光点）进行按键，而任务状态则要求在按键之前对光的颜色做出判断。在假设纯插入的情况下，条件之间的反应延迟差异可以解释为进行颜色辨别所需的时间。然而，按键所需的时间可能会受到决策过程本身性质的影响，从而违反纯插入的假设。更一般地说，任何行动背后的大脑状态都可能因引入额外的过程而发生改变。

Functional neuroimaging helped address the issue of pure insertion by employing the device of reverse subtraction. Thus, in certain circumstances, subtracting task state data from control state data revealed negative responses or task-specific deactivations (Gusnard & Raichle 2001). Investigators clearly showed, just as psychologists had suspected, that processes active in a control state could be modified when paired with a particular task. However, none of this work prepared us or anyone else for the experiment in which the control state was rest (i.e., simply lying quietly but awake in a scanner with eyes closed or visually fixating on a crosshair).
功能性神经成像通过采用逆向减法的方法帮助解决了纯插入的问题。因此，在某些情况下，从控制状态数据中减去任务状态数据揭示了负响应或任务特异性去激活（Gusnard & Raichle 2001）。研究人员清楚地表明，正如心理学家所怀疑的那样，控制状态下的活跃过程在与特定任务配对时可以被修改。然而，这些研究并没有为我们或其他人准备好控制状态为休息的实验（即，仅仅在扫描仪中安静地躺着但保持清醒，眼睛闭合或视觉集中在十字准星上）。

One of the guiding principles of cognitive psychology at that time was that a control state must explicitly contain all the elements of the associated task state other than the one element of interest (e.g., seeing a word versus reading the same word). Using a control state of rest would clearly seem to violate that principle. Despite our commitment to the strategies of cognitive psychology in our experiments, we routinely obtained resting-state scans in all our experiments, a habit largely carried over from experiments involving simple sensory stimuli (Fox et al. 1986) in which the control state was simply the absence of the stimulus (i.e., a resting state). At some point in our work, and I do not recall the motivation, I began to look at the resting-state scans minus the task scans. What immediately caught my attention was the fact that regardless of the task under investigation, the activity decreases almost always included the posterior cingulate and the adjacent precuneus.
当时认知心理学的一个指导原则是，控制状态必须明确包含与相关任务状态的所有元素，除了一个感兴趣的元素（例如，看到一个单词与阅读同一个单词）。使用静息状态作为控制状态显然似乎违反了这一原则。尽管我们在实验中坚持认知心理学的策略，但我们在所有实验中都常规地获取静息态扫描，这一习惯主要源于涉及简单感官刺激的实验（Fox 等，1986），在这些实验中，控制状态仅仅是刺激的缺失（即静息状态）。在我们工作的某个时刻，我开始查看静息态扫描与任务扫描的差异，虽然我不记得具体的动机。令我立即注意到的是，无论研究的任务是什么，活动的减少几乎总是包括后扣带皮层和相邻的前扣带皮层。

The first formal characterization of task-induced activity decreases from a resting state was a meta-analysis of 9 PET studies, involving 134 subjects, by my colleague Gordon Shulman (Shulman et al. 1997). This study generated an iconic image of a network of cortical areas that decreased in activity while performing various attention-demanding, largely non-self-referential tasks ( Figure 2 ). The unique identity of this network was confirmed a short time later by Jeffrey Binder and colleagues at the Medical College of Wisconsin (Binder et al. 1999<