静息态下自发思维的大脑表征形式
文献类型：学位论文
| 作者 | 李慧娴
|
| 答辩日期 | 2023-06 |
| 文献子类 | 博士 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 中国科学院心理研究所 |
| 其他责任者 | 严超赣 |
| 关键词 | 静息态功能磁共振成像 自发思维 出声思维 自然语言处理 表征相 似性分析 |
| 学位名称 | 理学博士 |
| 学位专业 | 认知神经科学 |
| 其他题名 | Brain representation of self-generated thought in the resting state |
| 中文摘要 | Resting-state functional magnetic resonance imaging (fMRI) is valuable in basic and clinical psychology due to its unique benefits. However, there is a critical question that current research on resting-state fMRI has overlooked: does the data obtained from "resting-state" fMRI only come from the "resting"/"empty" brain? Endogenously driven self-generated thought/spontaneous thought is a complex and heterogeneous phenomenon prevalent in people's daily lives, particularly during periods without external cognitive demands, such as during resting-state scans. Recent research has started to investigate the effects of ongoing mental activity on brain activity during resting-state scans, but progress has needed to be faster. While self-generated thought research provides insights, most has focused on task contexts, with limited research on resting states. The classical empirical sampling (ES) approach has made significant strides in self-generated thought research. However, existing research paradigms restrict it, and the direct representation of self-generated thought in brain activity remains unclear. Moreover, the content of self-generated thought is diverse, closely linked to its complex effects, and strongly associated with mental disorders. However, methodological constraints limit the exploration of self-generated thought content characteristics to a few single dimensions, requiring new research methods to expand exploration. Therefore, this paper focused on these two interrelated issues and delved deeply into the characteristics and neural representations of self-generated thought in a resting state, which refers to a state of mind with no obvious external tasks or stimulation. The paper focused on addressing limitations in methods used to investigate self-generated thought, and first explored the feasibility and effectiveness of using think-aloud as a more direct measure of resting-state self-generated thought. After establishing the feasibility of this method, the study integrated natural language processing (NLP) to quantitatively analyze the content of self-generated thought, calculating indicators of thought content divergence and expressions of sadness. Furthermore, the study investigated the relationship between these content features and individual traits, establishing that rumination is a sticky and negative form of self-generated thought. Overall, the study clarified the potential value of using the think-aloud method to investigate resting-state self-generated thought, as this method not only enables real-time data collection but also provides rich information about the content of self-generated thought, thereby facilitating research into the features of self-generated thought. Notably, researchers can achieve a more objective and quantitative analysis of thought content by combining the think-aloud method with NLP. To address the issues of neglecting sustained mental activity during individual resting-state fMRI scans and unclear neural representation of self-generated thought, Study 2 employed the think-aloud method to resting-state fMRI scans (Think-Aloud fMRI) to clarify the impact of sustained thought flow on brain activity during resting一state fMRI scans and further defined the neural representation of self-generated thought. This study first explored the brain activation patterns under the Think-Aloud condition. Then, the study examined the relationship between the divergence of thought content and brain activation and found correlations across multiple brain regions. Finally, the study adopted an innovative approach that combines NLP with representation similarity analysis (RSA) and identified the neural representations of self-generated thoughts during rest at three different scales: voxel-level searchlight analysis, region-level analysis using the Schaefer 400-parcel, and systemic level of the Yeo seven networks. The results showed that self-generated thoughts during resting-state were associated with a broad range of brain regions spanning all seven Yeo networks. The proportion of significant brain regions, from largest to smallest, is the default mode network, frontoparietal control network, visual network, ventral attention network, somatomotor network, dorsal attention network, and limbic network. This study emphasizes the importance of considering sustained cognitive activity in resting一state fMRI and clarifies the spatial patterns of sustained thought flow during resting-state fMRI scans, providing support for spatiotemporal neuroscience and preliminary methodological support for Think-Aloud fMRI. To further clarify the behavioral manifestations and their relationship with the brain activity of self-generated thought during the resting state with different content characteristics, based on the rich thinking content information provided by the think-aloud method, Study 3 explored the multidimensional content characteristics of self-generated thought during the resting state within and between groups from a behavioral perspective, furthermore, Study 4 used fMRI to investigate the corresponding brain activity patterns associated with different content characteristics of self-generated thought during resting-state, as described in detail below. Study 3 conducted a multidimensional evaluation of the thought content features of the thought events collected through the think-aloud method. This evaluation included dimensions such as perceptual orientation (interoceptive, internal, and external) and social dimension (self, other). Besides, we evaluated the emotional experiences associated with self-generated thought using eight negative and four positive emotion dimensions. Based on this multidimensional analysis of thought content features, this study explored the thought content features of healthy individuals, healthy men and women, and individuals with depression during the resting state. Meanwhile, this study compared various thought content features between healthy men and women, and between those with depression and the healthy controls. The study found that self-generated thought during the resting state in individuals with depression exhibited unique characteristics, which differed significantly from those of healthy controls in several thought content dimensions. Specifically, compared to healthy controls, self-generated thought in individuals with depression had less external focus, was more self-focused, less future-oriented, less attentive to positive events, more attentive to negative events, and exhibited lower levels of reflective thinking and higher levels of ruminative thinking. In terms of emotional experiences accompanying self-generated thought during the resting state, depression was associated with significantly higher levels of negative emotions, as evidenced by significantly higher levels of all eight negative emotions, and significantly lower levels of all four positive emotions, compared to healthy controls. Additionally, the study found significant differences in self-generated thought during the resting state between healthy men and women. Moreover, women exhibited similarities with individuals with depression in some thought content dimensions. Specifically, women exhibited higher levels of self-focus, greater attention to negative events, and higher levels of several negative emotions compared to healthy men. Finally, this study utilized NLP to develop regression models for multiple content assessments and a classifier for distinguishing between depression and healthy individuals. The findings validated the value and potential of the think-aloud method for clinical applications in MDD. Based on Think-Aloud fMRI data and behavioral data of thought content characteristics, Study 4 used two analytical approaches to reflect the relationship between self-generated thought content characteristics and brain activity from different aspects: (1) the comparison analysis of each thought content characteristic and its baseline; (2) the correlation analysis between changes in thought content characteristics and changes in brain activity. Both analyses indicated a close connection between the content characteristics of self-generated thought and brain activity, and different thought content characteristics corresponded to different brain representations. In terms of the emotional experiences accompanying self-generated thought during the resting state, the study found that the overlapping and differential brain activity corresponding to different emotional experiences were detectable by the BOLD signal, and the neural representation of internally driven thought and externally induced emotions showed consistency. This study further emphasizes the significant impact of ongoing psychological activities during the resting state on brain activity and highlights the importance of studying self-generated thought content characteristics in more detail from a neural perspective, providing insights into the neural representation of self-generated thought content characteristics. In summary, this paper focuses on the self-generated thought flow during resting一state and strives to advance research in this field, hoping to provide insights for resting-state fMRI research and self-generated thought domain. Given the limitations of current methods in the field, we first developed a method to directly measure self-generated thought during the resting state (Study 1). Subsequently, this method was applied in fMRI with NLP to clarify the representation of resting一state thought flow in large-scale brain networks using RSA (Study 2). Then, we systematically explored the behavioral performance of multidimensional content characteristics of self-generated thought within and between groups (Study 3) and investigated the corresponding brain activity patterns of different content characteristics of self-generated thought (Study 4). In short, this paper emphasizes the importance of investigating sustained thought flow during the resting state from multiple perspectives, revealing some content characteristics of self-generated thought during the resting state and obtaining a spatial map of brain activity corresponding to self-generated thought during the resting state. The theoretical and methodological support provided by this paper will help clarify the spatial structural characteristics of spontaneous brain activity and advance research in resting一state fMRI and self-generated thought domains. |
| 英文摘要 | 静息态功能磁共振成像技术因其独特的优势被广泛应用于基础心理学和临床心理学研究，但是目前针对静息态功能磁共振成像的研究一直都忽视了一个关键问题:所谓的“静息态”是完全来自于“静止”或者“空白”的大脑吗?内源性驱动产生的自发思维是复杂而异质的，并具有不确定性，它普遍存在于个体的日常生活中，特别是当个体处在没有任何外部认知负荷的休息状态，如静息态扫描过程中个体所处的状态。近年来研究者开始关注静息态扫描中持续的认知活动对大脑活动的影响，但研究进展较少。虽然我们可以从自发思维的研究领域获得启示，但是自发思维的研究主要关注于任务背景，基于休息状态/静息态的研究较少。经典的经验取样法虽然在自发思维研究领域已取得了很大的进展，但受限于现有研究范式，自发思维的直接大脑活动表征仍不清楚。其次，自发思维的内容具有多样性，与其复杂的影响密切相连，并与精神疾病具有紧密的联系。但由于方法的限制，对自发思维内容特征的研究仍局限于几个单一维度，亚需新的研究方法来扩展对自发思维内容特征的探究。因此，本论文针对这两个紧密联系的问题，聚焦于静息态(休息状态，无明显的外部驱动任务)下的自发思维，逐步深入地探究了它的特征及其大脑表征形式。 针对自发思维领域的方法限制问题，研究一首先探究了直接测量静息态自发思维更有效的方法一一出声思维法的可行性与有效性。在明确方法的可行性之后，研究融合自然语言处理对思维内容进行了量化分析，计算了思维内容差异性和悲伤情绪表达指标，并进一步探究了这两个内容特征指标与个体特质之间的关系，在明确了量化指标行为意义的同时证明了反当是一种粘性、负性倾向的自发思维。总之，研究一明确了出声思维法对静息态自发思维研究的潜在价值，该方法在实现实时思维收集之外，提供了丰富的思维内容信息，可以促进研究者对自发思维内容特征的研究。特别的是，该方法结合自然语言处理可以使我们对思维内容进行更客观的量化分析。 针对静息态功能磁共振成像领域对个体静息态扫描过程中持续心理活动的忽视问题与自发思维神经表征不明确的问题，研究二将出声思维法应用于静息态功能磁共振扫描中一一出声思维功能磁共振成像(Think-Aloud fMRI，在明确静息态功能磁共振扫描过程中个体持续思维流对大脑活动存在影响之外，更进一步明确了静息态自发思维对应的大脑表征形式。该研究首先探究了出声思维下的大脑激活模式。之后，研究计算了思维内容差异性指标和出声思维下的大脑活动之间的关系，发现思维内容差异性指标与多个大脑区域具有相关。最后，研究采用创新的自然语言处理与表征相似性分析相结合的分析方法，在全脑三个神经尺度上明确了静息态自发思维的神经表征:体素水平的全脑探照灯分析，基于Schaefer 400分区的区域水平全脑分析，以及基于Yeo七网络的系统水平分析。研究发现，静息态自发思维与广泛的大脑区域相关，涉及七个Yeo网络，显著脑区占比从大到小依次为默认网络、额顶控制网络、视觉网络、腹侧注意网络、感觉运动网络、背侧注意网络和边缘网络。该研究强调了在静息态功能磁共振成像中考虑持续进行的认知活动的重要性，明确了静息态持续思维流对应的大脑空间模式，为时空神经科学提供了支持，并为出声思维功能磁共振成像提供了初步的方法学支持。 为更进一步明确静息态下自发思维不同内容特征的行为表现及其与大脑活动之间的关系，基于出声思维法提供的丰富思维内容信息，研究三从行为学上探讨了静息态自发思维多维度内容特征在群体内的表现和群体间的差异，研究四利用功能磁共振成像技术探究了静息态自发思维不同内容特征对应的大脑活动模式，具体如下所述。 研究三对出声思维法收集到的思维事件进行了多维度思维内容特征评估，例如思维知觉导向维度(内感受、内部导向、外部导向)、社会维度(自我、他人)等，并对自发思维伴随的情绪体验进行了多维度评估，共包含8种负性情绪和4种正性情绪。基于多维度思维内容特征评估，该研究探究了健康人、健康男性、健康女性和抑郁症在静息态下各自的思维内容特征(维度内比较)，并比较了健康男女之间、抑郁症与健康对照之间多个思维内容特征上的差异。C1)研究发现，抑郁症静息态自发思维具有独特的特征，并在多个思维内容特征上与健康人具有显著差异，具体表现为:与健康对照相比，抑郁症静息态自发思维更少由外部导向、关注自我更多、关注未来事件更少;对积极事件的关注更少，而对消极事件的关注更多;思维反省的程度更低，而思维反当的程度更高。在静息态思维伴随的情绪体验方面，抑郁症具有明显的负性情绪色彩，具体表现在他们对8种负性情绪的体验程度均显著高于健康对照，而对4种正性情绪的体验程度均显著低于健康对照。C2)此外，研究发现健康男性和健康女性的静息态自发思维也具有一些显著差异，并且女性在某些思维内容条目上与抑郁症更相似，具体表现为:与健康男性相比，健康女性静息态自发思维对自我的关注更多、对消极事件的关注更多、对多个负性情绪的体验程度更高。最后，该研究结合自然语言处理训练了多个内容评估的预测模型和抑郁症与健康人的分类预测模型，并验证了模型的有效性，证明了出声思维法在抑郁症临床应用上的价值和潜力。 研究四基于出声思维功能磁共振成像数据和思维内容特征评估行为数据，采用两种分析方式从不同侧面反映了静息态自发思维内容特征与大脑活动之间的关系，具体为:C1)每个思维内容特征与其基线的比较分析;C2)每个思维内容特征变化与大脑活动变化之间的对应关系分析。两种分析均表明，静息态自发思维的内容特征与大脑活动具有紧密的联系，并且不同思维内容特征对应的大脑表征是不同的。在静息态自发思维伴随的情绪体验方面，研究发现静息态自发思维伴随的不同情绪体验对应的大脑活动之间具有重叠和差异性;内部驱动思维伴随的情绪体验与外部诱发情绪一样宽广，均可以通过B OLD信号进行检测，并且二者的神经表征具有一致性。该研究进一步强调了静息态持续进行的心理活动对大脑活动存在显著的影响，并再次从神经层面强调了对自发思维内容特征进行更细致研究的重要性，也为自发思维内容特征的神经表征研究提供了启示。 综上所述，本论文聚焦于静息态下的自发思维流，致力于开拓该领域的研究，希望为静息态功能磁共振成像研究、自发思维领域研究提供启示。针对目前领域方法的局限性，研究首先开发了直接测量静息态自发思维的方法(研究一)，随后将该方法应用于功能磁共振成像中，融合自然语言处理，采用表征相似性分析明确了静息态思维流在大尺度脑网络上的表征形式(研究二)。之后，我们逐步深入、系统地探究了静息态自发思维多维度内容特征的群体内和群体间的行为学表现(研究三)，探索了静息态自发思维不同内容特征对应的大脑活动模式(研究四)。总之，本论文从多个方面强调了探究静息态持续思维流的重要性，并揭示了静息态下自发思维的一部分内容特征，获得了静息态自发思维对应的大脑空间图谱，为明确自发脑活动的空间结构特征和脑自发活动的研究提供了理论和方法支持，同时将促进静息态功能磁共振成像研究、自发思维领域研究的发展。 |
| 语种 | 中文 |
| 源URL | [http://ir.psych.ac.cn/handle/311026/46176]  |
| 专题 | 心理研究所_认知与发展心理学研究室 |
| 推荐引用方式 GB/T 7714 | 李慧娴. 静息态下自发思维的大脑表征形式[D]. 中国科学院心理研究所. 中国科学院大学. 2023. |
入库方式： OAI收割
来源：心理研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。