生物运动多维度特征加工的脑机制
文献类型：学位论文
| 作者 | 王睿迪 |
| 答辩日期 | 2024-06 |
| 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 | 蒋毅 |
| 关键词 | 生物运动 多维度特征加工 功能磁共振成像 脑磁图 表征相似性 |
| 学位名称 | 理学博士 |
| 学位专业 | 认知神经科学 |
| 其他题名 | The brain mechanisms of multi-dimensional biological motion attributes processing |
| 中文摘要 | The human brain can rapidly perceive others' actions in complex environments and extract features such as walking direction, gender, emotion, and identity. These processes are commonly referred to as biological motion processing. The ability to process biological motion is crucial for human survival and adaptation, it has received extensive attention from researchers. Currently, there has been a broad exploration of the specific neural mechanisms underlying the processing of biological motion information compared to non-biological motion information. However, there has been less focus on the neural mechanisms involved in processing the various feature information contained in biological motion. This study systematically investigates the brain mechanisms for the multi-dimensional biological motion perception, including the interaction between multiple attributes and the attributes’ function. Firstly, attributes contained in biological motion span different hierarchical levels, such as walking direction belonging to physical attributes, gender belonging to biological attributes, and emotion belonging to social attributes. It remains unknown how the brain extracts and represents these attributes from multi-dimensional biological motion. To address this question, functional magnetic resonance imaging (fMRI) technology was used to study the brain representations of participants when viewing multi-dimensional biological motion stimuli. Using multiple regression representation similarity analysis, distributed brain networks representing the walking direction, gender, and emotion features of biological motion were identified. Results showed that these three sets of brain networks conform to a hierarchical structure of mutual influence between emotion, gender, and walking direction, and also exhibit mutual regulatory relationships. Multi-voxel pattern analysis (MVPA) further demonstrated that specific brain areas within these networks are specialized for encoding corresponding attributes. Additionally, analysis of the similarity between neural activity and behavioral results indicated that some brain regions show a correlation with subjective ratings from participants, revealing the brain's neural circuits involved in encoding biological motion attributes under interactions among different attributes. Building on the first study, the second study delved into the temporal processes of representing multi-dimensional biological motion attributes. Using magnetoencephalography (MEG), this study examined the processing of walking direction, gender, emotion, and identity attributes in biological motion. The results from the multiple regression representation similarity analysis showed that the brain first represents emotional features, followed by gender and walking direction, and lastly identity features. Furthermore, results showed that there is the interplay between different features, specifically, emotion features influence gender and orientation processing, gender features influence orientation processing but not emotion processing, and walking direction features do not affect the processing of other features. Additionally, combining the findings from studies One and Two, this study proposed a hierarchical processing model for the features of biological motion. Finally, Study Three investigated the neural mechanisms of biological motion attribute’s function, primarily focusing on the direction. Using fMRI and an improved central cue paradigm, the study explored the neural mechanisms underlying attentional shifts induced by walking direction in biological motion. Incorporating different types of social cues (i.e., eye gaze) and non-social cues (arrows), the study also investigated the specific neural mechanisms involved in social attention elicited by biological motion and eye gaze. MVPA revealed that the neural mechanisms of biological motion mediated attention effect include the posterior superior temporal sulcus (pSTS) and widespread dorsal and ventral attention networks. Although attentional orientation induced by biological motion can be decoded from extensive brain activity, only the right anterior and pSTS showed specific decoding for attentional orientation induced by social cues (biological motion and eye gaze). Cross-category MVPA further demonstrated that the right STS and right STG could decode attentional orientation induced by both biological motion and eye gaze cues, but not by arrow cues. These findings collectively confirm the presence of a dedicated social attention processing network in the human brain, with the right temporal regions being key neural mechanisms specifically involved in processing social attention. In conclusion, this study explores the temporal and spatial-specific neural mechanisms underlying the representation of multi-dimensional attributes of biological motion, as well as the neural mechanisms involved in attentional orientation induced by walking direction. The study expands our understanding of the mechanisms underlying the processing of biological motion perception, deepening our insights into the neural mechanisms behind attention induced by biological motion and eye gaze. This research provides avenues for optimizing computational models for biological motion feature processing and artificial neural network models for gait recognition, as well as insights into the specific neural mechanisms underlying social attention. |
| 英文摘要 | 人脑可以从复杂环境中快速感知他人的动作，并在此基础上提取动作包含的朝向、性别、情绪和身份等特征信息，这些加工过程通常被称为生物运动加工。生物运动加工关系到人类的生存和适应，因此得到研究者广泛关注。目前对于生 物运动信息相比于非生物运动信息加工的特殊机制已经得到较为广泛的探究，然而，对于生物运动所包含的各种特征加工的神经机制还少有研究者关注。本文从生物运动多维度特征提取角度，具体为特征加工间的相互影响以及特征的功能两个角度着手，系统探究人脑对生物运动特征的加工机制。 首先，生物运动包含的特征具有不同层级，如朝向信息属于物理属性，性别 信息属于生物属性，情绪信息属于社会属性。大脑如何从同时具备多维度特征的 生物运动中提取这些属性并进行表征仍然未知。为了回答上述问题，研究一使 功能性磁共振成像技术来研究参与者在观看多维生物运动刺激时的脑表征。通过 多元回归的表征相似性分析方法，发现了表征生物运动的朝向、性别和情绪特征 的分布式脑网络。结果显示上述三组脑网络一方面符合自情绪、经性别、至朝向 的相互影响的层级结构，另一方面三者存在相互调节关系。多体素模式分析进 步揭示了上述三组脑网络中的部分脑区专门用于编码对应属性。此外，与行为结 果的表征相似性分析还揭示了部分脑区神经活动存在与被试主观评分结果的关 联。这些结果揭示了在特征间相互作用的情况下，生物运动属性编码的大脑网络，并强调了大脑对于多维生物运动属性的处理是循环交互的。 在研究一的基础上，研究二进一步探究了生物运动多维特征在大脑中表征的时间进程。该研究采用脑磁图技术，考察生物运动朝向、性别、情绪和身份特征的加工。多元回归表征相似性分析显示，大脑首先对于情绪特征进行表征，其次是性别和朝向，最后是身份特征。并且不同特征之间的加工存在相互影响，情绪特征影响性别和朝向加工，性别特征影响朝向加工但不影响情绪加工，朝向特征不影响其他特征的加工。 最后，研究三探究了生物运动特征的社会功能，主要考察朝向特征的功能。采用功能磁共振成像和改进的中央线索范式，本研究探究了生物运动朝向诱发注 意转移的神经机制。此外，结合不同类型的社会性线索（即眼睛注视）以及非社 会性线索（箭头），本文探究了生物运动和眼睛注视共同诱发社会性注意的特异 性神经机制。多体素模式分析发现，生物运动朝向诱发注意的神经机制包括后部颞上沟和广泛的背侧和腹侧注意网络。尽管生物运动诱发的注意定向可以从广泛 的脑区神经活动中解码出来，但只有右侧前、后颞上沟可以特异地解码由社会性线索（生物运动和眼睛注视）诱发的注意定向。跨类别的多体素模式分析进一步显示，在右侧颞上沟和右侧颞上回脑区，可以跨生物运动和眼睛注视线索进行相互解码，但不可以跨箭头进行解码。这些发现共同证明了人类大脑存在专门的社会性注意加工网络，而右侧颞上沟、颞上回是专门用于加工社会性注意的关键神经机制。 综上所述，本研究从生物运动多维特征表征的角度出发，探究了不同特征的时空特异性神经机制和朝向特征诱发注意的神经机制，扩展了生物运动感知加工的机制模型，加深了对于生物运动诱发注意以及社会性注意背后的神经机制的理 解，为未来构建生物运动特征加工的计算模型和步态识别的人工神经网络模型提 供了思路，为社会性注意的特异性神经机制的探究方向提供了启示。 |
| 语种 | 中文 |
| 源URL | [http://ir.psych.ac.cn/handle/311026/48001]  |
| 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 | 王睿迪. 生物运动多维度特征加工的脑机制[D]. 中国科学院心理研究所. 中国科学院大学. 2024. |
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。