认知与发展心理学研究室知识库

人类作为社会性动物，准确感知其他个体的社会性信息以便于采取合适互动 方式，对于人类的生存和生活具有重要意义。性别作为一种社会性信息，对性别信息的知觉，是实现社会互动的重要前提，更是人类生存繁衍的关键。人类也确实表现出了在性别信息识别中的非凡能力。已有大量研究采用面孔刺激，证实人们能够迅速、轻松、准确地识别出性别信息。而在真实视觉世界中，人们能够从多种不同刺激中提取出性别信息，例如面孔、身体、客体。当前关于性别的研究， 尤其是性别加工的计算模型和神经机制的研究，多集中于面孔刺激。各类承载性 别信息的视觉刺激对于性别加工的全面理解都有重要意义，除面孔刺激外，其他 类别刺激的性别加工机制也亟待进一步探究。性别作为一种高级特征，能够从不同刺激中提取出来，这是否支持存在一种类别泛化的性别表征？换言之，性别信息表征是基于类别特异的机制还是具有跨类别共享的机制？目前，仅能根据有限的行为证据推断可能存在跨类别共享的性别表征，但尚未有明确证据支持这种共享的性别表征存在神经基础。本研究将从行为、计算模型和神经层面探究多种类别刺激性别信息表征的认知神经机制以及是否存在跨类别性别表征共享的机制。

本文共包含四个研究。研究一、二关注面孔及生物运动刺激的性别表征，旨在利用适应范式和同时对比范式分别考察时间和空间维度上的性别情境效应，从行为和计算模型层面探究同类别与跨类别性别信息表征的机制。研究一采用适用 范式，发现生物运动性别适应后效随着适应刺激真实感的降低（从真人的运动逐 渐变为假人的运动），呈现出先增大后维持不变的趋势。这一结果符合基于对抗 模型的计算编码机制的预测，与面孔性别信息加工基于相似的计算机制。但适应一种类别测试另一种类别时，却均未观察到跨类别的性别适应后效，暂不能为这 两类刺激的性别表征存在共享神经基础提供来自时间情境效应的证据。研究二采 用同时对比范式，发现对中央测试面孔或生物运动性别知觉会偏向与外周刺激性 别相反的方向，首次揭示了面孔与生物运动性别知觉存在同时对比效应。计算建模的结果也再次证实了面孔和生物运动性别表征均基于对抗模型的计算编码机 制；但仍未能观察到跨类别的同时对比效应，同样暂不能为面孔和生物运动性别表征存在共享神经基础提供来自空间情境效应的证据。小结研究一和研究二，两者均观测到面孔与生物运动性别表征存在情境效应，且基于相同的计算编码机制，但无论在时间及空间维度，彼此的性别情境效应均很难发生迁移，虽表明面孔与生物运动这两类刺激的性别表征以相同的计算机制工作，但暂未提供证据支持存 在性别表征的共享神经基础。

研究三、四聚焦探究面孔、身体、客体三类静态刺激类别内与跨类别性别信息表征的神经机制。研究三借助脑电技术，采用快速视觉周期呈现范式，利用基 础刺激频率和新异刺激频率分别标记类别内和跨类别性别信息，通过频谱分析 机器学习解码技术探究面孔与身体、面孔与客体的跨类别信息表征是否具有共享 的神经基础。频谱分析的结果显示，枕颞叶等区域同时编码了面孔和身体的性别 信息，两者可能具有共享的神经基础；类别内性别信息解码的结果显示，不同性 别信息引发的脑电响应模式存在显著差异，进一步利用全脑的脑电响应模式进行 跨类别性别信息解码，结果发现身体和客体的性别表征可以互相解码，提示二者 也可能具有共享的神经基础。研究四借助功能磁共振成像技术，记录了被试在观 看男性和女性的面孔、身体及客体图片时的血液动力学反应，探究类别内与跨类别性别加工涉及的具体脑区以及这些脑区间的功能连接。多变量模式分析结果显 示面孔、身体和客体类别内的性别信息表征主要位于枕颞叶，而右侧颞中回能够 跨类别解码面孔和身体、身体和客体、面孔和客体的性别信息，验证了研究三的 结论，表明存在跨类别性别表征的共享机制。心理生理交互分析的结果则针对能 够解码类别内性别信息的脑区，揭示了性别加工过程涉及的这些脑区间的功能连 接变化：身体性别知觉显著调节了距状皮层、楔叶、舌回和颞中回、颞下回、梭状回之间的功能连接，客体性别知觉显著调节了梭状回和距状皮层、舌回之间的 功能连接，而面孔性别知觉没有显著改变这些脑区之间的功能连接。但是，面孔 和身体性别加工涉及的脑区间功能连接模式表现出显著相关，表明了这两类性别 表征的脑网络具有较高的相似性。小结研究三和研究四，两者互相印证，共同证明了面孔、身体和客体性别表征存在共享的脑机制。 综上所述，本文从行为、计算模型和神经三个层面系统探究了人类性别信息表征的机制，丰富并深化了对于多种类别刺激的性别表征的认识，揭示了面孔、 身体、客体性别表征共享的神经机制，为全面理解人脑中的性别信息表征做出重 要的贡献。

As social beings, humans rely on accurately perceiving social cues from others to appropriately engage in interactions, which is crucial for their survival and life. Gender is a form of social information, and the perception of gender plays a vital role in social interactions and is essential for human reproduction. It has demonstrated that humans possess exceptional ability to perceive gender information. A large number of studies reveal that people can quickly, easily and accurately identify the gender of faces. In the real visual world, gender cues can also be extracted from various stimuli, such as faces, bodies, and objects. However, studies on gender perception, especially about the computational model and neural mechanisms, mostly focuses on face. All various of visual stimuli with gender information are of great significance to the comprehensive understanding of gender representation. Thus, besides face, it is essential to further explore gender perception from other categorical stimuli. Gender, as a high-level feature, can be extracted from different stimuli. However, does that mean, gender representation is category-generalized? In other words, is cross-category gender information representation based on category-specific mechanisms or cross-category shared neural mechanisms? At present, there are limited behavioral evidences that support the existence of shared gender representation across categories, but no clear evidence revealing the neural basis of such shared gender representation. This study, through behavioral, computational and neural levels, investigated the cognitive and neural mechanism underlying the gender representation in various stimuli, and explored the existence of a shared mechanism for gender representation across categories.

The research comprises four studies. The first two studies, focusing on face and biological motion, aimed to adopting the adaptation and simultaneous contrast paradigm to investigate the gender contextual effects in the temporal and spatial dimensions, which can explore the representation of gender information within and across categories at the behavioral and computational level. Study 1 employed the adaptation paradigm and observed that the gender adaptation aftereffect of biological motion increased first and then remained unchanged, when the perceived realism of adaptation stimulus decreases (the biological motion changing from real to unreal). This result was consistent with the prediction of computational encoding mechanism based on opponent model, similar to the computational mechanism of face gender representation. However, when one category was adapted and another category was tested, no significant cross-category gender adaptation effect was observed, which cannot provide supportive evidence for the shared neural basis of gender representation across these two categories from the temporal contextual effect. Study 2 adopted the simultaneous contrast paradigm and found that the gender perception of a central face or a point-light walker was repelled away from the gender of its surrounding faces or walkers, revealing simultaneous contrast effects of gender perception in both face and biological motion. Moreover, the gender representation of faces and biological motion is also excellently accounted by the opponent computational model. However, no significant cross-category simultaneous contrast effect was observed, which cannot provide supportive evidence for the shared neural basis of gender representation between faces and biological motion from the spatial contextual effect. In summary, Study 1 and Study 2 observed contextual effects in gender perception of faces and biological motion, which were based on similar computational mechanism, but neither the temporal nor the spatial contextual effect of gender can transfer between faces and biological motion. Although these findings indicate that the gender representation of faces and biological motion rely on similar computational implementation, there is currently no evidence to support the shared neural basis for gender representation.

Study 3 and 4 investigated the neural mechanism of the representation of gender information within and across categories, focusing on three static stimuli including face, body and object. In Study 3 the fast visual period stimulation paradigm with electroencephalogram was adopted to label intra-category gender by basic stimulus frequency and cross-category gender by odd stimulus frequency, and spectrum analysis and machine learning technology were used to explore the shared neural basis of gender information representation across faces and bodies, faces and objects. The results of spectrum analysis showed shared neural basis for gender representation of faces and bodies, mainly including the occipital and temporal region; the results of withincategory decoding showed that male and female stimuli of each category elicited significant different patterns of neural response. Further decoding the gender across categories using the whole brain response, the result showed that the gender can be decoded across bodies and objects, suggesting a shared neural basis for gender representation of these two categories. In Study 4, we recorded hemodynamic response in the brain with functional magnetic resonance imaging when participants viewed faces, bodies, and objects varying in gender, to further explored the specific brain regions and functional connectivity involved in gender processing. The multivariate pattern analysis showed that the gender information of faces, bodies, and objects could be decoded mainly in the occipitotemporal lobe, and gender could be decoded across categories in the right middle temporal gyrus, indicating a shared representation of gender across categories. The psychophysiological interaction analysis revealed the functional connectivity modulated by gender processing between brain regions that can decode with-category gender information: the gender perception from body significantly modulates the functional connectivity between the calcarine, cuneus, lingual gyrus and middle temporal gyrus, inferior temporal gyrus, fusiform gyrus, while the gender perception from object significantly modulates the functional connectivity between the fusiform gyrus and calcarine, lingual gyrus. By contrast, no functional connectivity has been significantly modulated when the face gender is processed. Moreover, the functional connectivity patterns for the face and body gender perception are significantly correlated, suggesting the similarity between the neural networks underlying gender processing of these two categories.

In summary, Study 3 and Study 4 confirmed each other and jointly substantiated a shared brain mechanism of gender representation of faces, bodies and objects. Moreover, the functional connectivity patterns for the face and body gender perception are significantly correlated, suggesting the similarity between the neural networks underlying gender processing of these two categories. In summary, Study 3 and Study 4 confirmed each other and jointly substantiated a shared brain mechanism of gender representation of faces, bodies and objects.