|其他题名||Localization and Characterization of an Occipital Word Sensitive Area ----- A new cerebral region for visual word processing|
|关键词||文字识别 枕叶 物体类别加工 正字法 功能磁共振|
|摘要|| 文字是人类文明的最伟大发明之一，快速、准确地识别文字是有效阅读的基础。因此，对文字识别及其神经机制的研究具有重要的学术意义和应用潜力。在人类腹侧枕颞皮层，对于特定物体类别的视觉加工通常存在有两个脑区，一个位于枕叶区域，另一个位于更靠前方的枕颞交界处，例如面孔识别系统的枕叶面孔区 （OFA）和梭状回面孔区 （FFA）,位置识别系统的枕叶位置区（OPA）和旁海马回位置区 （PPA）。但是对文字加工系统的研究却局限在单个区域（左 侧枕颞沟的VWFA），文字识别理应也遵循物体模式识别的层级加工理论模型，即有一个由简单特征到复杂特征、由低级脑区往高级脑区的逐级累加的加工过程，这就提示在枕叶也可能存在文字加工区域。本论文旨在检验这个假设，确定该区域精细位置并研究其功能特性。|
研究一中使用功能定位(汉字图片减去控制刺激图片在大脑中引起的激活)、解剖标志（大脑的沟回位置）、邻近的其它功能区和视网膜皮质映射图（视野中不同方位的刺激对应着枕叶视皮层的分区）等信息来确定枕叶文字敏感区 （我们命其名为OWA）的详细位置。在13名参加实验的中文为母语的被试上皆确定出了大脑双侧OWA,主要位于枕下回或枕中回，其平均Talairach坐标是左半球的[ 41，76，6 ]和右半球的[ 39，73，6 ]，OWA 和视觉词形区（VWFA）之间的空间距离约为22.9毫米。我们还进一步对其中5名被试进行了更详细的实验和分析，结果显示：OWA 相对早期视觉皮层拓扑区要更靠前，相对外侧物体区（LOC）更靠腹侧，相对VWFA 更靠后，相对于视觉运动区（hMT+）更靠下方。
研究二在研究一的基础上，考察10名被试OWA 对汉字和其他三类经典视觉刺激 (昆虫、椅子、车辆) 的反应特性，检验该区域对文字刺激的选择性，并与其它感兴趣区（VWFA、LOC 和左侧角回）作比较。单变量分析的结果显示：在平均反应强度上，OWA 与VWFA 一样，对汉字的反应强于其他三类刺激；左侧角回也对汉字的反应最强；而LOC 对椅子的反应最强。在对汉字的选择性指标上，左侧OWA 与左侧VWFA 以及左侧角回的差异不显著。多变量模式分析的结果则显示：左侧OWA 对于四类刺激的表征结构类似于VWFA，与LOC 不同；而在将汉字与其他刺激区分的能力上，左侧OWA 显著低于左侧VWFA。上述结果说明OWA 与VWFA 相似，对文字刺激有着类似的反应特性，但OWA 对文字的特异化程度不及VWFA，而且OWA 在对文字反应的选择性上具有一定的左侧化趋势。
研究三采用16类视觉刺激（汉字、英语词语、数字、音符、数学符号、像汉字的东巴文、像图形的东巴文、简笔小人、人脸、动物、房屋、树木、水果、抽象雕塑、家具、交通工具），进一步考察该区域对文字的选择性加工程度，并与VWFA 作比较。结果显示：以平均反应强度排序，左侧OWA 中排在前五强的刺激类别依次是英语词、汉字、动物、人脸、简笔小人；左侧VWFA 中排在前五强的刺激类别依次是英语词、数学符号、汉字、动物、数字。左侧VWFA 对英语、数字的反应显著高于所有其他类别，但是左侧OWA 对于其前五强刺激的反应差异不显著。这些结果说明，与VWFA 相比，OWA 对文字选择性相对较弱，可能它加工的是符号系统中更低层次的视觉特征。
研究四利用汉字刺激的视觉特性，构造出汉字真字（符合正字法且有读音），伪字（符合正字法但不可读），非字（不符合正字法不可读）三类刺激来考察OWA 对汉字正字法信息是否敏感，并与VWFA 和LOC 作对比。结果显示，在左侧OWA，非字的反应显著强于真字和伪字，而真字与伪字没有差异；左侧VWFA 和左侧OWA 对这三种刺激的反应趋势几乎相同；而左侧LOC 对物体图反应最强，对非字和伪字的反应差异不显著。这些结果说明，OWA 与VWFA 一样，都对正字法信息敏感。
|其他摘要||Writing word is one of the greatest inventions of human civilization and human beings acquire the rapid word recognition ability through reading. Therefore, it is of great academic value and clinical significance to investigate the neural mechanisms of word processing. In the human ventral occipitotemporal cortex, category-specific processing for visual objects typically involve pairs of cortical regions, often with one located in the occipital cortex and another more anteriorly in occipitotemporal junction, such as the Occipital Face Area (OFA) and the Fusiform Face Area (FFA) for face processing, the Occipital Place Area (OPA) and the Parahippocampal Place Area (PPA) for place/scene processing. However, the existing neuroimaging studies on word recognition almost exclusively focus on one single region, i.e., the VWFA in the fusiform area. Moreover, according to hierarchical computational theory, there are a series of layers processing increasingly more complex information — from simple features to converged complex features, flowing from early visual cortex to higher areas in word recogition, like perception of other objects. The above studies and theory lead us to hypothsize that there may be a word sensitive region in the occipital lobe. My thesis aims to test this hypothesis, and to localize this region and investigate its functional characteristics.|
In our first study, a new region in the occipital cortex that selectively responded to Chinese characters (we tentatively label it as the OWA) was identified. We then precisely localized this region using the following methods: activation maps produced by the contrast of Chinese character vs. scrambled object image, well-established regions in the ventral fusiform and occipital corteces, anatomical landmarks (gyrus and sulcus in brain), and retinotopic mapping. The OWA was identified and localized in all 13 native Chinese speakers. This region was located near the inferior occipital gyrus or the middle occipital gyrus. In normalized Talairach space, the coordinates of the OWA averaged across subjects were [-41,-76,-6] in the left hemisphere and [39,-73,-6] in the right hemisphere. And on average, the distance between the OWA and VWFA was 22.9 mm. Detailed retinotopic maps and human visual motion area (hMT+) were obtained from five of these subjects, which showed that the OWA was anterior to the retinotopic areas, more ventral to Lateral Occipital Complex (LOC) , posterior to the VWFA, and inferior to the hMT+.
In our second study, we examined the OWA’s response characteristics to Chinese characters and other categories of visual images (cars, chairs and insects) and compared it’s selectivity to words (Chinese character) with other regions (the VWFA, LOC and left Angular Gyrus (AG)) in 10 subjects who also participated in the first study. Univariate analysis showed that characters generated stronger responses than the other three types of stimuli in the OWA, VWFA, and AG, while chairs generated strongest responses among the four types of stimuli in the LOC. However, the difference of selectivity index for characters between left OWA and left VWFA was not significant. Multivariate pattern analysis showed that the representation structure for the four categories in left OWA was similar to that in left VWFA, but different to that of the LOC. The discriminability index for characters vs. other categories in left OWA was significantly lower than that in left VWFA. The results suggest that the OWA has similar mean response properties as the VWFA, but with a less degree of specialization for words and the OWA is left-lateralized in terms of its word-selectivity.
In our third study, we employed sixteen categories of visual images (Chinese characters, English words, numbers, musical notes, mathematical symbols, Chinese character-like Dongba characters, picture-like Dongba characters, human figures, faces, animals, houses, trees, fruits, abstract sculptures, furniture, vehicles) to examine the detailed response profiles to multiple types of stimuli in the OWA and compare the OWA to VWFA with regarding to their word-selectivity. Averaged response amplitudes from fifteen Chinese subjects showed that five categories (English words, Chinese characters, animals, faces, and human figures) in left OWA were the strongest among the 16 categories, but the response amplitudes of the five categories in left OWA were not significantly different. However, in left VWFA, responses to English words, mathematical symbols, Chinese characters, animals and Numbers were the strongest, and English words and numbers evoked stronger responses than all the other categories in left VWFA. These results suggest the word-selectivity of the OWA is weak relative to the VWFA, and the OWA probably plays a role of processing more basic visual features of the symbolic system.
In our fourth study, we used three types of stimuli images according to orthography of Chinese characters. They were real characters, pseudo-characters (with legal orthography regularity but no pronunciation) and non-characters (characters with
illegal orthography regularity and no pronunciation), to examine whether the OWA was sensitive to orthography in Chinese characters. The VWFA and LOC served as referenced regions. Analysis of the response amplitudes from 13 Chinese college students revealed that in left OWA, non-characterss generated significantly stronger responses than real characters and pseudo-characters, while no difference was found between real characters and pseudo-characters. The pattern of response amplitudes to the three types of stimuli in left VWFA was similar to that in left OWA. Although in left LOC, images of objects generated strongest response, the difference in response between non-charcters and pseudo-characters was not significant. Theses results suggest that like the VWFA, the OWA is also sensitive to orthographic regurality.
In sum, we conducted a series of fMRI experiments on the functional role of human occipitotemporal cortex in visual word processing. We firstly identified a word sensitive area in human occipital lobe, and it complemented the understanding of the neural mechanisms for word and object recognition.
|张波. 人脑枕叶文字加工区的确定、功能定位和反应特性研究[D]. 北京. 中国科学院研究生院,2015.|