FFA（Fusiform face area）和VWFA（visual word form area, VWFA）是位于梭状回上的对于面孔和书写词表现出选择性反应的区域。但是注意能不能改变FFA和VWFA等区域的类别选择性，以及相对于多种刺激类别FFA和VWFA对适宜刺激的反应是否仍具有特异性这两个问题的考察对于考察类别选择性区域的皮层特异性非常必要。本研究采用fMRI技术来考察这两个问题。 研究一通过MVPA的方法对高注意水平和低注意水平两种条件分别进行了条件内预测、条件间预测以及调整低注意水平条件的相对信噪（contrast-to-noise ratio, CNR）比之后的条件间预测。结果表明注意的作用体现在改变FFA和VWFA对刺激激活的CNR，而不能调节类别选择性区域对刺激表征的空间模式。研究二采用高空间分辨率fMRI扫描，通过信号变化的分析以及表征相似度分析（RSA, representation similarity analysis）考察FFA和VWFA以及比FFA和OWA更靠后的两个区域OFA（Occipital Face Area）和OWA（Occipital Word Area）对不同刺激类别在激活幅度上的差异以及激活模式上的相似程度。信号变化分析结果显示：在FFA和OFA，面孔的激活强度显著高于其他刺激类别的激活强度；在VWFA，汉字的激活强度显著高于其他刺激类别的激活强度；在OWA，汉字的激活强度显著高于除面孔外其他刺激类别的激活强度。RSA结果分析发现：在FFA，面孔刺激的激活模式与其他刺激出现分离；在OFA，面孔的激活模式除了与汉字的激活模式相关较高之外，与其他刺激类别的激活模式相关均较弱；在VWFA，文字的激活模式与线条图、工具的激活模式比较相似；在OWA，文字与工具的激活模式相关也较高。 研究结果表明：1）FFA和VWFA对文字的特异性反应不受任务引发的注意水平的调控；2）FFA无论从信号变化还是激活模式上都出现对面孔的特异性反应，OFA对面孔也具有选择性反应，但其选择性没有FFA强；3）VWFA和OWA从激活强度表现出对文字的特异性反应（OWA对文字的选择性没有VWFA强），但从激活模式上看，VWFA和OWA也参与了其他刺激类别的表征（尤其是线条图和工具）。; FFA（Fusiform face area）and VWFA（visual word form area, VWFA） are two brain regions in left occipito-temporal cortex showing selective responses to faces and written words. two of the necessary issues in examining response selectivity is, whether attention can modify the category selectivity of FFA and VWFA ; whether such specific responses still persist when multiple stimulus categories are selected. The present research used fMRI techniques to examine this two questions. In the first study, the MVPA method was used to investigate within-condition and across-condition prediction accuracy under high and low attention levels. The contrast-to-noise ratio (CNR) in the low attention level was also manipulated to see the impact on across-condition prediction accuracy. Results show that the effect of attention was to modify CNR but not the spatial pattern of the neural responses in brain regions showing stimulus category selectivity. In the second study, high spatial resolution fMRI scanning was used to conduct activation signal change analysis and representation similarity analysis (RSA). These analyses were meant to understand whether FFA and VWFA, and two posterior regions OFA (Occipital Face Area) and OWA (Occipital Word Area) , would show different level of activation for different categories of stimuli and to what extent the activation pattern was similar. The signal change analysis showed that, in FFA and OFA, the activation for face stimuli was significantly higher than that for other stimulus categories; in VWFA, the activation for Chinese characters was significantly higher than for other stimulus categories; in OWA, the activation to Chinese characters was significantly higher than for other stimulus categories except faces. The RSA analyses showed that, the activation pattern in FFA for face stimuli was distinctive from that for other stimulus categories. In OFA, other than a relatively high correlation with Chinese characters, the activation pattern of faces was not similar with that of other stimulus categories. In VWFA, the activation pattern of written words was similar to that of line drawings and tools. In OWA, the correlation in activation pattern was quite high between written words and tools. The research indicates that, 1) Category selectivity in FFA and VWFA to written words was not modulated by different levels of attention as induced by different tasks; 2) FFA showed response selectivity to faces both from BOLD signal changes and from the pattern of activation. Face-specific selectivity in OFA was not as strong as in FFA, showing strongest responses in signal changes to faces but similar pattern of activation to written words; 3) VWFA and OWA showed response selectivity in signal changes to written words, although such selectivity in OWA was weaker than that in VWFA. From the pattern of activation, both VWFA and OWA were involved in the representation other types of stimulus categories, in particular, line drawings and tools.