Abstract: Abstract This paper mainly studies three contents: first, whether different parts of face covering will effect face gender recognition for children with autism. Second, whether the types of expressions will effect face gender recognition for them. Last, if there is any difference between children with autism and other children when recognizing face gender. This study was a 3 (types of children: children with autism, children with mental retardation and typically developing children) × 3 (types of expression: happy, neutral and angry) × 4 (covered positions: eye, mouth, nose and none) mixed design. 63 children were enrolled in the study, including 21 children with autism, 21 children with mental retardation and 21 typically developing children. The stimuli were 240 face images. The experimental procedure of this study was completed by E-Prime. In the formal test, presentation would be shown first. And then presented the gaze point "+" in the middle of the screen. The stimulus would automatically transfer to the interface 1000ms later. We didn’t set time for stimulus, but when someone made key reaction, it would automatically transfer to the blank. The blank page would last for 1000ms. The next round of stimulus interface was followed by a random presentation of stimulus interface. Each of the 60 pictures was taken off during the experiment. The experiment mainly determined two correct analysis indexes of facial gender recognition: correct rate and correct reaction time. Variance analysis was used as the dependent variable with the type of test, and the types of expressions and the covering parts of face were as independent variables. The results showed that the main effect of the types of expressions is significant, F(2, 120)=145.056, p=.000, η2 p=.707. The main effect of types of children was closely significant, F(2, 60)=2.563, p=.086, η2 p=.079.When further compared, it showed that the accuracy of face recognition (M=.756, SE=.015) was higher in typical development children than in children with autism(M=.710, SE=.015). The interaction effect were significant between the types of children and the types of facial expressions ,F(4, 120)=4.641，p=0.002, η2 p=.134. The interaction effect was significant between the types of facial expressions and the parts of facial covering. Correct reaction time was used as the dependent variable with the type of test, and the types of expressions and the covering parts of face are as independent variables. The results showed that the main effect of types of children was not significant, F(2, 60)=1.921, p=.155. The main effect of the types of expressions was not significant, F(2, 120)=2.368, p=.098. The main effect of the parts of facial covering was not significant, F(3, 180)=.733, p=.533. The main conclusions of this study were as followed: First, the face recognition ability of children with autism was not as better as typical developmental children, and there was a significant difference between typical developmental children and mentally retarded children. Second, the processing of facial gender recognition for children with autism was affected by the expression types, which showed that the neutral expression was better than the happy expression and the happy expression was better than angry expression. Third, children with autism showed obvious local processing features in the process of facial gender recognition, which mainly relied on eye information for face gender recognition.

Key words: Key Words children with autism, gender recognition, face covering, facial expression

摘要： 摘 要 为探讨自闭症儿童在不同面孔遮蔽部位和不同面孔表情下的性别识别能力，本研究选取自闭症、智力障碍和普通儿童各21名对240张面孔表情图片进行性别识别。结果发现，自闭症儿童的性别识别正确率显著低于普通儿童，且表现出对中性和高兴表情的识别优于生气表情，同时在遮蔽眼睛时正确率最低。研究认为，自闭症儿童的总体性别识别能力偏低，且受面孔遮蔽部位和表情类型的影响显著，在面孔表情加工过程中体现出较为明显的局部加工特征。

关键词: 关键词 自闭症儿童 性别识别 面孔遮蔽 面孔表情