Faces carry a lot of social information and are more likely to capture attention than other objects. The purpose of this paper is to explore the social faces how to influence the three networks of attention. Previous studies had found that social faces positively affected attentional orienting and executive control, whereas reduced the efficiency of alerting, but the related studies are too little to form solid evidence. Besides, some researches showed that when participants were presented with the face flanker tasks in which faces replaced arrows, the flanker effect would disappear. However, the reason of this phenomenon is not known for us. Federico, Marotta, Adriani, Maccari, & Casagrande (2013) believed that it was the sociality of the target stimuli inducing the phenomenon. Because the social faces occupy more attention resources. But there has a problem that the cocial disturbed stimuli, over the two sides of the face, will disturb the attention at the same time. The present research employed the ANT-I paradigm (attention network test-interactions) to measure the three attention networks: alerting, orienting and executive control. Except the classic arrow flanker stimuli (arrow-arrow-arrow, AAA), the face flanker task (face-face-face, FFF) to examine the influence of social faces on attention was also developed. In order to examine the executive control of social information, there were two new mixed flanker tasks here: first was the central face as target stimuli and flanked by congruent or incongruent arrows (arrow-face-arrow, AFA); second was the central arrow as target stimuli and flanked by congruent or incongruent faces (face-arrow-face, FAF). Our hypothesis is that if classic flanker effect disappear in FAF and FFF conditions, the disappearance effect found in FFF condition is not caused by the central target face. Because in the FAF condition, the target stimuli is arrow without any sociality. Similarly, in AFA condition, if there is not any effect found in FFF condition, it also proves that the central target face plays no role in the effect. Because the target stimuli of both AFA and FFF are social faces. Conversely, the central target face cause it. The experimental procedure based on the study of Callejas, Lupiánez, Funes & Tudela (2004), replacing their flanker tasks through the E-Prime 2.0 software. A 2 (alerting: sounds, no sounds)×2 (orienting: right, wrong)×2 (executive control: congruent, incongruent)×4 (flanker type: AAA, FFF, AFA, FAF) repeated-measures ANOVA on reaction times was performed. The results show that: (a) the interaction between alerting and flanker type, and the interaction between orienting and flanker type do not reach significance, which suggests that social faces have no influence on alerting and orienting in the present tasks; (b) the interactive effect between executive control and flanker type is significant, F(3,44) = 78.5, p＜.01, η2 p = .64. Further simple-effect analysis reveals that when disturbed stimuli are arrows (AAA and AFA), the flanker effect appear, but when disturbed stimuli are faces (FFF and FAF), the flanker effect disappear. It suggests that the disappearance of the flanker effect described above might be caused by the sociality of disturbed stimuli. In summary, this study is firstly employing a mixed flanker task. And it not only extends the social information to be used in the ANT-I, but also illuminates the true reason of face flanker effect’s disappearance. This research offers a empirical support and theoretical perspective for the processing and control mechanism of social and nonsocial information.