Repetition suppression for faces, a phenomenon that neural responses are reduced to repeated faces in the visual cortex, have long been studied. However, the underlying primary neural mechanism of repetition suppression remains debated. In recent years, artificial neural networks can achieve the performance of face recognition at human level. In our current study, we combined human electroencephalogram (EEG) and the deep convolutional neural network (DCNN) and applied reverse engineering to provide a novel way to investigate the neural mechanisms of facial repetition suppression. First, we used brain decoding approach to explore the representations of faces and demonstrates its repetition suppression effect in human brains. Then we constructed two repetition suppression models, Fatigue and Sharpening models, to modify the activation of DCNNs and conducted cross-modal representational similarity analysis (RSA) comparisons between human EEG signals and activations in two modified DCNNs, respectively. We found that representations of human brains were more similar to representations of Fatigue-modified DCNN instead of Sharpening modified DCNN. Our results suggests that the facial repetition suppression effect in face perception is more likely caused by the fatigue mechanism suggesting that the activation of neurons with stronger responses to face stimulus would be attenuated more. Therefore, the current study supports the fatigue mechanism as a more plausible neural mechanism of facial repetition suppression. The comparison between representations in the human brain and DCNN provides a promising tool to simulate and infer the brain mechanism underlying human behaviors.