Humans are better at recognizing human faces than those of other species (e.g, monkeys) behaviorally. However, the neural basis of such phenomenon remains unknown. To elucidate this issue, a 128-ch high-density event-related potential were recorded in 13 healthy adults during viewing morphing human face into monkey face. In all stimuli, N170 and the late positive (LP) component (400-600 ms) were elicited as the major components in the temporo-occipital regions. The more the ratio of human face reduced, the smaller the N170 amplitude became. The latencies were also prolonged. The amplitudes of LP component for ambiguous face were larger than those of human and monkey face. These findings suggest that 1) N170 reflects species-specific face processing, and 2) the LP component processes the difference in complexity of face stimuli. Therefore, we conclude that the identification of species starts at around 170 ms, then detailed feature of faces are analyzed after 400 ms.