In the present study, we provided a cue (face or vase) beforehand to make subjects attend the target stimuli in the vase-face ambiguous figure, and compared the spatiotemporal cortical activation patterns underlying face (face-face) or vase (vase-vase) processing to the ambiguous figure using high-density (64-channel) event-related brain potential (ERP) recordings. Scalp ERP analysis found that the anterior N100, P160, N320 and posterior P100 and N160 were elicited by the face-face and vase-vase responses. The results of the ANOVAs showed that the anterior N100 and N320 elicited by the vase-vase response were more negative than the face-face response. The anterior N100 might reflect deployment of attention (conscious effort) to identify the target stimuli (face or vase) in early processing of the ambiguous figure, and the N320 might be the reversal negativity (RN) and was involved in involuntary perceptual reversals (from face to vase or reverse). Moreover, the mean amplitude (a late positive component: LPC) between 350 and 450 ms of the face-face response was larger than the vase-vase response on positive orientation over the front-central scalp regions. This result might support the view that the LPC reflect post-perceptual processing and indicated that the perceptual reversal of the vase-face illusion is influenced by top-down control.