Defective and faulty traffic lights are the major cause of traffic accidents. Insufficient prediction for the appearance of the amber light will lead to a rear-end collision if the lead driver take a severe brake. In order to make the amber light predictable, we provided an improvement tentative plan for the green traffic light, in which a rhythm flashing pattern was inserted in the latter presentation part of the green light. In this case, drivers can take advantage of temporal information given by the rhythm to predict the upcoming umber light. However the most appropriate duration and the frequency of the flashing part is still unclear. We used a rhythm prediction task to identify and compare the brief (500ms) and long (1700ms) interval prediction in fMRI. We observed left Insula and right middle frontal gyrus (MFG) showed significantly activated during brief interval prediction. Bilateral middle temporal gyrus (MTG), Superior frontal gyrus (SFG), and the right inferior temporal gyrus (ITG), caudate were more active in the long interval prediction. Overall, this research showed that reaction times for brief interval was shorter than long inter prediction, also suggested a distinct mechanisms recruited for the brief and long interval rhythm prediction.