We utilized the high temporal resolution of functional near-infrared spectroscopy to explore how sensory input (visual and rhythmic auditory cues) are processed in the cortical areas of multimodal integration to achieve coordinated motor output during unrestricted dance simulation gameplay. Using an open source clone of the dance simulation video game, Dance Dance Revolution, we investigated whether the speed of visual cues affect the temporal accuracy of dance steps and the regional brain activity at the middle temporal gyrus (MTG), the integration site of visual and auditory information. To this aim, visual cues of two speeds were employed: normal speed (Χ1) and slow speed (Χ0.5). However, the number of visual cues and the timing to respond to visual cues were the same between conditions. Auditory cues (music) remained constant at normal speed. Nine young-adult participants with experience playing this dance video game were recruited. Comparing the Oxy-Hb responses at the MTG in the normal speed condition with that in the slow speed condition, we found the signal peaked earlier in the slow speed condition (P < 0.05; paired t-test). We also found that the number of temporally accurate steps in the slow speed condition was smaller than that in the normal speed condition (P < 0.01; paired t-test). These results correspond to the results of our previous study that the time to peak amplitude in the MTG correlated positively with the temporal accuracy of dance steps. These results suggest that visual and auditory information may not be well integrated when the given speeds of multimodal cues are not temporally matched and that the sustained MTG activity can be a possible index of successful integration of multimodal cues. We propose that changes in these relationships can be monitored to gauge performance increases in motor learning and rehabilitation programs.