Multiple blades of a continuously rotating fan appear to “pause� when they pass behind those of an overlapped stationary fan, giving rise to the overall impression that the fan at the back is rotating jerkily (Peter, 1956). To account for this pause-and-go fan illusion, we propose that the blades of the rotating fan are “collective objects�, which tax the same pool of attentional resource used for extrapolating visual motion perception. This hypothesis was supported by results from a series of experiments using an adaptive staircase procedure. The participants were asked to judge the smoothness of the rotating fan, and the point of un-partial ambiguity calculated from the psychometric function was compared across different conditions. The results showed, as predicted, that the perceived jerkiness increased with the number of blades of the rotating fan, and when attention was distracted away from the blades by a visual onset. Whether this attentional modulation of object motion is modality-specific or is shared by different modalities was tested by replacing the visual onset with an acoustic one. The fact that similar results obtained in the visual- and acoustic-onset conditions suggest that cross-modal attentional modulation operates on visual motion of collective objects.