The visual system constructs perceptions based on ambiguous information. For motion perception, the correspondence problem arises, i.e., the question of which object went where. We asked at which level of processing correspondence is solved - lower levels based on information that is directly available in the retinal input or higher levels based on information that has been abstracted beyond the input directly available at the retina? We used a Ponzo-like illusion to manipulate the perceived size and separations of elements in an ambiguous apparent motion display. Specifically, we presented Ternus displays - for which the type of motion that is perceived depends on how correspondence is resolved - at apparently different distances from the viewer using pictorial depth cues. We found that the perception of motion depended on the apparent depth of the displays, indicating that correspondence processes utilize information that is produced at higher-level processes.