The Contribution Of Stereoposis And Motion Parallax To Depth Segmentation

Studies comparing how stereopsis and motion parallax inform depth yield mixed results, with some showing poorer performance from motion parallax. We propose that this stems from use of depth magnitude tasks, which bias observers towards using stereopsis. In Experiment 1, we used a depth segmentation task to evaluate performance of stereopsis and self-generated motion parallax individually and in combination. Observers performed equally well with individual cues but appeared to rely on stereopsis when both were available. Follow-up experiments showed that increasing the range of self-motion did not improve performance, but the use of object motion did. Subsequently, we replicated Experiment 1 using object motion with experts and non-experts and found segmentation from object motion was always better than stereopsis, and non-experts needed more stereopsis than experts to perform the task. Thus, we suggest that the ability to use depth information depends on several factors including task demands and observers’ experience.