Abstract
The role of features in resolving the correspondence problem (maintaining object identity as the object or the eyes move) is an open debate. Many studies have reported that features do not bias the motion seen in ambiguous split or quartet motion displays. In contrast, features strongly influence correspondence in the Ternus display (e.g., Hein & Moore, VSS 2009, VSS 2010), in particular over short distances in cycling displays (Hein & Cavanagh, VSS 2011). We hypothesized that the absence of a feature bias in previous studies was due to the larger displacements inherent in the split and quartet motion stimuli and the shorter presentation time of the displays. To test this, we compared the influence of distance and presentation time on feature biases in split motion and the Ternus display. We varied the horizontal offset between consecutive sets of discs as well as the number of presentation cycles and biased the percept toward one motion percept or the other by matching surface polarity and line orientation. Surprisingly, we found strong feature effects in both cycling displays, but this effect decreased with distance only in the Ternus display, whereas it remained robust at all distances in the split motion display. Furthermore, the presentation time of the display strongly affected the feature bias in split motion but not in the Ternus display. We speculate that in split motion the feature bias takes time to develop so that they become apparent in our repeating display whereas the single trials of previous studies did not. In the Ternus display, on the other hand, the feature effect is immediate.
Meeting abstract presented at VSS 2012