Abstract
Stimuli flashed around the time of saccades may be mislocalized, shifting towards the saccade target. The mislocalization pattern has been interpreted as peri-saccadic space compression (Ross et al., Nature 1997). To investigate the deforming effect of such “compression of visual space” on object size, we directly tested if objects would be perceived as smaller along the direction of saccades.
Horizontal bars of 1°, 5°, or 10° in length with centers at −15°, 0°, +9°, or +15° were flashed in a random order, when subjects (n=2) made saccades from a fixation on the left (−10°) to a saccade target on the right (+10°). Subjects reported the perceived location and size of bars by indicating the end-points with a mouse. Locations of the perceived bar centers followed the classic mislocalization pattern. Mislocalization occurred approximately within a time interval from −40 to 40ms relative to saccade onset, and localization error peaked (up to 10 degrees) near saccade onset time.
A piecewise linear model of mislocalization was established based on the data, with which the localization error for a point at any position and any onset time can be interpolated. The model was verified to be valid against mislocalization patterns from vertical flashing bar experiments. Predicted lengths of perceived bars were calculated by determining the mislocalization of end points based on the model. For 1 and 5° bars, actual perceived lengths were not correlated with predicted lengths (r[[lt]]0.12, p[[gt]]0.36). Perceived lengths of 10° bars were correlated with prediction (r=0.39, p[[lt]]0.001), but only shrank by about 27% of the predicted amount.
While perceived bar length could shrink under certain conditions, it seems that mislocalization of points might not imply a corresponding change in perceived object size.
Supported in part by NIH grants # EY12890 and EY05957 (EP) and EY017988 (MP).