Abstract
Real world scenes create a complex pattern of vertical disparities that provides information about the absolute distances to all surfaces in the scene as well as the current vergence state of the eyes. Rogers and Bradshaw (1995) showed that both vertical disparities and vergence contribute to depth scaling with relative weightings that depend on the size of the visual field. These results raise the question of whether a consistent but inappropriate pattern of vertical disparities could be used to recalibrate vergence. We used large field stereoscopic images of a real world scene which were presented in a modified Wheatstone stereoscope so as to reproduce exactly the vergence demands and the horizontal disparities of the original scene. However, the images were manipulated so that the pattern of vertical disparities was appropriate for a 3-D scene that was either closer to the observer (by the equivalent of 4.33 deg of convergence) or farther away, by the same amount. Observers viewed the stereoscopic images with inconsistent vertical disparity and vergence information for 12 minutes. To test whether the vergence system had been recalibrated, we presented stereoscopic images of a 3-D object which subtended < 10 × 10 deg, to render the pattern of vertical disparities ineffective for size and depth scaling (Rogers and Bradshaw, 1995). Observers were asked to adjust (i) the angular size of the 3-D object until it appeared to be the appropriate physical size and (ii) the horizontal disparities until the 3-D object appeared to have the appropriate depth. The results showed a significant recalibration (up to 25% of the discrepancy) of vergence signals in the direction of the inconsistent vertical disparity information. We also found significant recalibration of vergence signals after viewing a 3-D scene in which the pattern of vertical disparities was veridical but the vergence demand was appropriate for a scene that was either closer to the observer or farther away.