Assessing the Role of Vergence Changes in the Perception of Random-Dot Stereograms by Using Open-Loop Control of Vergence

Abstract

The fact that the 3-D shape of surfaces depicted by random dot stereograms can take several seconds or even tens of seconds to appear has been attributed to the failure to make appropriate vergence changes [B Julesz, 1971 Foundations of Cyclopean Perception (Chicago, IL: University of Chicago Press)]. Alternatively, the long latencies could be a consequence of the processing time needed to match the disparate images. To distinguish between these possibilities we measured perceptual latencies in a situation in which vergence changes had no effect on retinal disparities. To do this, horizontal eye movements were recorded with the aid of close-fitting scleral search coils in both eyes and the difference signal used to shift horizontally the two halves of a random-dot stereogram by equal and opposite amounts. When the amount of shift was equal to the magnitude of the vergence change, changes of vergence had no effect on the pattern of disparities—open-loop vergence. Three observers were presented with a sequence of stereograms depicting both ‘simple’ surfaces (a single square lying in front of the surround) and ‘complex’ surfaces, including spirals, ‘wedding cakes’, and saddle shapes under both normal and open-loop conditions. Under open-loop conditions, the complete 3-D shape was never perceived when the disparity range of the stereogram was large (>40 min arc), demonstrating the necessity of vergence changes, but the 3-D structure of ‘complex’ surfaces did build up over a period of several seconds indicating a separate disparity processing limitation.