Determinants of perceived depth: 1st- vs. 2nd-order contributions

Abstract

Recent investigations have shown that the visual system is able to localize objects in depth via processing of both luminance and contrast-based disparity information. (Wilcox and Hess 1995, 1996). Evidence to date suggests that the role of 2^nd-order stereopsis is to serve as a form of back-up to 1^st-order processing. However, this proposal has not been tested directly. Therefore, in these experiments we examine the hypothesis that 2^nd-order stereopsis serves primarily as a back-up system for depth localization when the 1^st-order disparity signal is unreliable or unavailable. We measured stereoacuity for Gabor patches in which the 1^st- and 2^nd-order components could be separately fixed at one of a range of disparities, while the other was varied in depth. Under these conditions, 1^st-order processing does not always determine perceived depth even when it does provide a valid signal. In fact, our results show that the contribution of 1^st- and 2^nd-order stereopsis to depth localization can be biased depending on the disparity at which the 1^st- and 2^nd -order information signals are presented. When the 2^nd-order stimulus is presented at large disparities and the 1^st-order signal is assigned small disparities, the 2^nd-order signal takes precedence. 1^st-order stereopsis only determines perceived depth when it and the contrast envelope are presented at small disparities. In this thesis we reveal that the range of disparities at which 1^st - and 2^nd-order processing operate depends on the ratio of the test disparity to the size of the component (i.e. the disparity/size ratio). Given that this ratio will determine which type of processing prevails, it is inappropriate to assume that 2^nd-order processing serves merely as a back-up system.