Figure 2. Slant estimates for three representative subjects.

Subjects estimated both the perceived surface slant that was dominated by the disparity cue and the perceived slant that was dominated by the perspective cue. Each data point represents one single estimate. The abscissa reflects the slant defined by disparity (left) and perspective (middle). From these estimates, we calculated the subjective level of incongruence (right). For subjects JX and LD, estimates are reliable, although somewhat underestimated, resulting in a correlation between objective and subjective incongruence. For subject RV, who is deliberately included because his stereovision proved to be poor, estimation of the disparity-defined slant is poor, and as a result so is the correlation between objective and subjective incongruence. Errorbars denote SD.