Abstract
It might be supposed that depth perception provides an ideal situation in which to study cue weighting — the end result is usually an unequivocal indication of depth and there are cues based on horizontal disparity as well as cues that we shall simply describe as “perspective “ cues. The latter include the subtense at the eyes of vertically orientated lines of equal length but different distance, the angles that horizontally orientated elements of closed figures make with the vertically orientated elements in the retinal projections of the rectilinear figures we use, and the angle between the projections of horizontal elements of stimuli that do not lie in a fronto-parallel plane. One major difficulty is that these factors cannot readily be measured in isolation (although monocular viewing gives some hope of measuring “perspective” cues in the absence of horizontal disparity). A second problem is not knowing the extent to which the weights that might attach to each class of cue are fixed (in early experience, say, and influenced, perhaps, by the extent of each observer's phorias) or remain plastic and change with the task, the stimuli, and the experimental requirements. (We have some evidence that any “weight” is not completely plastic from the fact that observers' performance is drastically affected by making perspective cues irrelevant; irrelevant cues should be, but appear not to be, ignored in our 2-AF-C tasks with feedback.) A third problem is not knowing the extent to which the weights are optimal. We report an attempt to solve some of these issues.