Chapter 11

Chapter summary

Depth computations are used to establish the depth ordering and intervals between the visible surfaces, to evaluate the distance to objects, and to estimate their shape.

A wide range of visual and non-visual cues provide information for depth processing, sub-divided into monocular cues (interposition, height in the visual field, perspective, size, texture gradient, blur, accommodation, motion, and shading), and binocular cues (disparity and vergence).

The disparity cue is based on small differences in the horizontal and/or vertical position of image elements in the two eyes, and has been intensively studied. Horizontal disparity is encoded by specialised neurons which are driven by both eyes and respond according to the disparity present in a region of the retinal image. The importance of vertical disparity has long been debated.

Different depth cues provide correlated information, but the cues vary in terms of their availability and precision in different images. So the visual system combines information from different cues in a complex, flexible and idiosyncratic way. Research indicates that perceived depth is based on a weighted average of the estimates provided by different cues. Cue weight varies according to stimulus conditions, task and observer.