P1.P.2Visual cues and constancies: monocular depth cues (height in plane, relative size, occlusion, linear perspective) and binocular depth cues (retinal disparity, convergence)

Although the retina is essentially a 2-D surface, we perceive a 3-D world. The brain uses depth cues — clues from the visual scene — to reconstruct distance. They split into two families.

Monocular cues (work with one eye)

Useful in pictures, photographs, and at long distances:

1. Height in plane (also called height in field). Objects higher in the visual field are usually further away. A tree painted near the top of a landscape painting reads as distant.
2. Relative size. If two objects of known similar size produce different retinal images, the smaller one is judged further away. A row of identical street lamps appears as a row of shrinking copies.
3. Occlusion (or interposition). If object A blocks part of object B, A is in front and B is behind. Painters use this routinely.
4. Linear perspective. Parallel lines appear to converge as they recede toward a vanishing point — railway tracks, road edges, building rooflines.

Other monocular cues sometimes mentioned: texture gradient (textures get finer with distance), aerial perspective (distant objects look hazy/blueish), motion parallax (closer objects sweep past faster — this links to Gibson's optic flow).

Binocular cues (require both eyes)

Most effective at close to medium distances:

1. Retinal disparity. Each eye, separated by about 6 cm, gets a slightly different view. The brain compares the two images and the amount of disparity signals depth — bigger disparity = closer object. Stereoscopic 3-D films exploit this.
2. Convergence. To look at a near object, the eye muscles rotate the eyes inward. The brain monitors how much the muscles converge — strong convergence signals a near object; eyes parallel signals a distant one.

Picture-book examples

In the famous Ponzo illusion, linear perspective plus height in plane make the upper of two equal lines look bigger — the brain assumes it is further away (and so must be physically larger to produce the same retinal size). This neatly demonstrates that depth cues operate even on flat 2-D pictures.

Common exam errors

• Saying "stereoscopic vision" without naming the cue (retinal disparity is the technical term).
• Treating linear perspective as anything to do with eye lenses; it is purely about image geometry.
• Forgetting occlusion because it sounds too obvious to be psychological — examiners want it named.