Constancy depends on scene cues, and stripping them away produces illusions

What it is

Size constancy and shape constancy work reliably only when the scene supplies enough depth and orientation cues for the visual system to run its joint computation. When these cues are artificially stripped out or replaced with misleading ones — placing an object against an empty, reference-free background, or deliberately offering a false perspective hint — the constancy mechanism does not simply switch off. It keeps running its joint computation using whatever cues (even false ones) are available, producing an illusion: a systematic deviation from the object's true size or shape. Classic demonstrations such as the Ames room and the Ponzo illusion work by deliberately constructing misleading depth cues so that the constancy mechanism "operates normally" and still arrives at a wrong conclusion.

Why it happens

The way the constancy mechanism operates leaves it unable to tell "genuine depth/orientation information" apart from "artificially constructed cues that look like depth/orientation information" — it simply carries out its fixed routine: combine whatever cue information is currently available, jointly infer the object's distance/orientation, and use that to correct the size/shape judgment derived from the retinal projection. This means an illusion is not a sign of the constancy mechanism malfunctioning — quite the opposite, it is exactly what this mechanism produces when operating as designed but fed untrue cues; the inference process itself is not wrong, the cues fed into it are. This is also why such illusions are so hard to dispel just by "knowing the truth": even after being told two figures are actually the same size, the constancy mechanism keeps producing a biased judgment based on the scene's cues, because this inference happens at a perceptual-processing stage that precedes conscious judgment.

Studying it

Classic illusions such as the Ames room and the Ponzo illusion are themselves the research paradigm: by systematically manipulating the strength or direction of perspective cues in a scene, researchers measure the size of the bias produced in participants' judgments of a target object's size or shape, examining the dose-response relationship between bias magnitude and cue strength; other studies compare illusion strength when cues are consistent versus when a single cue is swapped for a misleading version, isolating which cue contributes most to the constancy judgment. Common independent variables are the strength or type of the misleading cue and the ratio of genuine to misleading cues in the scene; common dependent variables are the magnitude of perceived size/shape bias and participants' confidence in their judgment.

Where it stops holding

These illusions require the scene to simultaneously lack other genuine cues that would contradict the misleading one — if a strong cue such as binocular disparity, which a single fabricated perspective cue cannot easily override, remains present, or if participants can move freely and gather motion-parallax information from changing viewpoints, illusion strength typically drops sharply or disappears. Classic illusion demonstrations mostly require participants to keep a fixed, often monocular, viewpoint, a precondition that usually does not hold in real environments where people move freely and view with both eyes.

Related

• Same group: A1.27.1 Size constancy keeps an object perceived as the same size across different distances · A1.27.2 Shape constancy keeps an object perceived as the same shape across viewing angles · A1.27.4 Insufficient rendering of lighting or viewpoint cues undermines object recognizability
• Nearby: A1.13.4 Depth cue conflict · A1.32 Visual illusions
• Site search: Ames room · Ponzo illusion · cue-impoverished scene · constancy illusion