Optical Illusions & Visual Perception

Illusions are evidence, not tricks

Your visual system does not record raw pixels the way a camera does. It actively constructs a three-dimensional interpretation of the world from an ambiguous, noisy, two-dimensional retinal image, using built-in statistical assumptions and edge-processing shortcuts that are normally correct and normally invisible to you. An optical illusion is simply a contrived arrangement of lines and shading where one of those shortcuts fires and gives a visibly wrong answer — which is exactly why vision scientists use illusions as precision tools for probing how the perceptual machinery actually works, rather than dismissing them as mere curiosities.

live demo · line and grid patterns distorting under context● LIVE

Size and context illusions: Muller-Lyer and Ponzo

In the Muller-Lyer illusion, two identical lines look different lengths because of the arrow fins added at each end — one popular explanation ties this to depth cues learned from a "carpentered world" of rectangular corners, where inward-pointing fins resemble a receding corner and outward-pointing fins resemble a projecting one. The Ponzo illusion uses converging lines to imply linear perspective and depth; two identical bars placed across the converging lines look different sizes because the brain assumes the one nearer the vanishing point must be farther away, and automatically scales it up to compensate. Both illusions make the same underlying point: perceived size is a relative, context-dependent judgement, not a direct measurement of retinal image size.

Lateral inhibition: the Hermann grid and Cafe Wall

Stare at a grid of black squares on white and ghostly gray dots seem to flicker at the intersections — the Hermann grid illusion. Retinal ganglion cells have center-surround receptive fields that respond to contrast, and at an intersection more of a cell's inhibitory surround is covered by white than along a single corridor, which dampens the perceived brightness right there. The Cafe Wall illusion makes perfectly parallel mortar lines between offset rows of light and dark bricks look sloped, an effect driven by the interaction of edge-detecting neurons with the brightness contrast at each brick boundary, subtly biasing the perceived tilt of the straight lines between them.

simplified center-surround response:
  response = centerBrightness - k * surroundBrightness

  along one grid line:   surround mostly white  -> weak inhibition
  at an intersection:    surround mostly white on ALL sides
                          -> stronger inhibition -> dot looks darker

Filling in what isn't there: Kanizsa Triangle and Rubin's Vase

The Kanizsa triangle shows a bright triangle with crisp edges and a surface brighter than the background — except no triangle is actually drawn. Three pac-man-shaped discs with matching notches are enough to trigger amodal completion, the visual system's tendency to infer an occluding shape from partial evidence; neurons as early as visual areas V1 and V2 have been recorded responding to these illusory contours as if a real edge were present. Rubin's vase works differently: it is a genuinely bistable figure, where the exact same contour is assigned alternately to a vase or to two facing profiles, and the brain flips between the two readings rather than blending them, because figure-ground assignment is a binary decision, not a continuous one.

Zollner and orientation illusions

In the Zollner illusion, a set of perfectly parallel lines looks tilted once short diagonal hatch marks are added across them. Orientation-tuned neurons in the primary visual cortex interact laterally with neighbouring neurons tuned to different angles, and this interaction biases the perceived orientation of a line away from the orientation of nearby distractor lines — a tilt-repulsion effect that has been mapped out in detail through psychophysics experiments and is consistent with known lateral-inhibition circuitry between orientation columns.

Frequently asked questions

Are optical illusions a flaw in the eye or the brain?

The brain, not the eye. The retina reports raw contrast and edges accurately; illusions arise in the visual cortex, where the brain applies statistical assumptions about the world (like consistent lighting, common perspective cues, and how surfaces usually behave) to turn ambiguous 2D input into a 3D interpretation. Those assumptions are normally correct and useful, and illusions are simply the rare geometric arrangements where they misfire.

Why do the gray dots in the Hermann grid disappear when you look straight at an intersection?

Retinal ganglion cells have center-surround receptive fields, and at a grid intersection more of the inhibitory surround is covered by white than at a point along one line, which dampens the perceived brightness there and creates a ghost gray dot. Your fovea has much smaller receptive fields than your peripheral vision, so looking directly at an intersection uses fine receptive fields that don't produce the same inhibition, which is why the dot vanishes exactly where you fixate.

What makes the Kanizsa triangle special compared to other illusions?

It shows the brain actively constructing edges and surface brightness that are not physically present in the image. Three pac-man shapes are enough to trigger amodal completion, and neurons as early as visual areas V1 and V2 have been recorded responding to the illusory contour itself, as if a real edge were there — direct physiological evidence that perception is an active inference process, not a passive camera.

Try it live

Everything above runs in your browser — open Optical Illusions and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.

▶ Open Optical Illusions simulation

Optical Illusions: Where the Geometry Fools the Visual System