Skip to main content
← Zpět na blog

A Grid of Black Squares. Ghost Dots at Every Corner. But Never Where You Look.

How many dots do you see?

You are looking at the Hermann grid, discovered by the German physiologist Ludimar Hermann in 1870 while reading a textbook on acoustics. A grid of black squares is arranged on a white background, with thin white corridors separating the squares. Your peripheral vision sees small grey spots at every corridor intersection. Focus on any one intersection directly · and the spot at that location disappears. Peripheral vision shows the spots; foveal vision does not. The spots are nowhere in the actual ink. They are a product of your retina’s lateral inhibition, applied to intersections where the local white area is larger than at any single corridor point.

What you are about to learn. What the Hermann grid is, how retinal lateral inhibition produces the phantom dots, why the spots are only visible in peripheral vision, the relationship between receptive field size and the spots’ visibility, and why the Hermann grid was one of the earliest successful applications of what we now call computational neuroscience to a perceptual phenomenon.

What the Illusion Looks Like

Draw a grid of black squares · say 4 by 4 squares, each about 1 to 2 centimetres on a side. Leave thin white corridors (maybe 3 to 5 millimetres wide) between adjacent squares, so the white forms a grid of intersecting corridors. At every intersection of the corridors, the local white region is the square corner · that is, a small white square where two horizontal corridors and two vertical corridors meet.

Look at the overall pattern. Your peripheral vision sees small grey spots at every one of these intersections. Focus directly on any single intersection; the spot there vanishes while the spots at other intersections remain visible. Move your gaze; the spot at the previously-fixated intersection reappears, and the one at the newly-fixated spot disappears.

The minimal recipe. A grid of dark squares on a light background, separated by thin light corridors. The corridor width should be small relative to the square size (typically 1 to 5 percent). At each corridor intersection, the local bright area (the intersection itself) is surrounded by more bright area than the midpoint of a corridor is. This asymmetry is what drives the illusion via retinal lateral inhibition.

Why It Works: Lateral Inhibition in Retinal Ganglion Cells

The Hermann grid is a consequence of lateral inhibition in retinal ganglion cells, which was one of the earliest computational phenomena identified in neural processing.

Step 1

Retinal ganglion cells have centre-surround receptive fields. Each on-centre ganglion cell responds to light in a small central region (centre) and is inhibited by light in a surrounding annular region (surround). The centre is about 10 times smaller than the surround. Off-centre cells do the opposite.

Step 2

The cells compute a local contrast signal. A ganglion cell’s output is proportional to (centre luminance) minus (surround luminance), weighted by the receptive-field sizes. This computes a local contrast: bright regions surrounded by dark give strong positive signals; bright regions surrounded by equally bright give weak signals.

Step 3

At the corridor intersections, the surround is brighter than at the midpoint. When a ganglion cell’s centre is on a corridor intersection, its surround includes parts of the corridors extending in all four directions (more bright). When the centre is on the midpoint of a corridor, the surround includes parts of the dark squares on two sides. The surround is therefore brighter at intersections than at midpoints · more inhibition, less signal, apparent darker appearance.

The retina computes before the cortex sees. The Hermann grid is generated before any signal has reached the cortex · it is a computational artefact of the very first visual processing stage. Your retina is already doing sophisticated spatial filtering on the image. Everything your brain sees has already been pre-processed by retinal lateral inhibition. The Hermann grid is a clean demonstration of that pre-processing at work, producing an illusion that is invariant across individuals · everyone with normal retinal structure sees it the same way.

Why the Spots Disappear at Fixation

The spots are only visible in peripheral vision and vanish at the fovea. This is because foveal ganglion cells have much smaller receptive fields than peripheral ganglion cells.

The receptive-field-size story. In the peripheral retina, ganglion-cell receptive fields are large (maybe 1 degree visual angle). These large fields can straddle multiple corridors at a typical Hermann grid, producing the darker-than-midpoint response at intersections. In the fovea, receptive fields are tiny (maybe 0.02 degrees). At fixation, a single corridor intersection is much larger than any individual foveal ganglion cell’s receptive field · the cells see only uniform white, and the lateral inhibition computation does not produce a darker signal. So the spots disappear when fixated. Walk the Hermann grid far enough away, and even foveal fields are smaller than the relative corridor width · the spots vanish everywhere.

A Harder Variant

Below is a Hermann grid at difficulty 3 · finer corridor lines, more squares. The phantom dots are vivid in peripheral vision.

How many dots do you see?

Common misconception: “the spots are in the image.” They are not. Sample any pixel at an intersection with a colour picker. The pixel is pure white, identical to the pixels along the midpoints of the corridors. The apparent grey of the spots is entirely generated by your retina. This is one of the clearest demonstrations that perception and image content are not the same thing · your visual system adds information to the raw input.

Limitations of the Lateral-Inhibition Account

The classical lateral-inhibition account of the Hermann grid has been revised in recent decades. Detailed measurements show that simple centre-surround inhibition does not quite fit the observed strength and geometry of the illusion.

The modified account. Baumgartner’s 1960 retinal-receptive-field explanation was a qualitative fit but not a quantitative one. Later work (Schiller, Spillmann, and others in the 1990s and 2000s) showed that the strength of the Hermann grid depends also on cortical processing · specifically on orientation-selective neurons in V1 · and that a purely retinal model underestimates the illusion’s flexibility. The modern account attributes the Hermann grid to a combination of retinal lateral inhibition and cortical simple-cell responses. Both contribute; the retinal part is the foundation and the cortical part is a refinement.

Hermann’s Accidental Discovery

Ludimar Hermann was reading John Tyndall’s 1867 textbook on acoustics in 1870 when he noticed that the book’s page-layout grid · which separated columns of text with white gutters and placed diagrams in neat rows and columns · produced these odd grey spots at every intersection of the white gutters. He wrote it up and published a short note. It became one of the most-cited demonstrations in vision science.

The pattern of discovery. Many classical optical illusions were discovered incidentally by scientists noticing something odd in everyday visual experience. Gregory and the Cafe Wall (tile pattern in a cafe). Necker and the cube (crystallography drawings). Hering and the radial starburst (patterned cloth). Hermann and the grid (textbook layout). Accidental observations are often the best way to discover illusions, because they guarantee the figure is something people might actually encounter in daily life.

Where the Hermann Grid Appears

Test Yourself on 50 More Illusions

The Hermann grid is one of more than 50 classical illusions on PlayMemorize. Each round draws a deterministic SVG scene and asks one grounded question: which is larger, which is brighter, which is actually parallel. The reveal overlay shows the true geometry plus a one-line “why it works” caption.

The takeaway. The Hermann grid is a demonstration that your retina is already computing before any cortical processing happens. Lateral inhibition in ganglion cells produces the phantom grey spots at corridor intersections, where the local surround is brighter than at corridor midpoints. The spots only appear in peripheral vision because peripheral ganglion cells have large receptive fields that can straddle the grid structure; foveal cells are too small. Ludimar Hermann noticed it in an acoustics textbook in 1870. Baumgartner explained it mechanically in 1960. Modern accounts integrate retinal and cortical contributions. A century and a half of research into a single phantom grid · still producing refinements, still teaching us how the retina thinks.

Připraveni hrát?
👁️

Illusions

Your eyes lie - the math knows the truth. Spot equal lengths, identical greys, and truly parallel lines across 57 classic optical illusions

Hrát nyní - zdarma

Bez registrace. Funguje na jakémkoli zařízení.