obrien.vision

Visual illusion

Hermann Grid

Easy 2–3 minutes

Interactive Hermann Grid demo with an accessible explanation and parameters.

The Hermann grid illusion — a clear, layered explainer

What you see

A Hermann grid is a field of dark squares separated by straight, bright “streets”. Most observers notice ghostly grey smudges at the intersections of the bright streets. The smudges vanish if you look straight at an intersection, and they’re most visible a little away from where you’re fixating. The effect works in reverse contrast too (bright smudges on dark “streets”) (Wikipedia, n.d.; Bach, n.d.).

How to make it pop (and how to kill it)

  • High luminance contrast between squares and streets strengthens the smudges; reducing contrast weakens them (Bach, n.d.).
  • Straight, uninterrupted, orthogonal bars are crucial; if you curve the bars or round the corners, the illusion fades dramatically. This “straightness factor” has been documented experimentally (Geier et al., 2008).
  • The effect is clearest with moderate street width (not too thin, not too thick), a regular lattice of intersections, and normal viewing distances (Bach, n.d.).

A short history

  • Often credited to Ludimar Hermann (1870), who noticed the effect while reading a book with matrix-like diagrams. Ewald Hering soon noted the reverse-contrast version; David Brewster likely reported a related observation even earlier (Wikipedia, n.d.; Illusions Index, n.d.).
  • In the 1990s, Lingelbach, Schrauf & Wist introduced the scintillating grid variant: grey bars on black with white discs at intersections. Observers perceive black “scintillating” dots that blink in and out, especially with small eye movements (Schrauf et al., 1997).

The classical account: lateral inhibition (retina)

The textbook explanation invokes centre–surround receptive fields in the retina (and LGN). A neuron whose receptive-field centre sits at a street intersection has more bright surround than a neuron whose centre lies along a straight street between squares. Surround activation inhibits the centre more at intersections, so the intersection is encoded as slightly darker — hence the grey smudges. Two standard corollaries:

  1. Smudges disappear on direct fixation: in the fovea, receptive fields are very small, so “intersection vs. along-a-street” makes little difference.
  2. Reverse-contrast version flips the sign (white smudges on dark streets) via OFF-centre cells.

This line of thinking goes back to Baumgartner (1960) and later reviews (e.g., Spillmann, 1994). It’s consistent with classic centre–surround physiology (e.g., Hartline et al., Limulus) and remains widely taught (Spillmann, 1994; Bach, n.d.).

Where the simple retinal story struggles

Several robust observations don’t sit neatly with a pure centre–surround account:

  • Curving or slightly wavering the bars (or rounding the junctions) abolishes the smudges, even though the local luminance configuration at a point is similar — suggesting orientation/geometry matters (Geier et al., 2008).
  • There’s a marked orientation dependence and the appearance of phantom bands when grids are tilted — again pointing to processes beyond circularly symmetric retinal receptive fields (SpringerLink, 2003).
  • The scintillating grid behaves differently (requires or is strongly enhanced by eye movements; the “dots” blink), and its strength depends on disc and bar geometry in ways that aren’t fully captured by a concentric Mexican-hat filter (Schrauf et al., 1997).

These issues led many researchers to add cortical, orientation-selective and contextual modulation mechanisms to the explanation. Reviews and re-examinations highlight that lateral inhibition is part of the story, not the whole (Schiller, 2005).

The contemporary view: multiple stages, spatial filters, and context

Today, most vision scientists treat the Hermann family as a multi-mechanism phenomenon:

  • Early spatial filtering (centre–surround, contrast gain control) plausibly creates a baseline lightness difference between intersections and corridors (Spillmann, 1994).
  • Cortical, orientation-selective channels (e.g., V1-like filters) and contextual interactions make the effect geometry-dependent (straightness, junction structure), explaining why curvature or irregular spacing can extinguish it (SpringerLink, 2003).
  • For the scintillating grid, temporal factors (fixational eye movements, microsaccades) plus the specific white-disc/grey-bar arrangement produce the blink of dark dots; many demonstrations note that scanning is needed for the full “scintillation” (Schrauf et al., 1997).

A useful practical summary (with demos) is Michael Bach’s compendium: he shows that adding curvature also kills the scintillating variant — a compelling nudge that global layout matters, not just local centre–surrounds (Bach, n.d.).

Parameter sensitivities (what matters, experimentally)

  • Bar/“street” width: intermediate widths maximise the effect; very thin or very thick streets reduce it (Bach, n.d.).
  • Corner geometry: rounded corners or curvy bars suppress the illusion (the “straightness factor”) (Geier et al., 2008).
  • Regularity & density: many evenly spaced intersections strengthen appearance; isolated crossings don’t generally produce scintillation (Wikipedia, n.d.).
  • Viewing & eye movements: stronger in peripheral vision; direct fixation reduces it. Scintillating dots are particularly tied to small eye movements (Schrauf et al., 1997).

Why it still matters

The Hermann grid is more than a party trick. It has been used to probe receptive-field organisation and contextual modulation in human vision and to test models that combine retinal and cortical processing (e.g., spatial filtering plus orientation-selective interactions). Continued work — including on tilt, curvature, and dynamic variants — keeps refining how early and mid-level vision interact to construct lightness and contrast (Spillmann, 1994; SpringerLink, 2003).

The scintillating grid: sibling, not clone

  • Construction: grey bars on black; white discs at intersections.
  • Percept: black dots appear to pop in and out at the discs (often darker than the background).
  • Key differences: stronger temporal dependence (eye movements); geometry of the discs/bars crucial; doesn’t reduce to the same parameters as the basic Hermann grid (Schrauf et al., 1997; Oxford Academic, n.d.).

Further reading (selected)

  • Spillmann, L. (1994). The Hermann Grid Illusion: a Tool for Studying Human Perceptive Field Organization. Perception, 23, 691–708. (classic review linking grid illusions to receptive-field organisation)
  • Schrauf, M., Lingelbach, B., & Wist, E. (1997). The Scintillating Grid Illusion. Vision Research, 37, 1033–1038. (original scintillating grid paper)
  • Geier, J., Bernáth, L., Hudák, M., & SĂ©ra, L. (2008). Straightness as the main factor of the Hermann grid illusion. Perception, 37, 651–665. (curvature/straightness challenge to the retinal account)
  • Schiller, P. H. (2005). The Hermann Grid Illusion Revisited. Perception, 34, 1375–1397. (re-evaluation; argues simple lateral inhibition is insufficient)
  • Hamburger, Dixon & Shapiro (2012). From Hermann’s grid to Spillmann’s weaves. In Perception Beyond Gestalt (overview of variants and mechanisms).
  • Orientation dependence and phantom bands with tilted/undulating bars: see review and experiments.
  • Accessible demos and notes (with references and parameter explorations):
    • Michael Bach’s pages on the Hermann grid and the scintillating grid.
    • MIT lecture notes summarising the textbook (retinal) account and its limits.
    • Wikipedia’s overview for quick orientation and bibliography pointers.

Bottom line

  • The Hermann grid’s grey smudges arise from early spatial interactions (centre–surround / lateral inhibition) shaped and sometimes overridden by orientation-dependent, cortical context effects.
  • The scintillating grid shares family resemblance but adds temporal dynamics and stricter geometric constraints.
  • No single mechanism elegantly covers all variants; the consensus view is multi-level processing (retina and cortex) with geometry and eye movements deciding how strongly you see the effect.