obrien.vision

Visual illusion

Rotating Snakes

Medium 2–3 minutes

Interactive Rotating Snakes demo with a detailed explanation of the peripheral drift illusion.

Heads up: Some viewers may experience visual discomfort. Take breaks.

The Rotating Snakes illusion — a clear, layered explainer

What you see

Rotating Snakes is a static image that appears to rotate—usually more strongly in your peripheral vision. Look slightly off the pattern, blink or move your eyes a touch, and the “snakes” seem to turn; stare directly, and the motion largely vanishes. The effect is a famous member of the peripheral drift illusion (PDI) family (Bach, n.d.; Wikipedia, n.d.).

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

  • Asymmetric luminance sequence is essential. The classic tangential order is Black → Dark → White → Light (or a close luminance-equivalent). Reverse the order and the perceived rotation flips; swap steps to make them symmetrical and the motion collapses (Bach, n.d.).
  • High contrast makes it stronger; low contrast weakens it. Colour works but only if luminance relationships follow the asymmetric sequence (Backus & Oruç, 2005).
  • You’ll see the motion most in the periphery (not at fixation), and brief transients—blinks, tiny eye movements—trigger episodes of motion (Murakami, 2006; Otero-Millan et al., 2013).
  • Variants like rings, arcs, or stripy “drifts” can all work; the geometry needn’t be snakes per se, but the luminance ordering is non-negotiable (Bach, n.d.).

A short history

  • Fraser & Wilcox (1979). Described illusory motion from stationary luminance gradients (“escalator illusion”). Faubert & Herbert (1999) named the broader class peripheral drift illusion, emphasising periphery and transients. Kitaoka & Ashida (2003) catalogued the phenomenal traits and introduced now-iconic artworks; Kitaoka’s “Rotating Snakes” (2003) became the best-known example (Wikipedia, n.d.).
  • In 2005, Conway, Kitaoka, Yazdanbakhsh, Pack & Livingstone reported direction-selective neuronal responses to static PDI displays, and framed the effect as a static cousin of four-stroke apparent motion; it made the Journal of Neuroscience cover (Conway et al., 2005; Journal of Neuroscience, 2005).

The classical ingredients

  1. Asymmetric luminance steps along the motion path (e.g., B → D → W → L) bias the visual system. Perceived motion tends to run from dark towards light along that sequence (Wikipedia, n.d.).
  2. Transient events (microsaccades, saccades, blinks, sudden onsets) “kick” the system so that the asymmetric code produces a net motion signal (Faubert & Herbert, 1999; Otero-Millan et al., 2013).

What’s going on under the bonnet?

Latency differences + standard motion detectors

A prevailing account says dark and light regions are processed with slightly different response timings (latencies). When a transient occurs, these timing offsets feed into ordinary low-level motion detectors, creating illusory motion energy in the dark→light direction (Conway et al., 2005; Backus & Oruç, 2005).

Eye movements (and why fixation kills it)

The illusion is episodic: it flares with microsaccades and blinks and fades during steady fixation. Quantitative work links the rate of microsaccades to perceived speed/strength; the peripheral emphasis also fits (Murakami, 2006; Otero-Millan et al., 2013).

A fresh twist: the pupil

Recent work proposes that pupil dilation transients contribute: brief luminance changes at the retina during pupil dynamics may help trigger episodes, and pinhole viewing (which removes pupil fluctuations) reduces the illusion (Laeng et al., 2024).

Modelling

Simple arrays of standard motion detectors with mild non-linearity reproduce robust rotation given the asymmetric luminance sequence—supporting the idea that no exotic machinery is required, just timing asymmetries + non-linear integration (Bach, n.d.).

Parameter sensitivities (practical notes)

  • Order matters most. Keep the tangential sequence Black → Dark → White → Light (or luminance-matched equivalents). Reversing the order reverses perceived rotation; equalising steps tends to kill it (Bach, n.d.).
  • Contrast helps. Higher global and local contrast strengthens the effect (Backus & Oruç, 2005).
  • Periphery & transients. Strongest when viewed off-centre and during blinks/eye movements; central fixation and prolonged stillness attenuate it (Faubert & Herbert, 1999).
  • Direction quirks. Most patterns spin the same way for most observers, but specific luminance mixes can flip direction, a phenomenon mapped in luminance-space measurements (Gekas & Mamassian, 2017).

Why it still matters

Rotating Snakes has become a testbed for how early temporal dynamics, fixational eye movements, and motion mechanisms interact. It bridges art and neuroscience, constraining models that meld contrast/luminance adaptation, timing asymmetry, microsaccades, and classical motion detectors (Conway et al., 2005; Backus & Oruç, 2005).

Further reading (selected)

  • Kitaoka’s original demo and notes (colour sequences, variants, direction): Akiyoshi’s illusion pages (Wikipedia, n.d.).
  • Michael Bach’s explainer: concise notes on the asymmetric luminance sequence, modelling, and myths (Bach, n.d.).
  • Conway et al., 2005 (J. Neurosci.): direction-selective neurons respond to static PDI displays; “static four-stroke” account (Conway et al., 2005).
  • Backus & Oruç, 2005 (JOV): contrast/luminance adaptation + timing explanation; demonstrations with movies (Backus & Oruç, 2005).
  • Faubert & Herbert, 1999 (Perception): foundational paper on peripheral drift, emphasising transients and periphery (Faubert & Herbert, 1999).
  • Murakami, Kitaoka & Ashida, 2006 (Vision Research): fixation instability ↔ illusion strength (Murakami et al., 2006).
  • Tomimatsu, Ito, Sunaga & Remijn, 2011; related: halt/recovery time-courses of static motion illusions (Tomimatsu et al., 2011).
  • Martinez-Conde et al., 2013/2019 reviews: microsaccades—rates, durations, perceptual roles (Martinez-Conde et al., 2013; 2019).
  • Pupil dilation hypothesis (2024, JOV): episodes track pupil dynamics; pinhole viewing reduces the effect (Laeng et al., 2024).

Bottom line

  • The snakes “move” because asymmetric luminance steps interact with transient eye events and timing differences in early visual responses, producing illusory motion energy—strongest in the periphery, typically dark→light (Conway et al., 2005; Backus & Oruç, 2005; Wikipedia, n.d.).
  • Microsaccades, blinks (and perhaps pupil transients) start and stop the episodes you feel as bursts of rotation (Otero-Millan et al., 2013; Laeng et al., 2024).
  • Good demos keep the luminance ordering intact and contrast high; the “snake” motif is decorative—the sequence is the star (Bach, n.d.).