Text from the interactive explanation
Butterfly wing patterns in flight create powerful illusory motion cues
George R. A. Hancock, Emmanuelle S. Briolat, Anna E. Hughes, Laura A. Kelley, Jolyon Troscianko
Original paperMotion dazzle · bird's-eye motion lens
Butterfly patterns and flapping combine to send a predator's motion detectors the wrong way. Look through a bird's-eye motion detector and see it happen.
From George R. A. Hancock, Emmanuelle S. Briolat, Anna E. Hughes, Laura A. Kelley & Jolyon Troscianko, “Butterfly wing patterns in flight create powerful illusory motion cues”, Nature 658, 958–962 (2026) · doi:10.1038/s41586-026-11062-w
Flapping: on each upstroke the wings swing backwards and fold inwards, so their edges briefly move the wrong way while the body keeps going forwards.
Single pale band · flapping · twin: averaged grey wings
Simplified live model — reproduces the logic of the paper's detector, not its numbers. A simplified re-implementation of the paper's approach, running in your browser: forwards confusion = (B − F)/(B + F) and sideways confusion = (S − F)/(S + F), with S = (L + R)/2, computed every frame from the rendered butterflies.
Patterned Single pale band
2-beat mean · trace: instantaneous forwards confusion
Twin averaged grey wings
2-beat mean · trace: instantaneous forwards confusion
Measuring…
1 · The detector
How a bird sees motion
Animals from flies to birds judge movement with elementary motion detectors . Each one compares the light at two neighbouring points of the eye a moment apart: if what the lower point saw an instant ago now arrives at the upper point, something moved up. Wire two such half-detectors back to back, subtract one from the other, and the sign of the output tells you the direction.
The authors used a bird-like version. Before any motion is computed, the video is blurred to the sharpness of a small songbird's eye (about 6 cycles per degree ) and smoothed in time to a 100 Hz flicker-fusion limit — “fluid motion vision” that removes the wagon-wheel artefacts a computer would otherwise see. Grids of vertical and horizontal detectors then report motion energy in four directions: forwards, backwards, left and right.
From those four numbers come the paper's two measures. Forwards confusion is the contrast between backwards and forwards energy; above zero, wrong-way motion outweighs forwards motion, which should make a predator misjudge speed and strike behind. Sideways confusion compares left–right energy with forwards energy, a cue for misjudged turning.
Detectors also suffer from the aperture problem : through a small window, a stripe sliding along its own length looks like nothing is moving at all, and an oblique stripe moving forwards looks as if it is moving sideways. Turn the stripes in the lab and watch the forwards signal vanish.
Energies relative to stripes crossing the direction of travel (90°). 0° = stripes parallel to the direction of travel (“vertical” on a head-up butterfly).
At 30°, part of the forwards motion leaks into the sideways detectors.
2 · Real butterflies
Filmed at 1,000 frames per second
Supplementary Video 1 from Hancock et al. (2026), CC BY 4.0 — trimmed, scaled and re-encoded.
The team filmed take-offs of five European and African species — seven morphotypes, 18 flights — then re-ran every frame with the wings repainted as the butterfly's own average grey , as black , or as white .
Higher contrast raised motion energy in every direction. But only the natural patterns shifted the balance, “reducing the relative amount of forwards motion”: across 2,391 frames per treatment, natural wings produced significantly more forwards and sideways confusion than any of the pattern-removed versions.
Natural pattern vs pattern removed measured · take-off recordings
3 · Behaviour
Be the predator
To test whether the model's confusion actually misleads a hunter, the authors built a 240 Hz touch-screen game. 100 volunteers tried to catch butterflies flying sinusoidal paths at about 40 cm per second — 3,000 strikes in all. Most landed behind the butterfly: on average 77.6 pixels, about 1.2 body lengths. And the more forwards confusion a butterfly produced, the more elongated and backward-shifted its strikes became; between the most and least confusing of the five species, the strike centroids differed by about 38% of a body length .
Try it. Press Start and tap each butterfly as it crosses the grey field. The five pattern types are drawn for this page; the ellipses on the map are the paper's measurements for the real species.
Press Start, then tap the butterfly as it flies.
Your strikes appear as dots, in butterfly-centred coordinates (heading up; sideways distance folded to the right).
measured ellipses: densest 25% of human strikes per species (SI Table 15), crosses = centroids; dashed line = mean strike, 1.2 body lengths behind.
4 · In silico evolution
Evolve a dazzler
The authors let a genetic algorithm design wings. Starting from random patterns, each generation kept the best third of 24 butterflies and bred the rest from them, for 20 generations on three wing shapes: 57,600 patterns in all. Selected for forwards confusion, the patterns converged towards real butterflies — closer to them than random patterns or random-selection lineages. Selected for forwards energy , they turned into immaculate white butterflies , like the Pieridae “whites”: a strategy expected to make predators overestimate speed and overshoot.
Run a miniature version below. Choose what to select for and watch the population change.
Population of 24, sorted by fitness once measured; outlined = the 8 survivors that breed.
Miniature re-run with the simplified detector and simple stripe/band genes; the paper evolved 57,600 reaction–diffusion patterns rendered in Blender.
5 · Across a continent
What makes a dazzler — across 757 European butterflies
Next the team built a virtual flight rig: 757 morphotypes of 397 European species, flapping and gliding under a bird's-eye camera at 2,000 frames per second. Random-forest models then ranked which features of a wing best predicted the confusion it created. simulated flight · 757 morphotypes · 397 species
Forwards confusion was maximised, in order of importance, by:
Higher forewing contrastBold light-on-dark (or dark-on-light) patterning on the forewing.
Resolvable vertical stripes on the hindwingStripes running parallel to the body, wide enough for a bird to resolve.
Uneven stripe distributionOften a single stripe running centrally across fore- and hindwing, or a contrasting edge.
Higher hindwing luminanceA bright hindwing behind a darker forewing.
Rougher hindwing shapeTails and scalloped edges.
Vertical stripes on the forewingSeveral stripes parallel to the body axis.
“Vertical” means parallel to the body axis when the butterfly is pictured head-up. “Show me” loads an illustrative genome into the lens above; the live model is simplified, so its numbers will not match the paper's ranking.
6 · Implications
Beyond butterflies
The dazzling stripes of zebras and snakes have long been suspected of scrambling a predator's motion perception, but the evidence has been equivocal — and earlier tests used rigid targets. By letting the patterns move the way animals actually move, this study offers empirical support for long-standing “motion dazzle” hypotheses.
7 · Source & credits
Source
Hancock, G. R. A., Briolat, E. S., Hughes, A. E., Kelley, L. A. & Troscianko, J. Butterfly wing patterns in flight create powerful illusory motion cues. Nature 658, 958–962 (2026). https://doi.org/10.1038/s41586-026-11062-w · article page. Open access under a Creative Commons Attribution 4.0 licence.
Video. Supplementary Video 1 from Hancock et al. (2026), CC BY 4.0 — trimmed (24.0–36.6 s), scaled to 1280×720 and re-encoded.
Figures and numbers. Measured values (take-off contrasts, strike ellipses, strike distances) and simulated results (757-morphotype analysis, genetic algorithm) are quoted from the paper and its Supplementary Information.
Illustrations. All butterflies on this page are original procedural drawings; the paper's illustrations from the Collins Butterfly Guide (Richard Lewington) are not reproduced.
About the live model. The lens, aperture lab and miniature evolution run a simplified re-implementation of the paper's approach: top-down 2D wing kinematics (flap elevation projected through a tilted hinge, fore–aft sweep, body surge at 1.93 body lengths per wingbeat); a single detector scale; an approximate avian luminance (0.45 R + 0.45 G + 0.10 B, a stand-in for blue-tit double-cone catch); spatial blur with a centre–surround and temporal smoothing standing in for the paper's fluid-motion filtering; Reichardt correlators on neighbouring cells; and a uniform background. The lens view follows the butterfly (as the paper re-centres its take-off frames) while the detector grid stays fixed in the world, and the energy maps are lightly smoothed for display. It reproduces the logic of the paper's measures, not the authors' numbers.
Topics in this iPaper
Biology
we suggest that butterfly wing patterns work in tandem with flight dynamics to create visual illusions that confuse predators
Abstract, page 958
Computer science
we used genetic algorithms to evolve 57,600 unique butterfly wing patterns in silico, selecting for maximal motion confusion
Digital evolution for motion confusion, page 960
motion dazzle
Our results offer empirical support for longstanding 'motion dazzle' hypotheses
Summary and implications, page 961
animal coloration
Animal markings have evolved to exploit limitations in the vision of their predators
Introduction, page 958
elementary motion detectors
Elementary motion detectors (EMDs) form the basic unit of the motion processing system across a wide range of taxa
Motion-confusion measures, page 958
predator prey interactions
We validated motion-confusion model predictions by measuring attack misdirection in human 'predators' (100 participants, 3,000 trials)
Motion confusion and human misdirection, page 959
genetic algorithms
we opted to simulate evolution and selection for motion confusion with genetic algorithms
Methods, In silico butterfly pattern evolution, pages 964–965
butterflies
We rendered the simulated flight (both flapping and gliding) of 757 unique butterfly morphotypes across 397 European species
Motion confusion across European species, page 959
visual ecology
have seen increasing use within visual ecology research, notably for animal camouflage research
Methods, In silico butterfly pattern evolution, page 965