## What to see

The pattern at the right is stationary – yet it seems to expand, especially if you don’t look directly at it, for instance when reading this text.

Not everybody perceives this illusion, for as yet ill-understood reasons.

## What to do

This is one of the many variants of the “Snake Illusion” as made famous by Akiyoshi Kitaoka. So what is special here? The controls in the present demonstration allow you to explore the figural components to optimise the illusion.

First thing to recognise: colour is nice, but not necessary. So for ease of control the figure switches to grey when adjusting the brightness.

The upper slider controls the relative brightness (technically: the luminance) of the medium bright stripes. As the diagram at the bottom right shows in a cross-section of part of the figure: there are narrow very dark (bluish) and very bright (yellowish) stripes, which I will call anchor stripes. In between these are wider stripes with a darker and a lighter grey. The double arrows indicate that these levels can be adjusted (with said slider). When they are equal (slider in the middle), there is no illusionary motion. When their brightness is exchanged, the illusory expansion changes to contraction.

The number stepper left of the sliders adjusts the numerosity of the star figure. As you may see, this does not play a major role. Below are color pickers where you can change the colours to your heart’s content. If you’re lost, click on the “reset” button.

• While adjusting the stripe brightness, a strong additional motion illusion occurs.
• Experimenting with this figure gives you the tools to construct your own snake illusion variant. All you need are these 4 different luminance levels. The ideal luminance can also be achieved via colour, blue tends to be rather dark: but then there are 3 dimensions to every colour: hue, luminance and saturation.
• We just calculated a simple model (preprint) showing that this illusion is a trivial consequence when you couple motion detectors into arrays with a little non-linearity.

We just (2020) calculated a simple model showing that this illusion is a trivial consequence when you couple motion detectors into arrays with a little non-linearity.

## Sources

The original “Rotating Snake”

Bach M & Atala-Gérard L (2020) The Rotating Snakes Illusion is a straightforward consequence of non-linearity in arrays of standard motion detectors, i-Perception 11(5), 1–9 [→PDF]

Clément Depoutre pointed out to me that he has a similar design (scroll to bottom). I may have seen it (a case of “source agnosia”?) when I programmed the above demonstration. His design from June 2007 has been “adopted” by many internet sites.

More literature on the companion page.