Look at the neighbouring blue lines. They move, yes, and what kind of shape do they represent? I would say they move in pairs.
Click the “Occluders vis(ible)” button. Now the blue lines suddenly outline a diamond, which moves in a circular fashion!
Please note: the motion of the blue lines is exactly the same, regardless whether the green occluders are green or have the background gray (=invisible). Our percept, however, undergoes a striking change.
When the occluder are on, the slider allows you to set the transparency of the occluders: just a little bit off background, and the blue line combine to a diamond. There is also some hysteresis here. There is more to try using the other buttons, or directly via the colour and transparency slider.
Technically the mechanism at work here is known as “motion binding”. When the edges of the diamond are covered by occluders with the same colour as the background (here, white) there is no information on the vertices of the square. Now the ends of the line become a property of the line and there is insufficient information to detect the circular movement.
The reappearance of the diamond (with occluders invisible) on eccentric viewing is probably caused by blurring the distracting line ends.
Adelson EH, Movshon JA (1982) Phenomenal coherence of moving visual patterns. Nature 300:523–525
Lorenceau J, Shiffrar M (1992) The influence of terminators on motion integration across space. Vision Res 32:263–275
Lorenceau J, Alais D (2001) Form constraints in motion binding. Nature Neuroscience 4:745–751
McDermott J, Weiss Y, Adelson EH (2001) Perception 30:905–923. I recently found their associated website with fine demos.
Mosca & Bruno have a similar demonstration in their nice collection on motion perception
Kanwar Plahar prompted me to draw attention to eccentric viewing, which had already been described by Lorenceau & Shiffrar (1992).