In the neighbouring figure you see three blue disks. And the first of the three jumps to the end and back to the beginning, ok?
Now move the slider to the right. This blanks the display between the movements for an adjustable interval, the so-called ISI (inter-stimulus interval). The perception progressively changes: at 100 ms ISI the entire group seems to move, doesn't it?
Thus changing ISI causes perception to switch from “element motion” (ISI < 25 ms) to “group motion” (ISI > 60 ms). The time values are from Pantle & Picciano (1976), here with less control over display timing anything above “0 ms” causes group motion for me.
Using the button “Toggle tags” adds individual colour tags to the disks, revealing that the left disk moves to the right. Still, at high ISI values we perceive group motion and rather accept the change in tag colours.
This is a case of “apparent motion” and a widely researched effect, without a full understanding of the mechanisms yet IMHO. It is usually called “Ternus Display” after the German psychologist Josef Ternus, but was first described 1917 by Julius Pikler (thanks, Thomas, for digging this up). Elaborate experiments (e.g. Otto et al., 2008) use the effect to decide whether a given kind of processing is tied to the retinal location or to a later stage tied to the grouping.
Pikler J (1917) Sinnesphysiologische Untersuchungen. Leipzig: Barth
Ternus J (1926) Experimentelle Untersuchungen über phänomenale Identität. Psychologische Forschung 7:81–136
Pantle A, Picciano L (1976) A multistable movement display: evidence for two separate motion systems in human
Otto TU, Öğmen H, Herzog MH (2008) Assessing the microstructure of motion correspondences with non-retinotopic feature attribution. J Vision 8:(7)1–15
Thanks to Evelina Thunell & Michael Herzog for kind advice.