“Magnetic Sand” or “Interactivity” Illusions

Poster Presentation: Saturday, May 18, 2024, 2:45 – 6:45 pm, Pavilion
Session: Multisensory Processing: Illusions, recognition

Shinsuke Shimojo1 (), Kensuke Shimojo2, Eiko Shimojo1, Daw-An Wu1; 1Caltech, 2University of California San Diego

We investigate a series of new illusions elicited when moving one’s hands closely over a dynamic random-dot (white-noise) display (VSS’23 Demo Night). (1) Draw a letter or a circle or any other patterns, and a trail (typically brighter/whiter, but darker for some) is vividly seen, decaying after 500 ms or so (nicknamed “iconic trace”). (2) Repeatedly open and close your fingers very close to the display as if picking up and releasing the dots (squeeze/move away and open/approach back to the display). Nearby random dots tend to be attracted to and then repulsed by the fingers in concert with your motion (“magnetic sand”). (3) Make small back-and-forth motions with your open palm facing the display. The dots near your palm appear to be “captured” in their movements (4) Make similar motions with your hand behind the display (thus not visible). Again, the corresponding dots appear to be captured. These illusory effects are robust across the range of parameters we tested at least: 10-30 Hz frame rate, from barely-detectable to the highest-possible luminance contrast. Several different mechanisms seem to be involved, such as dynamic occlusion-based contrast adaptation and de-adaptation in (1) and (2), dynamic occlusion-based biases in local motion distribution in (2), and action capture in (3) and (4). Several observations might help further investigation. First, whereas the iconic trace illusion (1) tends to be the strongest in most of the observers, there was no significant rank-order correlation in the strength of the different effects across observers (N=10), indirectly supporting the involvement of multiple mechanisms across illusions. Second, just observing other’s actions yields significant effects, though this tends to be weaker than in the own-action condition. Third and most intriguingly, all these effects are perceptually interactive (while not really in the engineering sense).

Acknowledgements: Kensuke Shimojo is supported by Masason Fundation. The Chen Institute at Caltech