Visual illusions of motion are a major source of interest in the vision research community. Explaining how a still image can produce an illusory perception of motion is important for a more complete understanding of our visual system.

A certain class of motion illusion, known as motion aftereffects, has a well-understood mechanism. When your eyes are exposed to continuous motion in one direction for a long time, motion detectors in your visual system become adapted to this motion. Once the motion is removed, the adapted motion detectors return to an altered baseline, producing relatively more activity in the unadapted direction (i.e. the direction opposite to the adapting motion). This is why after staring at a waterfall (moving down) may lead you to perceive upward motion when you look at the non-moving landscape nearby.

There is another type of motion illusion that does not require prior adaptation to actual motion. The peripheral drift illusion, first discovered by Fraser and Wilcox (1979) and later popularized by Akiyoshi Kitaoka (2014) presents observers with a static image that produces illusory perceptions of smooth drifting motion. See three examples of peripheral drift in a new paper by George Mather and Patrick Cavanagh, published last month in the Journal of Vision (click on the link and see Figure 1.)

The key to perceiving the peripheral drift motion illusion is to look at the images while moving your eyes back and forth between different regions, or by blinking. Every time you move your eyes or blink, your pupils “reset”, leading to small changes in their size over the course of a second or two. In their new paper, Mather and Cavanagh provide compelling evidence that these transient changes in pupil size are the underlying cause of the peripheral drift illusion.

To test this, four participants (including the two authors) participated in three experiments that involved observing each of the three images above under different instructions. In the “onset experiment”, one of the images would appear, and the participant was instructed to press a button whenever the illusory motion stopped (usually around one to two seconds after presentation). The “blink experiment” was similar, except each trial began with a prompted blink by the participant. In the “saccade experiments”, each trial began with a prompted “saccade” in which participants were asked to look at a different part of the image. During the study, changes in participants’ pupils were monitored with an eye tracker.

The data analysis showed a typical pupil response pattern at the beginning of each trial: a short constriction of the pupil (lasting a few hundred milliseconds) followed by a 1-2 second pupil dilation. By comparing the time course of each pupil dilation to the time it took participants to indicate that the motion had stopped, the researchers found astonishingly high correlations. Across the three conditions, the duration of the peripheral drift illusion was nearly perfectly predictable by the duration of the pupil dilation period.

This research confirms that pupil size changes play a major role in the peripheral drift illusion, but the exact mechanisms for this are still being explored. For example, the theory cannot explain exactly why certain changes in contrast or luminance result in illusory motion in opposite directions. Future research will address these questions, aiming to build a more complete model of how involuntary pupil changes affect our perception of motion.