The Pursuit of Illusory Motion

In 2006, researchers Peter Tse and Po-Jang Hsieh at Dartmouth College reported a striking new illusion, the infinite regress illusion, that went on to wow audiences at the Illusion of the Year Contest. The illusion (also known as the double-drift illusion or the curveball illusion) shows a series of Gabor patches (small disks with internal dark and bright stripes) moving straight up and down the screen. Meanwhile, the interior of the Gabor patches also contain motion; specifically internal motion that is orthogonal to the direction of global movement of the Gabor patches; that is, the internal motion of each patch might depict rightward motion, while the patches themselves continue moving straight up and down.

Tse and Hsieh discovered that when these Gabor patches are placed in our peripheral vision, they produce a compelling motion illusion in which the patches themselves seem to move diagonally, rather than up and down. The illusion is so strong that the illusory diagonal path appears to be influenced just as much by the actual global motion of the Gabor patches as by the internal motion contained within. In fact, the authors summarized the perceived movement as a linear combination of the actual global movement plus the internal movement of the patches. Critically this illusory motion effect only occurs in peripheral vision, as the true motion of the Gabor patches can be readily seen once the viewer fixates on the patches themselves. Here is a video demonstration of the Infinite Regress Illusion.

More recently, Matteo Lisi & Patrick Cavanagh (2015) showed that despite our misperception of the moving Gabor patches, our ability to make a saccade toward a patch seems to be unimpaired. That is, when we make a saccadic eye movement to fixate a Gabor patch, our eyes tend to correctly land on the target, suggesting that our saccadic system relies on immediate retinal information, rather than on time-integrated motion information, and as such is not susceptible to the illusion.

The stark dissociation between our perceptual system, which relies on integrating information from the past and present, and a saccadic system that relies only on current information left an important question open: What happens with smooth pursuit eye movements?

Smooth pursuit refers to eye movements that are continuous rather than saccadic—this occurs, for example, when we are following a moving dot on the screen with our eyes, or when we slowly rotate our head while maintaining fixation on a stationary object.

However, to investigate whether smooth pursuit eye movements are susceptible to the double-drift illusion, investigators had to devise a task where the smooth pursuit was based on an object that was not actually there. Otherwise, smooth pursuit would stabilize on the object and follow its exact path.

To do this, Dartmouth researchers Marvin Maechler, Nate Heller, Matteo Lisi, Patrick Cavanagh, and Peter Tse devised an experimental paradigm in which participants were asked to smoothly pursue the "inferred midpoint" between two moving Gabor patches. The stimuli in their recently published experiments looked like this.

The task was quite demanding, and their data are based on responses from just seven participants. Nevertheless, the results are compelling and clear: When smoothly pursuing the inferred midpoint between two Gabor patches, participants were systematically biased by the double-drift illusion. Whereas the true motion of the Gabor patches was a diagonal path that moved diagonally down-right (with no internal motion), and then diagonally up-left back to the original starting point (with rightward internal motion), participants' eye movements showed a different path. Initially, the eyes followed the true downward diagonal path of the Gabor patches, but on the upward diagonal path, the eye movements instead followed the illusory trajectory (up and to the right); that is, the smooth pursuit eye movements followed the illusory path induced by the internal motion in the Gabor patches.

This new research addresses the open question about the relationship between saccades and smooth pursuit eye movements. Whereas saccades can operate correctly on retinal information, ignoring the illusory effects of the double-drift illusion, smooth pursuit eye movements are as susceptible to the illusion as we are. These results provide evidence that smooth pursuit eye movements are driven by higher-level visual processes that are more closely aligned to our conscious perception than to the (more accurate) bottom-up retinal signal.