Abstract
The human eyes are never at rest. In the so-called “fixation” periods in between saccades, the eyes wander incessantly following seemingly erratic trajectories, a motion known as ocular drift. These movements shift the image on the retina across many receptors, yielding temporal modulations that enhance high spatial frequencies. Consistent with this effect, previous studies have shown that ocular drift is beneficial for high-acuity vision.
Ocular drift is widely believed to be an involuntary, random motion, presumably resulting from physiological limits in oculomotor precision. However, theoretical considerations indicate that changes in drift characteristics would be beneficial to emphasize different ranges of spatial frequencies. Do humans actively modulate their eye drifts depending on the visual task?
Here we compared the characteristics of eye drifts measured during free viewing of natural images (n=28), examination of faces (n=22), reading (n=13), testing of visual acuity (n=15), and sustained fixation on a marker (n=29). Eye movements were recorded by means of a high-resolution DPI eye tracker.
We report considerable changes in drift characteristics across tasks. In all cases, Brownian Motion (BM), a specific case of a random walk that well describes the motion of a particle in a fluid, proved to be a good model of ocular drift. However, the diffusion constants of motion
differed significantly across tasks and were considerably smaller in the tasks that required fine spatial discriminations. Furthermore, whereas BM provided an excellent fit in high-acuity tasks, small deviations from BM were evident in other tasks, where ocular drift was better modeled by
fractional Brownian Motion (fBM) with a Hurst index larger than unity. These results are consistent with predictions based on the frequency content of the luminance modulations delivered to the retina in the various tasks. They show that humans actively tune ocular drift according to the task at hand.