A lot of time and effort is spent to create the most visually pleasing virtual or augmented environments. Research focuses on how to reduce perceptual artifacts or increase image quality metrics. Inspiration for that often comes from vision science.For example one idea is foveated rendering: Here by mimicking the resolution of the visual system images are rendered with less resolution in the periphery.

However, while different display types, illumination patterns or image corrections are evaluated perceptually, could it be that they still impact our behavior?

The influence of display duty cycle on perceptual artifacts and oculomotor behavior

A critical characteristic of head-mounted displays is the refresh rate and the related duty cycle. Refresh rate defines how many frames are presented each second (e.g. 90 Hz = 90 Images), whereas the duty cycle define what percentage of each frame the display is actually illuminated. For the example of 90 Hz, one frame would last around 11 ms, but for a duty cycle of 10%, the display would actually only be turned on for 1.1 ms.

Current virtual reality head sets run with a rather low duty cycles, since otherwise during head or eye movements, the image smears across the retina and observers perceive motion blur. If the display is only illuminated for a short time, the amount of smear on the retina is reduced, and motion blur is less visible. However, also low duty cycles can lead to artifacts like the illusory perception of copies of a targeted stimulus, called phantom array.

In our study, we investigated the tradeoff between these artifacts and whether the characteristics of the display also affected our oculomotor behavior. For low duty cycles there was a large variability across observers in whether they could see the phantom array artifact, but there was a consistent change in their oculomotor behavior: The shorter the persistence of the display ( refresh rate * duty cycle = time display is actually illuminated), the higher the saccade error.

While an increase in saccade error of half a degree of visual angle seems to be negligible, since the resolution in the visual field exponentially decays towards the periphery, even that small change significantly affected the performance in a visual discrimination task.

Based on our psychophysical and oculomotor results, we could then model and weight the potential impacts on the observer given a specific duty cycle, to estimate the display characteristics that would limit negative impacts. Thus, demonstrating the value of psychophysical and oculomotor tasks in applied research.

Goettker, A., MacKenzie, K. J., & Murdison, T. S. (2020). Differences between oculomotor and perceptual artifacts for temporally limited head mounted displays. Journal of the Society for Information Display, 28(6), 509-519.