Abstract
Vision in daytime conditions relies on pointing one's fovea to successive targets of interest. Looking directly at visual targets may not be the best strategy to acquire information in nighttime conditions, however, because rods (which mediate night vision) are absent from the fovea. Here we explored the relationship between eye movements and perception in scotopic conditions, with a technique first proposed around 325 BC and still used today, known as "averted vision." Looking away from faint celestial objects, instead of directly at them, can improve their detection. The reason is thought to be that light falls on more rods as the objects approach ~20° of visual angle away from the center of gaze. No studies have determined the pattern of averted vision that best facilitates perception, however. We assessed averted vision while recording eye positions of dark-adapted human participants, for the first time. We used a CRT monitor and a series of three neutral density filters to create a centrally-placed dim artificial star, of apparent magnitude 4.2 (dimmer than Tau Ceti). After dark-adapting, participants (n=12) made eye movements to a series of fixation targets, viewed binocularly. Participants made a yes/no judgment about whether the star was visible at each fixation location. This provided a map of detection performance in all directions surrounding the star. If rod-cone distribution was the sole predictor of performance, we should have found peak performance 20° away from the star and a linear improvement as gaze approaches 20°. Contrary to this prediction, our participants' accuracy peaked around 10°, and decreased at greater distances. This work may provide a better understanding of rod function, and in turn lead to improved screening and assistive technologies for eye movement and night vision disorders.
Meeting abstract presented at VSS 2017