Abstract
Background: The sense that one is “looking at” a target is associated with stimulation of the fovea, and eye movements position or keep targets in the fovea. With recent advances in retinal imaging, in particular the AOSLO, we can now determine the absolute retinal position of a target during eye movements and visual direction judgments, and compare the retinal loci for each. Method: Subjects were imaged with the UC Berkeley AOSLO over a 2 degree field centered on the fovea. Stimuli were 10 arcmin square targets made by modulating the scanning laser. For eye movements, the target jumped randomly (saccades), followed a circular trajectory (pursuits) or remained stationary (fixation) in the raster. Subjects were instructed to hold gaze on the target as accurately as possible. For visual direction judgments, the target was flashed for one frame subjects indicated whether the target was above or below and left or right of their point of fixation. Videos were analyzed to extract the movement of the eye and to register all images to a common reference. Results: Pursuit and fixation precision was similar, with a standard deviation of about 5 arcmin and largely overlapping distributions of retinal loci. Target landing points following small saccades were more variable, and showed significant offsets compared to pursuit and fixation loci. Judgments of visual direction were typically twice as noisy as fixation, and in one amblyopic subject were three to four times as noisy. Discussion: There is no single cone on which a target must fall to be “looked at” but subjects are able to make absolute retinal position judgments with a precision of around 10 arc min. The various eye movement systems are as good or better than these judgments, and do not appear to share the same reference point as perceived visual direction.
Supported by the NSF Center for Adaptive Optics.