Abstract
I have recently reported a positive correlation between detection threshold for unreferenced motion and fixation instability (Murakami, in press), arguing that retinal image motions due to small eye movements can lower motion sensitivity. However, the analysis was based only on monocular data because stimuli were presented at only one location, partially covering the blind spot of one eye. Thus, the present study examined correlations between motion perception and eye movement in binocular data, by presenting the stimulus at various locations and by identifying the dominant eye for each observer. On a gray background, a random-dot field translated coherently for 0.85 s in one of eight radial directions. It was presented at one of six locations (left, left-down, down, right-down, and right at 8.5 deg eccentricity, and the center). For the center presentation, the fixation spot was either visible or not. Detection threshold speed was estimated from each observer's direction identification rates for each location. In a separate session, each observer's eye movements were recorded with an eye tracker, and the s.d. of instantaneous velocity during fixation was taken as an index of instability. Results revealed a positive inter-observer correlation between threshold and instability for all locations. Correlation coefficients did not change whether or not there was a fixation spot. Furthermore, unexpectedly, the correlation between threshold and the dominant eye's instability (r = 0.55−0.64) was much higher than between threshold and the other eye's instability (r = 0.26−0.39). These results provide more extensive evidence of the deteriorating effect of fixation instability on unreferenced-motion detection. In the current task, a fixation spot does not function as an effective reference frame. Visual performance near threshold presumably depends more on the inputs from the dominant eye, but at the same time, suffers more from velocity noise created by this eye's movements.