Abstract
With each saccade, visual objects cause motion streaks across the retina in the direction opposite of the movement. In the absence of pre- and post-saccadic stimulation, these intra-saccadic motion streaks are often visible. The investigation of their phenomenology provides the opportunity to assess how the visual system integrates retinal and eye position information to achieve world-centered perception as the eyes move. Participants (N=10) made horizontal saccades (16 degree of visual angle, dva), as we used a high-speed projection system (1440 frames per second) to present rapid, continuous, intra-saccadic stimulus motion: Starting at one of nine different locations on the screen, a low spatial frequency noise patch moved in one of 12 possible directions, traveling 3, 4.5, or 6 dva at a velocity of 240 dva/s. After each saccade, participants reported whether they had perceived a motion streak during the movement, and if so, reproduced its path on the screen using a computer mouse. Motion streaks were detected in 71% of all trials. Detection improved with lower retinal velocities and when the noise stimulus (on a given trial) happened to contain orientation information parallel to its retinal trajectory, thereby producing more pronounced streaks. Localizations in space were also best explained by the stimulus' retinal rather than world-centered position. In contrast, the perceived trajectory of motion—including its direction, length, and overall path—was more similar to the stimulus' world-centered path. Moreover, motion streaks that extended further in time and space were more likely to be reported in world-centered coordinates. Thus, while intra-saccadic perception emanates from processing in retinal coordinates, it is readily available in world-centered coordinates, suggesting the visual system has access to a high-fidelity representation about the position of the eyes even during saccades. Our results also provide evidence that motion streaks might well play a role in trans-saccadic vision.
Meeting abstract presented at VSS 2018