Item

Abstract

During everyday viewing we rarely keep our eyes still. Our visual system has to take these eye movements into account in order to create a veridical percept of object motion. When we make smooth pursuit eye movements, the perceived velocity of a moving object can be obtained by summing two signals, one estimating retinal image velocity and the other estimating eye velocity. Previous studies have shown that the gains of these two signals differ. Here we investigate whether they also differ in their latencies. Observers compared the peak velocity of sinusoidally moving dot patterns viewed during sinusoidal smooth pursuit eye movements and during fixation. The relative gains and phases of the two signals were estimated from the amplitude matches by fitting a simple linear model. At VSS2005, we showed that the model described the data well for most observers, but the estimated signal gains and phases showed considerable variability. Also, the gain ratio was very low for most observers, suggesting they ignored eye-velocity information and judged instead the relative motion in the display. Here, we tested whether removing the vertical edges in the stimulus window, using a large-field cylindrical screen, promoted head-centred judgements. Using this display, observers seem more able to judge head-centred sinusoidal motion consistently during smooth eye pursuit. Relative signal gain was comparable to that previously reported in the literature. Moreover, the results suggest that retinal motion signals lag eye-movement signals by a small amount.