Abstract
Mislocalisation phenomena (e.g. the Fröhlich, flash-lag, and representational momentum illusions) have sparked a debate into what mechanisms underlie the perceived position of moving objects. Proposed models include temporal integration and spatial extrapolation of positions. Findings showing that abruptly disappearing moving objects do not ‘overshoot’ have been used to argue against spatial extrapolation. We recently proposed that abrupt offsets lead to competing neural representations: one bearing an extrapolated position due to a cortical internal model, the other based on transient retinal signals (Desimone, 1998; Keysers & Perrett 2002; Maus & Nijhawan, 2006). Here we investigate a new effect predicted by this ‘competition model’.
Two bars, one directly above the other, moved horizontally across the screen. One bar disappeared abruptly (offset-bar), while the other bar moved on. Observers (n=6) judged the position of the continuously moving bar in relation to the position in which the offset-bar vanished (offset-position). Methods of adjustment and constant stimuli (2AFC) were used. As predicted, observers perceived the continuously moving bar as being ahead of the offset-position. In a control experiment we confirmed that the offset-position is perceived accurately (with a small undershoot). When the offset-bar reappeared after various short time intervals, intermediate positions were perceived.
A simple computational model of temporal averaging did not reproduce these results. Our findings are consistent with the notion that a moving object's perceived position is based on competing neural representations. We argue that the perceived position of continuously moving objects is based on cortical internal models, which spatially extrapolate positions to compensate for temporal delays in the visual pathway. However, abrupt offsets elicit strong retinal transients, carrying accurate position information, that usually win the competition for perception.