Abstract
Purpose. Recent work on visual and motor extrapolation of curved contours indicates that increased uncertainty due to prolonged occlusion leads to “flattening” - the curvatures are systematically underestimated (Singh & Fulvio, 2005, 2007; Mrotek & Soechting, 2007). We wondered whether other sources of uncertainty would have similar effects. We investigated how sampling the trajectories of moving objects affects observers' extrapolation through occluded regions for several trajectory curvatures.
Methods. A dot traveled at constant speed across a display and was occluded by a half-disk. Subjects judged whether the dot would reemerge above or below a variable “tick mark” located at the opposite, curved edge of the occluder, responding by keypress. During training, the dot traveled along straight trajectories. In the experimental session, subjects extrapolated lines and four circular-arc trajectories varying in curvature. Trajectory sampling ranged from continuous to less than 50% visible. We used an adaptive staircase to estimate the extrapolation PSE for each of 5 (curvature) by 4 (sampling) conditions. Three observers participated in the task.
Results. Across subjects, line path extrapolation accuracy was high during training with no significant effect of sampling (0.59–3.19 deg bias across conditions). In the curves session, the subjects' estimated points of emergence were biased in the direction consistent with “flattening”. Path curvature had a significant effect on bias (pConclusions. Consistent with previous work, increased curvature of dynamic trajectories leads to increased “flattening” in extrapolation. Variation in sampling rate across the range considered had no discernible effect.
NSF Graduate Research Fellowship (JMF) ONR N 00014-07-1-0937 & NIH EY015261 (PRS) NIH EY02668 (LTM).