Estimating the time to collision with a rotating nonspherical object.

We measured the precision (i.e. Weber fraction) and the accuracy with which the time to collision (TTC) with a simulated approaching object is estimated. We simulated a rigid spherical object and a rigid nonspherical object (an oblate spheroid whose longer axis was vertical). We used the following viewing conditions: that the available information about TTC was only monocular (M), only binocular (B) and monocular plus binocular (M and B). In addition to the approaching SPHERE condition, we used the following three simulation conditions for the oblate spheroid: (2) a slow rotation through 90 degrees (SIDE-END); (3) a slow rotation through 90 degrees (END-SIDE); (4) several complete rapid rotations (RROT). Weber fractions for discriminating TTC were similar for all 12 combinations of viewing and simulation conditions. When only monocular information was available, perceived TTC was longer in the M/SIDE-END condition than in the M/SPHERE condition, and shorter in the M/END-SIDE condition than in the M/SPHERE condition. As well, observers were strongly influenced by task-irrelevent variables in the M/SIDE-END condition so that they could not properly perform the task. The addition of binocular information considerably improved the accuracy of estimating TTC in simulation conditions SPHERE, END-SIDE and RROT, and allowed reliably accurate estimations to be made in the SIDE-END simulation condition. We suggest that, when binocular information is combined with monocular information about TTC, the two kinds of information are weighted roughly equally when the approaching object is a rigid sphere, but the binocular information is weighted more heavily when the approaching object is nonspherical and rotating.