Abstract
INTRO
Features play a fundamental role in motion perception because, unlike contour segments, their signal is unambiguous for rigid objecs. But what happens when two features signal different motions?
METHODS
Three invisible circular apertures (1.6° diameter) were separated by 1.2°. The outer two contained line terminators, the centre a featureless contour. The orientation of the line was vertical, and the motion of the central segment horizontal. In the first condition, the direction of one of the terminators was horizontal and that of the second terminator differed from this by 0°, ±10°, ±25°, ±45°. In a second condition, the direction of one terminator was fixed at either + or −45° to horizontal, while the other varied within ±45°. Different configurations are consistent with a physically shrinking, expanding or rigidly translating line. The perceived direction of motion of the centre segment was measured.
RESULTS
Vector averaging (VA) of the two terminators' velocities provides a reasonable overall fit to our data, but there are a few notable exceptions. When one of the terminators is moving in the same direction as the central segment, perception is biased away from VA, towards the second terminator's direction, if its motion is very different (45°) from that of the line segment. In contrast, if this terminator's motion is close to that of the line segment, perception is in the opposite direction: towards the motion of the central segment. When both terminators move in opposite directions from horizontal, their VA has to differ from the central segment's motion by more than ±10° before an effect is seen.
CONCLUSIONS
Our results suggest that the visual system typically implements a vector averaging solution for non-rigidly translating objects, but, depending on stimulus configuration, tends towards the motion of a featureless line segment or towards the motion of one of the features.
This research is supported in part by EPSRC Grant No. GR/S59239/01.