Abstract
When an observer moves through a scene, both stationary and moving objects project moving images onto the retina. For observer motion in a straight line, the image motion for stationary objects forms a radial pattern. How do moving observers detect a moving object in the scene? In theory, observers could identify objects whose speed or direction of motion differs significantly from that of the other objects in the scene. We tested how large an angular deviation is needed to detect a moving object in a radial optic flow field. We also examined observers' abilities to detect an object moving at an angular deviation from a deformation field, to test whether global pattern information contributes to this detection process. In each trial, observers viewed a scene that consisted of 25 white circles moving on a black background in either a radial or deformation flow pattern. In half the trials, one of the circles, the target, moved with an angular deviation from the radial or deformation pattern. Observers pressed a key to indicate whether or not this target circle was present in a trial. We measured the percentage of correct responses for angular deviations of 10, 15, 20, 25, 30 and 45 deg, and trial durations of 0.25, 0.5, 0.75 and 1.0 sec. For the radial pattern, the average threshold (75% correct) for the 9 observers tested was 20.7 deg for the 0.25 sec duration and decreased to 13.6 deg for the 1.0 sec duration. For the deformation pattern, the thresholds were much higher, at 35.4 deg for the 0.25 sec duration and decreasing to 29.0 deg for the 1.0 sec duration. Because the local rate of change of motion direction is the same for the radial pattern and the deformation pattern, this result implies that observers make use of global pattern information in addition to local motion cues when detecting a moving object within a radial optic flow field.
Supported by NSF grant #0196068