Abstract
A person moving through the world must be able to detect and identify moving objects to avoid collisions. In previous studies we examined how well people can detect a moving object within a radial optic flow field simulating observer motion. In this study, we examined the problem of moving object identification by a moving observer. Observers viewed a radial flow field, consistent with observer motion in a straight line. In every trial a target circle moved at an angular deviation from this radial pattern. Observers used the cursor to select the circle that they believed moved differently from the pattern. We measured the percentage of correct responses for angles of deviation between 8 and 40 degrees and for trial durations of .25, .5, .75 and 1 sec. We tested scenes containing 4, 9, 16, or 25 circles. The results for 6 observers show that as the trial duration increased the thresholds for identifying the target tended to decrease. Observers had nearly the same accuracy for 9, 16 or 25 circles, with average thresholds of about 12.5, 11.6 and 11.7 deg, respectively, for a 1 sec duration. These are very similar to the thresholds for simple object detection, which for 25 circles was 13.6 deg for a 1 sec duration (Royden, Connors & Mahoney, 2005). Observers showed much lower accuracy with 4 circles, with average threshold of 36.5 deg for 1 sec duration. This decrease in accuracy is similar to that seen with simple object detection. These results suggest that if an observer can detect that some object is moving within the flow field, then they can identify specifically which object it is.
Supported by NSF grant #0343825