Abstract
Previous research (Hahn, Andersen, and Saidpour, 2003) suggest that in addition to optical flow subjects can use static scene information for the perception of heading. In the present study we investigated encoding limitations in recovering information for estimating heading. Observers judged the direction of movement (left or right) using two sequentially presented static scenes that varied in interstimulus interval (ISI) to simulate the presence (50 ms ISI) or absence (1000 ms ISI) of apparent motion. The first frame was presented with 12 objects of varying shapes positioned on the ground, followed by a second frame simulating forward motion to the left or to the right of a vertical bar. To examine encoding limitations a subset of the objects (3, 6, or 9 objects) on the second frame were omitted. The heading directions were ± 5, 11, 17, and 23 deg from the center of the screen. Observers determined their heading direction relative to a reference object (a vertical pole at 8 deg from center). The results indicated that heading accuracy increased with an increase in the heading angle with greater accuracy occurring when apparent motion was present as compared to when it was absent. Heading accuracy decreased as more objects were omitted on the second frame. Surprisingly, the decreased performance for omitted objects occurred regardless of whether apparent motion was present. These results suggest that the visual system does not encode all information in a sparse scene for estimating heading. The importance of these results to models of optical flow will be discussed.
Supported by NSF BCS 0001758 and NIA grant 2R01AG013419-06A2.