Abstract
Our basic research interest is contributions of sensory inputs for path-integration-type navigation in our day-to-day life, for example, the navigation in a town of narrow streets or hallways inside buildings. While optic flow has been assumed to be the one and only visual input used for path integration as originally defined, sequences of views including information about upcoming path (i.e. the edges of streets and layouts of buildings or walls) would be also used for the navigation (Loomis, Klatzky, Golledge, & Philbeck, 1999). The present study explored contributions of two visual inputs in homing accuracy by a path completion task using visual stimulus with the optic flow and the sequences of views (Video), or only the latter one (Image). Participants remained seated and watched two videos that showed equivalent views from a person traveling inside a maze along paths with 3 and 7 turns & legs (Path3 and Path7), or successive still images that were processed from the videos. Right after the videos or the images ended, they were asked to indicate starting points. The judgments were recorded as directional data and analyzed by some tests for randomness to infer homing accuracy. In Path3, the data of both conditions were clustered around a correct angle rather accurately, and there was not a marked difference in homing accuracy between Video and Image. Compared to Path3, the judgments of Path7 tended to be inaccurate especially in Image. Whereas the data of Video were still considered to cluster around a correct angle, the data of Image were completely random. These results suggest that in a simple path, the sequences of views are independently sufficient for homing. On the other hand, in a complex path containing more turns and legs, the optic flow added to the sequences of views significantly improves the homing accuracy.