Abstract
Path integration is a navigation process by which moving animals integrate self-movement vectors using internal and/or external cues to estimate their current position and orientation relative to the origin. We used hallway-mazes in an immersive virtual reality Cube to examine whether the presence of landmarks can facilitate path completion tasks. In our maze, position changes (driving along the hallway) were purely visual, controlled by button presses. Orientation changes (turning at the intersection between two hallway segments) were controlled by physical turning along with optical flow information. Participants traveled along random paths with five segments, either with or without distinctive landmarks at segment intersections. Then they were instructed to return to the origin directly from the end of the path. Theoretically, the existence of the landmarks may improve path integration performance by providing salient anchors at the critical points of the path structure and therefore facilitate the construction of the pathway representation. However, landmarks may also introduce a cost because landmarks require the navigator to encode, remember and update more locations (both the origin and the landmark locations), and consequently increase working memory load. We found no landmark advantage for homing performance. In order to encourage participants to pay attention to the landmarks, a second experiment required participants to return to one of the landmarks in a subset of the landmark-present trials. The data again showed no evidence of improvement in homing performance with the landmarks than without. Instead, homing performance was slower in the landmark-present condition than in the no-landmark condition, possibly due to increased working memory load and/or task complexity.
Supported by NSF Grant BCS 03-17681 to R. F. Wang.