Piloting and path integration within and across boundaries.

Three experiments investigated whether navigation is less efficient across boundaries than within boundaries. In an immersive virtual environment, participants learned objects' locations in a large room or a small room. Participants then pointed to the objects' original locations after physically walking a circuitous path without vision. For participants who learned the objects in the large room, the testing position and the learning position were in the same room so that participants did not cross boundaries before testing; for participants who learned the objects in the small room, the testing position and the learning position were in 2 different rooms so that participants crossed boundaries before testing. Participants who learned the objects in the large room, during testing, either saw cues indicating the targets' locations (piloting group) or not (path integration group). Participants who learned the objects in the small room, during testing did not see any cues correctly indicating the targets' locations. The results showed that pointing accuracy was higher for those who learned the objects in the large room and in the piloting group than for those who learned the objects in the small room. However, this cross-boundary cost did not occur when we contrasted participants who learned objects in the large room and in the path integration group with participants who learned in a small room. These results suggested that navigation that relies on path integration only is not sensitive to boundary crossing, although navigation that relies on piloting is less efficient across boundaries than within boundaries.