Abstract
Previously, we created a non-Euclidean hedge maze that contained two "wormholes," which teleported walking participants between locations by covertly rotating the virtual maze. Participants were unaware of this radical violation of Euclidean structure and took shortcuts to the new wormhole locations of target objects, indicating "rips" and "folds" in spatial knowledge. To test whether additional orientation cues would reveal the maze rotation and preserved Euclidean structure, we subsequently added external landmarks. However, shortcut performance was unaffected. In the present experiment, we provide further orientation cues by adding directional cast shadows in the maze. Participants learned the locations of nine objects (places) while freely exploring the maze in three conditions: (1) no external landmarks or cast shadows, (2) external landmarks and cast shadows rotate with the maze, and (3) external landmarks and cast shadows remain fixed with respect to the laboratory. We then probed their spatial knowledge using a shortcut task. Despite the addition of cast shadows to the landmark cues, on probe trials participants still take shortcuts in the direction of wormhole targets rather than Euclidean target locations, consistent with our previous results. Spatial knowledge is consistent with a topological graph of the environment, supplemented with some coarse metric information, but does not appear to be integrated into a globally consistent Euclidean map.
Meeting abstract presented at VSS 2012