Human Place Learning in a Computer Generated Arena

We describe the development of a computer-generated arena within which one can study human place learning by asking subjects to locate an invisible target. A series of studies demonstrate that such learning is based on acquiring knowledge about the spatial relations among the distal cues presented in this arena. We show that (1) the presence of proximal cues does not prevent learning about the distal cues; (2) the removal of individual cues does not impair performance until all distal cues are removed; and (3) the re-arrangement of distal cues profoundly impairs performance. In further studies we demonstrate that learning can proceed even when the subject is placed on the target rather than having to navigate to it, and even if the subject merely watches someone else navigate to the target. Finally, we demonstrate that learning is impaired by traumatic brain injury, and that aged subjects do not perform as well as young adults. This paradigm should prove useful in investigations of spatial cognition in general, and the role of specific neural systems inparticular.