This study compares aspects of spatial perception in a physical environment and its virtual representations in a CAVE and Panorama, derived from recent research. To measure accuracy of spatial perception, participants in an experiment were asked to look at identical objects in the three environments and then locate them and identify their shape on scaled drawings. Results are presented together with statistical analysis. In a discussion of the results, the paper addresses the two hypothetical assertions - that depth perception in physical reality and its virtual representations in CAVE and Panorama are quantifiably different, and that differences are attributable to prior contextual experience of the viewer. The role of prior or tacit knowledge in these environments is established from the empirical data. It is concluded that the CAVE offers a higher potential for spatial experience and learning than the Panorama. The results also suggests that knowledge gained in physical contexts is more readily transferred to its virtual simulation, while that gained in virtual experience is not reliably transferred to its equivalent physical context. The paper discusses implications for spatial ability, learning and training in virtual environments, in architectural education, and participatory design processes, in which the dialogue between real and imagined space may take place in virtual reality techniques.