Specificity, transfer, and retention of spatial knowledge from navigation using maps, virtual, and real environments

The present research examined the degree to which spatial knowledge is subject to specificity of alignment and to the particular geometry, or layout, of a building. Specificity has been traditionally defined by starting alignment (e.g. Richardson, Montello, & Hegarty, 1999). Dynamic alignment instead captures the temporal aspect of navigation and is defined by the direction of travel along a particular path through an environment. A series of three experiments examined the effect of training participants using maps, virtual, or real environments to navigate along a route, or way find among landmarks, and learn the names and locations of landmarks. Dependent measures assessed both route knowledge and survey knowledge. After training, participants were tested in the real environment either immediately (Experiment 1) or after a two-week retention period (Experiment 2). For immediate and retention testing, performance after VR training was similar to that after real-world training, and was quicker and transferred better than map training to navigation in the actual world building. However, this was true only when the testing route was traveled in the trained direction of travel, or was dynamically aligned. In Experiment 3, participants were trained on a single route or three routes using a map or virtual reality (VR) model, half of which differed structurally from the actual building by altering the navigational affordances (e.g. locations of walls, doors, and stairs). Participants were subsequently tested using a way finding paradigm. When trained in a modified building and tested in the actual building, VR participants demonstrated attenuated specificity for route measures (navigation) after three-route training, suggesting that the hyper-specificity of VR training can be reduced by increasing participants' functional field of view (FOV). A final pilot experiment comparing training using normal and reduced FOV in the actual building supported this finding.