Theoretical Approach
The first is the notion of transfer of knowledge from one situation (the virtual environment)
to another (the real environment). Admittedly most psychological studies of transfer of
information have examined transfer of mathematical or verbal information, rather than
information about an environment, but nevertheless we see this work as highly relevant. Based
on about fifteen years of work , Holyoak and his colleagues have concluded that transfer
depends upon the development of a situational representation that is common both to the
original situation (the base) and the new situation (the target) (Holyoak & Spellman, 1993).
It follows, therefore, that if you want to predict whether or not a person has been trained to
the point at which transfer will be successful you should test that person to see if his or
her representation of the base situation is (a) in a form that potentially could be
transferred to the target and (b) is associated with cues that will be present when the target
situation is encountered.
This brings us to our second body of psychological knowledge; how do people represent real
environments? This has been the topic of extensive study. Siegel and White (1975) presented a
typology of representations that, with some modifications, is still quite useful (cf.
Gallistel, 1990, with respect to representations used by non-human animals).Siegel and White
state that when people first experience an environment they identify prominent landmarks. The
next step is to learn procedures (essentially turns and traverses) that are required to move
along specific routes from one location to another. This is called a procedural
representation. So long as people restrict their movements to familiar paths between set
points procedural representations are all that they really need to get along. However, with
experience, some people develop surveyor's representations of the space about them. These are
analogous to a map-like representation of the environment, although with certain clearly
defined distortions( Huttenlocher, Newcombe, & Sandberg, 1994 ;Tversky ,1981 ).
The study of these distortions and biases in people’s cognitive representation of a
large-scale space is the third body of literature which is directly relevant to our efforts.
It is known, for example, that people tend to remember object locations as being more aligned
than they actually are (Tversky, 1981). People also exhibit systematic errors based on
partitioning a space into subregions. For instance, no one has a surveyor's representation of
the entire city in which they live; instead, they have representations of neighborhoods and a
map-like representation of the relationships between the neighborhoods. Distances between
points in separate neighborhoods are usually overestimated (McNamara, Hardy, & Hirtle 1989),
apparently because of their semantic connection with the neighborhood. Somewhat surprisingly,
partitioning is also applied to reasonably small spaces, without landmarks, such as a sandbox
that can be encompassed in a single glance (Huttenlocher & Engberetson, 1995). We are
interested in how these biases and heuristics manifest themselves in spatial layouts learned
in virtual environments.
Developing surveyor level representations from real explorations is a difficult task. As Infield noted, there are very wide individual differences in related tasks. Individual differences are also noticed in actual environmental studies, outside of the laboratory. Moeser (1988) found that some people can have as much as two years daily experience with a building without developing a surveyor's representation of it! Lindberg & Garling (1983) conclude that this is because the development of a surveyor's representation requires a conscious, cognitive effort. It follows that some people, if they are satisfied with their procedural knowledge of an area, simply never take the trouble to develop surveyor's representations if a more easily acquired procedural representation is sufficient to let them go about their daily business.