hide
Free keywords:
-
Abstract:
The systematic investigation of spatial behavior and cognition requires generic description systems
for spatial form and configuration that capture behaviorally relevant properties of environments.
Isovists (i.e. viewshed polygons describing the the visual space from single observation
points, cf. [1]) have always been conceived as closely related to spatial behavior and, indeed,
case studies found correlations between isovist measurands and large-scale statistical pedestrian
dispersal [2] as well as navigation performance [3]. The relations between visuo-spatial
properties and small-scale locomotion behavior, however, are widely unknown. The study
presented here quantitatively explored correlations between isovist derivatives and behavior in
architectural environments at the level of trajectories.
In a navigation experiment, 16 participants were asked to find the place offering the best
overview in each of 16 virtual indoor spaces. Subjects were advised to approach this position
as accurately, directly, and quickly as possible. During this, their positional and orientational
trajectories were recorded. Participants experienced the virtual environments from an egocentric
perspective on a standard CRT offering a simulated field of view of 90x63 degrees; they
interacted with the simulation using a customary joypad.
The analysis tested for subjects’ performance with respect for finding the best overview
place. Mean characteristic values derived from the individual trajectories were compared to
task performance as well as to global isovist measurands of the corresponding virtual indoor
scenes. Several strong correlations were found. For example, subjects’ performance was significantly
correlated with the global isovist measurand ‘jaggedness’, describing aspects of the
shape complexity of spatial situations (r=-.66, p<.01). On the level of trajectories, jaggedness
was negatively correlated with average angular velocity during locomotion, i.e. subjects tended
to turn more slowly in more complex environments (r=-.72, p<.01). Furthermore, in scenes
featuring a high variance in visual stability, subjects showed a decreased tendency to stop and to
reorient when solving the navigation task (r=-.55, p<.05). Correspondingly, a similar negative
correlation with overall performance was found (r=-.56, p<.05).
Altogether, the results suggest that isovists and their derivatives have predictive power not
only for overall performance in the task tested, but also for spatial behavior at the level of
trajectories, and therefore promise novel insights into the perceptual basis of locomotion.