Spatial Cues in 3D Visualization - Ambient Intelligence for Scientific Discovery - page 123

Table 1. Summary of gender-based differences in the experimental performance mea-
sures. We observed definite gender biases for performing particular spatial tasks that
are largely consistent with existing literature. In our experiments, males excelled at
tasks involving mental rotation (i.e. object matching) and object movement (i.e. ob-
ject positioning), whereas females were more accurate resizing objects that were fixed
in location.

Object Task: Accuracy/Gender: RT/Gender: Rot. Dist./Gender:

Matching Males more accurate Males faster Not Applicable

Positioning Males more accurate No M/F difference Males use more

Resizing Females more accurate No M/F difference Males use more

more use of the rotational distance feature in both the positioning and resizing
tasks, evidently finding the ability to rotate the objects more useful than did the
females in attempting to position and resize the objects.
How do these findings suggest approaches to develop 'gender neutral' vi-
sual interfaces? One suggestion is to add meaningful landmarks and to diminish
the viewer's reliance on 'spatial presence' and, particularly, on mental rotation
ability. Visual interfaces that enable the stationary viewer to 'look right' or
'look left' (or up, down, backward, etc.) for familiar landmarks may be an effec-
tive, gender-neutral alternative to the use of typical motion cues, such as 'flying
around' virtual spaces, to extract information from a scene. The reliance on mo-
tion cues, in particular, to extract information from the visual features of objects
presented in a scene is especially problematic.

5 Conclusions and Discussion

What do these cumulative findings tell us about the relative ecacy of visual
cues in extracting spatial information from visual objects and features presented
in computer-generated scenery? Which depth cues are most important in this
regard? How can we make use of these findings so as to better enable the per-
ceptual discovery of information embedded and represented in 2D and 3D visual
scenery and user interfaces?
With respect to extracting spatial information, especially regarding the rela-
tive depths and placements of objects, stereoscopic viewing was the most power-
ful and useful depth cue among those examined in these particular tasks. Motion
cues are also useful in this regard, although there is evidence that motion is more
useful for male, as compared to female, viewers. Moreover, it is clear that the
relative ecacy of the individual depth cues tested are tied to the nature of the
task. For example, the cast shadows were very useful for positioning objects, but
not for resizing objects. So it is problematic to state that any one particular cue
will be universally useful in all visual tasks. The nature of the task itself must
also be considered.
Additional implications for the design of 3D user interfaces, particularly with
respect to applications that involve positioning and resizing objects, include: (1)
stereoscopic viewing is a powerful depth cue, superior to monoscopic viewing,