Exploring the Benefits of Immersion in Abstract

Information Visualization
Dheva Raja, Doug A. Bowman, John Lucas, Chris North

Virginia Tech
Department of Computer Science
Blacksburg, VA 24061
{draja, bowman, jolucas, chnorth1}@vt.edu

ABSTRACT Head tracking is the measurement of the user’s head

The benefits of immersion with regard to information position and orientation, which is then used to render the
visualization applications have rarely been explored. In this world from the user’s point of view in space. It provides an
paper, we describe a user study designed to explore a intuitive method of viewing from various perspectives in
variety of information visualization tasks in immersive and the VE. Taken together, we call these two characteristics
non-immersive 3D scatterplots. In the non-immersive immersion.
version the information was displayed using only one wall Immersion has been shown to be beneficial in applications
of the CAVE, while the immersive version used all four where spatial knowledge of an environment is useful. In an
walls. We also examined the effects of head tracking, information visualization application, physical immersion
giving a total of four conditions: four walls in the CAVE and head tracking could allow for more efficient
with head tracking, four walls without head tracking, one identification of trends in data, greater spatial
wall with head tracking and finally, one wall without head understanding of the entire data set, and easier
tracking. By separating the variables in this way, we can identification of single data points.. The CAVE [Cruz-
independently evaluate the effects of immersion and head Neira93] is unique in that it allows data to be seen in a wide
tracking. In general, we found the fully immersive field-of-view without moving one’s head and without
condition, (four walls with head tracking) to be most useful distorting the spatial relation of the data. By contrast, in a
in viewing the datasets and performing the tasks. head mounted display (HMD), the user is still immersed
Keywords but does not benefit from the extent of the peripheral vision
Immersion, Head tracking, CAVE, Virtual Environment, that one does in the CAVE.
Virtual Reality, Information Visualization, 3D Scatterplots Our goal is to understand the benefits of immersion for
1. INTRODUCTION information visualization. If immersion is proven to be
Numerous visualization applications have been developed beneficial, immersive VEs could become more popular for
for use in immersive virtual environments (VEs). These analyzing 3D data sets. Our results in this paper are limited
primarily are comprised of, but not limited to, scientific to 3D scatterplots, and thus are not necessarily applicable to
visualizations, architectural walkthroughs, and simulations all types of 3D data visualizations, but further studies could
of various devices in commercial and industrial enterprises. show immersion’s usefulness. Other 3D data visualization
As one might imagine, not all applications have an techniques that could benefit from immersion are: multi-
industrial or scientific purpose for their origin. Some quite dimensional function visualizations, 3D histograms, line
stunning artistic demonstrations, as well as quite a few graphs and surface maps.
games have been developed or ported for use in VEs. We have developed a CAVE-based information
Physical immersion and head tracking are the primary visualization application (CaveDataView), and conducted
characteristics that make these applications so compelling. an exploratory pilot study in which we evaluate the benefits
Physical immersion is the degree to which the virtual world of immersion for 3D scatter plots. The goal of this research
surrounds the user in space. A VE where the user feels a was to show that there are tangible benefits to viewing a
strong sense of “presence” generally has a high degree of generic information visualization application in a
physical immersion with a wide field of view. physically immersive, head-tracked environment such as
the CAVE.
________________________________________________
To be specific, our hypotheses were:
1. A high degree of physical immersion will allow
higher levels of task performance and greater
user satisfaction when visualizing datasets
represented by 3D scatterplots.
 2. Head tracking will allow higher levels of task The precursor to this research was the Wizard application
performance and greater user satisfaction [Datey01], an HMD-based immersive VE for exploring 3D
when visualizing datasets represented by 3D scatterplots. Wizard provided both a small hand-held
scatterplots. overview of the dataset and a larger version of the data
We begin with a survey of relevant related work, followed through which the user could navigate.
by a description of our immersive information visualization 2.3 Comparative Studies
application. Next we describe the design of our evaluation In the most closely related work to our study, a desktop
and its results. Finally, we conclude and present our ideas visualization toolset named XGobi was compared with a
for future work. C2 virtual environment system [Arns98]. Similar datasets
2. RELATED WORK were displayed on both devices, and a series of user tasks
were performed in order to see if there were any tangible
2.1 3D Information Visualization Techniques
3D information visualizations take a complex and abstract benefits to viewing datasets in an immersive environment.
dataset and organize it into a 3D visual representation, The authors hypothesized that viewing high-dimensional
which can be navigated and accessed by the user. Abstract statistical data would be more efficient in an immersive
properties of the data are mapped into perceptual qualities, VE. The authors found that some tasks requiring a good
such as position, orientation, size, shape, color, or motion, deal of spatial understanding were performed more quickly
and relationships between pieces of data are represented in the VE than on the desktop.
spatially. The resulting visualization can reveal patterns in Wickens [Wickens95] compared conventional 2D graphs
the data that may not be obvious from the original dataset. with 3D graphs for presenting 3D data. Users were asked
A number of 3D information visualization tools have been questions about the data that ranged from focused attention
developed. Dataspace [Anupam95] is a system for on a single data point, to questions that
interactive 3D visualization of large databases. IVEE integrated the entire data space. Here the authors found that
[Ahlberg95] is a 3D environment that uses a number of the 3D display resulted in the longest times for focused
techniques such as maps, star fields and query mechanisms attention tasks, but this diminished markedly as the
for visualizing a database. Work from Xerox PARC questions became more integrative in nature.
[Robertson93] provides additional examples of the use of
interactive 3D graphics for information visualization. The best example of a controlled experiment attempting to
Spotfire [Spotfire] is a commercial desktop application that quantify the benefits of immersion is the work done by
allows users to load in a dataset of their choice, and Pausch and his colleagues [Pausch97]. In this work, a
visualize it using 2D and 3D graphs. 3D scatterplots in comparison was made between an HMD with head tracking
Spotfire are very similar to the plots we display in and a stationary HMD with hand input for navigation and
CaveDataView. viewpoint control. They hypothesized that users would be
able to find a target faster in the head tracked condition, but
2.2 Information Visualization Applications in VEs did not find this to be the case. However, the head tracked
There are not many current applications for information HMD users were able to determine if a target was not
visualization in immersive VEs, and very few specifically located in an environment significantly more quickly. This
intended for use in the CAVE. One notable exception is the suggests perhaps, that the head tracked subjects built a
C2 statistical program, which was loosely based on a cognitive map of the space more quickly, and avoided
desktop tool called XGobi. [Arns98] redundant searching
VR Vibe [Hollands95] is an HMD-based application, 3. SYSTEM DESIGN AND DESCRIPTION
which creates a visualization of bibliographies for 3.1 CaveDataView
information retrieval. Users specify keywords in 3D space, CaveDataView is the VE application we developed to test
and representations of the documents are then displayed in our hypotheses. It can display scatterplots of 3D datasets in
the space according to how relevant each document is to a CAVE (figure 1). We developed the application in C++
each of the keywords. The position of a document depends using the DIVERSE application programming interface
on the relative importance of each of the keywords to it, (API) [Kelso02]. DIVERSE allowed us to create an
which is computed using document-matching algorithms. application that would work, with little to no modification,
The LEADS system developed at University of Nottingham on a CAVE, and HMD or a desktop.
[Ingram95] applies concepts based on urban planning for DPF, the Diverse interface to OpenGL Performer, was used
database visualization and abstract information domains. for the generation of the scene graphics. DPF and
The system uses a city metaphor based on districts; nodes Performer were powerful enough to allow us to render
and edges connected by paths and landmarks facilitate scenes with large numbers of data points. We tested our
formation of cognitive maps to avoid getting ‘lost’ in the application under DPF 2.3.1 which is available for both the
information space. The LEADS system shows how using Irix and Linux platforms. To interface with the tracking
an easily recognized metaphor simplifies information hardware, we used the Diverse Toolkit (DTK). DTK
visualization using immersive VEs. encapsulates the tracking system and places the needed
information into shared memory making it easily available tracking to display an immersive VE. The CAVE uses an
to the application. Diverse uses a Dynamic Shared Object Intersense IS900 VET tracking system that comes with a 6-
structure that allows application modules to be quickly DOF Head Tracker and a 6-DOF Wand control device. The
added or removed. This structure was invaluable in that it head tracker tracks the user’s movement through the
allowed us to develop on desktop computers and move environment and causes the scene to render the correct
relatively seamlessly to a CAVE display system. perspective for the user’s position and orientation. The
wand device has a small joystick and four buttons, and is
used for navigation and manipulation of the environment.
Wand navigation is used for moving quickly through the
virtual world; for fine movements the user can walk or just
move his head when head tracking is enabled. Figure 3
shows a user standing in the CAVE while CaveDataView is
running.

<FIGURE 1 - A 3D scatterplot in CaveDataView>

CaveDataView reads in tab-delimited files of three column
data. Each line of the file is a distinct data point. Once all
points have been read, the data space is scaled to fit in a
reasonably sized VE. Data points are represented by yellow
cubes laid out on a 3D grid. The grid itself is a 3D object
that stretches from the origin along the three coordinate
axes to a size three times that of the dataset. This is
important so that the user does not get disoriented while
navigating through and around the data. Lighting and <FIGURE 3 – CaveDataView in the CAVE>
shading are added to the scene to make the edges and We realized that in order to test our hypotheses we would
corners of the cubes easier to see. Shading also helps the need to give the user some ability, beyond basic navigation,
user orient themselves to the cubes as each side had to interact with the environment. More specifically, the user
different shading applied to it due to the direction of the would have to be able to point at individual objects of
light. interest to them or that were important to the tasks they
were asked to perform. To do this, we implemented simple
ray-casting [Mine95]. A virtual ray is projected from the tip
of the wand in the direction the wand is pointed. When the
ray intersects a data point, the ray shrinks to the point of
intersection on the side of the data point. This shrinking
provides a valuable visual cue to highlight the data point of
interest. To further highlight the data point of interest, the
color of the cube is changed to red when intersected by the
ray.
Aside from pointing, we required a method for the user to
report verbally which data point they had selected. To do
this each data point was assigned an ID number that had no
relation to its location or any of the attributes of the data.
We first attempted to put this ID number on the side of
each cube, but quickly exceeded the rendering power of the
SGI Onyx that renders the scene. The decision was made to
only label the currently selected cube with its ID number.
<FIGURE 2 – The Virgina Tech CAVE> As this ID number was difficult to read from a distance, we
also placed it at the start of the selection ray above the
Our CAVE (figure 2) is a four walled display, 3 sides and a
wand (see figure 4). This labeling method worked very well
floor, that uses stereo projection technology and head
as when close to the data points the labels on the sides of In an immersive projection display, such as the CAVE,
the cube were very readable, and the wand label was however, we can separate these two variables and examine
readable when selecting cubes from a distance. their effects separately. This has led to our experimental
design, shown in table 1. In the physically immersive
condition, we use all four walls of the CAVE, while in the
physically non-immersive condition, only the front wall is
used. Separately, we can control the use of head tracking
(on or off), leading to four possible combinations. Two of
these conditions are novel as compared to Pausch’s
experiment. In the physically immersive, non-head-tracked
condition, the virtual world surrounds the user, but the user
cannot rotate her head to view different parts of the world,
nor can she translate her head to get a different perspective
on the world. In the physically non-immersive, head-
tracked condition, the virtual world only appears in front of
the user, but she is free to turn or translate her head to see
the world from a different point of view.
Tasks involving different datasets were performed in these
<FIGURE 4 – Ray-casting and data point labeling in four conditions, and the times to complete the tasks were
CaveDataView> measured. Subjects provided difficulty ratings and
disorientation ratings on a seven-point scale after each task
4. EVALUATION was performed in a particular condition. Additionally, at
4.1 Experimental Design the conclusion of the subjects’ series of tasks, a
Quantifying the benefits of immersion is obviously an questionnaire was administered. The results were then
important challenge for the VE community. If attractive gathered and evaluated.
cost/benefit ratios can be proven for particular tasks and
domains, industries will be much more likely to invest in Head Tracking No Head Tracking
VE technology. However, it is also very difficult to design Non- One wall CAVE One wall CAVE, no
controlled experiments to measure these benefits. Previous Immersive Head Tracking
work has taken two approaches.
Immersive Four wall CAVE Four wall CAVE, no
First, some researchers have attempted “practical” Head Tracking
comparisons of immersive and desktop systems (e.g.
[Datey01]), in which the same application is used with an <TABLE 1 –Experimental Conditions>
immersive display such as a head-mounted display (HMD) Because of the exploratory nature of this pilot study, no
and a non-immersive display, such as a desktop monitor. statistical analysis was performed on the data.
The problem with these evaluations is lack of experimental
4.2 Tasks
control. There are many differences between the two
We developed 6 tasks with which to test our hypotheses.
conditions besides immersion, including input device,
The tasks themselves represented typical tasks that would
interaction techniques, resolution, brightness, field of view,
be performed when analyzing a data set. As noted above,
the user’s posture (standing or sitting), etc. Thus, if a
all tasks were timed, although for some tasks completion
difference between the conditions is found, it is not at all
time would not necessarily be the most important factor.
clear that the difference is due to immersion.
Tasks were chosen so that they would have only a single
The second approach is to control these additional factors correct answer.
by using the same display in both conditions. As we have
1. One Axis Distance
already seen, Pausch et al. [Pausch97] used an HMD along
with two ways for the user to rotate the viewpoint: via head One Axis Distance asked the subject to find the point with
turning and via hand turning. This study showed that the the highest Y value. This task tested their ability to judge
head-tracked condition was superior in some ways for a distances along one axis in the scatterplot. We felt this was
search task. We would claim, however, that there was still a an important basic task in that it is a key component in
confound between head tracking and physical immersion in gaining understanding of a single data point.
this study – in the head-tracked condition, the user was 2. Two Axis Distance
physically immersed (the virtual world appeared to
Two Axis Distance required the subject to locate the point
surround him in space), while in the hand turning
with the both the lowest X and lowest Y value. We were
condition, no physical immersion existed. In other words,
concerned about the subjects’ ability to orient themselves
in an HMD-based system, head tracking creates physical
so that they could compare distances along two axes at
immersion, and there is no way to separate the two.
once. This proved to be one of the more difficult tasks.
3. Trend Determination 4.4 Results
This section is organized by exploring the results pertaining
Trend Determination required the subject to get a general
to the first hypothesis regarding physical immersion,
sense of the layout of the data in order to spot trends.
followed by the second hypothesis regarding head tracking.
Subjects were asked to report the trend in this format: as A
increases/decreases, B increases/decreases. Subjects could Listed below are the four conditions. Four walls indicate a
also report that the data had no trend. All datasets used in high degree of physical immersion; one wall indicates a
the study exhibited some sort of trend. low degree of physical immersion.
4. Clusters Condition 1 = four walls with head tracking
The Cluster finding task asked subjects to locate clusters of Condition 2 = four walls with no head tracking
data points greater than 20 points. We felt this was another Condition 3 = one wall with head tracking
important task for data visualization, but as the definition of
“cluster” can be subjective, we had to loosely define what Condition 4 = one wall with no head tracking
we thought a cluster was, as well as use data that 4.4.1 Physical immersion
constrained what the user would identify as a cluster. In examining physical immersion, tasks completed under
5. Single Point Search condition one were compared with those completed under
condition three. Conditions two and four were not
The Single Point Search task had subjects locate a compared due to missing data. The average time to
differently colored point in a densely packed group of complete tasks one and two was 34.1 seconds under
points. The point was not visible from the subject’s starting condition one. The time to complete those tasks under
position and required them to navigate to find it. This task condition three was 30.5 seconds. For tasks four and five,
would be important in a real-world application where a data the time for condition one was 16.4 seconds. For condition
point is highlighted in another view and must then be three the time for those tasks was 25.7 seconds. For tasks
located in the VE view (e.g. a brushing-and-linking task). one and two, a high degree of physical immersion was a
6. Outliers few seconds slower, and for tasks four and five it was much
The Outlier task was designed specifically to answer a faster. Out of the four tasks that were completed under both
question about VE scatter plots. We asked the subject to conditions, three of them were faster in the physically
find the two data points that were furthest away from the immersive condition. From this data we can see a trend for
main group of data points. In the VE it was easy to miss greater task efficiency in a physically immersive
data points that were very far above the user or just out of environment when head tracking is also used.
the field of view. We expected the fully immersive The average perceived disorientation and usefulness levels
conditions to perform better for this task. of condition one were 1.75, (1-7, 7 being most disorienting)
and 5.5 (1-7, 7 being most useful), respectively. For
4.3 Subjects
condition three, the average disorientation and usefulness
Four subjects were recruited to participate in the study. We
ratings were 1.5 and 5.3. The average difficulty across
wanted to use experienced VE users in order to minimize
condition one for completing the tasks was 3.3 (1-7,7 being
problems associated with VE navigation or sickness. The
most difficult.) For condition three the average difficulty
subjects performed all 6 tasks once, and saw all four
was 3.2. These three metrics, therefore, did not indicate any
conditions at least once. Table 2 shows the design we used.
benefit for physical immersion in completing the tasks.
The entries in the table show the order in which the tasks
were performed. We ordered the tasks so that each task was All users stated that when four walls of the CAVE were
performed for each condition. Since this was an exploratory used, it was much easier to view large datasets. One subject
study, we felt four subjects were sufficient. We made sure comment read, “Four walls very useful in ability to view as
to get as much input as possible from each subject so that much of the data set as possible. Six walls would be even
we could refine our evaluation method. better!” This was reinforced by observing the subjects’
behavior when certain datasets were displayed in the one-
Condition wall condition. Some subjects turned to look at the side
1 2 3 4 walls expecting points to be displayed there, and when
none appeared, seemed a bit frustrated that they had to turn
back to the front wall to manipulate the dataset further.
Certain tasks seemed to benefit particularly from a high
Subject 1 1, 2 3, 4 5 6
degree of physical immersion. This might explain why
2 3, 4 5 6 1, 2 tasks four and five were completed significantly faster
3 5 6 1, 2 3, 4 under condition one than condition three. Task #4 involved
finding clusters, and in task #5, the subject had to find a
4 6 1, 2 3, 4 5 colored cube in a densely packed dataset. A wider field of
<TABLE 2 – Subject Task Order> view permitted a greater area of the dataset to be visualized
at once, resulting in markedly lower completion times.
4.4.2 Head tracking Several changes in the interaction methods might also
For head tracking, conditions one and two were compared. prove beneficial in the future:
Conditions three and four were not compared due to
• Display the numbers of multiple cubes close to the
missing data. For all tasks, condition one (with head
user. This could aid in the selection of a particular data
tracking) was much faster than condition two (no head
point without having to point directly at it.
tracking). The average disorientation and usefulness ratings
of condition one were, 1.75 and 5.5, respectively. For • Label the axes with the names of the attributes being
condition two those numbers were 2.75 and 4.25. The visualized. This could potentially reduce confusion and
average. difficulty for condition one was 3.3, for condition aid in orientation and understanding of the dataset.
two it was also 3.3. We see from this data a trend for • Introduce a different method of navigation and
greater efficiency, lower disorientation, and greater utility manipulation of the dataset to rotate the data about a
in a head-tracked immersive environment. point, in addition to the current method of navigating
Overall, head tracking was perceived to be beneficial in through it (first person), and add these two methods as
viewing the datasets. One subject commented, “Head controlled variables to find out the impact of the
tracking was very useful, it was much easier to view the navigation methods when looking for benefits of
data when moving around.” This was noted in several immersion.
subjects’ behavior during task completion. In particular, the 6. ACKNOWLEDGEMENTS
clusters task (#4) seemed to be quite well suited for head The authors would like to thank the subjects who
tracking. There was, however, one subject out of the four participated in this exploratory study.
who did not find head tracking useful.
7. REFERENCES
5. CONCLUSIONS AND FUTURE WORK [Ahlberg95] Ahlberg, C. and Wistrand, E. (1995). IVEE:
Due to the small number of subjects and design of the An information visualization & exploration environment.
experiment, we cannot make any definitive statistical Proc. IEEE Information Visualization, IEEE Computer
conclusions based on the data recorded. However, the data Press, 66-73.
do show trends in certain instances that are promising for
future study: [Anupam95] Anupam, V., Dar, S., Leibfried, T., and
Petajan, E.. (1995). Research report: DataSpace: 3-D
• A high degree of physical immersion resulted in visualizations of large databases. Proc. IEEE Information
generally lower times than a low degree of physical Visualization, IEEE Computer Society Press, 82-89.
immersion.
[Arns98] Arns, L., Cook, D., Cruz-Neira, C.(1998). The
• Head tracking showed a strong trend in favor of its use. Benefits of Statistical Visualization in an Immersive
This is apparent in not only task completion times, but Environment, Proceedings of IEEE VR'99, pp. 88-95
disorientation and usefulness ratings as well.
[Cruz-Neira93] Cruz-Neira, C., Sandin, D., and DeFanti, T.
• The combination of a high degree of physical (1993). Surround-Screen Projection-Based Virtual Reality:
immersion and head tracking seemed to yield the best The Design and Implementation of the CAVE. Proc. ACM
results, as completion time for most tasks across SIGGRAPH.
condition one were lower than for any other condition.
[Datey01] Datey, A. Experiments in the Use of Immersion
Based on the more subjective results and observations, we for Information Visualization. Unpublished Masters thesis,
can say with certainty that both physical immersion and Virginia Tech Computer Science Department. Available at
head tracking overall were perceived to be useful and http://scholar.lib.vt.edu/theses/available/etd-05092002-
beneficial, especially when viewing large datasets. 151043/
We believe this research to be an important first step [Hollands95] Hollands, J.G., Pierce, B., Magee, L.,
toward proving tangible benefits of immersion in Displaying Quantitative Information in Two and Three
information visualization applications, and look forward to Dimensions. Proc Human Factors and Ergonomics Society
continuing this work in greater depth. Our novel strategy 39th Annual Meeting, 1425-1429
for separating the effects of physical immersion and head
tracking (in itself a contribution) will be used in the future [Kelso02] Kelso, J., Arsenault, L., Satterfield, S., and Kriz,
for more formal experiments. R. (2002).DIVERSE: A Framework for Building
Extensible and Reconfigurable Device Independent Virtual
For a more formal study, we will use a full-factorial within- Environments. IEEE Virtual Reality, Orlando, FL,
subjects design, with replication, in order to achieve the
maximum statistical power. Additionally, a greater number [Lin02]Lin, J., Duh, H., Abi-Rached, H., Parker, Furness,
T.,(2002) Effects of Field of View on Presence, Enjoyment,
of tasks need to be designed that take full advantage of
head tracking then those currently used. The majority of the Memory, and Simulator Sickness in a Virtual Environment,
current set of tasks may be accomplished by viewing the Proceedings of the IEEE Virtual Reality Conference 2002,
p.164
datasets from afar, where head tracking is less significant.
[Mine95] Mine, M. (1995). Virtual Environment [Slater96] Slater, M., Usoh, M., Linakis, V., & Kooper, R.
Interaction Techniques. TR95-018, UNC Chapel Hill CS (1996). Immersion, presence and performance in virtual
Dept. environments: An experiment with tri-dimensional chess.
[Pausch97] Pausch, R., Proffitt, D., and Williams, G. Proceedings of 1996 ACM VRST, 163-172.
(1997), Quantifying immersion in virtual reality, [Spotfire] http://www.spotfire.com/products/decision.asp
SIGGRAPH'97. [Wickens95] Wickens, C., LaClair M., and Sarno K.(1995)
[Robertson 93] Robertson, G., Card, S., and Mackinlay, J. Graph Task Dependencies in 3D Data: Influence of 3D and
(1993). Information visualization using 3D interactive Color. Proc Human Factors and Ergonomics Society 39th
animation. Communications of the ACM, 36(4), April. Annual Meeting, 1420-1424