Journal of Autism and Developmental Disorders, Vol. 26, No.

6, 1996

Brief Report: Two Case Studies Using

Virtual Reality as a Learning Tool
for Autistic Children 1
Dorothy Strickland
North Carolina State University

Lee M. Marcus, Gary B. Mesibov, and Kerry Hogan

University of North Carolina at Chapel Hill School of Medicine

INTRODUCTION

In 1895 the Lumieres' documentary Train Entering Station reputedly made

audiences flee or duck for cover, fearful for their lives (Pirie, 1981). Film is
now widely recognized as a technology capable of making the human body
respond emotionally, as when adrenaline increases at a horror movie. Virtual
reality carries the fantasy a step further, immersing the user in an illusion which
responds as the real world does. The human interaction with the imaginary
world is not conducted through watching other peoples' motions but by con-
trolling the new reality through one's own actions. For this reason virtual reality
(VR) allows individuals to explore new ways of responding and learning.

Definition of Virtual Reality

The key difference between VR and typical computer programs is

the level of interaction with the computer-generated images. In conven-
tional systems, predetermined pictures and choices are presented to the
user. With VR, the computer generates a three-dimensional world, places

1Thanks to Division Inc. for their loan of the equipment and offices for the test, and to Ed
and Jim Bedford, Thea, Larry, and Lauren Gardner, Kayren McNight, Kay Flinn, and Libby
Webb for their help and ideas.

651
0162-3257/96/1200-065150950/09 1996 Plenum Pubfishing Corporation
652 Stricldand, Marcus, Mesibov, and Hogan

the user within it, and allows independently determined motion by the user
with appropriate environmental responses (Gregory, 1991). In its finest
form, VR produces computer-generated real life experiences.
There are several levels of VR, the most sophisticated of which is called
immersion. In this version the user wears a headset containing two small
video screens, one suspended in front of each eye. As the person moves, the
movement is tracked and used to reposition the user's location in the scene.

Previous Applications

Just how convincing are these computer-generated realities? Research

is just beginning, but early studies indicate the senses are susceptible to
this illusion. The sense of realism was convincing enough in an immersion
system to treat acrophobia with visually limited worlds (Rothbaum et al.,
1995). In claustrophobia studies, participants responded to an immersive
shrinking virtual room with claustrophobic symptoms when the room
reached a minimum size (Pyne, 1994). Flat-screen virtual displays were used
to transfer spatial information on building interiors to children with poor
mobility (Wilson, 1993) and teach shopping and monetary skills in a virtual
supermarket (Brown, Cobb, & Eastgate, 1993). Physically handicapped in-
dividuals are using fiat-screen virtual settings to learn to navigate with
wheelchairs (Kuhn, 1994; Powell & McJunkin, 1994; Swan & Stredney,
1994). Prototype immersion VR buses are being developed to train ado-
lescents with mental retardation to ride School buses (Mowafy, 1994).

Why Consider VR for Autism?

Several aspects of autism suggest that virtual reality might be helpful

with this population:
Sensory Problems. Many children with autism have difficulty with mul-
tiple sources of sensory input (Grandin & Acariano, 1986). Stimulation in
certain settings can be overpowering, causing difficulties and behavior de-
terioration. VR isolates specific stimuli from the environment and allows
subjects to control how much they will experience. Complex stimulus arrays
can be simplified.
Lack of Generalization. Difficulty generalizing behaviors learned in a
single setting to similar appropriate situations has frustrated treatment ef-
forts in autism. What is taught in one situation does not necessarily occur
naturally in related appropriate situations. VR makes generalization easier
because of the realism that it brings to treatment situations.
Virtual Reality as a Learning Tool in Autism 653

I/'tsual Thought Patterns. Many have observed that thinking in people

with autism is primarily visual (Grandin, 1992; Schopler, 1987). Intervention
techniques have been successful when they capitalize on these visual
strengths (Mesibov, Schopler, & Hearsey, 1994). VR, emphasizing visual
skills, seems to be an appropriate modatity for people with autism and
should give them an excellent opportunity for learning new concepts and
behaviors.
Individualized TreatmenL Although there are general characteristics that
all people with autism share, effective approaches must individualize their tech-
niques to meet the needs of individual clients (Schopler, Mesibov, & Hearsey,
1995). Computerized instruction using VR allows for constant readjustments,
based on the needs and skills of individual clients. Learning imagery can be
readjusted to compensate for individual styles and changing patterns.
Responsiveness with Computer Technology. Although computers have
not been adapted by special education programs as quickly as some would
like, there is increasing evidence that they represent an effective new ap-
proach to education and learning for children with developmental disabili-
ties (Howard, Buch, Watson, & Shade, 1991). Given the characteristics of
autism and the encouraging preliminary data, it appears reasonable to ex-
pand the use of computerized instruction for individuals with autism and
VR is a promising avenue for this extension.

DESIGN OF STUDY

This study was designed to determine if children with autism would

tolerate VR equipment and respond to the computer-generated world in
a meaningful way.

Project Team

The project was a collaborative effort between the North Carolina

State University Computer Science and Computer Engineering Depart-
ments plus the staff and two families from the Division for the Treatment
and Education of Autistic and other Communication Handicapped Chil-
dren (TEACCH) at the University of North Carolina at Chapel Hill. The
parents of the children acted as collaborators and cotherapists in the treat-
ment, an idea central to the TEACCH approach (Schopler, 1987). Teaching
the children how to cross the street was selected as the target VR scenario.
The design of the tests took place over approximately a 6-month period
with the actual tests conducted in 5 weeks.
654 Strickland, Marcus, Mesibov, and Hogan

Subjects

The children were selected as subjects based on the following criteria: (a)
unequivocal diagnosis of autism, based on early history and behavioral
observations, Childhood Autism Rating Scale (CARS) ratings (Schopler, Reichler,
& Renner, 1988), test results, and parent reports, (b) moderate level of skills
and ab'flities that would make participation and interest in the technology possible,
but a challenge; (c) strong interest in learning, but resistance to change and new
situations; (d) prior experience with computers; (e) parents and older siblings
who had the time and interest to work closely with the research stafL
S. is a 7 ~ year-old girl with very strong visual spatial skills, limited
verbal and language-based abstract reasoning skills. Recent cognitive test-
ing resulted in an IQ of 91 on the Leiter International Performance Scale.
She was able to do one subtest of the WPPSI-R with good success (Object
Assembly), but was unable to complete other Performance tests. On the
CARS (Schopler et al., 1988), she was classified as Mild to Moderately
Autistic with a score of 36.5.
R. is a 9-year-old boy with somewhat better language skills, but also
considered a visual learner. Cognitive testing resulted in an IQ of 62 on
the Merrill Palmer Scale of Mental Tests. He was able to do relatively well
on the Object Assembly test of the WPSSI-R, but unable to complete other
tests on this scale. On the CARS, he was classified as Mild to Moderately
Autistic with a score of 34.0.

VR Equipment and World

The computer we used was a ProVision 100 fully integrated VR sys-

tem, provided by Division Inc. of Chapel Hill, NC. Hard to render objects
such as people were avoided in favor of simplified, unfamiliar street settings
with different colored moving cars. For our tests, the children were asked
to watch a car moving at a moderate pace or follow images created by a
hand-controlled five-button 3D mouse (Figures 1 & 2). Since 20 minutes
is the point at which discomfort often appeared in previous tests with nor-
mal adults, we limited our immersion to 5 minutes at one time.
The interactive controls standard with most VR were difficult for the
children to use. Pointing also would not work because the computer cannot
track the hand motion unless it has a tracking device affixed, and it is dis-
concerting to point at something in front of your face and not see your
arm and hand. Because both children had some language skills, we pri-
marily used simple verbal responses such as "car," "red," "blue," and walk-
ing morion to indicate response to the world.
Virtual Reality as a Learning Tool in Autism 655

Fig. 1. S. using controls as she participates in the study.

Design of Sessions

The willingness of the children to wear a heavy, awkward helmet was

the initial major concern. To provide the greatest possibility of acceptance,
we followed the TEACCH concept of structured teaching which the children
656 Strickland, Marcus, Mesibov, and Hogan

Fig. 2. Street scenewith image seen in virtual world.

use in their classrooms at school. In this technique, physical organization,

schedules, individual work systems, and routines help to make situations
more understandable to the children, thereby making them more comfort-
able (Mesibov et al., 1994).

Physical Layout and Schedule

The physical layout involved consistent, visually clear areas and

boundaries for specific activities. Although the studies were done in the
offices of Division Inc. in Chapel Hill, we removed unnecessary items where
possible, closed connecting doors, and defined separate activity areas.
Schedules were used to explain which activities would occur and in
what sequence they would occur to help the children anticipate and predict
events. Each child had a basic routine of work first and then play with the
VR test between. In later tests the play period was dropped.
Virtual Reality as a Learning Tool in Autism 657

Preliminary Helmet Acceptance Efforts

The week before the tests began, R's mother had R. try several dif-
ferent helmets for familiarization. S. wears a helmet for horseback riding,
although she does not like it. Before attempting to place the helmet on a
child during the actual test, the older brother or sister of the child assisted
us by wearing the helmet and responding correctly to the VR scene. It was
hoped that seeing the sibling respond enthusiastically to the helmet would
make it more acceptable.

RESULTS AND DISCUSSION

S. and R. came in separately for approximately 30-minute to 1-hour

sessions. During that time both children repeated the work, helmet, play
cycle multiple times. S. had twenty-one 3- to 5-minute sessions over a 7-day
period, and R. had sessions of shorter duration over a 4-day period. The
goal was to teach the children to accept the helmet and pay attention to
the images. We initially asked each child to identify cars when they ap-
peared and say the car color if possible.
On the first attempt, both children accepted the helmet and immersed
themselves in the scenes. We started the children sitting in a swivel chair.
S. immediately followed the cars visually and identified the colors correctly.
At one point she said "ga" for gone when a car turned a corner. In the
first session S. stayed the full 5-minute maximum before the helmet was
removed. R. would not wear the helmet as long initially. He appeared more
concerned with understanding where the images were coming from. R.
would look at the scene in the helmet and then lift it up and look at the
same scene on the flat screen. Several times he tried to look in the front
of the helmet from the outside as if trying to find the images. We finally
removed the flat screen from his range of vision and R. immersed himself
in the helmet for longer periods. R. then followed the cars with his eyes
for short times and correctly identified the cars and their colors when
asked.
The number, color, and speed of the cars were modified repeatedly
and the children were placed in three different street scenes to determine
if they would generalize the learned responses across different scenes. Once
it became obvious they needed a more difficult task in the virtual world,
we tried a variety of experiments to measure the children's reaction to
changes in the VR task. We asked both children to walk in the virtual
world. We also asked each child to locate a movable stop sign in the scene
and walk toward it.
658 Strickland, Marcus, Mesibov, and Hogan

The results of these trials indicated:

1. These two autistic children accepted the virtual helmet.
2. The children repeatedly immersed themselves in the virtual scenes
to a degree that they verbally labeled objects and colors of objects.
3. Both children wore the helmet while standing and moved their bod-
ies while in the virtual scenes.
4. The children responded similarly to three different street scenes,
but more study needs to be done to determine if they were generalizing
across different surroundings.
5. Both children voluntarily took the hand controls at least once and
attempted to use them. R. is more computer literate and did this more
often.
Additional individual results were as follow:
1. Both children consistently tracked moving objects in a scene, with
both eyes, head, and body turning.
2. S. repeatedly turned and located a stop sign in a scene and walked
to it.
3. R. identified the stop sign, but would not consistently walk to
it.
The difference between these two children's ability to walk toward
the sign appeared to be related to their understanding of the VR image
as an interactive three-dimensional world. Although S. freely walked to the
sign to receive a reward, R., when told to "go to sign," would point to the
sign inside the helmet. R. appeared to respond to the image as though it
were on a typical fiat computer screen. S. appeared to understand that her
motions were translated into motions in the virtual world.

SUMMARY

The children complied with most requests. Some of our teaching

goals were limited by technology or space while others were limited by
the difficulty of presenting a task to the children in a way that was un-
derstandable within their environment. However, the opportunity to in-
troduce this technology to children was an important first step in exploring
the potential VR offers to understanding the perceptual processes in-
volved in autism.
Our results indicate that the children will accept a VR helmet and
wear it, identify familiar objects and qualities of these objects in their en-
vironment while using the helmet, and locate and move toward objects in
their environment while wearing the helmet.
Virtual Reality as a Learning Tool in Autism 659

More research is necessary to verify the potential in this area, espe-

cially to discover if learning experiences through VR generalize to other
environments, but it appears virtual reality may provide a useful tool for
furthering our understanding of autism and guiding efforts at treatment
and intervention.

REFERENCES

Brown, D., Cobb, S., & Eastgate, R. (1993). Close encounters, study at Nottingham University.
Special Children, 68, 31-33.
Grandin, T. (1992). An inside view of autism. In E. Schopler & G. B. Mesibov (Eds.),
High-functioning individuals with autism (pp. 105-126). New York: Plenum Press.
Grandin, T., & Acariano, M. (1986). Emergence: Labeled autistic. Novata, CA.: Arena.
Gregory, R. L. (1991). Seeing by exploring. In S. R. Ellis (Ed.), Pictorial communications in
virtual and real environments (pp. 328-337). London: Taylor & Fancil.
Howard, J. R., Busch, J. C., Watson, T. A., & Shade, D. D. (1991). The change-over to
computer-based technology in early childhood'special education. Journal of Research on
Computing in Education, 23, 530-544.
Kuhn, L. (1994). The virtual wheelchair: Overcoming real-world obstacles with cyberspace
technology. Second Annual International Conference on Virtual Reality and Persons with
Disabilities. Sponsored by IEEE and Center on Disabilities at California State University,
San Francisco.
Mesibov, G. B., Schopler, E., & Hearsey, K. A. (1994). Structured teaching. In E. Schopler
& G. B. Mesibov (Eds.), Behavioral issues in autism (pp. 195-206). New York: Plenum
Press.
Mowafy, L. (1994). Unpublished report. University of Dayton Research Institute and
Armstrong Lab, Mesa, Phoenix.
Pirie, D. (1981). Anatomy of the movies. New York: Macmillan.
Powell, G., & McJunkin, T. (1994). Autonomous wheel chair control by an approximate virtual
reality system. Second Annual International Conference on Virtual Reality and Persons
with Disabilities, San Francisco.
Pyne, F. (1994). Claustrophobia study. Unpublished report, Psychology Dept., University of
North Carolina at Chapel Hill.
Rothbaum, B., Hodges, L., Kooper, R., Opdyke, D., Williford, J., & North, M. M. (1995).
Effectiveness of computer-generated (virtual reality) graded exposure in the treatment
of acrophobia. American Journal of Psychiatry, 52, 626-628.
Schopler, E. (1987). Specific and nonspecific treatment factors in the effectiveness of a
treatment system. American Psychologist, 4~ 379-383.
Schopler, E., Mesibov, G. B., & Hearsey, K. (1995). Structured teaching in the TEACCH
system. In E. Schopler & G. B. Mesibov (Eds.), Learning and cognition in autism
(pp. 243-268). New York: Plenum Press.
Schopler, E., Reichler, R. J., & Renner, B. R. (1988). The Childhood Autism Rating Scale.
Los Angeles: Western Psychological Services.
Swan, J. E., & Stredney, D. (1994). The determination of wheelchair user proficiency and
environmental accessibility through virtual simulation. Second Annual, International
Conference on Virtual Reality and Persons with Disabilities, San Francisco.
Wilson, P. (1993). Nearly there, study at Leicester University. Special Children, 68, 29-30.