Abstract
The use of virtual-reality technology in the areas of rehabilitation and therapy continues
to grow, with encouraging results being reported for applications that address human
physical, cognitive, and psychological functioning. This article presents a SWOT
(Strengths, Weaknesses, Opportunities, and Threats) analysis for the field of VR rehabilitation
and therapy. The SWOT analysis is a commonly employed framework in the
business world for analyzing the factors that influence a company’s competitive position
in the marketplace with an eye to the future. However, the SWOT framework can also
be usefully applied outside of the pure business domain. A quick check on the Internet
will turn up SWOT analyses for urban-renewal projects, career planning, website design,
youth sports programs, and evaluation of academic research centers, and it becomes
obvious that it can be usefully applied to assess and guide any organized human endeavor
designed to accomplish a mission. It is hoped that this structured examination of
the factors relevant to the current and future status of VR rehabilitation will provide a
good overview of the key issues and concerns that are relevant for understanding and
advancing this vital application area.
1 Introduction
Virtual reality (VR) has now emerged as a promising tool in many domains
of therapy and rehabilitation (Weiss & Jessel, 1998; Glantz, Rizzo, &
Graap, 2003; Zimand et al. 2003; Rizzo, Schultheis, Kerns, & Mateer, 2004).
Continuing advances in VR technology, along with concomitant system-cost
reductions, have supported the development of more usable, useful, and accessible
VR systems that can uniquely target a wide range of physical, psychological,
and cognitive rehabilitation concerns and research questions. What makes
VR application development in the therapy and rehabilitation sciences so distinctively
important is that it represents more than a simple linear extension of
existing computer technology for human use. VR offers the potential to create
systematic human testing, training, and treatment environments that allow for
the precise control of complex, immersive, dynamic 3D stimulus presentations,
within which sophisticated interaction, behavioral tracking, and performance
recording is possible. Much like an aircraft simulator serves to test and train
piloting ability, virtual environments (VEs) can be developed to present simulations
that assess and rehabilitate human functional performance under a
range of stimulus conditions that are not easily deliverable and controllable in
the real world. When combining these assets within the context of functionally
Rizzo and Kim 119
relevant, ecologically enhanced VEs, a fundamental advancement
could emerge in how human functioning
can be addressed in many rehabilitation disciplines.
But we know that already. What we don’t know is:
What place will VR occupy in the future of rehabilitation?
Depending on who you ask, you’re likely to hear a
variety of responses to that question that might include
such words as: “Visionary!” “too expensive,” “just what
the field needs,” “but how will that impact the therapist’s
role?” “sounds like the Holodeck,” “need better
interfaces,” “hmm… interesting possibilities,” “can
they really do that?” and so forth. In essence, the view
that one takes of VR and its potential to add value
over existing rehabilitation tools and methods is often
influenced by such factors as one’s faith in technology,
economic concerns, frustration with the existing
limitations of traditional tools, fear of technology,
popular-media influences, pragmatic awareness of
current hardware limitations, curiosity, and healthy
skepticism.
For those working in the “trenches” trying to employ
VR in a meaningful way for rehabilitation purposes (or for
those just getting their feet wet), a more systematic strategy
for evaluating the state of the field could be of value
for informing one’s judgment, decision making, and
guesses as to what’s possible now and what lies ahead in
the future. Without applying a structured framework to aid
one’s thinking about the current status and future of VR
and rehabilitation, it is quite easy to regularly oscillate between
flights of wishful thinking and bouts of abject discouragement,
depending on the daily ebb and flow of provocative
data and system crashes. Perhaps our susceptibility
to this sort of bipolar “second-guessing” of VR could be
reduced if one is armed with a comprehensive yet intuitive
method for organizing the myriad factors that will serve
both to enable and to limit how well we can successfully
translate our virtual-reality rehabilitation vision into actual
reality! Such a strategy may also help us to identify realistic
goals and establish priorities regarding which clients are
the most appropriate candidates for VR and which technologies
are best suited to creating applications to meet
their needs. Although a high capacity to live with ambiguity
is a requirement for those who explore novel emerging
approaches in any discipline, a focused approach for guiding
our expectations could make that process more manageable
and productive in the long run.
In view of these issues, this paper will present a
SWOT analysis for the field of VR and the rehabilitation
sciences (see Figure 1). SWOT is actually an acronym
that stands for strengths, weaknesses, opportunities, and
threats, and is a commonly employed framework in the
business world for analyzing the factors that influence a
company’s competitive position in the marketplace with
an eye to the future. A classic success story for the value
of a SWOT analysis is Dell Computer Corporation’s use
of the framework to make the strategic decision to implement
mass customization, just-in-time manufacturing,
and direct Internet sales (Collett, 1999). However,
the SWOT framework can also be usefully applied outside
of the pure business domain. A quick check on the
Internet will turn up SWOT analyses for urban-renewal
projects, career planning, website design, youth sports
programs, evaluation of academic research centers, and
it becomes obvious that it can be usefully applied to
guide any organized human endeavor designed to accomplish
a mission.
Generally, a SWOT analysis serves to uncover the optimal
match between the internal strengths and weaknesses
of a given entity and the environmental trends
(opportunities and threats) that the entity must face in
the marketplace.
● A strength can be viewed as a resource, a unique
approach, or capacity that allows an entity to
achieve its defined goals (e.g., VR can allow for precise
control of stimulus delivery within a realistic
training or rehabilitation simulation).
● A weakness is a limitation, fault, or defect in the
entity that impedes progress toward defined goals
(e.g., the limited field of view and resolution in a
head-mounted display can limit usability and perceptual
realism).
● An opportunity pertains to internal or external
forces in the entity’s operating environment, such
as a trend that increases demand for what the entity
can provide or allows the entity to provide it more
effectively (e.g., tremendous growth in the interactive
digital gaming area has driven development of
120 PRESENCE: VOLUME 14, NUMBER 2
the high-quality, yet low-cost graphics cards needed
to make VR deliverable on a basic PC).
● A threat can be any unfavorable situation in the entity’s
environment that impedes its strategy by presenting
a barrier or constraint that limits achievement of
goals (e.g., clinical administrators’ and financial offi-
cers’ belief that VR equipment is too expensive to
incorporate into mainstream practice).
What has typically been found to be effective, based
on SWOT input, is a strategy that takes advantage of
the entity’s opportunities by employing its strengths
and by proactively addressing threats by correcting or
compensating for weaknesses. This paper will begin by
briefly reviewing the oft-discussed topics relating to VR
strengths and weaknesses, with illustrations from existing
work in the fields of rehabilitation and therapy. Due
to space limitations, the examples provided are not
meant as an exhaustive listing. The more challenging
analysis of opportunities and threats requires an examination
of scientific, technological, medical, marketing,
Strengths Weaknesses
● Enhanced Ecological Validity ● The Interface Challenge 1: Interaction Methods
● Stimulus Control and Consistency ● The Interface Challenge 2: Wires and Displays
● Real-Time Performance Feedback ● Immature Engineering Process
● Cuing Stimuli to Support “Error-Free Learning” ● Platform Compatibility
● Self-Guided Exploration and Independent
Practice
● Front-End Flexibility
● Back-End Data Extraction, Management, Analysis,
Visualization
● Side Effects
● Interface Modification Contingent on User’s
Impairments
● Complete Naturalistic Performance Record
● Safe Testing and Training Environment
● Gaming Factors to Enhance Motivation
● Low-Cost Environments That Can be Duplicated
and Distributed
Opportunities Threats
● Emerging Tech 1: Processing Power and Graphics/
Video Integration
● Too Few Cost/Benefit Proofs Could Impact VR
Rehabilitation Adoption
● Emerging Tech 2: Devices and Wires ● Aftereffects Lawsuit Potential
● Emerging Tech 3: Real-Time Data Analysis and
Intelligence
● Ethical Challenges
● The Perception That VR Will Eliminate the Need
for the Clinician
● Limited Awareness/Unrealistic Expectations
● Gaming-Industry Drivers
● VR Rehabilitation with Widespread Intuitive
Appeal to the Public
● Academic and Professional Acceptance
● Close-Knit VR Rehabilitation Scientific and
Clinical Community
● Integration of VR with Physiological Monitoring
and Brain Imaging
● Telerehabilitation
Figure 1. Summary of a SWOT analysis for VR rehabilitation and therapy.
Rizzo and Kim 121
and attitudinal trends that may well be open to varied
interpretations by readers depending on their experiences
and resources. This structured examination of the
factors relevant to the current status and future of VR
rehabilitation will be unlikely to produce a final answer
that readers will consensually agree upon. But we do
hope to stir the visionary pot enough to stimulate some
creative pondering of these issues that will linger on a
bit as you consider this vital and meaningful VR application
area!
In the following discussion, the term rehabilitation is
defined broadly as a general descriptor to refer to both
the assessment and treatment of impairments in human
physical, cognitive, or psychological functioning. Impairment
will refer to either: (1) a loss of existing ability
due to injury, a disease process, mental disorder, or in
some cases, the aging process; or (2) a failure to display
age-based normative abilities due to a specified developmental
or learning disability.
2 VR Rehabilitation Strengths
2.1 Enhanced Ecological Validity
Traditional clinical rehabilitation methods have
been criticized as limited in the area of ecological validity,
that is, the degree of relevance or similarity that a
test or training system has relative to the “real” world,
and in its value for predicting or improving “everyday”
functioning (Neisser, 1978). Adherents of this view
challenge the usefulness of analog tasks for addressing
the complex integrated functioning that is required for
successful performance in the real world. A primary
strength that VR offers rehabilitation is in the creation
of simulated realistic environments in which performance
can be tested and trained in a systematic fashion.
By designing virtual environments that not only “look
like” the real world, but actually incorporate challenges
that require real-world functional behaviors, the ecological
validity of rehabilitation methods could be enhanced.
As well, within a VE, the experimental control
required for rigorous scientific analysis and replication
can still be maintained within simulated contexts that
embody the complex challenges found in naturalistic
settings. Thus, VR-derived results could have greater
predictive validity and clinical relevance for the challenges
that patients face in the real world.
A number of examples illustrate efforts to enhance
the ecological validity of assessment and rehabilitation
by designing VEs that are replicas of relevant archetypical
functional environments. This has included the creation
of virtual cities (Brown, Kerr, & Bayon, 1998;
Costas, Carvalho, & de Aragon, 2000), supermarkets
(Cromby, Standen, Newman, & Tasker, 1996); homes
(Pugnetti et al., 1998; Rose, Attree, Brooks, & Andrews,
2001); kitchens (Christiansen et al., 1998; Davies
et al., 2002), school environments (Stanton, Foreman,
& Wilson, 1998; Rizzo, Bowerly, et al., 2002),
workspaces/offices (McGeorge et al., 2001; Schultheis
& Rizzo, 2002); rehabilitation wards (Brooks et al.,
1999), and even a virtual beach (Elkind, Rubin,
Rosenthal, Skoff, & Prather, 2001). While these environments
vary in their level of pictorial or graphic realism,
this factor may be secondary in importance, relative
to the actual activities that are carried out in the environment,
for determining their value from an ecological
validity standpoint. Since humans oftentimes display a
high capacity to “suspend disbelief ” and respond as if
the scenario were real, it could be conjectured that the
“ecological value” of a VR task may be well supported
in spite of limited graphic realism and less immersion
(such as in flat-screen systems). In essence, as long as
the VR scenario resembles the real world, possesses design
elements that replicate key real-life challenges, and
the system responds well to user interaction, then ecological
validity would likely be enhanced beyond existing
analog approaches. Evidence to support this view
can be drawn from clinical VR applications that address
anxiety disorders. While a number of the successful VR
scenarios designed for exposure-based therapy of specific
phobias would never be mistaken for the real
world, clients within these VEs still manifest physiological
responses and report subjective units of discomfort
levels that suggest they are responding as if they are in
the presence of the feared stimuli (Wiederhold & Wiederhold,
1998). As well, the extinction of the fear re122
PRESENCE: VOLUME 14, NUMBER 2
sponse that occurs in the VE is often seen to generalize
to the non-VR world, and thus provides evidence for
the ecological validity of this form of treatment (Glantz
et al., 2003).
2.2 Stimulus Control and Consistency
That Supports Repetitive and
Hierarchical Delivery
One of the cardinal strengths of any advanced
form of simulation technology involves the capacity for
systematic delivery and control of stimuli. In fact, one
could conjecture that the basic foundation of all human
research and clinical methodology requires the systematic
delivery and control of environmental stimuli and
the subsequent capture and analysis of targeted behaviors.
In this regard, an ideal match appears to exist between
the stimulus-delivery assets of VR simulation approaches
and rehabilitation requirements. Much as a
tank simulator can provide combat testing and training,
VEs can be developed to present simulations that assess
and rehabilitate human physical, cognitive, and psychological
processes under a range of stimulus conditions
that are not easily controllable in the real world. This
“Ultimate Skinner Box” asset can be seen to provide
value across the spectrum of rehabilitation approaches,
from analysis at an analog level targeting component
cognitive processes (e.g., selective attention performance
contingent on varying levels of stimulus-intensity
exposure), to the complex orchestration of more molar
functional behaviors (e.g., planning, initiating, and
physically performing the steps required to prepare a
meal in a chaotic setting). Although traditional flatscreen
computer-based testing and training tools also
offer these assets, it is our view that the added value for
using VR resides in the capacity for systematic stimulus
delivery embedded within immersive simulations of
functional real-world environments.
This strength can also be seen to allow for the hierarchical
delivery of stimulus challenges across a range of
difficulty levels. In this way, an individual’s rehabilitation
can be customized to begin at a stimulus challenge
level most attainable and comfortable for that person,
with gradual progression of difficulty level based on performance
gains. For example, on the analog level, Yang
and Kim (2002) created a motion training system in
which subjects started with a simple 1D translation
task and later moved to more complex 6-degrees-offreedom
motion tasks. With such hierarchical stimulus
challenges, subjects displayed performance improvements
equivalent to real-world training. For
more complex integrated behavior, the assessment of
driving skills following traumatic brain injury is one
example where individuals may begin at a simplistic
level (i.e., straight, nonpopulated roads) and gradually
move along to more challenging situations (i.e.,
crowded highways) (Schultheis & Mourant, 2001).
Such stimulus-control assets provide the opportunity
to identify, modify, and train individual performance
strategies at various hierarchical levels of challenge
within a VE. As well, the successful execution of
many everyday activities requires the integration of a
variety of cognitive and motor functions, and subsequent
component evaluation of these complex behaviors
is often challenging to clinicians and researchers.
With this powerful capacity for stimulus control
within a VE, the impact of specific patient assets and
limitations may be better isolated, assessed, and rehabilitated.
A good illustrative example of this can be
seen in Rizzo, Bowerly, et al. (2002), with a VR
classroom that was designed to systematically present
distractions while children diagnosed with attention
deficit hyperactivity disorder attempted to focus on a
vigilance task within the classroom. Substantial degradations
in attention performance and increases in motor
hyperactivity were seen in these children (compared
to normal controls) when stimulus distractions
typically found in real classrooms were systematically
introduced.
2.3 Real-time Performance Feedback
Performance feedback as to the status and outcome
of a response is generally accepted to be necessary
for most forms of learning or skill acquisition and is
equally essential to the learning process that underlies
Rizzo and Kim 123
rehabilitation (Sohlberg & Mateer, 2001). While feedback
can be presented in a VE to signal performance
status in a form that wouldn’t naturally occur in the real
world (e.g., a soft tone occurring after a correct response),
more relevant or naturalistic sounds can also be
creatively applied to support response calibration and
enhance the perceived realism of the scenario. For example,
in an Internet-delivered VR application designed
to help children with learning disabilities practice escape
from a house fire (Strickland, 2001), the sound of a
smoke-detector alarm raises in volume as the child gets
near to the fire’s location. As the child successfully navigates
to safety, the alarm fades contingent on her choosing
the correct escape route.
The potential value of virtual-performance feedback
for rehabilitation applications can be seen from applications
designed to support physical therapy in adults
following a stroke (Deutsch, Latonio, Burdea, &
Boian, 2001; Jack et al., 2001). These applications
use various glove and ankle interface devices that
translate the user’s movements into a visible and
somewhat relevant activity that is presented graphically
on a flat-screen display. For example, in one application,
as the user performs a prescribed hand exercise
designed to enhance fractionation (independence
of finger motion), the image of a hand appears on the
display, playing a piano keyboard, reflecting the actual
hand movements of the client. In a similar application,
the appropriate hand movement moves a
“wiper” that serves to reveal an interesting picture
along with a display of a graphic rendering of a performance
meter representing range of movement.
These features serve not only as mechanisms for providing
feedback regarding the ongoing status of targeted
movement, but also could serve as motivators.
Results from this lab with stroke patients, presented
in a series of seven case studies, reported positive results
for rehabilitating hand performance across
range, speed, fractionation, and strength measures
(Jack et al.). In one noteworthy case, functional improvement
was reported in a patient who was able to
button his shirt independently for the first time poststroke
following two weeks of training with the VR
hand interface. As well, by making the repetitive and
often boring work of physical therapy exercise more
interesting and compelling, patients reported enhanced
enjoyment leading to increased motivation.
2.4 Cuing Stimuli to Support
“Error-Free Learning”
The capacity for dynamic stimulus delivery and
control within a VE also allows for the presentation
of cuing stimuli that could be used for “error-free”
learning approaches in rehabilitation-training scenarios.
This asset underscores the idea that in some cases
it may not be desirable for VR to simply mimic reality
with all its incumbent limitations. Instead, stimulus
features that are not easily deliverable in the real
world can be presented in a VE to help guide and
train successful performance. In this special case of
stimulus delivery, cues are given to the patient prior
to a response in order to help guide successful errorfree
performance. Error-free training, in contrast to
trial-and-error learning, has been shown to be successful
in a number of investigations with such diverse
subjects as pigeons to persons with developmental
disabilities, schizophrenia, as well as a variety of CNS
disorders (see Wilson & Evans, 1996, for review).
This asset can also be harnessed to provide immediate
performance feedback to users contingent on the status
of their efforts. Such automated delivery of feedback
stimuli can appear in graded (degree) or absolute
(correct/incorrect) forms and can be presented
via any—or multiple—sensory modalities (though
mainly audio, visual, tactile are used), depending on
the goals of the application and the needs and sensory
capabilities of the user. For example, Brooks et al.
(1999) reported success with a severely amnestic
stroke patient using an error-free VR training approach
for wayfinding in a rehabilitation-ward VE
that produced positive transfer to the real ward. Harrison,
Derwent, Enticknap, Rose, & Attree (2002)
also reported mixed results with the use of cuing
stimuli in a VR system designed to train maneuverability
and route finding in novice motorizedwheelchair
users.
124 PRESENCE: VOLUME 14, NUMBER 2
2.5 Self-Guided Exploration and
Independent Practice
Independent self-assessment and “home-based”
skills practice by clients are common components of
most forms of rehabilitation. Generally, it is accepted
that having clients do “homework” will promote generalization
of skills learned in treatment proper to everyday
behavior. The widespread increase in access to
personal computing over the last decade has also encouraged
the autonomous use of computerized self-help
software by clients. As such, it is likely that the independent
use of VR will also become more common as access
to systems and software expands in the future. Notwithstanding
the potential for shoddy VR applications
to reach the marketplace with little evidence to support
their efficacy or value, the option for independent VR
use (when guided by an appropriate professional) can be
viewed as a strength for a number of reasons. When
compared with existing flat-screen computerized testing
and training formats, VR is distinguished by its capacity
to provide higher levels of both immersion and interactivity
between the user and the VE. These unique features
are seen to enhance the suspension of disbelief
required to generate a sense of presence within the VE.
When this psychological state of presence occurs, it is
conjectured to create a user experience that may influence
task performance (Sadowski & Stanney, 2002).
This user experience may produce behaviors that are
different from what typically occurs in persons undergoing
traditional testing and training, due to the user’s
attention being more occupied “within” the VE. As
well, the user experience may be less self-conscious due
to the perceived removal of the test administrator from
the immediate personal and attentional space. This experience
could provide a unique qualitative window into
how people perform tasks when operating in a more
independent and autonomous fashion. For example, if
clients were allowed to freely interact within a functional
VE (i.e., office, mall, home, etc.) that contained very
subtle test challenges, the recording, and later observation,
of more naturalistic client behaviors would be possible.
This could include observing how individualized
compensatory or problem-solving strategies are spontaneously
employed when challenged with a complex situation.
As well, this may also be of value for monitoring
decision making and risk taking in the assessment of
potentially hazardous real-world skills in a VE, such as
driving an automobile.
2.6 Interface Modification Contingent
on User’s Impairments to Support
Access to Rehabilitation
Limitations still exist in fostering truly naturalistic
VR interaction with current 3D user-interaction methods
(Bowman, Kruijff, LaViola, & Poupyrev, 2001).
However, for certain populations with sensorimotor
impairments, existing 3D UI tools can still support access
to rehabilitation in VR beyond what is possible with
traditional non-VR methods. Oftentimes in rehabilitation,
permanent impairments in one domain of functioning
may interfere with the testing and training of
another set of functions that potentially could benefit
from treatment. For example, one of the current challenges
in cognitive rehabilitation concerns the effective
adaptation of testing and training methods by clients
with significant sensory and/or motor impairments.
When such adaptations are attempted, the question often
arises as to how much a client’s performance reflects
centrally based cognitive dysfunction versus artifacts due
to more peripheral sensorimotor impairments. VR offers
two ways in which this challenge may be addressed in
the testing and training of cognitive and everyday functional
abilities in persons with sensorimotor impairments.
One approach places emphasis on the design of
adapted human-computer interface devices in a VE to
promote usability and access. The thoughtful integration
of adapted interface devices between the person
and VR system could assist those with motor impairments
to navigate and interact in rehabilitative VR applications
(beyond what might be possible in the real
world). Such interface adaptations may support actuation
by way of alternative or augmented movement,
speech, expired air, and tracked eye movement, and by
way of neurofeedback-trained biosignal activity (Barrett,
McCrindle, Cook, & Booy, 2002). These devices, while
Rizzo and Kim 125
admittedly “unnatural,” allow patients to interact beyond
what their physical impairments allow. One basic
example involves the use of a gaming joystick to navigate
in a VE that was found effective for teaching wayfinding
within a VE modeled after an amnestic client’s
rehabilitation unit (Brooks et al., 1999). These authors
partially attributed the observed positive training effects
to the patient’s capability for quicker traversing of the
VE using a joystick compared to what her ambulatory
impairments would allow in the real environment. This
strategy, by minimizing the impact of peripheral impairments
on performance, allowed for centrally based performance
components to be more efficiently trained.
A second approach to this challenge has been to tailor
the sensory modality components of the VE around the
needs of the persons’ impairments. The few efforts in
this area have mainly attempted to build simulated
structures for persons with visual impairments by the
use of enhanced 3D sound (Lumbreras & Sanchez,
2000) and haptic stimuli (Connor, Wing, Humphreys,
Bracewell, & Harvey, 2002; Lahav & Mioduser, 2002).
For example, Lumbreras and Sanchez, aiming to design
computer games for blind children, created a 3D audio
VR system referred to as “AudioDOOM.” In this application,
blind children use a joystick to navigate the
mazelike game environment exclusively on the basis of
3D audio cues (i.e., footstep sounds, doors that “creak”
open, echoes, etc.) while chasing “monsters” around
the environment. Following varied periods of time in
the audio VE, the children are then given Legos to construct
their impression of the structure of the layout.
The resulting Lego constructions are often noteworthy
in their striking resemblance to the actual structure of
the audio-based layout of the maze. Children using this
system (who never actually have seen the physical visual
world) often appear to be able use the 3D sound cues to
create a spatial-cognitive map of the space and then accurately
represent this space with physical objects (i.e.,
Legos, clay, sand). Examples of some of these constructions
are available on the Internet (http://www.dcc.
uchile.cl/mlumbrer/audiodoom/audiodoom.html).
While tested mainly with children, it is possible to conceive
of such 3D audio-based environments as providing
platforms for assessment and rehabilitation of persons
with visual impairments at any age.
2.7 Complete Naturalistic
Performance Record
The review of a client’s performance in any rehabilitation
activity typically involves examination of numeric
data and subsequent translation of that information
into graphic representations in the form of tables
and graphs. Sometimes videotaping of the actual event
is used for a more naturalistic review and for behaviorrating
purposes. These methods, while of some value,
are typically quite labor-intensive to produce and sometimes
deliver a less than intuitive method for visualizing
and understanding a complex performance record.
These challenges are compounded when the goal of the
review is to provide feedback and insight to clients
whose cognitive impairments may preclude a useful
understanding of traditional forms of data presentation.
VR offers the capability to capture and review a
complete digital record of performance in a virtual
environment from many perspectives. For example,
performance in a VE can be later observed from the
perspective of the user, from the view of a third party
or position within the VE, and from what is sometimes
termed a “God’s-eye view,” from above the
scene, with options to adjust the position and scale of
the view. This can allow a client or therapist to observe
the performance from multiple perspectives and
repeatedly review the performance. Options for this
review also include the modulation of presentation, as
in allowing the client to slow down the rate of activity
and observe each behavioral step in the sequence
in slow motion.
2.8 Safe Testing and Training
Environment, which Minimizes Risks
due to Errors
As alluded to in the VR driving example presented
above, when developing certain functionally
based assessment and rehabilitation approaches, one
must consider the possibility of safety risks that may
126 PRESENCE: VOLUME 14, NUMBER 2
occur during activities designed to test and train abilities
in the real world. Driving would probably represent
one of the more risk-laden activities that a client
with cognitive and/or physical impairments would
undertake in order to achieve functional independence.
However, even simple functional activities can
lead to potential injury when working with persons
having CNS-based impairments. Such potential risks
can be seen in the relatively “safe” environment of a
kitchen (i.e., burns, falls, getting cut with a knife) as
well as in more naturally dangerous situations such as
street crossing, the operation of mechanical/industrial
equipment, and driving a motor vehicle. Additionally,
the risk for client/therapist injury and subsequent liability
concerns may actually limit the functional targets
that are addressed in the rehabilitation process. These
“overlooked” targets may actually put the client at risk
later on as they make their initial independent efforts in
the real world without having such targets addressed
thoroughly in rehabilitation.
Thus far, this asset has served as a driving force for
VR-system design and research with clinical and “atrisk”
normal populations. Such applications include:
street crossing, with unimpaired children (McComas,
MacKay, & Pivak, 2002), with populations with learning
and developmental disabilities (Strickland, 2001;
Brown et al., 1998), and with adult traumatic-braininjury
groups with neglect (Naveh, Katz, & Weiss,
2000); kitchen safety (Rose, Brooks, & Attree, 2000);
escape from a burning house, with autistic children
(Strickland, 2001); preventing falls, with at risk elderly
(Jaffe, 1998); use of public transportation (Mowafty &
Pollack, 1995), and driving, with a range of clinical populations
(Rizzo, Reinach, McGehee, & Dawson, 1997;
Liu, Miyazaki, & Watson, 1999; Schultheis & Mourant,
2001). In addition to the goal of promoting safe performance
in the real world, some researchers have reported
positive results for building a more rational awareness of
limitations using a VR approach. For example, Davis &
Wachtel (2000), have reported a number of instances
where older adults, poststroke, had decided not to continue
making a return to driving a primary immediate
goal after they had spent time in a challenging VR driving
system.
2.9 Gaming Factors to Enhance
Motivation
Plato is reputed to have said, “You can discover
more about a person in an hour of play than in a year of
conversation” (cited in Moncur & Moncur, 2002). This
ancient quote has particular relevance for VR rehabilitation
applications. Observing and/or quantifying a person’s
approach or strategy when participating in a structured
game-based activity may provide insight into
functioning similar to what is acquired with traditional
(yet less meaningful) standard performance assessments.
The capacity for a person to become engaged in a gaming
task and become less focused on the fact that they
are being “tested” may provide a purer gauge of naturalistic
ability. As well, another more compelling clinical
direction may involve leveraging gaming features and
incentives for the challenging task of enhancing motivation
levels in clients participating in rehabilitation. In
fact, one possible factor in the mixed outcomes found in
rehabilitation research may be in part due to the inability
to maintain the clients’ motivation and engagement
when confronting them with a repetitive series of training
challenges, whether they be cognitive or physical
activities. Hence, the integration of gaming features in
VR-based rehabilitation systems to enhance client motivation
is viewed as a useful direction to explore.
Thus far, the integration of gaming features into a VE
has been reported to enhance motivation in adult clients
undergoing physical and occupational therapy following
a stroke (Jack et al., 2001; Kizony, Katz &, Weiss,
2003). As well, Strickland (2001) reports that children
with autism were observed to become very engaged in
the VR safety-training applications she has developed
that incorporate gaming features. Further anecdotal observations
suggest that children diagnosed with attention
deficit hyperactivity disorder often have a fascination
for the type of stimulus environments that occur
with computer/video games (Greenhill, 1998). Parents
are often puzzled when they observe their children focusing
on video games intently, while teacher reports
indicate inattention in the classroom. Additionally, in
the first author’s clinical experience, it was observed that
some of the young adult traumatic-brain-injury clients,
Rizzo and Kim 127
who had difficulty maintaining concentration on traditional
cognitive rehabilitation tasks, would easily spend
hours at a time playing the computer game SimCity.
These observations suggest that designers of rehabilitation
tasks can benefit from examining the formulas that
commercial game developers use in the creation of interactive
computer games. These formulas govern the
flow and variation in stimulus pacing that provide linkage
to a progressive reward and goal structure. When
delivered within a highly interactive graphics-rich environment,
users are observed to become extremely engaged
in this sort of game play. Neuroscience research
in the area of rapid serial visual presentation (RSVP)
may provide some scientific insight into the human attraction
to these fast-paced-stimulus environments. In
this regard, Biederman (2002) suggests that a gradient
of opiate-like receptors in the portions of the cortex involved
in visual, auditory, and somatosensory perception
and recognition drives humans to prefer experiences
that are novel, fast, immersive, and readily interpreted.
This may partly underlie the enhanced motivation that
is observed for the types of activities that are presented
in interactive gaming environments. While many reasons
may contribute to the allure of current interactive computer
gaming, a proper discussion of these issues is beyond
the scope of this article. However, the potential
value of gaming applications in general education and
training is increasingly being recognized. An excellent
presentation of these topics can be found in Prensky
(2001), along with an extensive gaming bibliography
that is available at the Digiplay Initiative (2002).
2.10 Low-Cost Functional
Environments That Can be Duplicated
and Distributed
Rather than relying on costly physical mock-ups of
rehabilitation environments, VR offers the capacity to
produce and distribute identical “standard” environments.
Within such digital rehabilitation scenarios, normative
data can be accumulated for performance comparisons
needed for diagnostics and for training
purposes. While the initial cost to produce an environment
may be high, this financial outlay could be dissipated
with cost sharing by professionals adopting the
environment. In view of this, the future evolution of VR
in rehabilitation will likely be driven by three key elements.
First, continuing advances in the underlying enabling
technologies necessary for VR delivery, along
with concomitant hardware-cost reductions, will allow
VR to become more available and usable by independent
clinicians and researchers. Second, this potential
for increased access and the impact of market forces will
result in further development of new VR applications
that target a broader range of clinical and research targets.
And finally, continued research aimed at determining
reliability, validity, and utility will help establish certain
VR applications as mainstream rehabilitation tools.
Contingent upon the occurrence of these events, it will
be possible that future rehabilitation professionals will
be able to purchase a VR system that provides them
with a suite of environments (i.e., home, classroom, office,
community, etc.) within which a variety of testing
and training tasks will be available. This has already occurred
in the area of VR anxiety-disorder applications,
with no fewer than three companies marketing systems
in this manner. Internet access to libraries of downloadable
VR scenarios will become a likely form of distribution.
Data-mining, scoring, and report-writing features
will also become available similar to what currently exists
with many standardized computer-administered
paper-and-pencil tests. As well, highly flexible “frontend”
interface programs will allow clinicians and researchers
to modify stimulus delivery/response capture
parameters within some VEs and tailor system characteristics
to more specifically meet their targeted purposes.
This level of availability could provide professionals with
unparalleled options for using and evolving standard VR
applications in the service of their clients and for scientific
aims.
3 VR Rehabilitation Weaknesses
3.1 The Interface Challenge 1:
Interaction Methods
Enhanced ecological validity has already been
raised as an important strength for VR-based rehabilita128
PRESENCE: VOLUME 14, NUMBER 2
tion. However, before this vision can be fully reached,
conceptual and technological advances need to occur in
the area of 3D user-interface devices and techniques.
From a human-computer interaction perspective, a primary
concern involves how to design more effective,
efficient, and easily learnable methods for human interaction
with such complex systems. Current methods are
still limited in the degree to which they allow users to
naturalistically interact with the assessment and rehabilitation
challenges presented in a VE. Rehabilitation developers
are often constrained to use existing hardware
that often falls short of the aim to foster natural interaction.
This is a significant problem, since in order for
persons with cognitive and/or physical impairments to
be in a position to benefit from VR applications, they
should ideally be able to easily learn how to navigate
and interact within a VE in a manner similar to how
they do it in the real world. Many modes of VR interaction
(i.e., wands, joysticks, 3D mice, etc.), while easily
mastered by unimpaired users, could present problems
for those with cognitive or physical impairments. A case
can be made that, short of fostering truly realistic naturalistic
interaction, perhaps interaction methods based
on “magic” that give the user “suprahuman” interaction
abilities are a viable option (Bowman et al., 2001).
However, these methods would need to be highly learnable,
and for some rehabilitation applications, devices
might also have to be custom-made to suit the special
needs and impairments of patients. This would require
extensive usability testing even before the clinical effi-
cacy of the scenario could begin to be evaluated. Even if
patients are capable of using a less natural interaction
method at a basic level, the extra nonautomatic cognitive
effort required to interact/navigate could serve as a
distraction and limit the assessment and rehabilitation
processes. In this regard, Psotka (1995) hypothesizes
that facilitation of a “single egocenter” found in highly
immersive interfaces would serve to reduce “cognitive
overhead” and thereby enhance information access and
learning.
This is representative of the larger general problem
of designing usable interfaces for all VR applications.
Despite some recent progress in interaction modeling
(Bowman et al., 2001), finding the best interface
method for a given application usually requires costly
and time-consuming usability testing. Other major obstacles
for designing usable interfaces for VR-based rehabilitation
include the rapid changes in hardware capabilities,
device availability, cost, and the lack of a mature
methodology in interaction design. The situation is often
compared to the case of 2D desktop interfaces,
where a mature interface methodology has emerged
over the last 30 years using devices that are standardized/fixed
(mouse, keyboard, monitor) (Preece, Rogers,
& Sharp, 2002). With the absence of such an established
design methodology, we are still limited to a trialand-error
exploratory approach to VR interaction
design. Such a method adds another layer of challenges
for rehabilitation professionals who have little familiarity
with usability testing. Additionally, managed-health-care
dollars are scarce for supporting this type of essential,
though less sexy, usability research with clinical populations.
This is an area that needs the most attention in
the current state of affairs for VR rehabilitation, and
better multidisciplinary collaboration in application development
may be a key element for future success.
3.2 The Interface Challenge 2: Wires
and Displays
Many devices that are required to operate a VR
system or to track user behavior require wires and various
connectors that are a source of distraction and inconvenience.
Wires constrain interaction, both mentally
and physically, and complex motion can result in the
user getting tangled in wires and cables that limit usability
and could create safety hazards. Similarly, after all
the years of advancements in the field of VR, an inexpensive
technology for head-mounted display (HMD)
systems with near-human vision quality has yet to arrive.
HMDs are still cumbersome, mostly tethered, and provide
only limited resolution and field of view. Provision
of stereoscopy within an HMD is still problematic, due
to the conflict between flat displays and human eye accommodation
and convergence factors that sometimes
results in user eyestrain, headaches, and other side effects.
Fully immersive projection displays (i.e., CAVE,
Powerwalls, Immersadesks) may offer better visual charRizzo
and Kim 129
acteristics, yet remain a very expensive option that reduces
pragmatic availability for rehabilitation uses. 3D
sound display technology has seen much progress in the
last 20 years and control of the azimuth (left and right)
of sound stimuli has become quite effective (Duda,
2004). However, the 3D sound quality produced from
the general-purpose sound cards is at its best with the
use of a headphone instead from a set of speakers (not
tethered). This is because the head-related transfer functions
used in the sound cards are usually sampled with
microphones located in the two ears of a human head
model. This in turn makes the use of multichannel
speakers difficult because the synthesized sound is based
on two channels (two-speaker configurations can be
used with less accurate effect). Moreover, it is still diffi-
cult to produce sound effects in terms of the elevation
and range. Even though inexpensive wireless headphones
are available, the fact that headphones (or earphones)
must be used at all in addition to the HMD is
sometimes a nuisance for users.
Progress has also been relatively slow in display technology
for addressing other sensory modalities such as
touch and olfaction. For example, force-feedback haptic
devices require a large mechanical structure, and this
poses many challenges in terms of usability, distraction,
and costs. The available commercial products in this
area also mainly target only proprioceptive sensation
with limited degrees of freedom using expensive exoskeletal
devices (Cybergrasp, 2004). “Phantom-type”
devices (Massie & Salisbury, 1994), while showing
some value for surgical simulation (MacFarlane, Rosen,
Hannaford, Pellegrini, & Sinanan, 1999), have such
constrained functionality that it is tough to justify the
costs in view of the limited rehabilitative targets that
they can address. However, researchers with the resources
for such force-feedback equipment have reported
successful implementation with clinical groups
using the Reach-In system (Broeren, Bjo¨rkdahl, Pascher,
& Rydmark, 2002) and with custom-built devices
for hand and ankle rehabilitation (Burdea, Popescu,
Hentz, & Colbert, 2000). For tactile sense simulation,
the research has primarily focused on techniques to recreate
the texture of virtual surfaces using special devices
and materials such as the piezo-electric elements (Hirota
& Hirose, 1995; Allison, Okamura, Dennerlein, &
Howe, 1998; Burdea, 1996; Ikei & Stiratori, 2002).
These systems typically are applied to a relatively small
skin area such as the fingertip and find very low utility in
actual system deployment. Proposals to use small vibratory
devices for tactile feedback applied to a larger skin
area have been made in several application contexts
(Tan & Pentland, 1997; Jang, Yang, & Kim, 2002),
but results on their utility and usability are still preliminary.
In general, most precision haptic/tactile
devices are still too costly to find their way into mainstream
rehabilitation, and the troubling thought is
that there does not seem to be a visible breakthrough
in these areas in the near future. This is unfortunate
since better touch simulation is especially desirable
for motor rehabilitation and for creating applications
for patients with visual and auditory impairments
(Lahav & Mioduser, 2002).
Tracking devices and sensors are one of the fundamental
technologies for any VR system, yet still present
considerable usability, cost, and accuracy limitations.
Even though wireless tracking methods have become
available, they are still expensive and not accurate
enough. Magnetic trackers, perhaps the most popular
type, either require a specially made magnetic-field-free
operating room or a cumbersome calibration process to
achieve a reasonable accuracy. Inertial trackers are prone
to drift and error accumulation after some extended
period of use. Ultrasonic and optical methods have lineof-sight
and occlusion problems. Accurate marker-based
tracking systems are very expensive, low in usability
(e.g., users must wear special suits), and thus are finding
use mainly for animation production and motion capture,
and not for everyday use as is often required in
rehabilitation. Real-time marker-less vision-based tracking
is still in early stages of development for 6DF applications
(Wren, Azarbayejani, Darrell, & Pentland, 1997;
Cohen, Li, & Lee, 2002), but may provide useful options
in the future. In essence, significant advances are
still needed to produce low-cost and accurate tracking
technology required for more effective VR rehabilitation
systems.
130 PRESENCE: VOLUME 14, NUMBER 2
3.3 Immature Engineering Process for
VR-Based Rehabilitation Systems
While often touted as the media of the next generation,
virtual reality still has not caught up with the
mainstream of content development, let alone for VRbased
rehabilitation applications. One of the major obstacles
is the lack of models, methodologies, and tools
for VR system/content development. Building, testing,
and maintaining a VR-based rehabilitation application is
undoubtedly a very complex process. Developers must
integrate disparate bodies of knowledge in both engineering
and rehabilitation that include such subareas as
tracking, displays, interaction, computer graphics, simulation,
human factors, biokinesiology, cognitive psychology,
and so forth. What makes VR rehabilitationapplication
development so difficult is that, on top of
having to tackle traditional computational and logical
challenges, one must also address usability concerns specific
to the application, its task components, and the
characteristics of the clinical user group. The requirements
for a VR application designed to treat fear of flying
in an otherwise healthy adult phobic will naturally
differ substantially from one to teach a Down’s Syndrome
teenager how to navigate a supermarket.
The system science for VR has not yet reached maturity
to efficiently address these challenges (Seo & Kim, 2002).
Recent object-oriented VR development tools provide
only abstractions for the system functionalities (i.e., device
handling, displays, etc.). Unlike ordinary programming
tasks, VR execution and development environments are
different, not just in the temporal sense, but also in the
physical sense. Many VR systems have usability problems
simply because developers find it costly and tiring to switch
back and forth between the development (e.g., desktop)
and execution environments (e.g., immersive setup with
HMD, glove, trackers, etc.), and fail to fully test and configure
the system for usability. Such issues make system
optimization and cost estimation for building a VR application
very difficult. Therefore, the perceived cost of a
VR application becomes quite high, and in most cases,
wrongly so, because with the knowledge and application of
system science, a cost-effective solution should be found.
Consequently, the inability to predict overall development
cost remains a major weakness for VR to be proliferated to
many different fields. In comparison to game development,
VR-based rehabilitation systems tend to be “oneoffs”
rather than marketed for a mass audience. Such a
nature is more of a reason to have a model for cost estimation
and for development. That is, game development may
justify a large investment due to its large market, while a
one-of-a-kind VR system usually cannot afford to do so
and requires a more targeted, systematic approach.
3.4 Platform Compatibility
In order for VR to come into mainstream rehabilitation
practice, one basic computational device needs to
be easily configurable to run a variety of applications in
a manner akin to the ease of installing and running both
MS Word and Adobe Acrobat on the same machine.
However, most virtual-reality applications are not interoperable.
This is primarily due to the relatively short
history of active VR-based application development. In
the early days (late 80s to mid-90s), most VR applications
were developed using Silicon Graphics (SGI)
graphics workstations that ran the IRIX operating system
(a variant of the UNIX operating system). SGI was
also instrumental in developing many useful tools for
VR application development, including the Performer,
OpenGL, and Inventor (SGI, 2004a). However, due to
their high cost and the PC graphics-card revolution,
many developers now have switched to the PC platform.
This created initial problems in that there were few
good development tools equivalent to those available on
the SGI workstations and it took years for VR engine
API vendors to come up with relatively stable and bugfree
versions of their software for PCs. For example, the
Windows version of the very popular Performer became
available only in 2003 (SGI, 2004b). Consequently,
many interesting and useful early SGI-based VR applications
need to be substantially modified in order to be
ported to the MS Windows environment, arguably the
most popular operating system in use today. The transition
from special-purpose graphics workstations to clustered
PCs for CAVE-type systems (which usually require
a form of a multiprocessing system) has only
recently begun as the major graphics-card vendors
Rizzo and Kim 131
started to offer the “Gen-Lock” feature, which can connect
graphics cards installed on different PCs to work
together in synchrony. Thus, the current situation is
such that, although we are starting to truly converge
toward using PC/Windows as the main development
platform, there are still many legacy applications that are
written for other hardware platforms and operating systems.
Similarly, while many developers are making a
case for the use of the LINUX operating system, very
few clinical settings are willing to adopt this OS as they
are already beholden to the Microsoft juggernaut!
However, it is not just the problem of different operating
systems and computing hardware, but more importantly,
the fact that applications are not written in a
flexible and reconfigurable manner. VR systems use
many different devices and software, including graphics/video-processing
cards, trackers, buttoned devices,
cameras, 3D sound hardware, voice-recognition software,
mono/stereo displays, and so forth. Applications
that have been developed for one particular configuration
of devices often fail to run (under the same computing
operating environment) if the proper devices
(and their drivers) are not available. In many situations,
it is difficult to duplicate the exact same operating environment
and as such, applications need to be able to
accommodate (through software control) similar devices
(e.g., between different graphics cards, trackers, driver
versions, display types). For rehabilitation applications,
the situation should exist where a number of different
device configurations can still work with similar effects
(e.g., using wired trackers or wireless vision-based trackers,
using tactile devices or without), without requiring
the constant scrutiny of a programmer on site. Surely,
this presents a problem in effectively maintaining VR
systems, which again illustrates the critical need for an
engineering-science approach to the whole lifecycle of
VR application development.
3.5 Front-End Flexibility
Rehabilitation therapists and professionals are often
not programmers. Consequently, in order to maximize
the usability and subsequent usefulness of a VR
rehabilitation program, great care needs to be placed on
building an intuitive front-end interface. We will go as
far as saying that the rule of thumb should be that the
menus, options, and so forth for adjusting stimulus parameters
should be no more complex than that found in
MS Powerpoint! Unfortunately that is often not the
case with much of the clinical VR software that is being
used in rehabilitation. Oftentimes a program is hardcoded
so that the clinician can use it only in a “one-sizefits-all”
manner, or else any adjustments require actually
going into the program files and changing lines of code.
As simple as code modifications are to a programmer,
requiring nonprogrammer clinicians to do this is often a
“deal-breaker.” To avoid this requires the application of
user-centered design strategies, whereby the “user” in
this case is the clinician. While the few commercial VR
rehab/therapy developers are now addressing this problem
(i.e., Virtually Better, Digital Mediaworks, Psychology
Software Tools, etc.), many university-based VR
packages build these functions in as an afterthought.
Sadly, many innovative scenarios do not get applied and
tested to their full potential after a clinician with little
computer-science tech support gets frustrated following
a challenging first start or subsequently when adjustments
cannot be made easily in the program to suit the
needs of a particular patient.
3.6 Back-End Data Extraction,
Management, Analysis, Visualization
As with the front-end problem mentioned above,
non-computer-savvy professionals need VR software
that spits out performance data automatically and in an
intuitive form. While this may add cost to the initial development
and consequently is often not well attended
to in clinical applications, it is an absolute requirement
for a VR rehabilitation application to be adequately used
and evaluated. Most VR rehabilitation applications that
record complex performances often produce very large
files of raw data that then require another program to
unravel the metrics that a clinician is interested in. While
those files of raw output of course have value, the next
step needed to make an application have real usable
clinical utility is to also deliver basic summary scores,
comparison statistics with accumulated normative data,
132 PRESENCE: VOLUME 14, NUMBER 2
and intuitive representations of the findings in standard
graphical and even 3D formats. It is our view that both
the front-end and back-end weaknesses seen in much of
the VR rehab software to date is the product of developers
with limited resources rushing to simply produce
a proof of concept, collect some quick clinical data, and
then apply for more grant money to fix usability problems
later. As we all know from other application areas,
this is usually a more costly approach and, even worse,
may lead to the program being abandoned in the early
stages of testing out of frustration by the clinical professionals
using the application.
3.7 Side Effects
In order for VR to become a safe and useful
tool for any application, the potential for adverse side
effects needs to be considered and addressed. This is
a significant concern as the occurrence of side effects
could limit the applicability of VEs for certain clinical
populations. Two general categories of VE-related
side effects have been reported: cybersickness and aftereffects.
Cybersickness is a form of motion sickness
with symptoms reported to include nausea, vomiting,
eyestrain, disorientation, ataxia, and vertigo (Kennedy,
Berbaum, & Drexler, 1994). Cybersickness is believed
to be related to sensory-cue incongruity. This
is thought to occur when there is a conflict between
perceptions in different sense modalities (auditory,
visual, vestibular, proprioceptive) or when sensorycue
information in the VE environment is incongruent
with what is felt by the body or with what is expected
based on the user’s history of real-world
sensorimotor experience (Reason, 1970). Aftereffects
may include such symptoms as disturbed locomotion,
changes in postural control, perceptual-motor disturbances,
past pointing, flashbacks, drowsiness, fatigue,
and generally lowered arousal (Rolland, Biocca, Barlow,
& Kancherla, 1995; DiZio & Lackner, 1992;
Kennedy & Stanney, 1996). The appearance of aftereffects
may be due to the user adapting to the sensorimotor
requirements of the VE, which in most cases
is an imperfect replica of the non-VE world. Upon
leaving the VE, there is a lag in the readaptation to
the demands of the non-VE environment, and the
occurrence of aftereffects may reflect these shifts in
sensorimotor response recalibration. The reported
occurrence of side effects in virtual environments in
unimpaired populations varies across studies, depending
upon such factors as the type of VE program
used, technical drivers (i.e., vection, response lag,
field of view, etc.), the length of exposure time, the
person’s prior experience using VEs, active versus
passive movement, gender, and the method of measurement
used to assess occurrence (Hettinger, 1992;
Regan & Price, 1994; Kolasinski, 1995). In a review
of this area, Stanney and colleagues (1998) target
four primary issues in the study of VE-related side
effects that may be of particular value for guiding
feasibility assessments with different clinical populations.
These include: “(1) How can prolonged exposure
to VE systems be obtained? (2) How can aftereffects
be characterized? (3) How should they be
measured and managed? (4) What is their relationship
to task performance?” (p. 6). These questions are particularly
relevant to developers of clinical VEs, as
these systems are primarily designed to be used by
persons with some sort of defined diagnosis or impairment.
It is possible that clinical users may have
increased vulnerability and a higher susceptibility to
VE-related side effects, and ethical clinical vigilance
to these issues is essential. Particular concern may be
necessary for neurologically impaired populations,
some of whom display residual equilibrium, balance,
perception, and orientation difficulties. It has also
been suggested that subjects with unstable binocular
vision (which sometimes can occur following strokes,
TBI, and other CNS conditions) may be more susceptible
to postexposure visual aftereffects (Wann &
Mon-Williams, 1996). These issues should be investigated
further in order to determine what effective
methods exist to reduce side effects that could limit
the feasibility of VEs for applications with clinical
populations. An extended discussion of clinical data
in this area can be found in Rizzo, Schultheis, &
Rothbaum (2002).
Rizzo and Kim 133
4 VR Rehabilitation Opportunities
4.1 Emerging Advances in VR
Technology 1: Processing Power and
Graphics/Video Integration
Providing visual realism is undoubtedly an important
component in developing an effective VR rehabilitation
environment. Although it is still difficult to exactly
quantify what constitutes a minimum level of
realism, the continuing revolution in consumer-level
computer-graphic technology has helped the cause of
VR for many applications, including the rehabilitation
domain. The level of possible detail in visual realism can
be indirectly measured by the capabilities of the polygon
and texture-processing power of the graphics hardware.
This figure, the number of polygons that can be rendered
in a scene in real time, has increased almost exponentially
during the past several years. For example, the
original PlayStation, released in 1995, rendered
300,000 polygons per second, while Sega’s Dreamcast,
released in 1999, was capable of 3 million polygons per
second. The PlayStation 2 renders 66 million polygons
per second, while the Xbox set a new standard, rendering
up to 300 million polygons per second (Xbox,
2004). Thus, the images on today’s $200 game consoles
rival or surpass those available on the previous decade’s
$50,000 computers (Laird, 2001).
Textures and images have been used effectively to
enhance visual realism and the graphics-processing capabilities
on the PC level have also increased dramatically
over the past few years, with this trend expected to continue
in the future. The days of up to 128 MB of dedicated
onboard texture memory (128 128-sized texture
is about 64 KBs, and 128 MB texture memory can
hold 2000 of these) are already looming and what we
once called the graphics card or VGA controllers are
now becoming a separate dedicated computing unit for
graphics. Such graphics processing units (GPUs) are
now programmable and thus pixel-and texel-specific
computations can be customized for enhanced realism
and real-time performance (Fernando & Kilgard, 2003).
This makes image-based rendering, such as view morphing
and environment maps (in addition to using traditional
textured 3D polygons) an attractive approach for
enhancing pictorial realism. Many other dedicated computational
modules on today’s graphics board will continue
to make real-time shadows and realistic light effects
(such as ray casting) possible. These light effects
are very important for constructing images tuned to the
human visual system. Stereoscopic support, multichannel
outputs, and multiboard synchronization are becoming
standard or popular features. With the popularity
of PC-based games, PC architectures and data bus
systems are being redesigned and customized for the
exchange of a large amount of model and image data
between the CPU, memory, and the graphics subsystem.
One of the promising trends in interactive
graphics is the resurgence of computer vision techniques,
an excellent alternative to using wired sensors to
track user behavior. Computer vision requires considerable
image processing, and the ever increasing power of
CPUs and GPUs is also making vision-based techniques
more viable.
4.2 Emerging Advances in VR
Technology 2: Devices and Wires
While the issue of cumbersome displays with limited
capabilities and the design of usable interfaces remain
a challenge for VR-application developers (and
researchers), two recent developments are very promising.
The first concerns the emerging advances in display
technology driven by general consumer markets. For
example, autostereoscopy makes stereo viewing possible
without the need to wear any special equipment (i.e.,
polarized or active shutter-type glasses), a feature that
can enhance the overall usability of VR systems. The
Sharp Corporation has recently unveiled an autostereoscopic
notebook computer based on an LCD parallax
barrier technology that is only $500 more than the ordinary
notebook computer. Moreover, such technology
can be combined with digital light projection (DLP)
technologies to be applied to larger projection display
systems. Digital television and new types of display
systems such as plasma TV, LCD, and DLP will eventually
make advanced digital imagery ubiquitous in
homes, with hopefully some “spillover” into clinical
applications.
134 PRESENCE: VOLUME 14, NUMBER 2
Another positive development is the strong trend toward
wireless technology. In the context of VR, there
are two visible thrusts in this direction. One is the already
mentioned resurgence of computer vision techniques
for tracking user motion and for monitoring and
inferring user state and intention. Computer vision
techniques generally suffer from long computation time
due to the inherent nature of image processing (having
to make a calculation for each pixel). Dedicated hardware
and faster/cheaper CPUs are practically making
this limitation disappear. For example, Sony has recently
introduced a camera-based interface for the PlayStation
2 called the EyeToy for under $50 (EyeToy, 2004).
Canesta Inc. has developed a new solid-state chip that
can sense 3D objects and process such data in real time
(the technology has been applied to implementing a
virtual keyboard by tracking human fingers) (Canesta,
2004). Cheap vision processing will one day enable
wireless whole-body tracking (Wren et al., 1997; Cohen
et al., 2002) that will have significant impact on the feasibility
of physical therapy VR applications. Real-time
vision processing will also enhance mixed/augmented
reality systems applied to rehabilitation. Augmented
reality systems would work quite well for certain rehabilitation
applications by allowing the user’s own body
parts to be seen when interacting with overlaid virtual
objects. This would be of value for many areas, but especially
for physical therapy, where this approach could
be leveraged to enhance the sense of proprioception and
promote hand-eye coordination. The second thrust with
the wireless-technology trend is the advent of ubiquitous
computing. Ubiquitous computing advocates the
distribution of processing and sensor elements around
everyday living environments so that various objects in
the environment can monitor and respond to us to provide
intelligent service (Weiser & Brown, 1996). While
such a society is still a long time in coming, efforts in
wireless communication are already making an impact
on wireless-sensor technology. In fact, using wireless
technology such as Bluetooth, Radio Frequency ID
(RFID) tags, ad hoc networks, and so forth, it is already
possible to unwire various current sensors used in the
VR context. Advances in this domain, if fueled by widespread
adoption with consumer technology, could result
in low-cost tracking sensors that would have significant
impact on the usability of VR in general, as well as for
rehabilitation applications.
4.3 Emerging Advances in VR
Technology 3: Real-Time Data Analysis
and Intelligence
One of the strengths observed for VR rehabilitation
is the capability for sensing 3D behavioral-action
data and then presenting feedback regarding ongoing
progress for enhancing learning, perhaps via an errorlesslearning
paradigm. Such useful feedback is possible with
the emergence of real-time motion analysis coupled
with direct acquisition of motion data (Baek, Lee, &
Kim, 2003). Traditional motion analysis used to be an
off-line and sometimes manual process because the raw
data is usually in video form and such video data processing
is very time consuming and difficult. For example,
a typical golf-swing motion analysis used to require
a golf professional to manually segment out (using a
mouse) the relevant features in the video (e.g., golf
club, shoulder line, etc.), and only then would the computer
algorithm be able to track motion and produce
automatic analysis data. Today, various motion sensors
that can directly collect specific motion profiles are making
their way into the motion and sports science arena.
This could serve as a starting point for designing VR
applications in motor rehabilitation that integrate gaming
features to enhance motivation! Going one step further
from simply analyzing patient data, AI techniques
could be applied to make intelligent rehabilitation suggestions
and perhaps even guide the pacing of stimulus
challenges within the VE, contingent on the patient’s
motor performance.
For rehabilitation domains with a social context, virtual
humans, realistic both in appearance and behavior,
will be very important. Realistic (in terms of appearance
and motion) virtual synthetic humans are already commonplace
in PC-level games. For example, current offerings
in the sports-themed digital gaming arena feature
texture-mapped real-life characters that can be
controlled by the user to produce lifelike behavior using
acquired motion data. VR applications for targeting psyRizzo
and Kim 135
chological processes related to social interaction will
require dynamic interaction with virtual human characters
that are intelligent and autonomous. For example,
such applications would be of value for addressing social
phobia in adults, as well as for helping children with
autism to learn appropriate social skills. This has been
the goal of artificial intelligence for a long time, and
even though producing a software agent with human
intelligence seems insurmountable at this moment, the
traditional AI and VR fields are merging to produce
virtual characters that are adaptive and convincing for
many applications. Laird (2001) has been applying a
general AI architecture called Soar to game engines
such as the Quake II to create agents that are challenging
opponents, not because of their superhuman reaction
times and aiming skills, but because of their complex
tactics or game knowledge. This will be an
important component in future VR rehabilitation applications,
and such approaches have the potential for creating
scenarios in which social interactions are targeted.
The Sims electronic game series provides an excellent
example of how social interactions can be the basis for
an engaging virtual arena (Sims, 2004). While development
costs and the lack of processing power prevent
games for today’s computers and consoles from exhibiting
a high level of artificial intelligence, that will soon
change. Currently, developers devote more resources to
advancing a game’s graphics technology than to enhancing
its AI. However, the recent USC/ICTdeveloped
game, Full Spectrum Warrior (ICT, 2004)
now devotes 60% of processor resources to the AI component.
It is likely that within two to three years, the
emphasis on graphics will have run its course as incremental
advances lead to only marginal improvements in
the game experience. At that point, the market drivers
to create games with higher AI “quotients” could have
significant impact on VR rehabilitation applications.
4.4 Gaming-Industry Drivers
There is no doubt that the recent growth in the
interactive gaming-industry arena will continue to drive
developments in the field of VR rehabilitation. The
gaming-industry-juggernaut’s growth is evidenced by
the fact that it has now surpassed the Hollywood film
industry in total entertainment market share, and, in the
USA, sales of computer games now outnumber sales of
books (Digiplay Initiative, 2002). While the 18–24-
year-old male population initially comprised the largest
group of users of commercial interactive computer gaming
applications, this popularity has also extended to
other age groups and to females at a rapid pace (Lowenstein,
2002). As such, it appears that gaming applications
have become a standard part of the “digital homestead”
as delivered on PCs and specific gaming consoles
(i.e., Playstation, Xbox, etc.). From this, interactive
gaming has become well integrated into the lifestyles of
many people who at some point may require rehabilitative
services. For this segment of the population, familiarity
with and preference for interactive gaming could
become useful assets for enhancing client motivation
and engagement when designing VR-based rehabilitation
tasks.
VR therapy and rehabilitation scenarios are also increasingly
being built off of a variety of software engines
developed by the gaming industry. Whether built from
the bottom up (DMW, 2004) or in the form of “mods”
whereby sections of a popular game title can be modified
to create a new scenario (Robillard, Bouchard,
Fournier, & Renaud, 2003), the graphics and audio
options available with these systems are fast becoming
the tool of choice for developers. Just as the advances in
game-industry-driven graphics cards have helped bring
high-quality VR rendering to the PC, we can expect
rehabilitation applications to benefit from the continued
growth of this field. With further advances, it is expected
that gaming hardware will soon become a common
low-cost platform on which many VR rehabilitation
applications will be built.
4.5 VR Rehabilitation Applications
Have Widespread Intuitive Appeal
to the Public
As much as developers and scientists working in
the field of virtual reality are cognizant of the challenges
and limitations that exist for creating usable applications,
the general public is still drawn to the idea of us136
PRESENCE: VOLUME 14, NUMBER 2
ing VR when it is linked to a pressing therapeutic target.
Regardless of whether the public interest and excitement
is due to realistic media and scientific reports or to
the illusions created about the state of VR as presented
in such movies as Lawnmower Man or The Matrix, it is
generally observed that people have an initial favorable
view of the concept of VR. An example of this can be
seen in the growing number of articles and TV reports
on VR health-related applications that appear in mainstream
media outlets (i.e., the NewYork Times; the
Washington Post; Tech TV; Discovery Channel, etc.).
For better or worse, reporters often choose what scientific
news to cover with the public interest in mind and
this has driven the increasing appearance of VR and
therapy/rehabilitation reports in both print media and
television. Another example that our research group has
noted is in the enthusiasm seen in participants attending
non-VR events, when an HMD demo is available at a
booth in an exhibition hall. We have recently set up
such demonstrations at a wide range of diverse scientific
and general non-VR-specific public events and have
consistently observed a steady stream of participants
waiting in lines, eager to try VR. While this public interest
should not be mistaken as a measure of the validity
or value of VR, this positive “vibe” can be seen as reflecting
a state of curious acceptance that bodes well for
future adoption of well-executed applications. However,
overhyped applications can as well lead to negative VR
impressions if an imbalanced “expectation to delivery”
ratio is created, as will be discussed later in the
“Threats” section of this paper.
4.6 Academic and Professional
Acceptance
There has recently been a noticeable positive shift
in the perception of VR applications in therapy and rehabilitation
by mainstream scientists and professionals.
While once viewed as an “expensive toy,” VR is gradually
being accepted as a “tool” that can provide new
options for how therapy and rehabilitation is done. Evidence
for this can be seen in the growing appearance of
VR articles in mainstream journals, along with theme
issues devoted to VR and associated technologies (i.e.,
Disability and Rehabilitation, Neuropsychological Rehabilitation,
Journal of Head Trauma Rehabilitation, Psychological
Inquiry, Psychotherapy: Theory, Research, Practice,
Training), acceptance of papers and symposia at
mainstream conferences (APA, RESNA, INS, NAN,
ACRM, AABT), and in the increased funding by such
organizations as NIH, NIMH, NIDA, NIDRR, and so
forth. Other indicators include such things as VR being
cited as “one of the emerging areas in the future of neuropsychology”
during the presidential address at the
National Academy of Neuropsychology in 2001, and in
the fact that in a poll published in the journal Professional
Psychology: Research and Practice (Norcross,
Hedges, & Prochaska, 2002), a group of 62 therapy
experts ranked VR and computerized therapies 3rd and
5th out of 38 interventions that are predicted to increase
in the next 10 years. Much of this shift in attitude
can be attributed to the highly visible initial clinical
studies yielding positive results using VR for exposure
therapy in persons with anxiety disorders. This application
area is intuitively appealing, well matched to the
assets available with VR, and solid research has consistently
produced clinical outcomes that support its added
value (Riva, 2002; Glantz et al., 2003; Zimand et al.,
2003). Further signs of mainstream acceptance can be
seen by the fact that the oldest and largest psychological
test publisher, The Psychological Corporation, has now
begun to support R&D for standardized VR test development
(DMW, 2004), along with interest growing in
similar rival test and therapy aid publishers. As VR technology
and application development continues to
evolve, this trend is expected to continue.
4.7 Close-Knit VR Rehabilitation
Scientific and Clinical Community
The “closeness” of the community of researchers
and clinicians that focus on VR rehabilitation has continued
to evolve. Perhaps this is partly due to an innate
human drive to form alliances with others of similar interests,
and especially so when the interest area is novel,
risky, and still seeking credibility from the mainstream.
Regardless of the underlying reason, evidence for this
growing close-knit community can be seen in the numRizzo
and Kim 137
ber of conferences that have formed to focus on this
area (i.e., MMVR, CyberTherapy, The International
Conference on Disability, Virtual Reality and Associated
Technology, The International Workshop on Virtual
Rehabilitation), the growth over the last five years in
membership on the VRPSYCH listserver (450 members
from 25 countries), and in the continued expansion in
multiuniversity VR collaborations.
4.8 Integration of VR with
Physiological Monitoring and
Brain Imaging
There is a rather compelling rationale for the integration
of VR with human physiological monitoring
and brain imaging for advanced research in areas relevant
to rehabilitation. Research in such areas as psychophysiology,
biokinesiology, and neuropsychology have
similar goals in that they aim to noninvasively record
bodily events or processes that are hypothesized to correlate
with some human mental and/or physical activity.
Examples of such efforts would include measuring galvanic
skin responses (GSR) while a person attends to
emotionally laden stimuli, electromyographic recording
of muscle responses as a person reaches for a target, and
functional magnetic resonance imaging (fMRI) of hippocampal
brain function while a person performs a wayfinding
task. While these monitoring technologies have
existed for some time, the stimulus-delivery media has
remained essentially the same for many years, relying
mainly on fixed audio and visual content. The use of VR
now allows for measurement of human interaction with
realistic dynamic content, albeit within the constraints
of the monitoring apparatus. Thus far, heart rate, GSR,
and other psychophysiological measures have produced
useful results within VR studies examining attention and
presence (Pugnetti, Mendozzi, Barberi, Rose, & Attree,
1996; Meehan, Insko, Whitton, & Brooks, 2002) and
have been used to enhance treatment effects using a VR
biofeedback paradigm for fear-of-flying clients in Wiederhold
et al. (2002). Even within the confines of a
3-tesla magnetic-imaging device with the user’s head in
a fixed position, humans can navigate and interact in a
VR world with specialized nonmetal displays and interface
devices. In fact a significant body of research has
emerged using fMRI to study brain function in normal
and clinical groups operating in virtual environments
(Maguire et al., 1998; Gron, Wunderlich, Spitzer, Tomszak,
& Riepe, 2000; Astur, Mathalon, D’Souza, Krystal,
& Constable, 2003; Baumann, 2005). The strength
of VR for precise stimulus delivery within ecologically
enhanced scenarios is well matched for this research and
it is expected that continued growth will be seen in this
area.
4.9 Telerehabilitation
The concept of delivering rehabilitation and therapy
to patients in remote locations for independent use
has been a popular topic since health care professionals
recognized the growing reach and power of the Internet.
The application of VR within a telerehabilitation
format is the next logical opportunity for considering
ways to improve access to this technology by a wider
group of potential beneficiaries. VR applications that are
Internet-deliverable could open up new possibilities for
home-based therapy and rehabilitation, which, if executed
thoughtfully, could increase client access, enhance
outcomes, and reduce costs. Future Internet distribution
of VR applications could also be supplemented by
maintaining connectivity between the remote client using
the system and a primary server at a rehabilitation
facility. In this manner, the client’s home-based performance
within the VR application could be tracked,
quantified, analyzed, and graphically represented in an
intuitively understandable format for analysis by key
rehabilitation professionals charged with monitoring
this information. In addition, continual updating of the
VR world and the actions and activities that it requires
of the client during rehabilitative exercises could be implemented
both by the monitoring therapist and via
“intelligent” systems on the main server. This functionality
would allow client progress to be efficiently
tracked, and this information could be used to evolve
the performance demands on the client in a manner designed
to foster effective and efficient rehabilitative outcomes.
An example of some early efforts to apply VR in
a telerehabilitation format for physical therapy are de138
PRESENCE: VOLUME 14, NUMBER 2
tailed in other articles in this issue (Holden, Dyar,
Schwamm, & Bizzi, 2005; Deutsch et al., 2005), However,
the possible benefits that could be accrued from
VR telerehabilitation applications are equally matched
by the enormous challenges that still need to be faced.
It would be unfortunate for clinicians to become enamored
with the obvious potential that exists with VR telerehabilitation,
yet lose sight of the sheer technical, practical,
clinical, and ethical complexities that still need to
be addressed (Rizzo, Strickland, & Bouchard, 2004).
5 VR Rehabilitation Threats
5.1 Too Few Cost/Benefit Proofs Could
Negatively Impact Mainstream VR
Rehabilitation Adoption
While both intuition and early research findings
suggest that VR offers many strengths for rehabilitation
purposes, the field lacks definitive cost/benefit analyses.
Such analyses must spell out both the clinical and economic
benefits, weighed against the costs for using VR
over already-existing traditional methods (Rizzo, Buckwalter,
& van der Zaag, 2002). This is a significant challenge
in view of the high initial development costs of
“one-off ” systems that have often characterized many
inspired VR applications thus far. Without such cost/
benefit proofs, health care administrators and mainstream
practitioners who are concerned with the economic
bottom line may have little motivation to spend
money on high-tech solutions if there is no expected
financial gain. This could result in a “Catch-22” scenario
whereby limited investment is put into R&D that
integrates enabling-technology advances into working
systems that might produce favorable cost/benefit
proofs at a later time. Favorable cost/benefit proofs for
VR use have been reported for military artillery-training
applications (Stone, 2003), where tangible metrics are
readily available (i.e., criterion artillery performance as a
function of costs for real ammunition and equipment
maintenance, etc., vs. VR hardware/software costs,
etc./over time). However, for rehabilitation applications
in the age of modern health care, the metrics are
often quite challenging for “experimental” treatments.
While improving “quality of life” may be a lofty goal to
target for a patient following a stroke, that type of successful
outcome is typically of less interest to third-party
payers, who are more focused on a “return to work”
metric. Ironically, problems may also arise when a VR
treatment shows too much value in attracting clients to
therapy who would ordinarily avoid treatment. This has
been observed in phobic clients who are willing to try
VR therapy after avoiding “standard” talk therapy for
many years. Health care payers are sometimes more interested
in having fewer numbers of patients seek treatment
if it negatively affects their bottom line (K. Graap,
personal communication, January 11, 2004). While this
complex topic is beyond the scope of this article, a fine
example of a health care cost/benefit analysis can be
found in Holder (1998).
5.2 Aftereffects Lawsuit Potential
As discussed in the “Weaknesses” section, aftereffects
may occur in users for various periods of time following
interaction in a VE. The potential for such aftereffects
following VR usage opens the possibility that a
developer, clinician, researcher, or supporting institution
could be held liable for damages in the event that
some injury should befall a patient upon leaving a VR
session. One can easily imagine an unfortunate scenario
in which a patient has a car accident while driving home
from a VR session and the potential legal difficulties that
might ensue if a case were made that the accident was
due to VR-induced perceptual aftereffects that impaired
depth perception. Kennedy, Kennedy, & Bartlett
(2002) present an excellent detailing of these issues in a
chapter on VEs and product liability. They suggest that
certain human-factors safety actions be undertaken that
include: “1) Systems should be properly designed; 2) Aftereffects
should be removed, guarded against, or warned
against; 3) Adaptation methods should be developed; 4)
Users should be certified to be at their pre-exposure levels;
5) Users should be monitored and debriefed” (p. 543).
Until we have better data on these issues, extra caution
may be needed, especially with clinical populations
having some form of central-nervous-system dysfunction
where readaptation to the real world could cause
Rizzo and Kim 139
them to operate differently from unimpaired populations.
For example, in one of our recent VR studies testing
visuospatial abilities with an elderly group (65
years old), since we couldn’t be confident regarding the
absence of potential perceptual aftereffects, we had
funding built into our grant to provide transportation
from the test site, thereby minimizing any possible risk
for altered driving behavior resulting from the VR exposure.
Concerns such as these must be addressed in order
to ensure a positive course for developing VR applications
for all persons and particularly for clinical groups.
5.3 Ethical Challenges
The feasibility of designing, developing, and implementing
VR rehabilitation applications has radically
advanced in the last five years and it is expected that this
evolution will continue into the foreseeable future.
Along with any technological advance in the health care
domain come ethical questions that require thoughtful
consideration. As professionals directly involved in the
application of this technology and as members of society
at large with a moral-ethical responsibility for the promotion
and maintenance of health, we are accountable
to consider and address incumbent ethical threats that
surround this emerging technology. This is especially
important in the rehabilitation sciences, where research
and clinical applications with patient populations require
a rational accounting of the potential risks and benefits.
Additionally, larger pragmatic and societal issues need
to be addressed for VR applications with unimpaired
users regarding the general human experience and its
impact on mental health. As in any area of ethical debate,
clear-cut answers that cover all dilemmas are rarely
found. For example, while immersion and interactivity
may enhance the realism of a VE, these same features
may also create difficulties for certain individuals with
psychiatric conditions or cognitive impairments that
produce distorted reality testing. Specifically, such conditions
could result in increased vulnerability for negative
emotional responses during or following VR exposure.
Although such incidents have yet to be reported in
the VR literature, insurance for monitoring responses
related to VR exposure becomes an ethical responsibility
for the professional. While current therapeutic uses of
VR still require a clinician to be present, future applications
may not have this requirement, and this potential
“opportunity” again underscores the need for advance
consideration of the ethical issues pertinent to the use of
VR as a clinical tool. As well, as potent VR tools become
more readily available for clinical purposes, some
clinicians may not have the qualifications or expertise to
deliver services in the area that the tool was designed to
address. Such inappropriate administration of treatment
by a tech-savvy but unqualified health care provider
could tarnish the credibility of the field in general and
create a public image of VR rehabilitation professionals
as high-tech charlatans! Since a full accounting of potential
ethical threats in VR rehabilitation is beyond the
scope (and page limits) of this article, the reader is referred
to Rizzo, Schultheis, et al. (2002) for a detailed
review of the use of VR in clinical practice and its potential
societal impact.
5.4 The Perception That VR Tools Will
Eliminate the Need for the Clinician
Discussion of the use of VR as a therapeutic tool
can sometimes raise concerns regarding the status of the
clinician when applying technology for therapeutic purposes.
One issue cited as a barrier for implementation of
computerized therapies is the potential impact on the
sanctity of patient-therapist relationship (Gould, 1996).
Ironically, a strong form of this argument can be seen in
writing on the use of computerized therapy by MIT AI
researcher Joe Weizenbaum, who wrote a languageanalysis
program called ELIZA that was initially designed
to imitate a Rogerian psychotherapist. Weizenbaum
(1976) concluded that it would be immoral to
substitute a computer for a human function that “involves
interpersonal respect, understanding, and love”
(cf., Howell & Muller, 2000). While admittedly his
view emerged out of the “shock” he experienced upon
learning how seriously people took the ELIZA program,
even basic automated computer applications in
“drill and practice” cognitive remediation were met
with criticism from some professionals who argued that
the introduction of computers was equivalent to the
140 PRESENCE: VOLUME 14, NUMBER 2
removal of the therapist (cf. Robertson, 1990). Although
supporters of VR therapy and rehabilitation are
quick to point out that these applications are simply
tools that extend the therapist’s expertise, there still exists
a view in some clinical quarters that any technology
serves to subvert the clinical relationship. The impact of
this threat will likely increase as more-believable human
agents begin to populate VR applications.
The issue of therapist acceptance will likely be surmounted
in time if VR applications continue to demonstrate
benign added value, but similar impressions on
the part of potential patients also need to be considered.
For example, will greater access to VR applications via
the Internet encourage individuals to undertake selftreatment
without feeling the need for professional
guidance? Or will slick marketing of costly VR rehabilitation
programs that lack evidence for effectiveness entice
both desperate family members and naı¨ve clients to
self-diagnose and self-administer treatment, yet deliver
no tangible benefit? These practical clinical issues need
to be considered in advance of widespread availability
and adoption of VR rehabilitation applications.
5.5 Limited Awareness/Unrealistic
Expectations
This threat is a counterpoint to the “opportunity”
cited above regarding widespread intuitive appeal of VR
to the general public. All first-time VR users bring into
the situation a set of expectations. Oftentimes, these
expectations are based on overhyped media representations
of what VR can deliver. If someone’s expectation
is the Holodeck, it is likely that they will be disappointed
when they are required to don an array of encumbering
devices that support primitive interaction in
an imperfect replica of the real world. Perhaps the best
remedy for this (aside from continuing to innovate) is in
the conduct of research that tests (and hopefully supports)
the validity and added value of a given VR application,
along with the honest acceptance that we still in
fact have a long way to go. Our lab’s rule of thumb for
VR demos is to usually spend some time preexposure to
brief the person on the rationale for a VR application,
discuss some of the methods that we have been previously
limited to with traditional non-VR tools, and have
posters in the lab from previous conferences that present
efficacy data from the application being demonstrated.
On the other hand, modulating expectations that
continue to be unrealistic in those who are familiar with
VR can serve another purpose. Mainstream researchers
and clinicians are beginning to take notice of this
emerging technology and it is important that it is not
viewed as a panacea for all rehabilitation concerns. The
history of medicine provides numerous examples of the
overexploitation of new technologies based on overwrought
expectations and little data. The medical use of
electricity serves as an example of a commonly accepted
procedure in the nineteenth century, during which time
faulty expectations combined with an urgency to implement
a potential “miracle” cure resulted in many clinicians
with minimal to no experience in the physics of
electricity misapplying the technique to patients (Whalley,
1995). Similarly, with today’s growing interest in
the use of VR for clinical research and treatment, it is
essential that hype does not displace reason in both the
presentation and use of the technology.
6 Conclusions
The view that emerges (to us) from this SWOT
analysis suggests that the field of VR rehabilitation is
still in an early phase of development characterized by
successful “proof of concept” systems, encouraging initial
research results, and a few applications that are finding
their way into mainstream use and clinical practice.
Many VR strengths are specified that will continue to
provide a justification for evolving existing applications
and creating new ones. Weaknesses exist, particularly
with certain limitations in areas of interface and display
technology, but do not threaten the viability of the field
in light of recent and expected opportunities in the
form of advances in the underlying VR-enabling technologies.
With thoughtful system design that targets
clinical and research applications that are well matched
to current technology assets and limitations, it is predicted
that VR rehabilitation will continue to gradually
grow and gain acceptance as a mainstream tool. Threats
Rizzo and Kim 141
to the field do exist, but none are “fatal” and all are
likely addressable with the high motivation and
thoughtfulness that seems to exist with the many researchers,
clinicians, and general proponents of this
field. From this analysis, our general prescription for
advancing VR rehabilitation can be summed up succinctly.
Applications need to be developed with strong
multidisciplinary collaboration and with continuous
user-centered input/evaluation methods. This is needed
to ensure the soundness of the rationale for the application
and to create something that is usable within the
current limits of technology. Researchers need to initially
collect incremental data over numerous small-scale
parametric studies to test and evolve usability, usefulness,
and access, especially with targeted user groups.
During this time, in advance of deployment of a system,
ethical, professional, and cost/benefit issues need to be
considered and specified. And along the way, share your
applications with other interested researchers to see if
they can independently operate your system and replicate
your results! While the field is not without its challenges,
as the technology continues to catch up with the
vision, it is our view that VR will have a significant positive
impact on the rehabilitation sciences.