Milton P. Huang, MD (1), Joseph Himle, PhD (1), Dr.-Ing Klaus-Peter Beier (2), Norman E. Alessi, MD (1)

The fast developing technologies of virtual reality provide the user with the convincing experience (or illusion) of being immersed in an artificial, three-dimensional world. Since this world is computer-generated, virtual environments are ideal for creating controlled conditions for observing the psychological reactions of people to their environment. People with psychiatric illnesses who exhibit symptoms when faced with a particular real life situation will presumably demonstrate these symptoms when placed in a virtual environment that simulates the situation accurately enough. The fact that the virtual environment can be controlled suggests the possibility of using VR for the treatment of many psychiatric illnesses [1]. This is illustrated most clearly in the treatment of phobias. Traditional treatment of phobias involves a process of progressive desensitization. The patient creates a hierarchy of feared situations and is exposed to each one, starting with the easiest and working to the most difficult. Thus, someone with acrophobia (fear of heights) would be taken from the first floor of a building to the second floor, then to the third and so on, slowly increasing the difficulty until they are able to master their anxiety in each situation. VR environments have been used in exposure treatment of fear of flying [2, 3], fear of spiders [4], and acrophobia [5]. The studies of these treatments have shown the effectiveness of virtual reality exposure when compared to no-treatment conditions. Further work remains to be done to compare virtual reality exposure to actual or "in-vivo" exposure, which is the current "gold-standard" for acrophobia treatment.

The purpose of our work is to compare how people react to virtual reality and how they react to reality, using subjects suffering from acrophobia. This first involves minimizing the differences between the VR and in-vivo exposure environments. We attempt to make the VR experience similar to the in-vivo experience by creating a VR model to duplicate the actual environment as closely as possible and by minimizing the intrusiveness of the VR equipment. We attempt to make the in-vivo experience similar to the VR experience by placing restrictions on the in-vivo participants so they will suffer some of the same handicaps that currently intrude upon all VR experiences. To compare individual reactions, we plan to use both subjective and objective measurements, as well as observe for differences in the effectiveness of treatment and persistence of improvement with exposure therapy.

The Anxiety Disorders program of the University of Michigan has traditionally treated patients suffering from acrophobia by exposing them to progressively greater heights looking out the windows of the East Elevator shaft of the University of Michigan main hospital building. We created a virtual reality model of the different elements of this experience, including walking in the lobby space, using the elevator, and examining the views out the window at different floors. This model is experienced in a CAVE (CAVE Automatic Virtual Environment) as developed by the University of Illinois at Chicago and now commercially sold by Pyramid systems. The CAVE is a projection-based VR system that surrounds the viewer with four 9x9 foot screens, arranged to form three walls and the floor of a cube. The viewer wears Stereographics' CrystalEyes liquid crystal shutter glasses and an Ascension Technologies Flock of Birds six-degrees-of-freedom head-tracking device. As they move inside the CAVE, a Silicon Graphics Onyx computer calculates the correct stereoscopic perspective projections for each wall, based on the model. A hand held wand with a second tracker and push buttons allows interaction with the virtual environment. The realism of the model has been increased by taking photographs of the actual environment and applying these as texture maps to the rendered surfaces.

The sophistication of the CAVE environment helps to minimize differences between VR exposure and in-vivo exposure. The shutter glasses are lightweight and less intrusive that heavier traditional head mounted displays. Their transparency allows one to visualize one's own body in the virtual environment, which is critical in acrophobia as one anticipates their approach towards a visual cliff. Combined with large projection screens, this system provides a wider field of view than any other VR system. We attempt to further control for the presence of VR equipment by having the in-vivo participants wear simulated shutter glasses. We instruct both groups to avoid touching the walls, and we operate the elevator buttons, as subjects in our virtual environment are unable to feel any of the objects they see around them. Finally, we also attempt to control for the time of day, which has an impact on biologic measures of stress response.

Our experimental protocol compares treatment in the CAVE to treatment in the actual environment. Subjects are selected as having diagnosis of specific phobia [6] using the Structured Clinical Interview for DSM-IV. Initial measures include use of the Acrophobia Questionnaire [7], Marks and Mathews' Fear Questionnaire [8], global ratings of severity [9], and the Origins questionnaire [10]. A behavioral approach test in the elevator lobby establishes a uniform measure of their ability to tolerate the in-vivo situation. This test takes the subject up floor by floor with less than a minute on each floor until the subject refuses to go higher. They are then randomized to a treatment condition of either in-vivo exposure, CAVE exposure, or a control condition of relaxation training. This last treatment is useful for general treatment of anxiety, but has not shown benefit in the treatment of specific phobia. In each condition, a therapist takes the subject through a standardized training for ninety minutes, accompanied by physiologic monitoring of heart rate, respiration rate, and galvanic skin response. Using a modified ten-point Subjective Units of Discomfort Scale (SUDS) [11], subjects will be asked to report their level of anxiety approximately every five minutes. As their SUDS decreases, they will move to the next floor as specified by protocol. The types of instructions and verbal encouragement that the therapists use are also set by protocol. After completion of exposure training, each subject is returned to the in-vivo situation for a repeat behavioral approach test. This will allow a direct comparison of improvement among all conditions, as well as examine the ability of VR exposure to generalize to real life. A subsequent behavioral approach test one week later will check for the maintenance of therapeutic benefits. Besides physiologic measurements, subjective measurements of the impact of these exposures are made through questionnaires. We will be able to compare rapidity of treatment as well as total improvement and maintenance of that improvement.

As we have not yet obtained permission for the use of human subjects, no experimental data is yet available. Our current process of modeling and experimental design has been instructive, teaching us lessons in two areas. The first relates to technical limitations of VR and the virtual environment design process. The second relates to limitations of psychological knowledge of individual responses to virtual environments.

Virtual reality has several technical limitations that impair its ability to simulate reality. Some of these have already been well discussed in the literature. Disparity between movements of the head and corresponding changes in the virtual environment results in "simulator sickness" [12, 13]. Flat projections create an inability to adjust focal depth as related to the supposed depth of the virtual object [14]. Directional sound is often ignored in VR environments [15], despite its degree of contribution to increasing perceived realism [16]. Our current model suffers to a degree from all of these problems, although use of the CAVE and faster computers solves most of these difficulties. In fact, the ability of the CAVE to eliminate many traditional problems of VR [17] makes this experimental design unique.

Other limitations arise from the design process itself. Reality is rich with many small details that must each be examined for their importance and impact. An ideal virtual environment would reproduce all of these. We intentionally left out some parts of the experience because of the difficulty in accurately replicating them. We do not simulate reflections off of the window, window tinting, or dust, which all commonly serve as cues that a barrier is present. Missing outdoor elements include a lack of people walking along the sidewalk, birds flying above or below, wind and other forms of weather, or the consistency of cast shadows, although we plan to add some of these elements to the final model. The feel of the hard waxed tile floor in the hospital can not be replicated in the CAVE, and we do not attempt to simulate the shiny, reflective aspect of the surface since this appearance would be incongruous with the feel of the CAVE floor. In general, these choices have also been guided by a subjective perception that they are not critical to the experience. Unfortunately, we have little certainty about which elements are psychologically important. Our clinical experience suggests that each individual is different and is sensitive to different cues. This sensitivity is related to the sum of experiences both conscious and unconscious that are psychologically linked to their fear. Someone with a fear of losing control and throwing themselves out a window might react more strongly to the window cues. Someone else whose fear of heights is tied to unconscious thoughts of hurting themselves and others might react more strongly to the presence of people outside.

Research into the psychological interplay with virtual reality is still at a very early stage. Our review of the literature suggests that no studies have yet been done which compare VR treatment to an equivalent treatment in-vivo. Studies usually use a no-treatment condition as a control, demonstrating that VR treatment is effective, but not comparing it with treatment in reality. Furthermore, the literature provides little guidance to suggest what variables make someone sensitive to a VR environment. Perhaps an individual's ability to visualize themselves in other situations, their ability to suspend disbelief, or their tendency to dissociate from their environment have an impact on how they react to a virtual world. Our experimental design allows for the exploration of some of some of these questions.

Our current experimental procedure is aimed at determining where the differences exist between virtual reality and reality in the context of acrophobia treatment. It compares the treatment of subjects in a real location to treatment in a VR simulation of that real location. Use of minimally intrusive technology and appropriate protocol design allow us to make treatment in these two settings as similar as possible, permitting us to examine VR itself as an independent variable.

The process of attempting to replicate the real world in a virtual world highlights the weaknesses of VR and potential areas for change. New technologies like the CAVE overcome many of the technical difficulties, although they leave the problem appropriate environmental design. There will always be practical limits on the number of detailed environmental aspects we choose to add to a simulation. As we do not know what particular aspects will be most important in any particular application, the selection of which elements to add and which to discard is difficult. The attempt to define and rank the value of such aspects is an important activity. In our work, we purse defining what environmental attributes are the most psychologically challenging for sufferers of acrophobia, and how the level challenge for each attribute varies for each individual. This concept could be extrapolated to the idea of creating a more general library of elements for virtual environments. Such a library could then serve to improve the process of VR design by creating standardized elements that could be tested for psychological impact, allowing more appropriate selection during environment design.

Our process of using VR for psychiatric purposes also suggests lessons for needed study on the interaction of psychological factors and virtual reality. We have little research on the psychological factors that influence the impact and effectiveness of VR. Many papers have investigated the concept of the subjective sense of "presence" or "being there" in a virtual environment, discussing issues of meaning and measurement [18] and environmental factors which alter it [19,20]. They fail, however, to examine psychological factors that are important to presence or even if presence is important in its psychological impact. In some applications, strong presence is detrimental [21]. In this study, our goal is to reduce a patient's fear of heights with a stable improvement in their abilities. It is possible that in some cases, a high fidelity of replication of reality will slow or otherwise limit this process and that cartoon-like buildings would work better for an individual than the texture mapped ones in our model. It is also possible that exposure to extreme or even clearly impossible situations might be beneficial. We do not yet know the answer to these questions. Further research needs to be done. We hope that work using the paradigm we describe and our own future work will elucidate what individual psychological factors influence the individual's reaction to virtual environments as this will help us define the limits and potential of VR in the treatment of psychiatric conditions.

[1] Glantz K, Durlach NI, Aviles WA: Virtual Reality (VR) and Psychotherapy: Opportunities and Challenges. Presence: Teleoperators and Virtual Environments. Vol 6 (1), pp. 87-105, February 1997.
[2] Rothbaum, B.O., Hodges, L.F., Watson, B.A., Kessler, G.D. and Opdyke, D. Virtual reality exposure therapy in the treatment of fear of flying: a case report. Behaviour Research and Therapy. Vol 34,5/6, pp. 477-481, 1996.
[3] North MM, North SM, Coble JR. Virtual Environments Psychotherapy: A Case Study of Fear of Flying Disorder. Presence: Teleoperators and Virtual Environments. Vol 6 (1), pp. 127-132, February 1997.
[4] Carlin, A.S., Hoffman, H.G. and Weghorst, S. Virtual Reality and Tactile Augmentation in the Treatment of Spider Phobia: A Case Study. Behaviour Research and Therapy. February 1997.
[5] Rothbaum, B.O., Hodges, L.F., Kooper, R., Opdyke, D., Williford, J. and North, M.M. Effectiveness of computer-generated (virtual reality) graded exposure in the treatment of acrophobia. American Journal of Psychiatry. Vol 152 (4), pp. 626-628, April 1995.
[6] American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Washington DC.: APA Press, 1994.
[7] Cohen DC. Comparison of self-report and overt-behavioral procedures for assessing acrophobia. Behavior Therapy. Vol 8, pp. 17-23, 1977.
[8] Marks IM, Mathews AM. Brief standard self-rating for phobic patients. Behaviour Research and Therapy. Vol 17 (3), pp. 263-267, 1979.
[9] Michelson L. Treatment consonance and response profiles in agoraphobia: the role of individual differences in cognitive, behavioral and physiological treatments. Behaviour Research and Therapy. Vol 24 (3), pp. 263-275, 1986.
[10] Menzies RG, Clarke JC. The etiology of fear of heights and its relationship to severity and individual response patterns. Behaviour Research and Therapy. Vol 31 (4), pp. 355-365, May 1993.
[11] Wolpe J. The practice of behavior therapy. Pergamon: New York, 1969.
[12] Hettinger LJ: Visually induced motion sickness in virtual environments. Presence: Teleoperators and Virtual Environments. Vol 1, pp. 306-307, 1992.
[13] Kennedy RS, Lane NE, Lilienthal MG, Berbaum KS, Hettinger LJ: Profile analysis of simulator sickness symptoms: Application to virtual environment systems. Presence: Teleoperators and Virtual Environments. Vol 1, pp. 285-301, 1992.
[14] Wann JP, Rushton SK, Mon-Williams M. Natural Problems for stereoscopic depth perception in Virtual Environments. Vision Research. Vol 19. pp 2731-2736, 1995.
[15] Max ML, Gonzalez JR. Blind Persons Navigate in Virtual Reality (VR); Hearing and Feeling Communicates "Reality." Medicine Meet Virtual Reality. K.S. Morgan et al. (Eds.) IOS Press, pp. 54-59, 1997.
[16] Hendrix C, Barfield W: The Sense of Presence within Auditory Virtual Environments. Presence: Teleoperators and Virtual Environments. Vol 5 (3), pp. 290-301, Summer 1996.
[17] Wann J, Mon-Williams M. What does virtual reality NEED?: human factors issues in the design of three-dimensional computer environments. International Journal of Human-Computer Studies. Vol 44, pp. 829-847, 1996.
[18] Sheridan TB. Further Musings on the Psychophysics of Presence. Presence: Teleoperators and Virtual Environments. Vol 5 (2), pp. 241-246, Spring 1996.
[19] Welch RB, Blackmon TT, Liu A, Mellers BA, Stark LW. The Effects of Pictorial Realism, Delay of Visual Feedback, and Observer Interactivity on the Subjective Sense of Prese: Teleoperators and Virtual Environments. Vol 5 (3), pp. 263-273, Summer 1996.
[20] Hendrix C, Barfield W. Presence within Virtual Environments as a Function of Visual Display Parameters. Presence: Teleoperators and Virtual Environments. Vol 5 (3), pp. 274-289, Summer 1996.
[21] Ellis SR. Presence of Mind: A Reacation to Thomas Sheridan's "Further Musings on the Psychophysics of Presence." Presence: Teleoperators and Virtual Environments. Vol 5 (2), pp. 247-259, Spring 1996.