Wilson (1996) suggests that many new technologies undergo some form of public backlash and that this can be because of a number of reasons. The technology may be being oversold, the impact on the environment and the lives of people is generally unpleasant (e.g. the motor car) or there is a backlash due to the frequency and duration of use (e.g. the office computer workstation). Both he, and Howarth (1994) draw parallels with the introduction of Visual Display Units (VDUs) in the early 1980s and the health speculation surrounding the use of VR equipment.
With VR systems, similar problems appear to have come to light. The introduction of VDUs was said to cause physical, physiological and psychological problems and these were exacerbated by the technological limitations of the screens at that time.
Herein, lies one of the biggest problems in defining problem areas as VR technology is new and is constantly being improved and developed by different manufacturers. One only has to look at the range of HMDs on the market to appreciate the different approaches to the design of one facet of VR technology. Therefore, one must consider the attributes of the VR system that is being used.
Similarly, one must expect the user population to extremely diverse. This is an important consideration when one considers the design of equipment that must be placed on the body. Custom fitting may be relatively easy to achieve in the workplace where there may be limited users but the use of VR for public space entertainment applications means that the equipment is available to a potentially large user base who may have little or no knowledge of the health and safety issues involved and little knowledge as to how to use the equipment correctly.
Finally, one must consider the demands of the task itself. For example, a target tracking task which induces considerable head movement may be more uncomfortable than a task which requires little head movement such as reading a document.. However, one must consider the complex interplay of the various factors. If system lag is high and the headset is heavy, tracking may be uncomfortable. If system lag is low, and the headset is light such tracking tasks may not be such a problem. Kolasinski (1995) identifies the number of potential factors that may be associated with the onset of simulator sickness symptoms in a VE. These factors are reproduced in Table 3.1 below.
When one considers that these factors are implicated in the onset of simulator sickness alone it becomes clear that the genesis of symptoms may be due to an extremely complex interaction between the factors in each of the three areas represented above. This may be the case for all suggested symptoms, and the current focus of much research is in trying to determine the most important factors in the onset of particular symptoms.
Ultimately, there is the question of which side effects DO actually occur, and which are more fanciful notions brought about by the general hype surrounding the technologies, applications and symptoms. There has certainly been rigorous research, and no one involved in VR disputes that fact that unpleasant symptoms can occur but little is known. Currently unanswered questions include issues such as the duration of symptoms, predicting individual susceptibility, coping strategies adopted by VR users and the impact upon the VR task and consequent activities such as driving or operating heavy machinery.
The possible side effects that have been suggested to date can be divided into three main areas. Table 3.2 below, outlines these effects.
It must be stressed that these are suggested symptoms and it must also be made clear that certain symptoms may only occur with certain types of VR implementation. The following sections will discuss each category in turn, highlighting the possible effects in each category that can be expected with the different implementations of VR.
It is important to consider the physical issues involved in the use of VR equipment. In some cases, the interaction techniques used are unique to the system and may present specific problems (such as the weight of some HMDs). In other cases, the demands may be well understood but the user must be aware of the type of problems that may arise. The following sections detail some of the physical issues that must be considered when one uses VR equipment.
The postural demands of desktop VR systems can be considered to be identical to using a standard PC in that one would expect the user to be sedentary at a desk or worktop. There is much literature on the subject of VDT workstation design (Grandjean, 1987; Lueder, 1986) and indeed in this country, regulations have been laid down as to safe working practices with such installations (Health and Safety Executive, 1983).
Posture issues with semi-immersive implementations may be a little more complex. One would expect that using a large monitor would provide a similar situation to that proposed above but it is quite possible that users of large screen installations may be standing, wearing shutter glasses and using some form of interaction device. In this case there may be some abnormal postural demands as the user interacts with the VE. However, it must be noted that shutter glasses are relatively lightweight (CrystalEyes weigh 3.3 ounces), as are the interaction devices, so the postural demands will be determined primarily by the interaction with the VE and the demands of the task.
Probably of biggest concern are the postural demands of fully immersive implementations. Immersive HMDs tend to be considerably heavier than shutter glasses (see also 3.1.3) and therefore provide additional load on the body. In the authors own research experience, users in an immersive headset were often observed propping up the weight of the headset using one hand, and interacting with the VE using the interaction device with the other. The long term effects of this unnatural posture are difficult to quantify, but user discomfort is often reported while using HMDs.
Repetitive strain injury is now relatively well known due to a number of high profile court cases such as the damages case lost by BT in December 1991. RSI injuries result from carrying out prolonged repeated activities using rapid carpal and metacarpal movements. Injuries such as tendonitis, fibrous tissue hyperplasia and ligamentous injury have been reported in association with using standard input devices such as mice, joysticks and keyboards and can leave the sufferer functionally disabled.
One would therefore expect RSI to be an issue with desktop VR systems and semi-immersive systems using standard interface devices. More difficult to determine is whether users of such devices as 3D mice or datagloves will suffer in a similar way. Howarth (1994) argues that the more natural gestures allowed by such devices may alleviate RSI problems although, again, this is dependent on the design of the interface. Any interaction technique that requires continual repetitive movement is undesirable (as it would be for a real world task), so it is important that VE developers appreciate this when interaction strategies are formulated.
The issue of headset weight is restricted to fully immersive systems utilising HMDs. Due to the limitations of current display screen technology many HMDs are still relatively bulky in terms of both size and weight, with some HMDs on the market weighing in at approximately 2kg (8lbs). Improperly fitting HMDs can cause discomfort and additional strain can be placed on the neck by large masses. So (1994) suggested that additional strain could be placed on the neck if the user remains relatively still in a VE and one must also consider the additional inertia caused by the HMD as the user executes a head movement. All these problems are exacerbated by heavy, poorly fitting and poorly balanced headsets and users should be made aware of the correct fitting techniques. Future developments will almost certainly mean that headsets will reduce in weight if they are to be commercially acceptable, or developments such as the BOOM system may be considered where the weight of the headset is not supported by the wearer. Ultimately, this issue will cease to be a major drawback of HMDs, but currently this problem is one of the biggest stumbling blocks for fully immersive systems.
Gupta, Wantland and Klien (1996) suggest that much of the peripheral equipment used in VR are potential fomites. A fomite is a harmless object that is able to harbour pathogenic organisms and as such, may serve as an agent for the transmission of infections. They go on to suggest that airborne pathogens and skin flora thrive in environments similar to those of HMDs and hand controller devices. An additional consideration here is that HMDs are often of enclosed design and generate a considerable amount of heat in powering the displays. This can often lead to some sweating for the user particularly if the immersive task demands a certain amount of physical activity.
To put these concerns into context, one must consider that traditional keyboards and mice can be regarded as fomites and their popularity is evident. This issue does become pertinent however, when one considers the use of fully-immersive systems in public space applications. Some HMD manufacturers now provide removable, washable pads positioned where the HMD cradle is in contact with the skin of the wearer.
Both Gupta et al. (1996) and Viirre (1993) suggest there may also be a risk of injury while the user is using a fully immersive HMD. As Viirre suggests, when a user is wearing an HMD, they are functionally blind in real world terms. This can lead to problems due to collision with real world objects or possibly the VR system cabling and even if the user has some external vision, the compelling immersive scene may distract attention from the outside world. Additionally, many HMDs also provide sound cues for the user that effectively cut off aural stimulation from the real world.
Effectively, the situation then occurs where the user is actively involved in a VE that may require a certain amount of movement and yet is receiving little or no input from real world cues. Consequently, it is important that the user is kept in a safe area, preferably protected with railings. A good example with this in mind is the safety barrier utilised in public space applications, where the user is enclosed in an limited area protected by a padded, circular railing during immersion.
Physiological issues are probably the most well documented and at present, well researched sickness issues currently attributed to VR systems. Indeed some reported physiological side-effects such as simulator sickness have been researched for some time. Of the possible physiological side-effects, visual symptoms and motion sickness type symptoms appear to cause the most concern. Consequently, much of the research into physiological effects has been concentrated in these areas.
The visual presentation of the virtual environment is extremely important. The processing and organisation of visual input involves the use of a larger portion of the brain than for any other sense and North (1993) estimated that for a complex task such as driving, 90 per cent of the received information is visual. It is therefore not surprising that manufacturers go to great lengths to provide a compelling visual environment.
The implementation of such visual environments for VR applications have not been without problems. Although the health and safety issues associated with monoscopic displays are relatively well understood, the issues associated with stereoscopic images are more complex.
Monoscopic implementations of desktop and semi-immersive systems essentially provide a visual scenario that is common. The visual environment created using a monoscopic desktop VR system is more or less identical to that experienced when one uses a standard office computer. Much research has been carried out in this area (National Research Council, 1983; Dain, McCarthy and Chan-Ling, 1988; Daum, Good and Tijerina, 1988; Shen, Chiu, Wang and Lo, 1988; Howarth and Istance, 1985) and a large body of literature is available. The principle visual side-effects of this implementation are asthenopia (eyestrain) and visual fatigue. Regulations for safe use of VDU displays have been laid down in this country in the Health and Safety (Display Screen Equipment) Regulations (1992) and internationally in ISO 9241-Part 3: Visual display requirements. Similarly, monoscopic semi-immersive systems provide a viewing scenario similar to a cinema, that of viewing a large screen at some distance. This could be considered the least visually taxing of all the implementations, as the viewing distances involved are usually close to optical infinity (approx. 4m or greater), and thus reduce the accommodative (focus) demand for the user.
As suggested however, stereoscopic semi-immersive systems may have additional side-effects. One of the main reasons for the genesis of side-effects is suggested to be the dissociation of accommodation and convergence in the visual system.
The dissociation of accommodation and convergence when using stereoscopic displays
In 1993, Mon-Williams, Wann and Rushton reported physiological symptoms in a number of subjects following immersion in an HMD. Of the 20 subjects who took part in their experiment, 12 complained of symptoms such as headache, eyestrain and nausea and 4 demonstrated a transient reduction in binocular visual acuity. The subjects also demonstrated signs of binocular stress including changes in heterophoria and an increase in near point of convergence. Mon-Williams and Pascal (1995) suggested that these signs of visual/binocular stress were linked, not only to poor image quality and close working distance of the screens, but more fundamentally with the discrepancy between accommodation and convergence demand when using a stereoscopic HMD. This problem will occur in any stereoscopic system where the main image is produced on a flat screen and stereo images are provided by displaying slightly different images to each eye.
In the natural environment, accommodation and convergence are intrinsically linked. If one accommodates (focuses) on a near object, the eyes will automatically converge. Similarly, if focus is changed to a distant object, the eyes will automatically diverge slightly (see Figure 3). When using stereoscopic display devices such as shutter glasses or HMDs this is not the case. In this situation, the accommodative demand is always constant but the convergence demand changes as the user regards objects at different geometric depths in the virtual world. This accommodation/convergence is not a natural occurrence and has been said to result in visual stress.
Figure 3. The relationship between accommodation and convergence in the real world (solid line) and when using a stereoscopic display (dashed line). The shaded area is the zone of clear, comfortable, binocular vision. (from Howarth, 1996)
More recent research (Peli, 1995) seems to suggest that although some symptoms occur due to this problem, the issue may not be as important as it was once thought. This may be due to slack within the visual system that can overcome such demands. However, Howarth (1996) suggests that if one is using a stereo display, it is desirable that geometric images are produced closer than the screen focusing plane as this technique produced less discomfort in subjects.
Additional visual considerations with HMDs
A further issue that must be considered with HMDs is the relationship between the inter-screen distance (ISD), inter-ocular distance (IOD) of the lenses and the inter-pupilliary distance (IPD) of the user (Howarth, 1997). The ISD refers to the distance between the centres of the two images on the display screen, the IOD refers to the distance between the optical centres of the lens systems installed in the HMD and the IPD refers to the distance between the centres of the pupils of the user. North (1993) indicated a mean IPD for UK men of approximately 64mm and a mean IPD for UK women of approximately 62mm. Initially, it was suggested that a mismatch between the users IPD and the IOD of the HMD may give rise to visual discomfort symptoms, transient heterophorias or muscle imbalances in the users eyes. A study by Regan and Price (1993b) reported that the greater the mismatch between the two measures, the greater the reported side-effects.
Figure 4. The relationship between ISD, IOD and IPD in a head-mounted display. The shaded grey box represents the HMD casing (adapted from Howarth, 1997).
The reason for the symptoms has been reported to be that viewing of the lens system on an off-centre axis effectively leads to distortion of the image as a convex lens can be regarded as two prisms. The consequences of prism adaption are well reported (Maddox, 1893; Sethi, 1986; Schor, 1986) and include eyestrain and visual discomfort. Recent research (Howarth, 1997) would appear to indicate that the relationship between the IOD and ISD of the headset is the most important consideration as this affects the image position for the viewer. Some HMD manufacturers now provide IPD adjustability on their HMDs, but this is more important in determining the portion of the image that can be seen in stereo (stereo overlap) of the image.
An additional consideration is that of the HMD users visual ability. Most people are not perfectly sighted meaning that the design of the HMD (or indeed shutter glasses) must allow the user to wear their normal optical correction or provide some sort of adjustability that allows each individual user to tailor the HMD to their own specifications. Some HMD manufacturers now provide focus adjustment on their models. However, providing this functionality does not mean that the user will adjust it correctly without adequate instruction and incorrectly-adjusted optics may be more detrimental to the user as optics with no adjustment at all.
Finally, one must consider the quality of HMD displays. Currently, the most inexpensive displays use LCD displays which tend to have low resolution, poor contrast and low levels of illumination. An informal study by Robinett and Rolland (1992) estimated users vision of a virtual snellen chart to be 6/60 but this situation is improving. Howarth (personal communication, 1997) estimated users vision of a later model low-cost, commercially available HMD to have improved to 6/24 but this still represents a considerable degradation of normal vision (6/6).
A review of the literature in this area certainly provides compelling evidence that physiological changes occur in the visual system following HMD immersion and that subjects do report specific symptoms. In our own studies, we re-measured visual changes at 5 and 10 minute intervals post-immersion. We found that for the vast majority of subjects any measured visual changes had reverted back to the original readings, suggesting that such changes are transient, and no evidence of long term effects has yet been published. Furthermore, physiological changes in the visual system (such as transient heterophorias) have also been shown to occur following more mundane tasks such as reading hardcopy text (Pickwell, Jenkins and Yekta, 1987) and using a VDT display (Howarth and Costello, 1996).
Research continues in this area, both in determining the appropriate optical set-up and in designing adjustability in to the HMD itself. Peli (1996) has recently published preliminary recommendations for optical tolerances in HMDs and manufacturers continue to develop the HMD to provide in-built adjustability in the areas of focus, convergence and adjustability.
Simulator sickness is by no means a new phenomenon. It is similar to motion sickness, which has existed for as long as humans have used additional modes of transportation, but can occur without any actual motion of the subject. The first documented case of simulator sickness occurred in 1957 and was reported by Havron and Butler in a US Navy helicopter trainer. The most common identifiable symptoms are general discomfort, nausea, drowsiness, headache and in some cases vomiting.
A number of recent research projects have identified symptoms in users of off-the-shelf VR systems. Regan and Price (1993a) reported that 61% of 146 subjects reported some symptoms following a 20 minute immersion in a VE. Costello and Howarth (1996c), in a study of one non-immersive and three fully immersive commercial VR systems, indicated statistically significant increases in reports of disorientation for all three immersive systems and significant increases in reported nausea for two of the three immersive systems following a 20 minute immersion period. No significant increases in these symptoms were reported by the same subjects using the non-immersive system.
Kennedy and Frank (1985) suggest that simulator sickness is both polygenic (has many sources) and polysymptomatic (induces many symptoms) indicating how difficult it is to predict individual susceptibility and to measure the effects on the user. The research is not all bad news, however, as in a later paper Kennedy, Berbaum, Lilienthal, Dunlap, Mulligan and Funaro (1987) claim that only about 30% of individuals will become ill even under the worst simulator conditions.
Effects of simulator sickness symptoms on the task in a VE
The underlying problem with such symptoms is that they affect the performance of the user both in the VE/simulator and in some cases, afterwards. Simulator sickness symptoms may distract the users attention from the task in hand and, in more severe cases, can actually cause the user to leave the simulation environment (Costello and Howarth, 1996c). This reduces the effectiveness of the simulator and can cause a reluctance on the part of the operator to use the VE/simulator again. The impact on consequent tasks has also caused some concern. This is due to the fact that the symptoms developed in a simulator/VE appear to last far longer than those developed in motion sickness. There are several reports of longer term symptoms and delayed after-effects in the literature. Kellogg, Castore and Coward (1980) reported after effects following simulator use occurring 8-10 hours after leaving the simulator and experimentation carried out by the author has produced anecdotal reports from subjects that simulator sickness symptoms following immersive VR use have persisted for up to two days in some cases.
This has obvious consequences for other tasks such as driving and Pausch, Crea and Conway (1992) indicate that many US Army aviation units have adopted a policy that prohibits flying an aircraft within 6 hours of a simulator flight.
Of the number of possible theories that have been proposed as the cause for simulator sickness (more detailed reviews of the various theories can be found in literature (Pausch et al. (1992), Kennedy and Frank (1985), Kolasinski (1995)), the theory of sensory conflict is currently regarded as the most popular. This theory was proposed by Steele in the 1960s under the name perceptual conflict and is also referred to in various texts as cue conflict. Conflict occurs when signals from the various spatial senses, the eyes, the balance organs and the non-vestibular position senses are in conflict with one another and do not correlate with signals received in past experience.
There are a number of ways in which this sensory conflict can occur when using large screen semi-immersive and HMD based VR systems.
Using a large screen semi-immersive system is similar to viewing an IMAX type cinema screen in that the viewing area takes up a considerable portion of the viewers visual field. This gives rise to a phenomenon known as visually induced motion sickness (VIMS) (McCauley and Sharkey, 1992). Here, conflict occurs if there is apparent motion of the image on the screen. The users visual senses indicate that the body is in motion but the balance organs and non vestibular senses indicate that the body is in fact static (assuming there is no motion base). This conflict can lead to the onset of the symptoms described above.
In an HMD based system, there may be an additional cause of sensory conflict. This is related to the time lag experienced when one uses a tracked HMD. In all VR systems, there is a delay between the execution of a head movement and its representation on the screen. In 1994 Bolas suggested that immersive systems with passive LCD screens may take in the order of 200 milliseconds to recalculate and redraw the screen image. This provides further potential for sensory conflict to occur. If one executes a head movement and the screen image is not changed for 200 ms the eyes will detect no movement but the balance organs indicate that the head is moving. Similarly, the instant that a head motion stops, it will take 200ms for the screen image to stop moving. Here the eyes indicate movement and the balance organs indicate that movement has ceased causing further sensory conflict. Although the increase in computational power will mean that such lags will be reduced, the contribution of such lag to motion sickness symptoms remains a controversial area. (see Kolasinski, 1995). Furthermore, it must be noted here that the symptoms stll occur in the absence of some putative, causal factors, e.g. accommodation/convergence conflict and tracker lag (Howarth and Costello, 1997).
Although a number of psychological effects have been proposed in Table 3.1, there appears to be little research carried out in this area to date. Previous research carried out into attitudes to computer games (Shotton, 1989) suggests that people may become "hooked" and one could postulate that in many ways the same could be said of VR. Indeed it could be argued that VR has greater potential to hook users due to a more compelling experience but there is little evidence to support this at present. However, as Howarth (1994) argues, people often become obsessed with their hobbies and it is important to view the use of VR equipment in this context.
Further behavioural effects that have been suggested (Wilson, 1996) include hallucinations, dissociation, literalisation and retreat from reality and a number of papers in a special health and safety issue of the VR journal, Presence, deal with these issues in more detail.
Similarly, there has been concern over ethical issues such as the construction of violent or pornographic worlds. Here again, there are similarities with the computer games industry where there has been some concern as to the violent content of computer game material. This does not seem to have been particularly damaging to this industry although there have been calls to censor or restrict certain games to particular age groups.
Gupta et al. (1996) also suggest that additional problems caused by VR may include anxiety and claustrophobia. Although it is possible that these effects may occur, there appears to be little evidence to date of any lasting effects.
Both Howarth (1994) and Wilson (1996) point out that as well as looking for problems, it is important to recognise that VR techniques may also prove beneficial in many applications. Currently there is much research work being carried out in the VR field that will be of benefit to users.
In terms of physical issues, more natural interaction techniques may reduce static posture problems, the use of LCD screens may minimise vision problems associated with CRT screens and physical loads associated with keying (Wilson, 1996).
VR also provides a much improved technique for health and safety training, though as Howarth (1994) suggests, this role is largely hidden. VR techniques can be used in ergonomic assessment of workspace layout, for rapid prototyping of control interfaces, for the simulation of potentially dangerous environments such as nuclear plant maintenance and in education and training of users in areas such as the maintenance of complex machinery. As Howarth says, the fact that the use of VR has helped an operator avoid an accident or react correctly in the event of a crisis is largely unseen.
VR also has numerous applications that can be directly related to health care. In a white paper on the use of Virtual Environments for Health Care, Moline (1995) indicates several areas where patient care can be assisted by VR techniques. These include:
North, North and Coble (1996) provide an overview of current work in the use of VR techniques to reduce phobias in their book VR Therapy.
Graphics Multimedia Virtual Environments Visualisation Contents