Immersive Virtual Reality for reducing experimental ischemic pain.

Hunter G. Hoffmana,b, Azucena Garcia-Palaciosa,c, Veronica Kapaab, and Jennifer Beecherab, Sam R. Sharar

aHuman Interface Technology Laboratory, University of Washington

bDepartment of Psychology, University of Washington

cUniversidad Jaume I (Castellon Spain)

dDepartment of Anesthesiology, University of Washington School of Medicine

*Corresponding author: Hunter G. Hoffman, Human Interface Technology Laboratory, University of Washington, Seattle, WA., 98195

phone number: (206) 616-1496

fax number (206) 543-5380

in press in the International Journal of Human Computer Interactions.

Hoffman, H.G., Garcia-Palacios, A., Kapa, V.A., Beecher, J. & Sharar, S.R. (2003). Immersive Virtual Reality for reducing experimental ischemic pain. International Journal of Human-Computer Interaction, 15, 469-486.



keywords: pain, virtual reality, distraction, analgesia


This study explored the novel use of immersive virtual environments as a non-pharmacologic pain control technique and whether it works for both females and males. Fourteen female and eight male students underwent pain induced via a blood pressure cuff ischemia lasting 10 min or less. Pain ratings increased significantly every two min during the no distraction phase (zero to eight min) and dropped dramatically during the last two min period when participants were in the virtual environment (a 59% drop for females and a 41% drop for males). Five visual analog pain scores for each treatment condition served as the primary dependent variables. All 22 participants reported a drop in pain in the virtual environment, and the magnitude of pain reduction from the virtual environment was large (a 52% drop) and statistically significant. This is the first study to show immersive virtual environment distraction is also effective for females. The results show that virtual environments can function as a strong nonpharmacologic pain reduction technique, showing the same pattern of results obtained from recent clinical studies using virtual environments with burn patients during physical therapy. Practical applications of virtual environment pain reduction, and the value of a multidisciplinary approach to studying pain are discussed.

Immersive Virtual Reality for reducing experimental ischemic pain.

Morphine-related drugs (i.e., opioids) are effective for treating the pain of burn patients resting in bed. In sharp contrast, while opioids are essential, they are usually inadequate for procedural pain of severe burn patients. The majority of burn patients receiving standard analgesic pharmacologies rate their pain during burn wound care procedures as severe to excruciating (Carrougher et al., submitted; Perry, Heidrich, & Ramos, 1981, see also Melzack, 1990). Managing pain from severe burns is particularly difficult because of the high frequency and intensity of painful procedures. Dressings are changed and wounds cleaned daily to prevent infection. Extreme pain increases the time and effort required to complete wound care, and is difficult for both patients and nurses. In addition, patients must undergo aggressive physical therapy (skin stretching and muscle building) to counteract contraction of the injured or newly grafted skin, and muscle atrophy from inactivity during the healing process. Intense pain during physical therapy can discourage patients from completing their exercises (Ehde, Patterson & Fordyce, 1998). Opioids have common side effects such as constipation, nausea, delirium, reduced respiration, and the less common but ever-present possibility of respiration failure. These factors complicate pain management based solely on pharmacologies. This fact has important medical implications because the amount of pain reported during hospitalization has been associated with postdischarge physical and psychologic recovery (Ptacek, Patterson, Montgomery, & Heimbach, 1995). For example, non-compliance with physical therapy can make additional surgery necessary, and/or can limit the final post-healing maximal range of limb motion, exaggerating burn-related physical disability (see Ward, 1998). New pain control techniques are needed that can be used in addition to traditional pharmacologies. The present study explores the use of immersive virtual environments to help reduce pain.

There is a strong psychological component to pain perception, and non-pharmacological psychological pain control treatments used adjunctively (in addition to) opioid analgesics can be effective (Patterson, 1992; 1995; Patterson, Everett, Burns & Marvin, 1992; Everett, Patterson & Chen, 1990). Such results are often interpreted within the context of a gate control mechanism. According to the gate-control mechanism proposed by Melzack and Wall, (1965), an incoming pain signal of a given neurological intensity can be interpreted as more painful or less painful, depending on what the patient is thinking and/or attending to at the time. Previous experience, expectation, culture, focus of attention and anxiety are psychological factors that can contribute strongly to the subjective experience of pain (Melzack, 1998). Such psychological influences are thought to modulate, inhibit, or modify the nociceptive signals at the spinal cord, which serves as a gate to control the intensity of pain signals ever reaching the higher cortex.

At a less theoretical level, regardless of the mechanism, the effectiveness of cognitive interventions involving distraction for pain reduction is already supported in the literature. In a meta-analysis of adjunctive treatments, Fernandez and Turk (1989) found that cognitive/behavioral strategies significantly reduced ratings of pain in 85% of the 47 studies analyzed, and studies that involved distraction were among the most effective. A growing number of recent laboratory and field studies from several disciplines are consistent with this conclusion (see Tan, 1982 for a review). Pain tolerance to laboratory-induced pain increased significantly for participants viewing humorous or repulsive movies (Weisenberg, Tepper, and Schwarzwald, 1995). Kozarek, Raltz, Neal, Wilbur, Stewart, and Ragsdale (1997) found that distraction with movies improved pain tolerance to gastrointestinal procedures in 82% of patients. Cartoon distraction paired with coaching reduced children’s distress during immunizations (Cohen, Blout and Panopoulos, 1997) and a program combining behavioral training with a video cartoon/movie distraction technique allowed 9 of 11 child patients to avoid traditional sedation for daily cancer radiation treatments (Slifer, 1996). Most relevant to the present project, Miller, Hickman and Lemasters (1992) found that burn patients shown scenic movies and music during burn wound dressing changes rated their pain 13% lower than a "no distraction" control group. Further controlled evaluations of pain distraction during burn wound care are needed, and a technology capable of creating a more dramatic reduction in pain than videos might be more widely adopted in hospital practice.

Immersive virtual environments (i.e., involving a head mounted display that occludes viewing the real world) is state-of-the-art technology for capturing attention. A VR computer system sends video output to 2 miniature TV LCD screens inside a wide field-of-view VR helmet. Position sensors attached to the head and hand keep track of the user’s head and hand position and orientation and feed this information into the virtual environment (VE) computer. When patients move their heads (e.g., look up toward the ceiling), the computer quickly updates the visual images in the artificial environment accordingly. These real time changes in sensory input, in response to their actions in the VE, give patients the illusion that they are inside the computer-generated environment, a sensation referred to as "presence." Presence is the essence of virtual environments (Laurel, 1995). In our preliminary case report (Hoffman, Doctor, Patterson, Carrougher, & Furness, 2000) patients saw a virtual kitchen complete with kitchen countertops, a window overlooking a partly cloudy sky, as well as three-dimensional cabinets, and doors that could be opened and shut. Patients could pick up a teapot, plate, toaster, plant, or frying pan by inserting their cyberhand into the virtual object, and clicking a grasp button on their 3-D mouse. Each patient also physically picked up a virtual plate possessing solidity and weight, using a mixed-reality force feedback technique (Carlin, Hoffman, and Weghorst, 1997; Hoffman, 1998). The multi-sensory, highly interactive, verbal, spatial, visual, auditory, and tactile nature of VEs makes the experience difficult for the brain to ignore, since it draws upon multiple attentional resources (e.g., Wickens, 1992) each of which has a limited capacity. Distractions can draw attention away from patients’ mental processing, thereby decreasing the amount of pain consciously experienced by the patient (McCaul and Malott, 1984; Farthing, Venturino, and Brown, 1984). VEs may prove to be an exceptionally attention-grabbing experience distraction. The cognitive/attentional state of the participant could be changed by drawing their minds inside a 3-D, immersive, computer-simulated VE.

Although case reports are inherently inconclusive (Campbell & Stanley, 1963), they can sometimes be revealing. Hoffman, Doctor, et al., (2000) measured pain levels of two pre-adult patients undergoing burn wound dressing changes, wound cleansing, and staple removal from skin grafts while adjunctively being distracted by a VE and by Nintendo64 (order counterbalanced) during a single wound care session. VE distraction was dramatically more effective as an analgesic than the Nintendo64 video game. Interestingly, that study also suggests why the VE works better; Presence in the Nintendo games was much lower than the illusion of presence in the VE. More recently, Hoffman, Patterson and Carrougher, (2000) found a statistically significant reduction in pain while in a VE in a controlled within-subject clinical study of severe burn pain during physical therapy (VE vs. no distraction). While in the VE, patients reported a 47% drop in the amount of the physical therapy session they spent thinking about their pain/wound care, a 33% drop in pain unpleasantness (an emotional component of pain), and a 22% drop in pain intensity.

In the present analog study, experiment 1 is a conceptual replication of Hoffman, Patterson and Carrougher (2000) in a controlled laboratory environment testing healthy uninjured undergraduates. Pain was induced by placing a tourniquet (blood pressure measuring cuff) on the participants’ arm for up to ten min. Every two min the participants made pain ratings. Pain studies using the tourniquet technique consistently show a steady increase in pain over a ten min ischemia (Maixner, & Humphrey, 1993; Hamalainen & Kemppainen, 1990; Segendahl, Ekblom, & Sollevi, 1994; Lorenz & Bromm, 1997). For example, Hamalainen and Kemppainen (1990) reported average subjective assessments of ischemic pain on 100 mm VAS ratings of 5 mm pain at 2 min, 10 mm pain at 4 min, 25 mm pain at 6 min, 45 mm pain at 8 min, and 62 mm pain after 10 min of ischemia.

Using a within-subjects design, pain reported during a no distraction condition (conventional treatment) was compared to the pain experienced while the participant was in the VE. After eight min of wearing the tourniquet with no distraction, participants went into an immersive VE, interacted with a virtual spider and ate a virtual candy bar (Hoffman, Holander, Schroder, Rousseau, & Furness, 1998). Participants rated pain on five 100 mm visual analog scales (VAS) and one rating of anxiety for each condition with the blood pressure cuff still on. A VAS is a line, usually l00 mm in length… with anchors at each end to indicate the extremes of the sensation under study." (Gift, 1989). The participant makes a mark on the line to indicate the amount of sensation experienced, and the experimenter measures the number of mm from the low end of the scale to the participants mark (Gift, 1989). Based on the 10 min ischemic pain literature consistently showing a steady increase in pain the longer the ischemia lasts, with no distraction, the last two min period of the 10 min ischemia would be the most painful segment of the study. Yet a drop in pain during the last two min when participants went into virtual reality was predicted in the present study.

Prior to the present study, only one female burn patient had been treated with a VE for pain control. One goal of the present study is to determine whether VE pain control is also effective for females. Another goal was to replicate a finding by Hoffman, Hollander et al., (1998) that allowing participants to physically touch and physically eat mixed reality virtual objects (using tactile augmentation) increases the user’s sense of presence in the virtual world.

Experiment 1.



Twenty-two healthy undergraduate Psychology students from the University of Washington participated individually for extra credit. Fourteen of the participants were female, eight were male. Studies were conducted under both written and verbal informed consent using protocols reviewed and approved by the University of Washington’s Committee on the Rights of Human Subjects. Students with a history of extreme susceptibility to motion sickness were excluded, and students were fully informed prior to volunteering/signing up that the experiment was designed to study pain, and involved a 10 min pressure cuff induced ischemia.


Ischemic arm pain was induced using a tourniquet (e.g., Hamalainien & Kemppainen, 1990; Fillingim, Maixner, Girdler, Light, Harris, Sheps, & Mason, 1997; Maixner, Gracely, Zuniga, Humphrey & Bloodworth, 1990). The participant elevated his/her left forearm for 60 seconds., after which a standard blood pressure measuring cuff was inflated (by the participant) to 250 mmHg on the lower bicept, (200 mmHg for females) and the forearm was returned to the horizontal position, resting on a table. This signaled the official beginning of the ischemia. No muscle exercises were performed.

Participants were told that it was important to let the experimenters know if they needed to take off the pressure cuff prior to the 10 min completion (e.g., if it became too painful), and experimenters would immediately begin slowly deflating and removing the pressure cuff. Participants gave pain ratings every two min during the ischemia. The experimenter also verbally asked the participant if they were alright after each two min period. Participants did not wear the VR helmet during the no distraction phase. Participants able to tolerate eight min of the ischemia with no distraction were put into the VE for the last two min of the ischemia, and afterwards asked to rate how much pain they had been in during the last two min (when they were in the VE). To minimize the occurrence of excessively high pain levels, the duration of the ischemia was reduced for any participant whose "average pain" measure for the most recent two min period with no distraction reached 50 mm or higher on a 100 mm VAS subjective pain rating at any pain rating prior to the eight min mark. Such participants were immediately placed into the VE for the next two min period, participants filled out current pain ratings for that two min segment in the VE, and the pressure cuff was removed. After answering their last pain ratings while wearing the cuff, the pressure cuff was slowly deflated for each participant, and the participant placed their left hand in their lap, and filled out the post-experimental questionnaire (e.g., simulator sickness and presence ratings). They were then engaged in conversation until six min had passed since the end of their ischemia, at which point they filled out pain ratings once more, to make sure their pain was back to zero, or very near zero. Most (n = 16) participants had returned to near zero pain within six min of ending the ischemia and all other participants (n = 6) had returned to near zero pain after 12 min. It was interesting to measure how quickly pain subsided, and the experimenters were obligated to monitor the participants’ full recovery. Furthermore, participants went on to participate in Experiment 2, which required that participants were no longer in pain.

A Silicon Graphics Octane MXE with Octane Channel Option1 coupled with a VR helmet was used to create an immersive, 3-D, interactive, computer-simulated environment. A Polhemus FastrakTM motion sensing system with 6 degrees of freedom sensors was used to measure the position of the user’s head and hand position. Participants experienced SpiderWorld, a modified version of Division LTD’s DVS-3.1.2 KitchenWorld2 complete with countertops, a window, a sink, a stove, and 3-D cabinets. Participants were randomly assigned to one of two groups: Ordinary VE or VE with tactile augmentation. No sounds were used in either condition. Participants in both groups were placed in the middle of the virtual kitchen, and did not have a navigation device. Instead, they stood in the virtual kitchen and could "pick up" virtual objects (e.g., a spider and a candy bar) with their cyberhand. Using tactile augmentation (Hoffman, Groen, Rousseau, Hollander, Winn, Wells, & Furness III, 1996; Carlin, Hoffman & Weghorst, 1997; Hoffman, 1998; Hoffman Doctor et al., 2000; Garcia-Palacios, Hoffman, Carlin, Furness, Botella-Arbona, in press, Hoffman, Garcia, Carlin, Furness, & Botella, in press), a real world toy twin of the virtual brown spider made it possible for participants to "physically touch" the virtual spider. A palm-sized toy replica of a Guyana bird-eating tarantula was attached to a position sensor held by the experimenter. As the participant reached out with their cyberhand to touch the virtual spider, their real hand explored the position-tracked toy spider. The objective was to provide the subjective experience of a virtual spider that felt furry, had weight (cyberheft), and solidity. Any movement of the toy spider led to a similar movement by the virtual spider. Similarly, participants in the tactile augmentation group could physically eat a virtual candy bar linked via a position sensor attached to its real world twin. They saw a Hershey’s chocolate bar in their cyberhand in the virtual kitchen. When they held the virtual candy bar up to their mouths, they took a bite out of the real position tracked candy bar. The experimenter pushed a button changing the graphics such that when the participant pulled the virtual candy bar away from their mouth, there was a bite missing in the virtual candy bar. This led them to have the illusion of physically eating the virtual candy bar as they stood in the virtual kitchen. Participants in the ordinary VE condition were handed an ordinary virtual spider (a position sensor with no toy spider attached), and imagined taking a bite from a virtual candy bar with no real candy bar attached (their position tracked hand lifted their cyberhand and candybar up to their mouth). The experimenter pushed the button to make the candy bar show a bite mark when participants imagined taking a bite. This software was chosen by Hoffman and colleagues for treating burn patients during physical therapy (Hoffman, Patterson and Carrougher, 2000) because it had been shown to consistently give users a relatively high sense of presence.

Pain, the primary dependent variable, was measured immediately after each experimental condition (ischemic pain with no distraction, and ischemic pain during VR). Participants completed five retrospective subjective pain ratings using 100 mm Visual Analog Scales (i.e., VAS, Huskisson, 1974; Gift, 1989). The VAS has concurrent validity and test-retest reliable (Gift, 1989). Such self-report scales provide valid reflections of pain experience across patient populations (see review by Jensen, 1997). Self-report is the most valid method for assessing pain experience (cf. Hilgard and Hilgard, 1994; Jensen, 1997). With respect to the most recent two min of the ischemia, participants rated A) how much time they spent thinking about their pain and/or their arm (endpoints labeled zero min, the entire time) B) their WORST PAIN (no pain, worst pain), C) their AVERAGE PAIN (no pain, worst pain), D) how much their arm BOTHERED them (not at all bothersome, the most bothersome), E) how UNPLEASANT they found the ischemia (not at all unpleasant, the most unpleasant), and F) their ANXIETY (no anxiety, highest anxiety). These measures are designed to assess the sensory component of pain (worst pain and average pain in this study) and the affective component (unpleasant and bothersome in this study). Sensory and affective are two separately measurable and sometimes differentially influenced components of the pain experience (Melzack & Wall, 1965; Gracely, McGrath & Dubner, 1978). Time spent thinking about pain, and bothersomeness are new measures of procedural pain recently introduced by Hoffman, Doctor, et al., (2000). After the ischemia, participants were asked the following ratings using visual analog scales: 1) To what extent (if at all) did you feel nausea as a result of experiencing the virtual environment? (none, very much) 2) While experiencing the virtual environment, to what extent did you feel like you went into the virtual world? (I did not feel like I went into the virtual world at all, I went completely into the virtual world) 3) How real did the objects in the virtual world seem to you (completely fake, indistinguishable from a real object). Hendrix and Barfield (1995) describe several studies showing the reliability of a similar subjective measure of presence.


First, results were analyzed with males and females combined. Alpha is set at .05 for F-tests. For t-tests, a Bonferroni (Keppel, 1982) correction factor (dividing alpha by the number of t-tests) was used (.05/6 = .008). According to 100 mm VAS pain ratings (see Figure 1), mean pain ratings increased significantly every two min during the no-distraction phase (0 to 8 min). There was a significant rise in pain from 2 min to 4 min F(1,21) = 9.42, p = .006, MSe = 49.02, from 4 to 6 min, F(1,21) = 10.60, p = .004, MSe = 32.92 from 6 to 8 min (the control condition), F(1,21) = 10.91, p = .003, MSe = 80.95, and a significant and large (53%) drop in pain between 8 and 10 min, when participants went into virtual reality, F(1,21) = 29.16, p < .001, MSe = 293.413.

As shown in Figure 2, comparing all participants in the control condition (6 to 8 min period, or the next to the last 2 min session) to the VE condition (8-10 min or the last 2 min session), pain while in the VE dropped significantly for each of the five VAS pain ratings, time spent thinking about their pain, t(21) = 4.79, p < .001, SE = 7.43, unpleasant, t(21) = 5.02, p < .001, SE = 5.44, bothersome, t(21) = 5.48, p < .001, SE = 5.28, worst pain, t(21) = 4.93, p < .001, SE = 5.10, average pain, t(21) = 4.41, p < .001, SE = 5.10. Although anxiety dropped approximately 50% when participants were in VR, the drop was not significant, t(21) = 2.80, p = .01, SE = 4.88.

The results are now analyzed for females and males separately. It is not appropriate to make gender comparisons in VE analgesic effectiveness because only half of the females but 3/4ths of the males completed the entire 10 min ischemia, and the amount of pressure used was lower on females (to keep pain levels from getting too high)3. This difference in the way males and females were treated precludes a meaningful statistical comparison of gender effects. First, the analyses above were conducted again with females only (see Figure 1) to determine whether females show significant reductions in pain while in VEs. According to 100 mm VAS pain ratings, mean pain ratings of females increased significantly every two min during the no-distraction phase (0 to 8 min). There was a significant rise in pain from 2 min to 4 min F(1,13) = 5.09, p = .04, MSe = 41.05, a significant rise in pain from 4 to 6 min, F(1,13) = 7.18, p = .02, MSe = 44.77, a significant rise in pain from 6 to 8 min (the control condition), F(1,13) = 9.34, p = .009, MSe = 91.57, and a significant and large (59%) drop in pain between 8 and 10 min, when participants went into the virtual environment, F(1,13) = 19.36, p = .001, MSe = 390.33.

As shown in Figure 2, analyzing only females, pain while in the VE dropped significantly compared to the control condition for all of the five VAS pain ratings, time, t(13) = 3.91, p = .002, SE = 10.92, unpleasant, t(13) = 4.51, p = .001, SE = 7.62, bothersome, t(13) = 4.33, p = .001, SE = 7.70, worst pain, t(13) = 3.84, p = .002, SE = 7.33, average pain, t(13) = 3.45, p = .004, SE = 7.46, but females showed no significant drop in anxiety, t(13) = 1.95, p = .07, NS.

Next, the same analyses were conducted again with males only (see Figure 1). According to 100 mm VAS pain ratings, mean pain ratings of males increased every two min during the "no distraction" phase (zero to eight min). There was a marginally significant rise in pain from two to four min, F(1,7) = 4.01, p = .09, MSE = 67.99, from four to six min (F(1,7) = 4.88, p = .08, MSE = 12.02, a non-significant rise in pain from six to eight min, F(1,7) = 1.85, p = .22, MSE = 60.73, and a significant and large (41%) reduction in pain for males while in the VE, F(1,7) = 16.85, p = .005, MSE = 87.50.

As shown in Figure 2, analyzing males only, pain while in the VE (eight to ten min or the last two min of the ischemia) dropped significantly compared to the control condition (six to eight min or the last two min of no distraction) for two of the five VAS pain ratings: time, t(7) = 4.0, p = .005, SE = 5.82, bothersome, t(7) = 4.37, p = .003, SE = 4.83, but the drop was not significant for unpleasant, t(7) = 3.28, p = .01, SE = 4.63, worst pain, t(7) = 3.42, p = .01, SE = 5.81, average pain, t(13) = 3.28, p = .01, SE = 5.11, or anxiety, t(7) = 2.54, p = .04, SE = 5.07.

Mean nausea ratings on a 100 mm scale were low (6 mm for females, mean = 10 mm for males). On 100 mm scales, mean presence in VR was 60 mm for females, 48 mm for males, and mean realism of the virtual objects was 55 mm for females, 41 mm for males. Although mean presence levels were slightly higher for all participants in the tactile augmentation condition (as predicted), the increase in presence was not statistically significant. One question asked participants the extent to which they felt like they went into the virtual world, (mean rating = 51 mm for ordinary VE and 60 mm for VE with tactile cues), t(20) < 1, NS. The other presence question asked participants how real the virtual objects seemed (mean rating = 46 mm for ordinary VE and 54 mm for VE with tactile cues), t(20) < 1, NS.


Immersive virtual reality reduced the amount of pain reported and the amount of time participants spent thinking about their pain and/or their arm during the VE segment of the ischemia. The experimental effect size was large and statistically significant. Since without the VE, pain would likely have continued to rise during the eight to ten min period, the present calculations of analgesic effectiveness (using the six to eight min period as the control condition) likely underestimate the magnitude of how much the VE reduced pain. These results provide the first available evidence from a controlled laboratory study that virtual environments can serve as a useful pain reduction technique, conceptually replicating a similar pattern of findings in a clinical study with severe burn patients during physical therapy (Hoffman, Carrougher & Patterson, 2000).

Because burn-injured patients are predominantly male, only one female burn patient was available to participate in the study by Hoffman, Patterson and Carrougher, (2000)4. The present study addresses for the first time the scientific issue of whether females can also experience VE pain reduction. Females showed large, statistically significant reductions in pain in VE compared to the no distraction control condition (as did males).

Simulator nausea, a component of sickness (Kennedy, Lane, Lilienthal, Berbaum, & Hettinger, 1992), is a form of motion sickness that can be elicited in some people by some VE experiences. Nausea was not a problem with this virtual world, but no problem would be expected with such a short treatment duration. Simulator sickness (or at least nausea) should be monitored closely in any medical use of VEs, especially since some manipulations intended to increase presence could unintentionally increase simulator nausea. Consistent with previous findings (Hoffman, 1998; Hoffman Doctor et al., 2000; Hoffman, Carrougher & Patterson, 2000), participants in the present study indicated that they felt a moderately strong illusion of being in a place (the sensation of "presence") while in KitchenWorld, and found the virtual objects somewhat realistic. Adding tactile cues to the virtual objects did increase the mean presence ratings, but unlike previous studies not involving pain (Hoffman, 1998), the increase in presence in the present study was not statistically significant. Pain is likely the reason presence did not increase significantly with the additional of tactile cues. Hoffman et al., 2000 speculate that VEs and pain are in a tug-of-war over the participants’ limited attentional capacity. That is, the VE reduces the amount of attention available to process pain signals, but pain also reduces the amount of attention available to be drawn into the VE (see also Hoffman, Prothero, Wells and Groen, 1998). Previous studies have shown that pain is attention demanding (Lorenz & Bromm, 1997). Participants likely feel less presence in the VE when in pain than when not in pain, and it may be harder to increase people’s sense of presence in the VE (e.g., by adding tactile cues) when the person is in pain. These are important theoretical issues for improving our understanding of how VE pain reduction works, and they are important practical issues. It seems likely that people in exceptionally high pain will require unusually attention grabbing virtual worlds in order to experience pain reduction.

Some shortcomings of Experiment 1 deserve mention. Because both the participants and the experimenter were aware of which condition the participants were in (i.e., knew when the participant was in the VE and when they were not), demand characteristics could have contributed to the drop in pain witnessed. That is, participants could have responded in a way they hoped would please the experimenter. Also, the ischemic pain paradigm was less than ideal as a source of pain, because there was large variability in how long the students could tolerate the pressure cuff, and the order of presentation of the two conditions (VE vs no VE) was not counterbalanced. The VE condition was always conservatively put last, because the literature consistently shows that for a ten min ischemia, the last two min are always the most painful. Future studies should include better controls for demand characteristics (e.g., participant-blind or even double-blind experimental designs), a different pain paradigm (e.g., thermal pain), and ideally, should include a control group that gets no VE at all (to replicate that finding in the literature with the student population under study).

Experiment 2

Experiment 2 begins to explore the mechanism by which VEs reduced pain. As in Experiment 1, the VE was predicted to reduce pain, based on the assumption that both pain and VEs require conscious attention, and that going into the VE would draw attention away, leaving less attention available to process information from the real world. This assumption is tested in Experiment 2. Experiment 2 explored the hypothesis that the attention attracted by the computer-generated world in the VE condition would lead to poorer performance on a working memory/divided attention task in comparison to performance in a "no VE" condition (order counterbalanced). Such working memory/divided attention tasks are a common way to assess mental work load and relative allocation of attention to a task such as monitoring a simulated flight control display (Wickens, 1992; Isreal, Wickens, Chesney, & Donchin, 1980). Divided attention tasks have been used to study the nature of human memory in general (Craik, Govoni, Naveh-Benjamin, & Anderson, 1996), and of age-related changes in memory (Anderson, Craik, & Naveh-Benjamin, 1998), differences between conscious and unconscious thought processes (Jacoby, Woloshyn & Kelly, 1989; Jacoby, Toth, and Yonelinas, 1993) and a number of other important cognitive processes. To our knowledge the present study is the first to use a divided attention task to study thought processes associated with VEs. The objective of Experiment 2 was to demonstrate attentional involvement in the virtual environment by showing that a concurrent task requiring attention to the real world could not be performed as efficiently in the VE as when not in the VE.


The twenty-two undergraduate Psychology students from the University of Washington who participated in Experiment 1 also participated in Experiment 2 for extra credit. Fourteen of the participants were female, eight were male.

Procedure. After completing Experiment 1, and after their pain had returned to zero, students participated in Experiment 2, which did not involve any pain or ischemias. Participants monitored a string of numbers presented auditorily as they interacted with the computer-generated world for two min (the VE condition) and then with no distraction (the two min "no VE" condition). The virtual world was the same one used in Experiment 1 (see Figure 3). Condition order was randomly counterbalanced (e.g., half of the participants received the no-VE condition first). Participants heard a continuous string of numbers between one and ten. They listened for 3 odd numbers in a row (e.g., 3, 5, 9,) at 1.5 sec intervals, from a tape player in the laboratory (Craik, 1982; Jacoby, Woloshyn, & Kelley, 1989). Participants’ task was to say "now" any time they heard three odd numbers in a row. The experimenter followed along the list of digits on paper, and made a mark each time the participant said "now" for later scoring. The participant also performed the same task with the VE helmet turned off for two min ("no VE" condition). The participant still wore the helmet in this condition, but could not see anything. Participants wore the helmet during the no VE condition in case the VE helmet impeded the user’s ability to hear the random string of numbers played by the tape recorder. Prior to each condition the experimenter emphasized to participants that their job was to monitor the numbers and to make as few mistakes as possible. Comparing performance (e.g., percent of odd triplets correctly identified) on the digit monitoring/divided attention task in the VE condition to performance in the "no VE" condition gave an indication of how much attention (if any) participants devoted to the virtual world. Approximately half of the participants were randomly assigned to receive ordinary virtual reality, and the other half received tactile augmentation. Participants assigned to the tactile group in Experiment 1 remained in the tactile group in Experiment 2 etc.

After each two min session (one with VE, one with no-VE, order counterbalanced), participants answered the following visual analog scale rating about the last 2 min period. "The task you just completed required you to monitor a list of numbers and to indicate when you heard three odd numbers in a row. In your estimation, approximately what percentage of these strings of odd numbers did you correctly identify?" (0% = got them all wrong, 100% = got them all right). After the second two min session in the VE, participants were also asked the following questions: 1. How much time did you spend thinking about or attending to the numbers being played to you by the tape recorder during the session when you were in the virtual environment? (zero min, the entire time). 2. How much time did you spend thinking about or attending to the numbers being played to you by the tape recorder during the session when you were NOT in the virtual environment? (zero min, the entire time)."


Participants were significantly more accurate at monitoring the numbers in the control condition than in the VE condition (95% vs. 74% respectively of the triplets that occurred were detected), t(21) = 5.93, p < .001, SE = 3.51. Without any feedback, participants were aware that they were less accurate in the VE. They estimated that their accuracy had been 91% for the control condition, and 65% for the VE condition, t(20) = 6.17, p < .001, SE = 4.43. And they estimated that the amount of time they had been able to attend to the numbers was higher in the control condition than in the VE (96% vs. 66% of the time, respectively), t(21) = 5.21, p < .001, SE = 5.26.


In Experiment 2, on a separate task not involving pain, young adult students showed a significant reduction in performance on a divided attention task (monitoring a string of numbers from a tape recorder for 3 odd numbers in a row) while in a VE compared to no VE (monitoring numbers while wearing a VE helmet that was turned off). The VE likely distracted attention away from the student’s primary task of monitoring the numbers in the real world. These results justify the assumption that the VE is attention grabbing and implicate the contribution of an attentional mechanism for VE pain reduction.

General Discussion.

Hoffman, Patterson and Carrougher, (2000) recently found that the VE dramatically reduced the pain of severely burn-injured patients during physical therapy. Using undergraduate volunteers, the present laboratory analog study conceptually replicates that clinical finding, and shows that VE pain reduction generalizes to a new pain etiology, pain originating from a blood ischemia rather than a burn injury. Study 1 shows for the first time that VE pain reduction is also effective for females.

The original rationale for trying VEs for pain control was based on the assumption that VEs are attention grabbing. Study 2 explicitly tests this assumption using a divided attention task. As predicted, participants were less accurate when in the virtual environment than when the helmet mounted display was turned off (no VE).

The conclusions that can be drawn from analog studies are limited without converging evidence from clinical populations. The severity of pain experienced during wound care cannot be replicated in a laboratory environment, and/or it would be unethical to induce such severe pain. Also, the context is dramatically different for the burn patient than for the undergraduate. Burn patients are dealing with pain for weeks or months, with concomitant physical, social and psychological adjustments, whereas the undergraduate’s pain experience lasted 10 min or less, the pain was completely gone within 16-22 min of the time it began, and the undergraduate’s life returned to normal when they walked out of the laboratory. While encouraging, the results of the present laboratory study alone are no guarantee that the technique would work in clinical conditions. Fortunately, the converging evidence from the present study and the results of Hoffman Patterson and Carrougher (2000) combine to form a complimentary multidisciplinary package, integrating clinical and basic research. One advantage of this widely advocated multidisciplinary approach (Dubner, 1997) is that an ischemic pain analog laboratory experiment can be performed very efficiently, with data collection completed after only one or two weeks, whereas the same study with clinical patients (e.g., burn patients) may take many months. The number of burn patients treated at Harborview is large, but is much smaller than the number of undergraduate Psychology students, and the population of burn patients is predominantly male. Analog studies such as the present study can dramatically accelerate our understanding of how VEs influence the processing of pain signals, and whether it is also effective for females, an understanding that will help us exploit the full potential of using VEs as a pain reduction technique.

Now that there is evidence that single session VE pain reduction treatments can reduce pain for short physical therapy sessions, it is important to find out if VEs continue to be effective when used more than once, and for longer treatments. Hoffman, Patterson, Carrougher, Nakamura, Moore, Garcia-Palacios, and Furness, (2001) recently addresssed this issue in a case study. The patient was a "32 year old male" with deep flash burns on his face, neck, shoulder, chest, and legs, covering 42% total body surface area. He performed physical therapy exercises to stretch the healing skin of his left shoulder, arm and hand under a physical therapist’s direction and had previous difficulty tolerating his pain during physical therapy. The patient reported less pain when distracted with VR, and the magnitude of pain reduction from the VE did not diminish with repeated administration (5 sessions on 5 different days) and longer treatment durations (up to 15 min of physical therapy in the VE). Simulator sickness remained negligible even for the longer sessions, for this patient. That case study has recently been replicated with seven severely burn-injured patients for 3 separate treatment sessions (Hoffman, Patterson, Carrougher, and Sharar, 2002). Additional empirical studies will be needed to verify whether VE pain reduction remains effective during multiple treatments of longer durations. Such studies will help determine whether virtual environments can become a viable form of nonpharmacologic pain control technique in everyday medical practice.

The potential impact of this research is not limited to burn patients. The present results encourage speculation that VE pain reduction will also generalize to other acute pain populations/etiologies (e.g., general dentistry, oral and other ambulatory surgery, chemotherapy, radiation therapy, childbirth labor pain, etc.). As a beginning, Hoffman Garcia-Palacios, Patterson, et al, (2001) recently showed preliminary evidence that VR can reduce dental pain. It may also allow patients to have short "breaks" from chronic pain. Because physical therapy can help reduce chronic pain and chronic fatigue syndrome. VR may help chronic pain and chronic fatigue patients indirectly by allowing such patients to tolerate physical therapy, and perhaps to motivate them to perform their therapy. Considering the magnitude of the pain reductions for people in VR, and the widespread need for better procedural pain control, additional research and development is warranted.

Acknowledgments. NIH grant GM42725-07, NIDRR grant #H133A970014, HD37684-01A1, (HD40954) and the Paul Allen Foundation for Medical Research. Thanks to Professor Thomas Furness III for valuable comments and generous hardware support, and for lending us Konrad Schroder. Special thanks to the participating undergraduate Psychology students, to Ross Chambers for fundraising and Ian Dillon from SGI for an equipment loan.


Anderson, N. D., Craik, F. I. M., & Naveh-Benjamin, M. (1998). The attentional demands of encoding and retrieval in younger and older adults: I. Evidence from divided attention costs. Psychology and Aging, 13, 405-423.

Campbell, D. T., & Stanley, J. C. (1963). Experimental and Quasi-Experimental Designs for Research. Boston: Houghton Mifflin Company, 84 pp.

Carlin, A. S., Hoffman, H. G. & Weghorst, S. (1997). Virtual reality and tactile augmentation in the treatment of spider phobia: A case study. Behaviour Research and Therapy, 35, 153-158.

Carrougher, G.J., Ptacek, J.T., Sharar, S.R., Wiechman, S., Honari, S., Patterson, D.R. Engrav, L.H., Heimbach, D.M. (submitted). A comparison of patient satisfaction and self-reports of pain in adult burn-injured patients.

Cohen, L.L., Blout, R.L., and Panopoulos, G. (1997). Nurse coaching and cartoon distraction: an effective and practical intervention to reduce child, parent and nurse distress during immunizations. J Pediatr Psychol. 22:355-370.

Craik, F. I. M. (1982). Selective changes in encoding as a function of reduced processing capacity. in F. Klix, J. Hoffman, & E. van der meer (Eds.), Cognitive research in psychology (pp. 152-161), Berlin: Springer-Verlag.

Craik, F. I. M., Govoni, R., Naveh-Benjamin, M., & Anderson, N. D. (1996). The effects of divided attention on encoding and retrieval processes in human memory. Journal of Experimental Psychology: General, 125, 159-180.

Dubner, R. (1997). Three decades of pain research and its control. Journal of Dental Research 76, 730-733.

Ehde, D. M., Patterson, D. R., Fordyce, W. E. (1998). The quota system in burn rehabilitation. The Journal of Burn Care & Rehabilitation, 19, 436-40.

Everett, J. J., Patterson, D. R., & Chen, A. C. (1990). Cognitive and behavioural treatments for burn pain. Pain Clinic, 3, 1133-45.

Farthing, G.W., Venturino, M., and Brown, S.W. (1984). Suggestion and distraction in the control of pain: test of two hypotheses. J. Abnormal Psychol;93:266-276.

Fernandez, E. & Turk, D. C. (1989). The utility of cognitive coping strategies for altering pain perception: a meta-analysis. Pain, 38, 123-135.

Fillingim, R. B., Maixner, W., Girdler, S. S., Light, K. C., Harris, M. B., Sheps, D. S., & Mason, G. A. (1997). Ischemic but not thermal pain sensitivity varies across the menstrual cycle. Psychosomatic Medicine, 59, 512-20.

Garcia-Palacios, A., Hoffman, H.G., Carlin, C.., Furness, T.A. III, Botella-Arbona, (in press). Virtual reality in the treatment of spider phobia: A controlled study. Behaviour Research and Therapy.

Gift, A. G. (1989). Visual analogue scales: measurement of subjective phenomena. Nursing Research, 38, 286-288.

Gracely, R. H., McGrath, P. A. & Dubner, R. (1978). Ratio scales of sensory and affective verbal pain descriptors, Pain, 5, 5-18.

Hamalainen, O., & Kemppainen, P. (1990). Experimentally induced ischemic pain and so-called diaphase fix current. Scandinavian Journal of Rehabilitation Medicine, 25-27.

Hendrix, C., & Barfield, W. (1995). Presence in virtual environments as a function of visual and auditory cues. Proceedings of VRAIS '95, Research Triangle Park, NC. IEEE Computer. Society Press. Los Alamitos, CA, 74-82.

Hilgard, E. R., & Hilgard, J. R. (1994). Hypnosis in the relief of pain. New York: Brunner/Mazel.

Hoffman, H.G. (1998). Physically touching virtual objects using tactile augmentation enhances the realism of virtual environments. Proceedings of the IEEE VRAIS ‘98, Atlanta GA, IEEE Computer Society, Los Alamitos, California, 59-63.

Hoffman, H. G., Doctor, J.N., Patterson, D. R., Carrougher, G. J., & Furness, T. A. III. (2000). Use of virtual reality for adjunctive treatment of adolescent burn pain during wound care: A case report. Pain, 85, 305-309.

Hoffman, H.G., Garcia-Palacios, A., Carlin, C., Furness, T.A. III, Botella-Arbona, (in press). Interfaces that heal: Coupling real and virtual objects to cure spider phobia. International Journal of Human-Computer Interaction.

Hoffman, H. G., Garcia-Palacios, A., Patterson, D. R., Jensen, M., Furness, T. 3rd, & Ammons, W. F. Jr. (2001). The effectiveness of virtual reality for dental pain control: a case study. Cyberpsychology and Behavior. Aug;4(4):527-35.

Hoffman, H.G., Groen, J., Rousseau, S., Hollander, A., Winn, W., Wells, M., & Furness III, T. (1996). Tactile Augmentation: Enhancing presence in inclusive VR with tactile feedback from real objects. Presented at the meeting of the American Psychological Sciences, San Francisco.

Hoffman, H. G., Holander, A., Schroder, K., Rousseau, S. & Furness, T. A., III (1998). Physically touching and tasting virtual objects enhances the realism of virtual experiences. Virtual Reality: Research, Development and Application, 3, 226-234.

Hoffman, H.G., Patterson, D.R., Carrougher, G.J. (2000). Use of virtual reality for adjunctive treatment of adult burn pain during physical therapy: a controlled study. Clin J Pain, 16(3):244-50.

Hoffman, H.G., Prothero, J., Wells, M. & Groen, J. (1998). Virtual Chess: The role of meaning in the sensation of presence. International Journal of Human-Computer Interaction, 10:251-263.

Hoffman, H.G., Patterson, D.R., Carrougher, G.J., Nakamura, D., Moore, M., Garcia-Palacios, A., & Furness, T.A. III (2001). The effectiveness of Virtual Reality pain control with multiple treatments of longer durations: A case study. International Journal of Human-Computer Interaction, 13(1), 1-12.

Hoffman, H.G., Patterson, D.R., Carrougher, G.J., & Sharar, S. (2001). The effectiveness of virtual reality based pain control with multiple treatments. Clinical Journal of Pain. 17:229-235.

Huskisson, D. C. (1974). Measurement of pain, Lancet, 2, 1127-31.

Isreal, J. B., Wickens, C. D., Chesney, G. L, & Donchin, E. (1980). The event-related brain potential as an index of display monitor workload. Human Factors, 22, 214-224.

Jacoby, L L., Toth, J. P., & Yonelinas, A. P. (1993). Separating conscious and unconscious influences of memory: Measuring recollection. Journal of Experimental Psychology: General, 122, 139-154.

Jacoby, L. L., Woloshyn, W., & Kelley, C. M. (1989). Becoming famous without being recognized: Unconscious influences of memory produced by dividing attention. Journal of Experimental Psychology: General, 118, 115-125.

Jensen, M. P. (1997). Validity of self-report and observation measures. In T.S. Jensen, J.A. Turner, & Z. Wiesenfeld-Hallin (Eds.), Proceedings of the 8th World Congress on Pain. Progress in Pain Research and Management, Vol. 8, Seattle: IASP Press, pp. 637-661.

Kennedy, R. S., Lane, N. E., Lilienthal, M. G., Berbaum, K. S., & Hettinger, L J. (1992). Profile analysis of simulator sickness symptoms: Applications to virtual environment systems. Presence: Teleoperators and Virtual Environments, 1, 295-301.

Keppel, G. (1982). Design and Analysis: A Researcher’s Handbook. 2nd Ed. Englewood Cliffs, NJ., Prentice-Hall.

Kozarek, R.K. Raltz, S.L., Neal, L., Wilbur, P., Stewart, S., and Ragsdale, J. (1997). Prospective trial using Virtual Vision as distraction technique in patients undergoing gastric laboratory procedures. Gastroenterol Nurs, 20,12-4.

Laurel, B. (1995). Virtual Reality. Scientific American, p. 90. September issue.

Lorenz, J. & Bromm, B. (1997). Event-related potential correlates of interference between cognitive performance and tonic experimental pain. Psychophysiology, 34, 436-445.

Maixner W., Gracely, R. H., Zuniga, J. R., Humphrey, C. B. & Bloodworth, G. R. (1990). Cardiovascular and sensory responses to forearm ischemia and dynamic hand exercise. American Journal of Physiology, 259, 1156-63

Maixner, W. & Humphrey, C. (1993). Gender differences in pain and cardiovascular responses to forearm ischemia. The Clinical Journal of Pain, 2, 16-25.

McCaul, K.D., and Malott, J.M.. (1984). Distraction and coping with pain. Psychological Bulletin;95:516-533.

Melzack, R. (1990). The tragedy of needless pain. Scientific American, 262, 27-33.

Melzack, R. (1998). Psychological Aspects of Pain: Implications for Neural Blockade, in (Cousins, M.J, & Bridenbaugh, P.O., Eds.), Neural Blockade: In clinical anesthesia and management of pain, (pp. 781-792), Philadelphia, P.A: Lippincontt-Raven.

Melzack, R., & Wall, P. D. (1965). Pain Mechanisms: A new theory. Science, 150, 971-979.

Miller, A.C., Hickman, L.C., and Lemasters, G.K. (1992). A distraction technique for control of burn pain. J. Burn Care Rehabil. 13,576-80.

Patterson, D. R. (1992). Practical applications of psychological techniques in controlling burn pain. The Journal of Burn Care & Rehabilitation., 13, 13-18.

Patterson, D. R. (1995). Non-opioid-based approaches to burn pain. The Journal of Burn Care & Rehabilitation, 16, 372-6.

Patterson, D. R., Everett, J. J., Burns, G., & Marvin, J. A. (1992). Hypnosis for the treatment of burn pain, Journal of Consulting and Clinical Psychology. 60, 713-717.

Perry, S., Heidrich, G. & Ramos, E. (1981). Assessment of pain by burn patients. The Journal of Burn Care & Rehabilitation, 2, 322-327.

Ptacek, J. T., Patterson, D. R., Montgomery, K. B. & Heimbach D. M. (1995). Pain, coping and adjustment in patients with burns: preliminary findings from a prospective study. The Journal of Pain and Symptom Management, 10, 446-455.

Segerdahl, M., Ekblom, A. & Sollevi, A. (1994). The influence of adenosine, ketamine, and morphine on experimentally induced ischemic pain in healthy volunteers. Anesthesia and Analgesia, 79; 787-91.

Slifer, K.J. (1996). A video system to help children cooperate with motion control for radiation treatment without sedation. J Pediatr Oncol Nurs., 13,91-7.

Tan, S. (1982). Cognitive and cognitive-behavioral methods for pain control: a selective review. Pain, 12, 201-228.

Ward, R. S. (1998). Physical Rehabilitation. (Carrougher, GJ, Ed.). Burn Care and Therapy, (pp. 293-327). NY.NY: Mosby.

Weisenberg, M., Tepper, I, and Schwarzwald, J. (1995). Humor as a cognitive technique for increasing pain tolerance. Pain;63,207-212.

Wickens, C. D. (1992). Engineering psychology and human performance. New York, NY, HarperCollins Publishers.


1. Silicon Graphics, Inc.13810 S.E. Eastgate Way, Suite 300, Bellevue WA., 98105, (425)746-7563,

2. Division Incorporated,

3. For purposes of analyses, the VR session was analyzed with the 8 to 10 min segment of the ischemia, and the control condition (the two min session immediately prior to the VR session) was analyzed with the 6 to 8 min segment, for participants whose ischemias did not last the full 10 min. For example, in the unusual event that a participant’s average pain reached 50 mm (e.g., 52 mm) after only 2 min of ischemia with no distraction, then the first two min were considered the control segment and analyzed as if they were the 6-8 min segment, and (if willing) the participant went into VR for two more min (analyzed as 8-10 min), after which the ischemia was ended early. In this example, the participant received missing data points for the 0-2, 2-4, and 4-6 min segment of the ischemia. To make sure our solution to the missing data problem was not misleading, results from the 13 participants who completed all 10 min of their ischemia was calculated. The mean pain ratings for 0 to 2 min = 31.14, 2-4 min = 34.45, 4-6 min = 39.22, 6-8 min = 42.32 (control), and 8 to 10 min (VR) = 23.88. A t-test comparing pain during VR to control pain for this subset of participants was significant, t(12) = 4.17, p = .001, SE = 4.42. Consistent with previous findings showing lower pain tolerance for females than males (e.g., Maixner & Humphrey, 1993), only 7 out of the 14 females (50%) completed the 10 min ischemia whereas 6 out of the 8 males (75%) completed the 10 min ischemia. For this reason, no attempt is made to make conclusions from these data about whether there are gender differences in VR analgesic effectiveness (though that is an interesting topic for future research).

4. Consistent with the findings from the present study, Hoffman et al.’s female patient showed a reduction in pain while in VR during physical therapy for her burn wound: She showed a drop of 50 mm for time spent thinking about her pain during VR compared to no distraction, a drop of 32 mm for pain unpleasantness, a drop of 34 mm for bothersomeness, and a drop of 29 mm for average pain while in VR.

Figure Captions

Figure 1. Mean pain ratings of males and females, taken every two minutes.

Figure 2: Pain and anxiety ratings of females only (n = 14).

Figure 3: Experimenter moves a wiggly legged virtual spider closer to the participant by moving the position sensor closer to the participant. For the "tactile augmentation" condition, a furry toy spider is attached to the position sensor, allowing participants the illusion of physically groping the virtual Guyana bird-eating tarantula when they reach out to touch the virtual spider with their cyberhand. Copyright Mary Levin, U.W.