Experiencing stress in a virtual warzone by playing with visual stressors
SUBMITTED IN PARTIAL FULLFILLMENT FOR THE DEGREE OF MASTER OF SCIENCE
NICOLIEN
BERKERS
6181724
MASTER
INFORMATION
STUDIES
GAME STUDIES
FACULTY OF SCIENCE
UNIVERSITY OF AMSTERDAM
August 14, 2015
1st Supervisor 2nd Supervisor 3rd Supervisor
G.C. van de Boer-Visschedijk Dr. A.H. van der Hulst Dr. O. Binsch
Experiencing stress in a virtual warzone by playing with
visual stressors
Nicolien Berkers
Science Park 904 1098 XH Amsterdam 00316-23094823nicolien.berkers@student.uva.nl
ABSTRACT
The military field of work encompasses stressful events that can influence immediate performance and advance the development of Post Traumatic Stress Disorder (PTSD). Training is necessary to prepare soldiers for encountering such events. Incorporating Virtual Reality (VR) techniques in military stress training is relatively new and not yet widely integrated. VR might provide a realistic environment and therefore effectuate skill and knowledge transfer from training to the operational field. Current literature investigates how VR games could induce a stress response. This study compares stress responses between two displays, a Head Mounted Display (HMD) and a desktop monitor. HMDs could increase the sense of presence in a virtual environment, therefore intensify the user’s response. The Oculus Rift DK2 and the game Battlefield 3 were used to examine stress responses and presence. Nervous system activity was measured in addition to threat and challenge level, stressful game elements, and cyber sickness. Results indicated that self-reported stress, presence and dizziness were higher for Oculus Rift gameplay compared with playing on the monitor. The mission was more challenging than threatening, with no differences between gameplay conditions. The visual stimuli ‘explosions’ and ‘enemy fire’ were more stressful with Oculus Rift gameplay. No differences were found in nervous system activity between the conditions. These results partly confirm the hypothesis that HMDs induce more stress during gameplay. Differences were found in subjective experience, but not in physiological response.
General Terms
Measurement, Performance, Design, Experimentation, Human Factors.
Keywords
Military training, stress, virtual reality, head mounted display, Oculus Rift DK2, FPS, game, Battlefield.
1. INTRODUCTION
Military personnel must undergo intense training in order to prepare for working in the operational field. Training under stressful circumstances is key, because soldiers will encounter stressful situations and training for this will increase their performance [6]. Stress has a major impact on performance, processes such as decision making and information processing are influenced by it [12]. It also has an impact on health on the long term. For example, soldiers might develop Post Traumatic Stress Disorder (PTSD) when there had not been sufficient opportunity to recover from stressful experiences [8].
New training paradigms are constantly being developed as new technologies arise. One of the most promising technologies is the use of virtual reality (VR) in serious gaming or simulation environments. VR incorporates head mounted displays (HMDs) that allow a stereoscopic view. The potential benefit for stress or resilience training seems apparent; a virtual world in which a trainee is completely immersed could be perceived as more realistic, and thus more stressful. But, as with all new training concepts, there is the issue of efficacy. To examine this, research is needed to analyze the behavior and mental/physical condition of trainees before, during and after training. The goal of this study is therefore to examine a method used to present stressors and evoke a stress response during military training in a VR environment. VR techniques are currently used in therapies, for PTSD for instance [17], but are not often seen in training procedures. HMDs are already used in gaming, and the possible benefits of incorporating them in training and research are being investigated more often. Games are occasionally included in military training procedures, but the use of computer screens is standard and HMDs are rarely seen. HMDs may provide a more realistic virtual environment in terms of immersion and presence and may be beneficial for the transfer of skills from training to the operational field. Presence in gaming and VR is described as the feeling of being physically present in a virtual world. This experience can be influenced by the level of immersion that is offered by the techniques used. Adding to traditional computer monitors and HMDs, a Computer Automatic Virtual Environments (CAVE) provides yet another level of immersion. CAVEs are usually set up in a room that the user can freely walk around in. The virtual environment then is projected upon a number of walls, the floor and ceiling included.
Immersion and presence may also be beneficial for the transfer of skills and knowledge from training to the operational field. A good transfer between video game training and the application of acquired skills to the operational environment is necessary. As described by Alexander [2], there are four factors that determine this transfer: immersion, presence, fidelity and buy-in. There are a number of benefits that video games offer for trainings that need to have a higher transfer: video game engines can be adapted for specific scenarios by the trainer, games can be easily distributed, games have the capacity to represent the real world, games can be implemented with lower costs, video games support the acquirement of cognitive skills (e.g. attention) [2].
To examine the stress responses to simulation techniques, both subjective and objective measurements are used. Objective measurements include physiological parameters that indicate sympathetic nervous system activity. This part of the autonomic nervous system is activated during a stress response, examples of activity are increased heart rate and sweat production. Heart rate
(HR) and heart rate variability (HRV) are used to measure changes in cardiovascular activity and galvanic skin resistance (GSR) and skin conductivity response (SCR) are used to measure changes in sweat production.
The objective and subjective responses to gaming and simulation techniques that aim to induce stress are investigated increasingly. These studies are also relevant for Defense personnel, because altering training paradigms might increase performance in the field and reduce development of PTSD. TNO is one of the organizations that investigates the applicability of gaming and simulation techniques for the Department of Defense in The Netherlands. The possibilities of gaming and simulation techniques for military training are currently investigated through a large scale project at TNO, and this pilot study is a start of that exploration.
2. RELATED WORKS
To this point, few researchers have investigated the relationship amongst VR stressors, stress response and subjective experience. There are some that look into the emotional changes that occur with the use of VR, the feeling of presence in VR settings, and the effect on various physiological parameters. These studies have not addressed the effects of VR specifically for military simulations. Juan and colleagues [11] studied the level of presence and anxiety in a virtual acrophobic task (fear of heights) by questionnaires. The non-acrophobic participants were security guards and had to look for suspicious objects in a virtual room. This task was completed once in a four-wall CAVE and once with an HMD. It was found that the HMD did not induce a high sense of presence, and the CAVE induced significantly more presence than the HMD. The CAVE also induced more anxiety [11]. This study shows that the level of immersion increases subjective experiences of anxiety and presence. Unfortunately, this experiment did not incorporate a desktop setting with the task presented visually on a monitor.
Another study that looked at emotional responses for a specific stressor compared the following three stimuli: real food, food photographs and virtual food presented with an HMD [9]. The participant group included females that were either healthy, anorexia patients or bulimia patients. Anxiety and presence were assessed with questionnaires. Heart rate, respiration rate and skin conductivity were measured to specify the physiological response. Real food and virtual food elicited a comparable emotional response in anorexia and bulimia patients, and both were higher than the response for food photos. For the VR condition, higher presence was correlated with more anxiety. These differences were not found in healthy participants [9]. This study shows that virtual stimuli can elicit a response that is similar to the response to real stimuli. Secondly, these results indicate that stressors specific to patients elicit a higher response compared with the response in healthy subjects. But it is necessary to note that the VR condition was the only one that included interactivity. The participants had to remove a lid from the food plates themselves, but this was done automatically during the other conditions. This makes it more difficult to compare the conditions.
The effect of immersion was also studied in combination with narrative. Gorini and colleagues had participants play a virtual hospital experience either on a screen or with an HMD, and with or without narrative [10]. The HMD condition was evaluated as more immersive than the screen condition, and it was found that
the virtual environment and objects in it were perceived as more real. Presence was measured with a questionnaire and HR was recorded to indicate the emotional response. HR was found higher in the group that experienced the virtual hospital with the HMD in combination with a narrative, compared with the same group without narrative. Immersion had no effect on heart rate [10]. Taken together, the results of this study show that a combination of high immersive techniques and narrative elicit the highest emotional response and subjective experience of presence. This experiment found no differences in HR between the groups with and without HMD [10].
Another study examined participants’ responses to a virtual modification of the Stroop task with incongruent stimuli, presented with a monitor, HMD and six-wall CAVE [13]. In addition to measuring anxiety, arousal, presence, GSR and SCR, simulator sickness was also evaluated. The results show that the CAVE and the HMD elicit more arousal than the monitor. SCR was higher for the CAVE compared with both HMD and monitor conditions. The HMD resulted in more sickness compared with the CAVE and monitor. Presence was highest in the CAVE compared with both HMD and monitor conditions. The HMD resulted in more sickness compared with the CAVE and monitor [13]. This study used a well-known task that evokes a stress response, namely the Stroop task. It is noteworthy to find differences in response for this task depending on the visual presentations used. The HMD elicited negative emotions during the commonly used stress task, but has a higher risk of inducing sickness compared with CAVE and monitor conditions.
Simulator sickness is also called cyber sickness and was studied by Sharples and colleagues as well [18]. Participants were instructed to complete a search task in a virtual factory, either on a computer screen or with an HMD. The HMD evoked more disorientation and sickness compared with the computer screen [18]. Measurements of stress and presence were however not included in this study.
To study VR induced stress more closely, Pallavicini and colleagues examined responses to VR with a HMD that involved technical breakdowns [15]. This condition was compared with video, audio and text conditions. The goal of this research was to test if VR still evokes stress responses when technological breakdowns occur. Physiological measurements, HR amongst others, indicated that VR with technological breakdowns induced less stress compared with the other conditions [15]. Unfortunately, the script that was used included actors in a real-life setting, and not an animated one with models as it would be used in a VR game. The VR condition was also not tested without technological breakdowns.
The studies discussed until here did not include virtual stressors that are relevant specifically for the military field of work and work related experiences. Virtual scenes that do include these kinds of stressors can be found in first person shooter (FPS) games. Two games from this genre were used in a study by Bouchard and colleagues [4], that also involved military personnel as participants, namely Afghanistan soldiers. The commercial games Left4Dead and Killing Floors were played on two different sized screens with 3D glasses, and in a four-wall CAVE environment. The soldiers also completed the Trier Social Stress Test (TSST). Immersion and stress were measured with questionnaires and heart rate was included as an indication of physiological response. It was found that the TSST and the games
Table 1: Studies examining VR tasks that induce stress.
Q = questionnaire, HR = heart rate, HRV = heart rate variability, GSR = galvanic skin resistance, SCR = skin conductivity response
Article Monitor HMD Presence
Q Stress Q HR HRV GSR/SCR
Military Stimuli
Military
Subjects Sickness
Juan et. al. [11]
Gorini et. al. [9]
Gorini et. al. [10]
Kim et. al. [13]
Sharples et. al. [18]
Pallavicini et. al. [15]
Bouchard et. al. [4]
Current study
elicited stress similarly throughout the different conditions, based on HR values. Screen size and selected FPS did not affect stress responses and immersion in any way [4]. The most noteworthy result of this experiment is that these commercial FPS games can induce stress in military personnel, similar to a stress task that is frequently used in stress research. Because there was no monitor condition without stereoscopic view and there was no HMD used, comparison amongst immersive conditions cannot be discussed. In sum, not all previous experiments have similar outcomes. It seems that some experiments show differences in physiological parameters between monitor and HMD conditions, and some do not. This could depend on what immersive conditions exactly were compared and if the stimuli used were specific for a disorder or not. Additionally, the response to stressors that was elicited in participants was defined differently throughout the articles. It had been called emotional response, stress response, anxiety, negative emotions or (emotional) arousal. Either way, responses were generally evoked by stressful stimuli and are measured with physiological data and questionnaires. The studies also used the terms immersion and presence differently. For instance, the FPS study with military personnel assessed immersion with a questionnaire [4]. But the terms immersion and presence have different definitions therefore it is not acceptable to use the terms interchangeably. The terms were used correctly in the study with the virtual hospital experience, in which self-reported presence results confirmed the different immersive conditions [10]. To examine how different levels of immersion affect the stress response to a virtual task in a military setting, it is necessary to include both subjective and objective measurements. An overview of the discussed literature is presented in Table 1, in which the checkmarks represent what task conditions and measurements were included in the experiment. As can be seen, all studies except the ones by Juan [11] and Bouchard [4] included a monitor and an HMD condition. It must be noted though, that the monitor condition in the study by Gorini [9] did not represent a world in which the player could move around, merely photographs were shown on the screen. This was also the instance for the study by Pallavicini [15], a monitor was used here to present a video of the script. This eliminated any control by the participant. The study by Bouchard [4] only used monitors in combination with 3D glasses and no HMD at all. Just three studies have included questionnaires for both presence and stress [4, 9, 11], but none of
them included measurements for two objective parameters. To obtain a complete view on the stress response it is also important to examine the sickness participants experience during the task. HMDs have a higher possibility to induce sickness [13, 18]. It is therefore necessary to examine if sickness itself is a stressor in a certain experimental protocol. Only one of the studies that examined sickness included both subjective and objective measurements for stress [13], but this study did not use stressors specific for the military work environment. The one study that used a commercial FPS for inducing stress in soldiers did not include an HMD condition, and the stressors were not realistic at all because the games were horror themed [4].
To fill the gap that exists in current literature, the goal of the current study is to compare the responses to a FPS game in a warzone setting that is played on a monitor compared with an HMD. Presence and stress are assessed with questionnaires, and the physiological response is assessed by measuring the HRV and SCR (Table 1 ‘current study’).
3. RESEARCH QUESTIONS 7
HYPOTHESES
The main research question of this study is:
Is the stress response higher when playing a FPS game with an HMD, compared with playing on a 24-inch monitor?
This main goal can be reached by answering the following sub-questions:
1. What is the self-reported stress level? 2. What are the HRV and SCR responses?
3. What is the self-reported experience of presence? 4. What is the self-reported level of sickness?
5. What is the threat and challenge assessment of the game?
6. What are specific stressful elements in the game? It is hypothesized that stress responses will be higher when stress is evoked by playing the game with an HMD. This means that self-reported measures of stress will be lower for playing the game on a traditional monitor. HRV decreases are associated with an increase in stress response, as there is less time between
subsequent beats. It is therefore expected that HRV will be lower when the game is played with the HMD. Increases in SCR are associated with an increase in stress response, so SCR is expected to be higher when the game is played with the HMD. To confirm the two levels of immersion, this study measured the level of self-reported presence induced by the two different immersive conditions. It is theorized that playing the game with an HMD is more immersive, and will therefore induce a higher sense of presence compared with the monitor. Based on previous results it is expected that the HMD will induce more sickness than the monitor [13, 18]. The self-reported assessment of threat and challenge is hypothesized to be higher for the HMD gameplay. Lastly, it is expected that visual elements will be more stressful when the game is played with the HMD.
4. MATERIALS & METHODS
4.1 Participants
Sixteen male participants were selected, of which fourteen completed the experiment and were included in analysis (N=14, Age: M=22,21 SD=3,40). Participants were mostly students (N=13) and did not receive compensation for taking part in any way. Selection was based on age (between 18 and 32) and low susceptibility to sickness (measured on a 7-point Likert scale, Appendix 2, items 1 and 8-10). The average susceptibility for nausea was M=1,43 SD=0,514, for headaches it was M=1,57 SD=0,76 and for migraines it was 1,00 throughout. Participants were asked about their gaming experiences over the last six months prior to the experiment (see appendix 2, items 2-5). The average virtual gameplay experience was between four and six hours per week and FPS games were played between zero and four hours per week. Battlefield 3 and other games from the series were played up to two hours per week over the last six months. Respondents who took medication that influences sickness, just like respondents with eye problems, stress or anxiety related disorders were not included in the experiment. Three participants wore lenses and two wore glasses, all fourteen participants were randomly assigned to a group: one group played the mission on the monitor first and with the HMD second, the other group played with the HMD first and the monitor second.
4.2 Game mission
The game that participants played is the eighth mission of Battlefield 3, called ‘Fear No Evil’ [3]. This mission was chosen because of the scripted nature, which ensures that every participant has a similar game experience. Battlefield 3 is an earlier game in the series, and can be played on a traditional monitor and with the Oculus Rift DK2, an HMD that is often used in VR gaming (Figure 1). None of the researches discussed in section 2 used the Oculus Rift during experiments.
The player controls consisted of a keyboard and a mouse. The mouse was used to look around, shoot and zoom in. The keyboard was used to drive a tank, with WASD keys and C for thermos vision. Keys 1 and 2 were used to select a weapon. In the Oculus Rift gameplay, the players’ head movements could also be used to look around and aim. The mouse was still available for looking around, so players could choose what they preferred to use. Fear No Evil starts with a short introductory video (see Figure 2), which explains the background story for the mission and describes the situation of the playable character. When the mission starts, the player is situated in a tank with only the ability to look around and shoot. For the Oculus Rift condition, this first part of the
Figure 1: Oculus Rift DK2. Obtained from [14].
mission is a good way to get used to the head movement controls in the game. After some moments, the game comments that the player can now take control of the tank. The player drives the tank a few streets before being ambushed by enemy fire. During the next stage of the mission the player drives the tank around the streets, even right through a building, and eliminates enemies and hostile tanks. The player then drives the tank to a bank building, where a brief cut scene starts. The player cannot drive the tank anymore, because the next goal is to defend a stationary evacuation helicopter from enemy fire. During the last couple of minutes of the mission, waves of enemies pose a threat to the player and at the end of the mission, the player is captured by an enemy that climbed on top of the tank. The mission takes between five and eight minutes to complete (if the player is not loitering around), the average time was 6:46 minutes.
Figure 2: Fear No Evil introductory video. This picture shows
one participant playing the mission on the desktop monitor, the screenshot shows a scene from the introductory video. The sheets
of paper on the left have the instructions written on them, they were explained by the experimenter as well. White wires can be seen on the top of the left hand fingers, these are electrode wires
4.3 Physiological measurements
Most of the previous studies described in section 2 included HR measurements to indicate sympathetic nervous system activity as a response to stressors (Table 1). This study recorded HRV measurements instead, because they provide information about the variability in times between beats. HRV was derived from an ECG measurement by calculating the time in milliseconds (ms) between R peaks. The RMSSD method (root mean square of successive differences between adjacent R peaks) was then used to calculate the HRV variable.
To look more closely at the sympathetic response, a second parameter was measured. Sweat production increases in response to stressors, hence skin conductivity was measured to indicate changes. Conductivity was recorded in micro Siemens (µS) and was chosen instead of resistance because it accounts for differences in conductivity for each individual sweat gland. As can be seen in Table 1, some previous studies also analyzed skin resistance or conductivity.
Skin conductivity and ECG were measured with the Equivital Hidalgo system [1]. Participants wore a chest strap that recorded ECG, and two electrodes were connected to this belt. These electrodes were placed upon the top of the second phalanges from the top of the middle finger and index finger of the left hand (Figure 2). The belt was worn under the participants’ shirts, covering the belt with the electrode wire coming out of the left sleeve (Figure 2).
4.4 Subjective measurements
Self-reported stress Similar to [9], a visual analogue scale (VAS) was used to measure the subjective experience of stress. Participants were asked to cross a horizontal 100mm line with a vertical stripe to indicate how stressful they felt. Drawing the line further to the right indicated a higher level of stress experience. The distance from the left of the line to the vertical mark was measured in mm. The difference between the rest task and the gameplay task was then calculated to construct variable ΔVAS.
Self-reported presence Similar to a previous study [9], the UCL presence questionnaire was used to learn about the self-reported sense of presence. The questionnaire consists of three statements about the feeling of being present in the virtual world (Appendix 4, items 17-19). Statements were assessed on a 7-point Likert scale, with higher scores indicating a stronger sensation of presence. Because only the gaming conditions took place in a virtual world, this questionnaire was not included for the baseline measurements.
Self-reported sickness The gameplay questionnaire included items about headache, nausea and dizziness to find out if cyber sickness occurred (Appendix 4, items 20-22). The items were statements about feeling sick and the participants had to indicate how much they agreed with them on a 7-point Likert scale, with ‘strongly agree’ to the right.
Threat and Challenge This study used an additional questionnaire to inspect how threatening and challenging the game was. The threat-challenge questionnaire from Drach-Zahavy and Erez [5] was translated to Dutch and participants filled in this questionnaire for the game conditions only (Appendix 4, items 5-16), not for the baseline. Items were answered on a 7-point Likert scale, with ‘strongly disagree’ all the way left and ‘strongly agree’ to the right.
Stressful game elements To look more closely at specific stressful elements of the game mission, participants were also asked to fill in a questionnaire about these elements. They were asked to indicate how stressful these items were on a 7-point Likert scale. Examples of these elements are visual ones, for instance smoke and driving the tank through a building. Some non-visual elements were included as well (Appendix 4, item 2). The elements were assessed on a 7-Point Likert scale, with ‘strongly disagree’ all the way left and ‘strongly agree’ to the right. Some elements were not always present during gameplay, participants could indicate this by selecting the ‘not available’ option. These elements are: camera recording, thermovision use and friendly fire. This section of the questionnaire was only included for the gameplay, not for baseline.
4.5 Set Up
The experiment was conducted at Science Park Amsterdam, a location part of the University of Amsterdam. A desktop computer and the Oculus Rift DK2 with lens set A was used (Appendix 1). Even though the room was used solely for the experiment, participants wore headphones to exclude any external sounds from distracting them. The part of the room that was used by the experimenter was situated behind the gameplay computer. Additional screens were used to eliminate external visual input. Participants used two stations: one table that was used for the rest task and to fill in the questionnaires, and one table that was used for playing the game (Figures 3 and 2). The game was played with Origin Online on a desktop computer with a 24 inch monitor. VorpX was needed as additional software to play Battlefield 3 with the Oculus Rift [19]. VorpX was set to ‘Virtual Cinema Mode’, otherwise the screen appeared zoomed in and the edges of the screen were not visible.
Figure 3: Rest station. This table was used by the participants to
complete the rest task and fill in questionnaires. Relaxing music was played through the headphones. The front of the table was
blocked by a black screen to eliminate external visual stimuli.
4.6 Procedure
Figure 4 shows an overview of the experiment procedure, the individual phases are explained in this section.
INTRO The experiment was firstly explained to the participants, after which they were asked to sign a consent form. A separate area was available for the participants to try on different sizes of Equivital chest straps. The experimenter verified the functionality of every chest strap before testing began.
Figure 4: Experiment procedure overview. After explaining the experiment during the introduction, every participant goes through two
blocks of testing. The first block is indicated by the purple frame and the second block is indicated by the orange frame. Every block consists of a rest task and a game task, indicated by the blue and red arrows.
REST In order to measure baseline values for HRV and SCR, participants had to complete a rest task for four minutes. The objective was to stare at an A4 sheet of paper that showed six furniture items in a 3D perspective (Figure 5). They were not allowed to look anywhere else. The headphones played relaxing music [16].
Figure 5: Rest task stimuli.
SELF RAPPORT Participants filled in a questionnaire during
this phase depending on the task that was performed just before. The questionnaire following a rest phase only included the VAS for subjective stress experience (Appendix 3). The questionnaire following the game task included the VAS also, in addition to items about stressful game elements, threat, challenge, presence and sickness (Appendix 4). There were no time restricts for this phase.
GAME The “Fear No Evil” mission was played during this phase. Depending on the group they were assigned to (see section 4.1), participants played the game first with the desktop monitor or Oculus Rift. The second time during this phase they used the medium that was not used the first time. It started with an explanation of the controls and by providing the participants with a short overview of the mission. They were informed that the mission starts with a video, and that this part could be used to try out the Oculus Rift head tracking (if the Oculus Rift was used during this phase). Participants were instructed to alarm the experimenter in the situation of technical
failures, because the Oculus Rift input could not be mirrored on the desktop monitor.
5. RESULTS
5.1 Subjective results
None of the variables were normally distributed, therefore Wilcoxon Signed Rank tests were executed with IBM SPSS Statistics 22 to analyze differences between gameplay conditions and groups.
One participant was excluded from VAS analysis, because he did not report any stress at all. When there is no difference calculated between variables the data is excluded from ranking [7]. As can be seen in Table 2, subjective stress levels significantly increased between baseline and gameplay for the monitor and the Oculus Rift. The increase in stress experiences is also statistically different between the conditions, with the Oculus Rift increasing more stress compared with the monitor.
Table 2: VAS Analysis.
B = baseline, GP = gameplay, Δ = delta. Variables were measured in millimeters (mm).
Variable Mean SD Significance
(P<0,05) z-Score Monitor B 18,77 19,20 0,033* -2,133 Monitor GP 26,19 21,26 Oculus B 13,81 13,58 0,013* -2,482 Oculus GP 34,00 20,71 Δ Monitor 7,42 11,95 0,033* -2,133 Δ Oculus 20,19 19,71
The results for presence are displayed in Table 3. The first two statements of the UCL presence questionnaire indicate a significantly higher presence level for the Oculus Rift, compared with the monitor (Appendix 4, items 17 and 18). Average scores calculated from all three statements are also significantly higher for the Oculus Rift condition.
The gameplay questionnaire included three statements about sickness: nausea, headache and dizziness. As shown in Table 4, there were no differences found in nausea and headache scores between the two conditions. For dizziness however, analysis
reveals that the Oculus Rift induced more dizziness compared with the monitor.
Table 3: UCL analysis.
1 = statement 1, 2 = statement 2, 3 = statement 3. Total scores are average scores calculated for all statements combined. Variables were measured on a 7-point Likert scale.
Variable Mean SD Significance
(P<0,05) z-Score Monitor 1 3,00 1,41 0,039* -2,069 Oculus 1 4,57 1,70 Monitor 2 2,36 1,34 0,023* -2,268 Oculus 2 3,71 1,68 Monitor 3 2,43 1,28 0,263 n.s. -1,119 Oculus 3 3,14 1,88 Monitor Total 2,60 0,93 0,006* -2,755 Oculus Total 3,81 1,20
Threat and challenge mean scores were calculated by averaging the scores from the six corresponding statements for threat and challenge separately. Table 5 shows that for both immersive conditions, challenge scores are significantly higher than threat scores. Comparing threat scores between conditions did not reveal differences (p = 0,779 z = -0,281). Challenge scores also did not differ between conditions (p = 0,384 z = -0,871). Figure 6 shows threat and challenge average scores per participant.
Additional analysis were performed on stressful game elements data, by comparing mean scores for every element between conditions. Analysis revealed that elements ‘enemy fire’ and ‘explosions’ were assessed as more stressful during the Oculus condition (Table 6).
Table 4: Sickness analysis.
N = nausea, H = headache, D = dizziness. Variables were measured on a 7-point Likert scale.
Variable Mean SD Significance
(P<0,05) z-Score Monitor N 1,07 0,27 0,317 n.s. -1,000 Oculus N 1,14 0,36 Monitor H 1,07 0,27 0,59 n.s. -1,890 Oculus H 1,50 0,941 Monitor D 1,07 0,27 0,038* -2,070 Oculus D 1,57 0,76
Table 5: Threat and challenge analysis.
Variables were measured on a 7-point Likert scale.
Variable Mean SD Significance
(P<0,05) z-Score Monitor Threat 1,65 0,72 0,002* -3,173 Monitor Challenge 3,83 0,79 Oculus Threat 1,70 0,67 0,001* -3,301 Oculus Challenge 4,05 0,69
Figure 6: Threat and challenge scores for both gameplay conditions. Mon = monitor, ocu = Oculus Rift DK2. Scores are
average scores for six 7-point Likert scale items.
Table 6: Stressful element analysis.
EF = enemy fire, EXP = explosions. Variables were measured on a 7-point Likert scale.
Variable Mean SD Significance
(P<0,05) z-Score Monitor EF 2,64 1,82 0,041* -2,332 Oculus EF 3,43 1,56 Monitor EXP 2,79 1,67 0,020* -2,332 Oculus EXP 3,71 1,90
5.2 Physiological results
HRV and SCR means were compared within and between the conditions with Wilcoxon signed rank tests as well. Tables 7 and 8 show the results of this analysis. HRV means did not differ between baseline and gameplay tasks for both immersive conditions. There was no difference found in HRV between monitor and Oculus conditions either. No differences were found in SCR means between baseline and gameplay task for both conditions. Additionally, no differences in SCR were shown between the monitor and Oculus.
Table 7: HRV analysis.
B = baseline, GP = gameplay. HRV was measured in milliseconds (ms)
Variable Mean SD Significance
(P<0,05) z-Score Monitor B 148,22 99,80 0,826 n.s. -0,220 Monitor GP 156,18 66,13 Oculus B 154,84 74,48 0,583 n.s. -0,549 Oculus GP 153,10 79,25 ∆Monitor 7,96 66,96 0,875 n.s. -0,157 ∆Oculus -1,74 73,78
Table 8: SCR analysis.
B = baseline, GP = gameplay, Δ = delta. SCR was measured in micro Siemens (µS).
Variable Mean SD Significance
(P<0,05) z-Score Monitor B 4,77 2,19 0,480 n.s. -0,706 Monitor GP 4,61 2,20 Oculus B 4,69 2,37 0,875 n.s. -0,157 Oculus GP 4,70 2,60 Δ Monitor -0,16 1,89 0,784 n.s. -0,275 Δ Oculus 0,01 0,88
6. DISCUSSION
To answer the main question of this study, both subjective and objective measurements were recorded. As shown in Table 1, some previous research did not involve more than one parameter for both types of measurement. Similar to previous studies though, the current experiment included self-reported stress measurements. Analysis indicated a higher self-reported stress response in the HMD condition compared with the monitor, a result that was also found by [9, 13].
Another result of this study is that self-reported presence was higher for gameplay with the HMD, in comparison to the monitor. Previous studies that examined presence as well show the same difference [9, 10]. On the other hand, Kim and colleagues [13] did not find any differences in presence between HMD and monitor conditions. This may be because the task that was used, the Stroop task, was not developed for a three dimensional virtual world. In contrast, the study did show a higher sense of presence for the CAVE condition [13]. Being able to physically walk around in the environment might be the reason that presence was higher during this task. Pallavicini and colleagues [15] also showed no differences in presence between HMD and monitor conditions, this is perhaps because of the intentional technical breakdowns that occurred during the tests.
Previous literature shows that sickness is higher in participants who completed a task with an HMD compared with a monitor [13, 18]. This finding is partly confirmed by the current study. Headaches, nausea and dizziness were tested and only dizziness was higher when the mission was played with the HMD. The Oculus Rift did not induce headaches and nausea, which might be because participants were selected based on a low susceptibility to these symptoms. This result can also be explained by looking at the type of mission and the player controls used in the current study. Players were able to sit still in a chair and were instructed to use the head tracking to look around and aim whenever they wanted, but it was not necessary at all. Gameplay tests with the Oculus Rift that took place before the experiment did effectuate nausea in some players. After discussing the players’ experiences it seemed that other factors might have influenced sickness sensations, for instance: susceptibility, mood, mental and physical state.
To use a simulation paradigm for stress induction, it seems reasonable that the task used should be threatening to the users in some way. This was tested by the threat and challenge questionnaires in this study. It was found however, that the game mission from Battlefield 3 was perceived as more challenging than threatening, independent of condition. Several participants actually stated during the experiment that they were excited about
being able to play a FPS game with the Oculus Rift. Combining this with the fact that the participants were not military personnel, it may be that the mission was seen as merely entertainment. The missions’ threat level might be assessed differently by military personnel in this case.
An additional analysis of the stressors in Fear no Evil revealed that two types of visual stimuli, ‘enemy fire’ and ‘explosions’ were evaluated as more stressful during Oculus Rift gameplay compared with the monitor. It was expected beforehand that visual type stimuli might be evaluated differently between conditions. Other visual stimuli that were not more stressful included thermovision use, friendly fire and smoke.
This study also looked at sympathetic nervous system activity as a response to virtual stressors. Two parameters were measured, namely HRV and SCR. In contrast to a number of previous studies, the current experiment did not show different sympathetic nervous system activity based on the two parameters. Various previous experiments also do not show differences in heart rates (not heart rate variability) between monitor and HMD conditions [4, 9, 10]. One study examined heart rate variability as well, based on this it was concluded that less stress occurred in the HMD condition compared with video [15]. As explained before, this study only tested technical breakdowns in VR, not without technical issues. Skin conductivity was similar between monitor and HMD conditions in previous literature [9, 13]. Correspondingly, the current study also did not show differences in skin conductivity between the two immersive conditions. The absence of differences in physiological stress response may be because of errors in the experiment procedure. Participants reported to feel more aroused before first time gameplay. Additionally, some baseline physiology measurements were higher (not significantly) compared with gameplay. For instance, baseline SCR values were higher than the gameplay SCR values in the monitor condition. Baseline measurements were possibly too short, so that participants could not fully relax in between tasks. This would indicate a reduction of stress for the gameplay compared with baseline, if the results were statistically significant.
7. CONCLUSIONS & FUTURE WORK
The main goal of this study was to examine if players had a higher stress response when playing a FPS game with the Oculus Rift compared with a desktop monitor. This was measured by letting participants play a Battlefield 3 mission on a monitor and with the Oculus Rift DK2. Results show that participants reported to experience higher stress levels, more presence and more dizziness during the Oculus Rift gameplay. The game mission was seen as more challenging than threatening, independent of condition. Two visual elements in the game, ‘enemy fire’ and ‘explosions’ were seen as more stressful during Oculus Rift gameplay. Physiological data including HRV and SCR did not reveal differences between the gameplay conditions. The main conclusion of this study is that even though there is no change measured in sympathetic nervous system activity, playing a FPS game in a warzone setting is experienced as more stressful than playing on a traditional desktop monitor. For upcoming TNO experiments it might be interesting to examine the experimental set up that is used in this study even further, most importantly by involving military personnel as participants. It might also be interesting to examine if subjective experiences alone will be sufficient for an effective transfer of training to the operational field, because changes in physiology were not found here. Additional results will thensupport the development of virtual reality techniques that can be used in military stress training.
8. ACKNOWLEDGMENTS
Appreciations are in order for all the people involved that helped to realize this paper:
Anja van der Hulst (UvA, TNO) Gillian van de Boer-Visschedijk (TNO) Olaf Binsch (TNO)
Rudy Boonekamp (TNO) Andrea Jetten (TNO) Pierre Valk (TNO) Isabelle Lamers Ronald Jongen Mina Biparva Sander Bakkes (UvA)
9. REFERENCES
[1] About Equivital: http://www.equivital.co.uk/about. Accessed: 2015-07-26.
[2] Alexander, A.L. et al. 2005. From Gaming to Training : A Review of Studies on Fidelity , Immersion , Presence , and Buy-in and Their Effects on Transfer in PC-Based Simulations and Games. November (2005), 1–14. [3] Battlefield 3 Fear No Evil:
http://www.ign.com/wikis/battlefield-3/Fear_No_Evil.
Accessed: 2015-05-04.
[4] Bouchard, S. et al. 2012. Modes of immersion and stress induced by commercial (off-the-shelf) 3D games. The
Journal of Defense Modeling and Simulation: Applications, Methodology, Technology. (2012).
[5] Drach-Zahavy, A. and Erez, M. 2002. Challenge versus threat effects on the goal-performance relationship.
Organizational Behavior and Human Decision Processes. 88, 2 (2002), 667–682.
[6] Driskell, J.E. and Johnston, J.H. 2005. Stress Exposure Training. Making Decisions Under Stress. 191–210. [7] Field, A. 2009. Discovering Statistics Using SPSS. [8] Fulton, J.J. et al. 2015. The Prevalence of Posttraumatic
Stress Disorder in Operation Enduring
Freedom/Operation Iraqi Freedom (OEF/OIF) Veterans:
A Meta-Analysis. Journal of Anxiety Disorders. 31, (2015), 98–107.
[9] Gorini, A. et al. 2010. Assessment of the emotional responses produced by exposure to real food, virtual food and photographs of food in patients affected by eating disorders. Annals of general psychiatry. 9, (2010), 30. [10] Gorini, A. et al. 2011. The role of immersion and
narrative in mediated presence: the virtual hospital experience. Cyberpsychology, behavior and social
networking. 14, 3 (2011), 99–105.
[11] Juan, M.C. and Pérez, D. 2003. Comparison of the levels of presence and anxiety in an acrophobic environment viewed via HMD or CAVE. International Journal. 4, 4 (2003), 379–456.
[12] Kavanagh, J. 2005. Stress and Performance. [13] Kim, K. et al. 2014. Effects of virtual environment
platforms on emotional responses. Computer Methods
and Programs in Biomedicine. 113, 3 (2014), 882–893.
[14] Oculus Rift DK2:
http://i1-news.softpedia- static.com/images/news2/New-and-Improved-Oculus-Rift-Dev-Kit-2-Revealed-Ships-in-July-433129-2.jpg.
Accessed: 2015-08-12.
[15] Pallavicini, F. et al. 2013. Is virtual reality always an effective stressors for exposure treatments ? some insights from a controlled trial. BMC Psychiatry. 13, 1 (2013), 1.
[16] Relaxing music:
https://www.youtube.com/watch?v=TAH_0ZwmHO8.
Accessed: 2015-06-18.
[17] Rizzo, A. et al. 2006. A Virtual Reality Exposure Therapy Application for Iraq War Post Traumatic Stress Disorder. IEEE Virtual Reality Conference (VR 2006). (2006).
[18] Sharples, S. et al. 2008. Virtual reality induced symptoms and effects (VRISE): Comparison of head mounted display (HMD), desktop and projection display systems. Displays. 29, 2 (2008), 58–69.
[19] VorpX: https://www.vorpx.com. Accessed: 2015-07-06.
Columns on Last Page Should Be Made As Close As
Possible to Equal Length
Appendix list:
1.
Software and hardware2.
Questionnaire general information3.
Questionnaire rest taskAppendix 1: Software and hardware
Computer Windows 8.1 Pro
Intel® Core™ i7-5930K CPU @ 3.50 GHz 16 GB RAM
NVIDIA GEFORCE GTX TITAN BLACK Microsoft wired keyboard 600
ASUS VG248QE 24.0” 16:9 monitor CM STORM devastator mouse
Display 1 = Rift DK2, 1920x1080, 60Hz refresh rate Display 2 = ASUS VG249, 1920x1080, 60 Hz refresh rate Eye – screen distance 72cm
VorpX VorpX version 0.8.1
Virtual Cinema Mode scene ‘none’ Show on VG248 (monitors were switched) FOV = 120°
Oculus Rift DK2 Runtime version 0.5.0.1
Firmware version 2.12
Rift Display Mode = ‘Extend desktop to HMD’
Battlefield 3 FOV = 90°
Appendix 2: Questionnaire general information
1. Wat is je leeftijd (vul een getal in):2. Hoe veel uren speel je gemiddeld per week een virtuele game (denk aan de afgelopen 6 maanden) 3. Hoe veel uren speel je gemiddeld per week een First Person Shooter (denk aan de afgelopen 6 maanden) 4. Hoe veel uren speel je gemiddeld per week een game uit de Battlefield serie (denk aan de afgelopen 6 maanden) 5. Hoe lang is het geleden dat je Battlefield 3 gespeeld hebt, of informatie erover hebt verkregen
6. Dominante hand (omcirkel de juiste optie)
7. Beschrijf enige belemmerende oogproblemen (bijv. kleurenblindheid, bril/lenzen) 8. Vatbaarheid op misselijkheid (kies één cijfer)
9. Vatbaarheid op hoofdpijn (kies één cijfer) 10. Vatbaarheid op migraine (kies één cijfer)
11. Slik je medicatie die de vatbaarheid op hoofdpijn, misselijkheid or migraine verhoogt (omcirkel de juiste optie) 12. Heb je een angst- of stress gerelateerde stoornis of aandoening (omcirkel de juiste optie)
Appendix 3: Questionnaire rest task
In hoeverre voelde je je gestrest tijdens de rust taak? Zet een verticale streep op de lijn hieronder, daar waar het overeenkomt met je ervaring. Hoe verder de streep naar rechts staat, hoe hoger je stress ervaring.
Appendix 4: Questionnaire gameplay task
1. In hoeverre voelde je je gestrest tijdens de taak? Zet een verticale streep op de lijn hieronder, daar waar het overeenkomt met je ervaring. Hoe verder de streep naar rechts staat, hoe hoger je stress ervaring.
2. Geef bij de onderstaande onderdelen aan in hoeverre ze voor jou stressvol waren tijdens het doen van de taak. Omcirkel per onderdeel één cijfer, of kies ‘niet van toepassing’ mocht het onderdeel niet zijn voorgekomen.
Camera opname Video in begin Geluidseffecten
Met tank door gebouw rijden Bank veiligstellen (missie deel) Helikopter verdedigen (missie deel) Overmeesterd worden aan het einde Vijandelijk vuur
Explosies
Gesproken communicatie Setting van missie (omgeving) Dreiging van snipers
Vuurgevechten Thermovisie gebruik Friendly fire Rook
(On)zichtbaarheid van vijanden
3. Wist je van te voren dat je overmeesterd zou worden aan het einde van de missie? Omcirkel de juiste optie. JA / NEE 4. Heb je de missie in één keer uitgespeeld? JA / NEE, namelijk pas na ____ pogingen
5. Deze taak was een dreiging voor mij
6. Ik ben bezorgt dat deze taak mijn zwakke plek heeft blootgelegd 7. Deze taak was lang en saai
8. Ik denk dat deze taak mijn zelfvertrouwen aan heeft getast 9. In het algemeen denk ik dat ik niet succesvol was in deze taak
10. Ik ben bezorgd dat ik de capaciteiten niet heb om deze taak succesvol uitgevoerd te kunnen hebben 11. Deze taak was een uitdaging voor mij
12. Deze taak gaf mij de mogelijkheid mijn cognitieve vaardigheden te laten zien 13. Deze taak gaf mij de mogelijkheid om obstakels te overwinnen
14. Deze taak gaf mij de mogelijkheid om mijn zelfvertrouwen te versterken 15. Over het algemeen denk ik succesvol te zijn geweest in het uitvoeren van de taak 16. Ik denk dat ik de capaciteiten heb voor een succesvolle prestatie
17. Hoe sterk was het gevoel dat je je echt in de virtuele omgeving bevindt? 18. In hoeverre waren er momenten dat de virtuele omgeving werkelijkheid was?
19. Als je terugdenkt aan je virtuele ervaring, denk je dan aan de beelden die je zag of aan de plaatsen die je bezocht? 20. Ik was misselijk tijdens het uitvoeren van de taak
21. Ik had hoofdpijn tijdens het uitvoeren van de taak 22. Ik was duizelig tijdens het uitvoeren van deze taak
