1 Introduction

Deployed United States Army Soldiers are frequently required to make critical, time-sensitive decisions in complex and dynamic operational environments. According to U.S. Army Regulation 350-1, the Army will train units and staffs in their core competencies under conditions that accurately and “realistically portray the operational environment” [1]. The extreme tactical environments our Soldiers face in contingencies overseas are unpredictable, mentally and physically challenging, and often insurmountable. These environments pose risk levels that are unacceptable to reproduce for training purposes. Therefore, it is critical to develop training tools that can be used and tested within a safe setting that can simulate extreme environments. To best simulate the experiences of Soldiers in theater, a training environment must be immersive in that it takes into account the emotional factors [2], creating the illusion of “being there,” thus providing a heightened level of arousal and encouraging the desire to perform well within the training.

The military uses simulations to train many diverse tactical and social skills, and to test the integration of new systems for future force use. However, current virtual simulations rarely mimic the actual harsh environmental conditions that our military encounters, such as sound, light, smell, or threat of return fire. Thus, because we cannot quantify the stress levels faced in theater, it is difficult to know the extent of comparability that simulations provide in terms of inducing stress in Soldiers.

The U.S. military does employ live training that is able to more closely approximate the battlefield environment than current simulation systems. Survival, Evasion, Resistance, and Escape (SERE) training is both physically and psychologically demanding, and is designed to parallel the stress experienced during real war, captivity, and other combat missions. This type of large-scale, field-based training offers a more realistic experience, but is more amenable to observational studies [3] than controlled experimentation. Immersive Virtual Environment systems that simulate the field-like operational environments allow for good experimental design and control. If these virtual systems were able to provide a more realistic immersive experience, they could provide a low-cost, low-risk way to experimentally measure Soldier response to specific training interventions and lead to more effective military training.

The introduction of stressor threats, such as return fire, in a simulation-based training scenario may induce higher levels of stress that more closely approximate those experienced in battlefield situations. If this is the case, then this newly stressful environment should produce behavior and performance that is more representative of how Soldiers respond in real-world environments. In order to understand the potential effectiveness of such additional stressor elements, arousal must be objectively measured while Soldiers perform real-world tasks in both simulated and live training events.

In order to best match a training environment to situations encountered on the battlefield, Soldiers must be put under stress. One component of the stress response is arousal, which can be caused by both physical and psychological stress [4]. Arousal is linked to learning, and is believed to be a determinant of one’s mental capacity to handle the stress of a given situation; that is, arousal levels that are too high or too low work against the trainee [5]. The effects of arousal on cognition and performance follow an inverted-U, such that a certain level of arousal, at the peak of the U, actually improves or even optimizes performance, but arousal beyond that optimal level will begin to hurt performance [6]. Lazarus and Folkman define stress as a state produced when stressors (environmental or social) tax or exceed an individual’s adaptive resources [7]. Fatkin and Patton extended Lazarus and Folkman’s definition to include that stress is a, “multifaceted, dynamic, and interactive process with psychological and physiological dimensions” [8]. Because stress has this multifaceted nature, psychophysiological stress measures show how a person is responding to both simulated and real events. The goal of the present study was to assess whether a simulated threat, a return-fire shock, is effective in producing a stress response that may more closely represent stress experienced on the battlefield. We examined arousal, as a proxy for stress and cognitive engagement, through heart rate variability.

Psychophysiological measures have been linked to psychological stress and cognitive function [9]. One such measure, heart rate variability (HRV), provides a non-invasive measure of the autonomic nervous system (ANS). Here, we report on interbeat interval (IBI), which measures the peak-to-peak interval of heartbeats, and is linked to cognition, such that a reduction in IBI is indicative of cognitive arousal [10, 11]. We also examined variability in IBI to determine if arousal levels were more variable as a result of a stressor. Intraindividual variability has been shown to be a marker of neural noise [12], and variability in electrical activity in the brain has been shown to be related to attention regulation in a cognitively demanding task [13].

The present study was designed to assess physiological measures of arousal during a simulated shooting task. The psychological data from this study are described in a previous report [14]; the current report focuses on the physiological responses during two types of feedback (shock and lifebar loss). The lifebar is a form of visual feedback designed to simulate the typical feedback given in the gaming industry, and the shock is a newer type of feedback that is meant to simulate hostile return fire. This effort hypothesized that shock feedback would induce higher levels of arousal than a lifebar loss, as indicated by measures of heart rate variability.

2 Methods

2.1 Participants

A total of 18 male current military, police, and special reaction team personnel volunteered to participate. The age range of participants was 27–48 years (M = 34, SD = 6.5). After obtaining written informed consent, participants completed a health screening form to identify issues precluding participation, such as pace makers or heart conditions. No participants were excluded. All participants were informed that they could withdraw from the study at any time without penalty; however, all completed the experiment. This study was conducted in accordance with IRB requirements (32 CFR 219 and DoDI 3216.02).

2.2 Design

The independent variable was Condition (Stress or Life Bar). The dependent measure was IBI. Baseline physiological measures were collected before the experimental sessions began. Analysis of HRV was performed using VivoNoetics VivoSenseTM software, which follows the European Society of Cardiology and the North American Society of Pacing Electrophysiology [15] procedures set for HRV standards. HRV was derived from a two-channel electrocardiogram. IBI was examined by condition, session, and in an event-related manner, examining IBI at each shock or lifebar loss. Intraindividual variability analyses were also run to provide information about the characteristics of a person’s response to stimuli over time [16].

2.3 Tasks and Stimuli

Each condition used all five screens in the Immersive Cognitive Readiness Simulator (ICoRS), a 300° immersive simulator [17] to provide 300° of visibility. Target pairs (friend/friend or friend/foe) were presented in various locations within each scene (e.g., behind a car, wall, building, natural terrain, rocks). The foe targets pointed and fired an M-9 pistol at the participant. The friend targets performed actions such as offering a soda, pulling out a wallet, or making an “I surrender” gesture. Target pairs were presented at the same time on the same screen for 2 s. Participants were instructed to only shoot at the foe targets. Based on subject matter expert input, the 2 s presentation of targets was used to induce a hasty decision. The interval between target pair displays varied between 2, 4, and 6 s. This inter-trial interval was used to minimize a pattern effect.

Based on shooting performance, up to 64 friend/friend and 64 friend/foe target pairs could have been presented. In the Shock and Life Bar conditions, a shock or lifebar loss occurred when a foe was not hit and a minimum of 30 s had passed since the last shock or change in the lifebar status. Therefore, participants could only experience up to 15 feedback events in each condition. Participants were told that even if they shot a foe, it was possible that they might receive a small shock or lose a lifebar because the target had the potential to fire while falling to the ground.

Before target presentation, an indicator sound was activated from the screen on which targets would appear, functioning as a virtual partner, and indicating where the participant should focus his attention. During the Shock condition, participants received a small shock if they missed a foe target. Similarly, in the Life Bar condition, this error caused a lifebar to turn from green to red. A scenario ended when the last (15th) shock was administered or the last lifebar turned red. Each shock and lifebar loss was considered to wound rather than inflict a lethal hit on the participant until the 15th and final feedback presentation. The end of each scenario was indicated by a message presented on the center screen, “the scenario has ended.”

2.4 Procedures

For full details of instrumentation and procedures, see [14]. All experimentation was conducted in ARL’s Cognitive Assessment and Simulation and Engineering Facility (CASEL) ICoRS [17]. The ThreatFireTM [18] safe return fire system was used to in the Shock condition to induce arousal. The ThreatFireTM system is a wearable belt that uses a rechargeable battery pack to deliver a 50- micro-amp, 200-ms electric shock to simulate the pain of hostile return fire. A modified M4 carbine rifle, fitted with a laser in the barrel and a magazine specially designed to hold CO2 was used to simulate a similar amount of recoil as one would experience when shooting live rounds on a training range. The LifeShirtTM was used to collect ECG signals during all aspects of the experiment. The shirt is a lightweight (8 oz.), machine-washable shirt with embedded sensors connected to electro pads placed on the wearer’s body. Respiratory function sensors are woven into the shirt and provide measurements of heart rate, heart rate variability (IBI ECG signals), as well as a number of other physiological measures. The sensors cycle at 200 Hz every second.

To begin, the participant was asked to stand quietly for 10 min to collect Baseline physiological measures. Next, the participant completed training in the simulator to become familiarized with the equipment and task and to allow aiming adjustment techniques. Each participant completed both conditions (Shock and Life Bar), and was informed of each condition just before it began. Following each condition, participants sat quietly filling out questionnaires; these times made up the Post-Shock and Post-Life Bar sessions.

For the Shock condition, the researcher placed the ThreatFireTM belt on the participant at the waistline. The order of conditions was randomized for each participant.

2.5 Data Analysis

The independent variable was Condition, examining Shock and Life Bar. The dependent variable was IBI. Repeated measures analyses of variance (ANOVAs) were run for all analyses. When assumptions were violated in the analyses of variance, a Greenhouse-Geisser correction was reported.

We compared IBI in the Shock and Life Bar conditions by running a one-way repeated measures ANOVA. We hypothesized that there would be an effect of Condition, with Shock causing reduced IBI.

We investigated recovery from arousal induced in each condition by running a repeated measures ANOVA. We examined Baseline, Shock, Life Bar, and Post-Shock and Post-Life Bar, referred to as Sessions. Paired t-tests were performed to examine if IBI levels returned to Baseline in each Post-Condition session. We hypothesized that IBI would return to near Baseline levels following the two Conditions, in both the Post-Shock and Post-Life Bar sessions.

To investigate if IBI was different in response to the two feedback types (shock and lifebar loss) in the two Conditions, we performed a repeated measures ANOVA on event-related IBI data, looking at the effects of Condition and Event. We hypothesized that IBI would differ not only by Condition, but also by Event, such that the IBI response would change over the course of feedback events.

To determine whether there were differences in, intraindividual variability (IIV) between the two Conditions, we calculated the coefficient of variation. Coefficient of variation is a measure of an individual’s variability that accounts for their mean [16]. A one-way repeated measures ANOVA was performed, and we hypothesized that there would be more IIV in the Shock than in the Life Bar condition.

3 Results

IBI was analyzed at both the scenario level and in an event-related manner. At the scenario level, we hypothesized that there would be lower IBI, indicative of more arousal and cognitive engagement in the Stress than the Life Bar condition. Results showed a significant effect, F(1,14) = 23.94, p < .001, such that IBI was lower in the Shock than in the Life Bar condition (Fig. 1a).

Because physiology was continuously measured, we were also able to examine IBI both during and Post-Scenarios, with the hypothesis that IBI would return to near Baseline levels after a scenario, indicating recovery. A repeated measures ANOVA was run on Baseline, Shock, Post-Shock, Life Bar and Post-Life Bar sessions, and results showed a significant main effect, F(2.25,29.30) = 43.53, p < .001 (Fig. 1b). Because we were interested in the IBI levels returning to near Baseline after the scenarios, paired t-tests post hoc analyses were run, and showed that Baseline significantly differed from the Shock (t(14) = 7.50, p < .001) and Life Bar (t(14) = 5.69, p < .001) sessions, but did not differ from Post-Shock (t(13) = −0.59, p = .568) or Post-Life Bar (t(14) = −1.18, p = .259) sessions. Additionally, IBI in the Shock and Life Bar sessions significantly differed from IBI Post-sessions (Shock: t(13) = −9.57, p < .001, Life Bar: t(14) = −10.65, p < .001). As can be seen in Fig. 1b, IBI increased in the two Post- sessions, returning to near Baseline levels, which supported our hypothesis of recovery after each scenario.

In the event-related analysis, we were interested in whether the physiological response, IBI, differed between the Shock and Life Bar conditions, and furthermore, if the relationship between the two conditions was sustained across the 15 feedback events during each session. Therefore, we ran a repeated measures ANOVA, which showed main effects of Condition (F(1,9) = 18.04, p = .002) and Event (F(3.07,27.67) = 7.74, p = .001). Visual inspection of the data (Fig. 2) suggests that IBI was lower in response to shocks than lifebar losses throughout most of the sessions.

IIV was measured using the coefficient of variation. Coefficient of variation is a measure of an individual’s variability that accounts for their mean [16]. A repeated measures one-way ANOVA was run to compare IBI in the Shock and Life Bar conditions. As can be seen in Fig. 3, there was significantly more IIV in the Shock than in the Life Bar condition, F(1,14) = 10.86, p = .005.

Fig. 1.
figure 1

a. Interbeat interval (mean ± SEM) examining the effect of Condition. b. Interbeat interval (mean ± SEM) examining recovery to Baseline Post-Shock and Post-Life Bar.

Full size image
Fig. 2.
figure 2

Interbeat interval (mean ± SEM) in the Shock and Life Bar scenarios by feedback event.

Full size image
Fig. 3.
figure 3

Coefficient of variation (mean ± SEM) in the Shock and Life Bar scenarios.

Full size image

4 Discussion, Limitations, and Future Directions

In the present study, we examined different analyses of HRV to assess arousal in a stressful simulated environment. Immersive environments have been shown to produce physiological responses similar to a real word environment [19, 20], however, this study was different because it used a 300-degree immersive environment versus a head mounted display, a standard desktop computer environment or gaming system. The experiment required full body participation, holding a modified real weapon, moving around the environment, and reacting to threat of return hostile fire via shock. The introduction of stressor threats to simulated environments may more closely represent the real operational environment military personnel encounter.

This report focused on physiological data collected during decision making in two conditions with different types of feedback (shock and lifebar loss). The results showed that the Shock condition resulted in higher arousal than the Life Bar condition. These results are consistent with previous literature, showing that a stressor causes physiological arousal, as seen through heart rate variability measures [21]. Analyses also showed that IBI returned to near Baseline levels after both conditions. There are two possible explanations for this. First, there was recovery from the stress of the task, which may indicate resiliency [22]. Second, this change from during to Post-scenario may be reflective of arousal related to what is referred to as breaks in presence (BIP) [23]. More research is needed in this area to replicate and better understand the finding.

We hypothesized that the Shock condition would cause reduced IBI compared to the Life Bar condition. Consistent with the literature, the IBI data reported here show that there was less heart rate variability during the Shock compared to the Life Bar condition [24, 25]. Thompson et al. reported similar findings, such that HRV was reduced during a tactical pistol-firing event [26]. To determine whether there were differences in IIV between the two Conditions, we calculated the coefficient of variation. The intraindividual variability analyses showed that IBI was more variable in the Shock condition than in the Life Bar condition. Therefore, though the Shock condition showed reduced IBI, individuals showed more variability within it, which may reflect an attempt to regulate the arousal response, or more engagement in the task, as has been suggested previously from electroencephalography data [13]. Future studies should attempt to replicate these findings.

We were additionally interested in how HRV changed in response to individual feedback events. We hypothesized that IBI would differ not only by Condition, but also by Event, such that the IBI response would change over the course of feedback events [27]. The event-related analysis showed that the Shock feedback reduced IBI more than the Life Bar feedback.

The present paper discussed the results of two different ways of examining IBI, session-level and event-related, as well as IIV in response to different types of feedback, in order to assess stress and arousal in a simulated environment. This study showed that shock is an effective method of inducing higher stress and arousal than typical feedback, like a lifebar. However, because simulated environments rarely mimic the actual stress and the dynamic cognitive demands of a battlefield, attempts should be made to collect these measures during a live high stress training environment for comparison of stress and arousal levels.

The present results show the importance of objectively measuring physiology to examine the effects of heightened arousal during Soldier-relevant tasks in a simulated environment. Examining the extent to which Soldiers experience arousal, which can often be a proxy for stress, we can know how immersive or stressful an environment is, and therefore its potential effectiveness as a realistic pre-deployment training environment. Future studies should attempt to identify performance metrics of Soldier-relevant tasks and examine how they relate to physiological measures.