We conducted experimental sessions between June and July 2020. Before we started collecting the data, we pre-registered the research protocol on 5/19/2020 on the website https://aspredicted.org/ with the number #41329. Before participation, subjects were informed that they were participating in a VR helicopter study investigating the prevention of VIMS. However, we did not inform them about the type of prevention investigated in the study until the very end of the experiment. As compensation, the subjects were offered the opportunity to participate in a raffle (three 20 Euro vouchers for different shops) or to receive course credit. They were also informed that participation was voluntary and that they could decide to discontinue the study at any time without giving a reason and without consequences. As an additional precaution, we chose a rating above 15 on the Fast Motion Sickness Scale (FMS) as a cut-off value to abort the experiment. All subjects included in the study signed an informed consent. The research protocol was approved by the institutional ethics board (HSD Hochschule Döpfer University of Applied Sciences) and was conducted in accordance with the declaration of Helsinki.
We assigned the subjects into three groups (control, peppermint, or ginger group) using a stratified randomization approach. During the assignment process, we stratified for gender and age as these may have an influence on VIMS (see, e.g., Keshavarz et al. 2018; Shafer et al. 2017).
In a between-subjects design, subjects completed a 15-min VR helicopter flight in the position of a crewmember, either without chewing gum, chewing a peppermint-flavored gum, or chewing a ginger-flavored gum. For some analyses, we also included time (pre–post) or the whole time course as within-subjects factor to examine the interaction between time and group. We assessed VIMS as our dependent variable with self-report measures.
All questionnaires were created with a survey tool (LimeSurvey) and filled in on a tablet in an offline version. VIMS was measured twice, before and after exposure, using the Simulator Sickness Questionnaire (SSQ) (Kennedy et al. 1993), and every minute during simulation using the FMS (Keshavarz and Hecht 2011). The SSQ contains 16 symptoms (i.e., headache, nausea, and eyestrain) rated on 4-point Likert scales with the choice of selecting none (0), slight (1), moderate (2), or severe (3). Values were weighted and summed for the total score and the subscales nausea, oculomotor distress, and disorientation according to the instructions of Kennedy et al. (1993). A study by Bouchard et al. (2007) found Cronbach's alpha to be 0.87. We applied the SSQ before and after VR exposure to ensure that the groups did not differ in baseline scores.
Additionally, we used the FMS (Keshavarz and Hecht 2011) as a single-item scale to continuously monitor VIMS symptoms every minute of VR exposure. The scale ranges from 0 (no sickness at all) to 20 (frank nausea). Peak FMS scores had been highly correlated with the SSQ subscales nausea (r = 0.83), disorientation (r = 0.80), oculomotor (r = 0.61), and the total score (r = 0.79) (Keshavarz and Hecht 2011). We used the FMS in addition to the SSQ, because it provides a broader range of response options and the ability to assess VIMS during exposure. According to the FMS instructions, we asked subjects to focus on general discomfort, nausea, and stomach discomfort, and to ignore other feelings such as excitement, fatigue, boredom, and nervousness.
In addition, we used the Motion Sickness Susceptibility Questionnaire (MSSQ) to ensure that the groups did not differ at baseline in terms of participant’s individual susceptibility to motion sickness (Golding 2006). The short form of the MSSQ, which was applied in our study, asks for the previous sickness occurrences in cars, buses, trains, aircrafts, small boats, large ships, swings, carousels in playgrounds, and leisure park attractions. Subjects can rate their experiences by selecting from not applicable/never traveled (coded with t), never felt sick (0), rarely felt sick (1), sometimes felt sick (2), and frequently felt sick (3). It asks separately for childhood experiences before the age of 12 and the experiences over the last 10 years. Calculation of the total scores followed the instructions of Golding (2006). In a validation study of the MSSQ, predictive validity for motion sickness showed a median of r = 0.51. Cronbach's alpha was 0.87 and the test–retest reliability was around r = 0.90 (Golding 2006).
The experimental groups were asked additional questions about the taste and duration of the chewing gum flavor. A custom bipolar item was used to ask how pleasant the taste of the chewing gum was perceived, ranging from very unpleasant (1) to very pleasant (6). Subjects were instructed to pay attention only to the taste of the chewing gum and not to its consistency. Furthermore, we asked the subjects how long they perceived the taste of the chewing gum to last during the simulation (not at all, only at the beginning of the simulation, until the middle of the simulation, close to the end of the simulation).
In addition to the aforementioned questionnaires, after the experiment, we also collected questionnaire data for another research project and administered the Multidimensional Assessment of Interoceptive Awareness (MAIA) (Mehling et al. 2012), Somatic-Symptom-Scale 8 (SSS-8) (Gierk et al. 2014; German version: Löwe and Voigt 2015), Measure of technology commitment (Neyer et al. 2012), and Igroup Presence Questionnaire (IPQ) (Schubert 2003).
According to a-priori power analysis with G-Power version 18.104.22.168 (Faul et al. 2009), a sample size of n = 74 would be sufficient (with alpha = 0.05, power = 0.80) to detect effect sizes (ηp2 = 0.099 corresponds to Cohen’s f = 0.33) similar to those reported by Keshavarz et al. (2015) for the interaction of time and odor, using the FMS (Keshavarz and Hecht 2011). Anticipating some dropouts and early aborts, we recruited 90 participants using an email list of the HSD Hochschule Döpfer University of Applied Sciences and social media. The participants were assigned to one of the three groups (control, peppermint, and ginger).
Exclusion criteria were known health issues like damages of the vestibular organs as well as diseases of the eyes that restrict vision and cannot be corrected-to-normal vision (e.g., through glasses or contact lenses). A necessary precondition was a normal or a corrected to normal vision, which was tested beforehand with an EN ISO 8596/7 vision chart. With regard to the chewing gum and its ingredients, we screened for fructose and/or sorbitol intolerance. At the time of the study or before, no motion sickness medications should have been consumed. In addition, extreme fear of heights was an exclusion criterion, as we used a helicopter simulation to induce VIMS.
Apparatus and stimuli
During the experiment, the subjects were seated on a stationary chair without armrests, had a presenter remote control in their hands, and wore a VR-HMD via which the virtual helicopter flight was displayed. The simulation was implemented using VBS 3 (Bohemia Interactive Simulations, n.d.), an environment for generating virtual 3D trainings for emergency personnel. The PC we used contained an Xeon CPU E5-1620 0 (3.60 GHz) processor (Intel, Santa Clara, United States), 32 GB (DDR3) of RAM and a GeForce GTX 1080 graphics card (with 8 GB GDDR5X memory) (Nvidia, Santa Clara, United States). The operating system was Windows 10 Pro (version 10.0.18363). As VR-HMD, we used the Vive Cosmos (HTC, Taoyuan), which offers a resolution of 1440 × 1700 pixels per eye with a 90 Hz refresh-rate, a diagonal field of view of 110° (HTC, n.d.), and a mechanism for adjusting the interpupillary distance (IPD). Due to the flip-up visor, the Vive Cosmos can be worn with glasses. Helicopter sound was delivered via integrated on-ear headphones and the volume was set to 70 in the windows settings during the VR exposure.
We created a simulation inspired by a search and rescue training for helicopter crews, where the crew scans the landscape for injured or missing people. The subject was flown in a helicopter using the autopilot mode along a fixed route around the coast of a peninsula. Waypoints were set to implement the route, so that the helicopter flew the same route along the waypoints for each subject. To complete a visual search task, they looked out of the right rear door of the helicopter and scanned the landscape for signs with Landolt rings on them. The task was to press a "yes" button on a presentation remote control when they recognized a Landolt ring with an opening to the top in a set of 14 Landolt rings, and to press a "no" button when they did not (see Fig. 2).
We used this task to ensure that subjects were looking out of the helicopter and not into it, which would generate no or insufficient VIMS. The entire simulation contained 30 trials, each with 14 Landolt ring signs embedded in the environment, which were visible for 6 s. Subjects completed ten consecutive trials and, after a break, the next ten. During the break, the helicopter continued to fly through the landscape and the subjects did not know when the next set of Landolt rings would appear (see Fig. 2). To enable the subjects to recognize the Landolt rings, the helicopter flew slower during the trials and faster during the breaks. This resulted in a mixture of smooth and shaky movements during the flight. We conducted a preliminary study with seven subjects (Mage = 28.43 years; SD = 3.64) to test whether the simulation evoked sufficient VIMS. We found that this was the case with an FMS mean peak score of 6.29 (SD = 5.02).
Selected chewing gums
To select chewing gums for the present study, we compared three chewing gums in a pre-test with 11 subjects (Mage = 26.89 years; SD = 3.45): a supposedly neutral mastic chewing gum, a peppermint chewing gum, and a ginger chewing gum. The subjects rated the mastic gum as unpleasant, not neutral in taste, and much tougher in consistency than the other chewing gums. For this reason, and also because commercial chewing gums are usually flavored, we decided to include only a peppermint and a ginger chewing gum of the brand Simply Gum (New York City, United States) in our study. These chewing gums are plastic-free, biodegradable, without synthetic content or added sweetener (Simply Gum, n.d.). According to the package information, the chewing gums contained real peppermint and ginger essential oils, respectively. The amount of peppermint and ginger oil could not be determined from the website or the packaging. For the study, the chewing gums were removed from the original packaging and repackaged in neutral brown packets, so that the type of gum was not identifiable. For identification, the packages were marked with a code. Our pre-test confirmed that the chewing gums were similar in terms of consistency and volume.
Prior to the experiment, subjects were informed about the procedure and about the possibility to terminate the study at any time without consequences. All subjects signed a written informed consent, successfully passed a vision test, and completed a demographic questionnaire, the pre-SSQ, and the MSSQ. Subsequently, the experimenter explained the tasks to be performed during VR exposure, presented the FMS in written form, and then handed out the chewing gum. Subjects were asked to place it in their mouth and chew it throughout the simulation. Then, they donned the VR-HMD and began the virtual helicopter flight in the position of a crewmember. Right at the beginning, they were told that they could adjust the IPD using the wheel on the side of the VR-HMD if the image was not sharp. During the 15-min virtual helicopter flight, subjects were exposed to sickness-inducing visual motion and completed the visual search task (Landolt rings). Meanwhile, they were asked to verbally rate their sickness every minute, using the single FMS item. The experimenter visually checked whether subjects chewed the gum and reminded them to do so when necessary. They then completed the post-SSQ and the questions on how pleasant and how long-lasting they perceived the taste. Finally, subjects were debriefed and left the laboratory once symptoms experienced during the experiment had subsided. Note that you can find information on COVID-19 precautions in the supplementary materials.
Statistical analysis and design
All statistical analyses were performed using SPSS (version 25) or JASP (version 0.13.1, 0.14). The a-priori significance level was set to p < 0.05. Although most of our data were not normally distributed (Shapiro–Wilk. p < 0.05), we chose parametric over nonparametric tests, because ANOVAs were shown to be relatively robust against violations of the normality assumption (see, e.g., Blanca et al. 2017). We conducted a mixed 2 × 3 ANOVA including the within-subject-factors time (pre–post) and the between-subjects-factor group (peppermint gum, ginger gum, and no gum) for the SSQ data. For the FMS data, we calculated a between-subjects-ANOVA with the factor group using FMS peak scores, i.e., the highest score a subject reported during VR exposure. Additionally, we performed a mixed 16 × 3 ANOVA to analyze the interaction of FMS time course and group. It included all 16 FMS scores (within-subjects factor time course), and group as between-subjects factor. For post hoc analyses, we used Helmert contrasts, comparing the control group against the two chewing gum groups and the peppermint and ginger groups against each other. Finally, we performed a one-tailed correlation analysis for pleasant taste and VIMS (SSQ, FMS peak scores) using the Spearman Brown Formula, since the collected data are not normally distributed.
After collecting all data, we excluded six subjects from the data set because of extremely high SSQ total pre-scores, which were identified as outliers in an SPSS boxplot analysis (1.5 interquartile ranges from median, which were SSQ scores of 44.88 or higher). After the experiment, the subjects were asked to comment on their experience and to state whether they had followed all instructions. We noted all reasons for non-compliance. Subsequently, two independent reviewers (without access to the data) judged the severity of the bias and decided which subjects were to be excluded from the analysis. Here, three subjects admitted to have made false statements regarding pre- and post-SSQ, two subjects had complaints due to excessive heat generated by the VR-HMD, one subject indicated reactance and a resistance to comply because of the military setting of the VR simulation, and one subject did not chew the gum. Thus, these seven subjects were excluded from the data analyses. According to our a-priori power analysis, a sample size of 74 would be sufficient. Thus, with 77 subjects remaining in our sample, the power is still adequate.