A Vestibular Training to Reduce Dizziness

Featured Application: Feasibility study of vestibular training to reduce dizziness in healthy participants. Abstract: Many situations can induce dizziness in healthy participants, be it when riding a carrousel or when making head movements while wearing a head-mounted display. Everybody—maybe with the exception of vestibular loss patients—is prone to dizziness, albeit to widely varying degrees. Some people get dizzy after a single rotation around the body axis, while others can perform multiple pirouettes without the slightest symptoms. We have developed a form of vestibular habituation training with the purpose of reducing proneness to dizziness. The training consists of a short (8 min) exercise routine which is moderate enough that it can easily be integrated into a daily routine. Twenty volunteers performed the training over the course of two weeks. We measured subjective dizziness before and after each daily session. We also performed several vestibular tests before (pre-test) and after (post-test) the two-week training period. They included exposure to a rotating and pitching visual environment while standing upright, as well as a physical rotation that was abruptly stopped. The results show that the dizziness induced during a given daily session decreased over the course of the two weeks. The dizziness induced by the rotating visual stimulus was significantly less after completion of the training period compared with the initial pre-test. Also, postural stability and post-rotatory spinning sensations had improved when comparing the post-test with the pre-test. We conclude that a short regular vestibular training can significantly improve proneness to dizziness.


Introduction
Dizziness is well researched in patients who suffer from one vestibular disorder or another or who are severely challenged in their daily activities, even if the exact cause for their dizziness has not been found.The dizzy patient (for an overview of a clinical assessment of dizziness, see Goebel [1]) who suffers from vertigo, frequently resulting in motion sickness during everyday activities, often benefits from habituation training, which does not pose a challenge to a healthy person.For instance, regular training which involves walking across a series of pillows or standing on one leg has been shown to be of benefit [2].Likewise, vertigo could be improved by rehabilitation training which involves changes in posture, turning the head, or watching a moving target while standing upright [3].Several diseases are associated with dizziness, such as Ménière's disease or multiple sclerosis.The former is a vestibular disorder which causes severe dizziness and requires extensive balance training (for a review, see Hansson [4]).For the latter, a variety of training protocols have been explored to improve the suffering of multiple sclerosis patients, often with success.Among them were resistive exercises, standing and reaching tasks, or active console games involving the Wii balance board (for a review, see Gunn et al. [5]).It is common for all of these rehabilitation measures that they seek to improve the symptoms in more or less severely handicapped patients and reduce dizziness encountered in everyday situations.
They do so by training the balance-related aspects of everyday actions, but they would typically not pose much of a challenge to a healthy person.
We are concerned with dizziness in healthy people, which may arise in situations which are particularly challenging.This is becoming more and more relevant, as potentially dizzying visual or vestibular stimuli play an increasingly important role in our daily lives.For instance, see-through displays are becoming more popular, be it in the shape of headup displays in automobiles or head-mounted goggles which augment our visual world with virtual information.The augmented or extended reality provided by such modern devices may compete or even be in conflict with our normal viewing experience, which can produce dizziness and ultimately symptoms of nausea in healthy users [6].Dizziness also arises when we perform repeated body rotations, such as when riding a carousel or when performing pirouettes while dancing or ice skating.People who easily suffer from dizziness in such situations usually avoid challenging rides in amusement parks or other such situations and never have to experience the negative consequences.It is thus not surprising that dizziness in healthy people has rarely been explored.
Studies addressing the issue of dizziness in healthy observers can be found in the context of professional sports, where the goal is to improve balance and postural control.For instance, there is evidence that a visual feedback method informing a person about slight shifts in their center of gravity while standing still can improve their postural control [7].In this context, studies have explored techniques to avoid dizziness, such as the spotting technique used by professional ballet dancers to maintain their spatial orientation during pirouetting [8].However, dizziness has not been intentionally induced in prior studies.
In the present study, we have done so, and we investigated whether people can adapt to complex whole-body rotations in order to reduce or eliminate dizziness.To this end, we designed a type of vestibular training for healthy observers.If such an adaptation is feasible, it should be possible to design training to reduce the general susceptibility to dizziness.In other words, if vestibular training leads to adaptation, then it should in principle be transferable to other contexts and reduce dizziness symptoms in general.
We designed and evaluated a form of vestibular training involving complex body rotations.We were able to build upon a pilot study previously conducted in our laboratory.It involved full-body rotations which were inserted into a long-distance run performed on a standard 400 m running track [9].The sports students who volunteered for the study were asked to incorporate at least 12 special rounds into their track training for a period of three weeks.They trained twice a week and performed five smooth rotations, alternating between clockwise and counter-clockwise each time they navigated a corner and amounting to 240 rotations per training day.They were asked not to stop but to keep running while spinning around at each of the four corners of the track.Before and after the three-week period, subjective dizziness after being spun in a rotating chair, dynamic visual acuity, and postural stability were tested.We found that this training was able to improve subjective dizziness in some participants.
In the present study, we sought to expand on this finding.We designed a type of vestibular training which addressed all three possible rotation planes.Since vestibular adaptation tends to be highly specific to the rotation plane addressed in training and does not easily generalize to other planes [10], we decided to address all three rotation planes in our vestibular training.To achieve this, we asked the participants to assume different postures while spinning.The participants rotated quickly around their vertical body axes while in an upright posture, and they also spun in different postures, which they assumed by bending their bodies in terms of pitch and roll and which they maintained while rotating around the yaw axis.
We hypothesized that this complex full-body rotation training would have a positive effect on subjective dizziness caused by fast physical rotations even if they deviated from those performed during the training.After all, the training would only be worth its while when generalizing to some extent.We also hypothesized that an equivalent visual stimulation, which is normally dizziness-inducing, would also appear less provocative after the vestibular training.Such equivalence of visual and vestibular stimulation has, for instance, been investigated by Bijveld et al. [11].For this purpose, we used a head-mounted display (HMD), which can simulate a versatile rotating drum.We measured, among other factors, postural stability after passive visual rotation and post-rotatory sensations, which should improve after the training.To give the complex adaptation a good chance to succeed, we decided to ask for 14 short daily training sessions rather than longer, less frequent exposures.

Participants
A total of 21 participants of normal susceptibility to dizziness were recruited.On a self-assessment scale ranging from 0 to 10 (0 = "not at all"; 10 = "extremely"), their average rating of vulnerability to dizziness in everyday life was 2.63 (SD = 2.56).All of them completed the pre-test and the training sessions.For the post-test, one participant was excluded because several weeks had passed between the last training session and the second test, leaving 20 participants (17 female, 3 male) with a mean age of 25.9 years (SD = 7.14) to be included in the analysis.The participants received partial course credit for their participation.Before the testing, they received information about the procedure and gave written informed consent regarding participation and data anonymization.This study was conducted in accordance with the Declaration of Helsinki [12].

Design
To test the effects of vestibular training, we utilized a pre-post design.Before and after self-administered training, we measured the participants' susceptibility to dizziness in a laboratory, subjecting them to visual and vestibular stimulation and recording their ability to balance.Between these laboratory sessions, the participants were asked to perform 14 daily vestibular training sessions, ideally within two weeks.Two thirds of them managed to complete the training period within the intended two weeks and completed the post-test exactly 14 days after the pre-test.The other participants spread the training sessions across a maximum of three weeks.The latest post-test was conducted 21 days after the pre-test.In the following, we first describe the different tasks performed in the laboratory sessions and then the structure of the vestibular training.

Visual Stimulation
The participants stood on a force plate (Nintendo Wii 2 balance board) which recorded their center of pressure (CoP) over time (see Figure 1).Cumulative CoP displacement over time will be referred to as the path length.The participants wore a head-mounted display (HMD; HTC Vive Pro 2 with a horizontal field-of-view of approximately 117 • (vertical approximately 96 • ) and a resolution of 2.448 × 2.448 pixel per eye) through which the visual stimuli were presented.The stimuli provided uncoupled visual motion; that is, the visual world rotated and swayed independently of the participants' actual body movements [13].The participants underwent two one-minute trials, each featuring a different visual stimulus.The HMD showed a black background against which white dots rotated clockwise, as if in a rotating drum.In the first condition, the dots rotated at a speed of 120 • per second, in the second condition, the dots additionally moved in a pitch motion of 20 • /s upward and downward.The pitch excursion ranged from + 15 • to −15 • .The condition without pitch was always presented first.Both conditions were presented for 60 s.The participants were instructed to move as little as possible during the stimulation and leave their eyes open while wearing the HMD, involuntary blinks notwithstanding.After a break of about 2-3 minutes, the participants were exposed to the condition with added pitch.Before each condition, after 30 s of stimulation and immediately after each condition, the participants provided verbal ratings for dizziness (using the FMS-D scale [6]).
The two dependent variables of the visual stimulation test were thus the path length of the CoP and FMS-D.
conditions were presented for 60 s.The participants were instructed to move as little as possible during the stimulation and leave their eyes open while wearing the HMD, involuntary blinks notwithstanding.After a break of about 2-3 minutes, the participants were exposed to the condition with added pitch.Before each condition, after 30 s of stimulation and immediately after each condition, the participants provided verbal ratings for dizziness (using the FMS-D scale [6]).The two dependent variables of the visual stimulation test were thus the path length of the CoP and FMS-D.

Balance Test
To test the ability to balance, we used two out of the six exercises of the Balance Error Scoring System (BESS) [14].For both exercises, we instructed the participants to balance on a foam pad (MAX XIVA, 48 × 39 × 6 cm) with eyes closed and hands on hips for 20 s.For the first exercise, the participants stood on one foot, and for the second, they assumed a standing position inspired by the Romberg test [15].They aligned their feet in a straight line, with the heel of the front foot touching the toes of the rear foot.The dependent variable was the time the participants were unable to maintain proper balance.The two experimenters independently measured for how many seconds the participant deviated from the intended position, either by stumbling, opening the eyes, releasing the hands from the hips, lifting one foot off the foam pad, or by strongly swaying laterally to keep their balance.

Vestibular Stimulation
The participants sat on a rotating chair with their eyes closed (see Figure 1).They lightly squeezed a tennis ball between their chin and chest, leading to a head pitch of approximately 45-60°, depending on the individual physiology.A second experimenter rotated the chair in a clockwise direction as evenly as possible for 30 s, aiming for an angular velocity of 120° per second.To facilitate an accurate rotation speed, a metronome with a frequency of 60 beats per minute was used.After 30 s, the experimenter abruptly stopped the rotation of the chair, and the participants opened their eyes but remained in a seated position with the head pitched downward.The participants reported the exact

Balance Test
To test the ability to balance, we used two out of the six exercises of the Balance Error Scoring System (BESS) [14].For both exercises, we instructed the participants to balance on a foam pad (MAX XIVA, 48 × 39 × 6 cm) with eyes closed and hands on hips for 20 s.For the first exercise, the participants stood on one foot, and for the second, they assumed a standing position inspired by the Romberg test [15].They aligned their feet in a straight line, with the heel of the front foot touching the toes of the rear foot.The dependent variable was the time the participants were unable to maintain proper balance.The two experimenters independently measured for how many seconds the participant deviated from the intended position, either by stumbling, opening the eyes, releasing the hands from the hips, lifting one foot off the foam pad, or by strongly swaying laterally to keep their balance.

Vestibular Stimulation
The participants sat on a rotating chair with their eyes closed (see Figure 1).They lightly squeezed a tennis ball between their chin and chest, leading to a head pitch of approximately 45-60 • , depending on the individual physiology.A second experimenter rotated the chair in a clockwise direction as evenly as possible for 30 s, aiming for an angular velocity of 120 • per second.To facilitate an accurate rotation speed, a metronome with a frequency of 60 beats per minute was used.After 30 s, the experimenter abruptly stopped the rotation of the chair, and the participants opened their eyes but remained in a seated position with the head pitched downward.The participants reported the exact moment when the subjective post-rotatory feeling of motion in the opposite direction (counterclockwise) had ended.The time between the stoppage of rotation and the end of the post-rotatory motion impression was measured via a stopwatch by the first experimenter.Additionally, before and immediately after the rotation, the participants again rated their dizziness using the FMS-D.Dizziness during the rotation was not assessed, as the rotation only lasted 30 s.The two dependent variables were thus the FMS-D and the duration of the post-rotatory motion sensation.

Training Protocol
Each of the 14 training sessions consisted of an 8 minute "vestibular workout".The workout included eight types of full-body rotations (four postures with two rotation directions) performed for 20 s each.Each rotation interval was followed by 40 s of stretching.In the first rotation exercise, the participants kept their torsos upright and arms outstretched.This was followed by rotations with the torso bent to the left (arms relaxed alongside body), then bent to the right, and then stooped forward (arms dangling).Each type of rotation was first performed clockwise and then counterclockwise.The rotation plane of the head was therefore alternately approximately in the yaw, pitch, and roll planes.The participants were instructed to perform the rotations continuously and quickly around their vertical body axes.A metronome set the beat to one second.For the upright posture, the participant should have completed a full rotation within two beats, corresponding to an angular velocity of 180 • /s.Extensive prior piloting revealed that angular velocities of up to 180 • /s were tolerable for an untrained person in an upright posture, whereas they were somewhat too demanding in the conditions with torso tilt.Thus, we reduced the rotational speed to 120 • per second for the rotations with a tilted torso.In addition, the participants completed a short questionnaire for each training session.This included an FMS-D rating before and after training as well as 10 point rating scales on the fitness level and fatigue after training.
The participants received verbal instructions for the workout as well as a written explanation on how to perform the different rotations and stretches.Additionally, they received a workout video to guide them through the exercises, with a trainer performing the workout and a voiceover explaining the exercises.A timer indicated when to move on to the next rotation or stretch.A demonstration video of the training, as well as additional Supplementary Materials can be found online at https://osf.io/69xw7/(accessed on 3 August 2024).

Procedure
We conducted the pre-and post-tests in a laboratory room of the local department of psychology.The pre-test took approximately 40 min, and the post-test took approximately 25 min.The pre-test took longer due to filling out a consent form and reading through the instructions.Apart from that, the laboratory sessions were identical.For practical reasons, two participants were instructed in parallel, which allowed potential dizziness symptoms to subside while wating for their turn.While one participant completed a given subtask of the pre-test, the other one was waiting outside the lab room.This was a precautionary measure to ensure that the ratings of one participant were not revealed to the other.Two experimenters were present at all times.The visual stimulation was administered first, followed by the balancing task and the vestibular stimulation.At the end of the pre-test, the participants received instructions for the training.They were debriefed at the end of the post-test.For a detailed timeline of the procedure, see Figure 2. The "vestibular workouts" were supposed to be carried out daily at no specific time or place.After the last workout, the participants answered a final questionnaire to indicate age, gender, whether they felt susceptible to dizziness during everyday tasks like driving, and whether they felt any subjective changes in their susceptibility to dizziness after they completed the training.

Results
We first report the results of the measurements taken in the laboratory before starting (pre-test) and after completing (post-test) the 14 day training protocol.We then report the data collected during the individual training sessions at home.

Results
We first report the results of the measurements taken in the laboratory before starting (pre-test) and after completing (post-test) the 14 day training protocol.We then report the data collected during the individual training sessions at home.

Laboratory Sessions (Pre-Test and Post-Test)
We first report the data from the tasks which measured susceptibility to dizziness by visual stimulation and then the data for balancing and vestibular stimulation.We expected the vestibular training to reduce the dizziness induced by visual and vestibular stimulation.It should have also improved the ability to balance.Prior to the analyses and after visual inspection of the data, we excluded one participant since that person did not complete the post-test.Due to technical issues with the HMD, two participants could not be tested with the visual stimulation task for pitch and were thus excluded from this analysis.We decided to keep the participants in the data set even if they did not complete all 14 training sessions but merely skipped one day or the other.We considered the minimum number of sessions which anyone had performed, namely five sessions, to be sufficient for potential training effects and abstained from excluding them.In total, we excluded only 1 of the 21 volunteers.

Results
We first report the results of the measurements taken in the laboratory before starting (pre-test) and after completing (post-test) the 14 day training protocol.We then report the data collected during the individual training sessions at home.

Laboratory Sessions (Pre-Test and Post-Test)
We first report the data from the tasks which measured susceptibility to dizziness by visual stimulation and then the data for balancing and vestibular stimulation.We expected the vestibular training to reduce the dizziness induced by visual and vestibular stimulation.It should have also improved the ability to balance.Prior to the analyses and after visual inspection of the data, we excluded one participant since that person did not complete the post-test.Due to technical issues with the HMD, two participants could not be tested with the visual stimulation task for pitch and were thus excluded from this analysis.We decided to keep the participants in the data set even if they did not complete all 14 training sessions but merely skipped one day or the other.We considered the minimum number of sessions which anyone had performed, namely five sessions, to be sufficient for potential training effects and abstained from excluding them.In total, we excluded only 1 of the 21 volunteers.Figure 4 shows the mean FMS-D ratings before, during, and after the visual stimulation in the pre-test and post-test.We calculated a repeated-measures analysis of variance (rmANOVA) of these ratings with the factors of visual rotation (without or with pitch) × time (before, during, or after exposure) × laboratory session (pre-test or post-test), with Greenhouse-Geisser correction for the degrees of freedom where indicated.The rmANOVA conducted on the FMS-D data of the 19 participants who completed both pitch conditions showed significant main effects for the time (F(2, 36) = 37.96, p < 0.001, η p 2 = 0.68, ε = 0.76) and laboratory session (F(1, 18) = 4.53, p = 0.047, η p 2 = 0.20), specified by a significant laboratory session × time interaction (F(2, 36) = 8.74, p = 0.004, η p 2 = 0.33, ε = 0.65) in combination with a session × time × visual rotation three-way interaction (F(2, 36) = 5.23, p = 0.010, η p 2 = 0.23).The remaining effects were not significant (F ≤ 2.31, p ≥ 0.146).Overall, we found higher FMS-D ratings during and after exposure compared with the baseline measurement before exposure, which shows that our visual stimulus was adequate to induce dizziness.This increase in dizziness was more pronounced in the pre-test compared with the post-test, with some additional variation as a function of the visual rotation condition.Taken together, this indicates that our participants were less prone to visually induced dizziness after they completed the training period.

Visual Stimulation
stimulation in the pre-test and post-test.We calculated a repeated-measures analysis of variance (rmANOVA) of these ratings with the factors of visual rotation (without or with pitch) × time (before, during, or after exposure) × laboratory session (pre-test or post-test), with Greenhouse-Geisser correction for the degrees of freedom where indicated.The rmANOVA conducted on the FMS-D data of the 19 participants who completed both pitch conditions showed significant main effects for the time (F(2, 36) = 37.96, p < 0.001, ηp 2 = 0.68, ε = 0.76) and laboratory session (F(1, 18) = 4.53, p = 0.047, ηp 2 = 0.20), specified by a significant laboratory session × time interaction (F(2, 36) = 8.74, p = 0.004, ηp 2 = 0.33, ε = 0.65) in combination with a session × time × visual rotation three-way interaction (F(2, 36) = 5.23, p = 0.010, ηp 2 = 0.23).The remaining effects were not significant (F ≤ 2.31, p ≥ 0.146).Overall, we found higher FMS-D ratings during and after exposure compared with the baseline measurement before exposure, which shows that our visual stimulus was adequate to induce dizziness.This increase in dizziness was more pronounced in the pretest compared with the post-test, with some additional variation as a function of the visual rotation condition.Taken together, this indicates that our participants were less prone to visually induced dizziness after they completed the training period.

Balance Tests
The mean ratings for the one-legged stance and for the Romberg balance tests are depicted in Figure 5.Each participant's balancing performance was rated by two experimenters using the balance error scoring system (BESS).The mean interrater correlation was r = 0.89 (r_min = 0.78, r_max = 0.96).Descriptively, our data show reduced errors after the training period in both balancing tasks, with consistently more errors in

Balance Tests
The mean ratings for the one-legged stance and for the Romberg balance tests are depicted in Figure 5.Each participant's balancing performance was rated by two experimenters using the balance error scoring system (BESS).The mean interrater correlation was r = 0.89 (r_min = 0.78, r_max = 0.96).Descriptively, our data show reduced errors after the training period in both balancing tasks, with consistently more errors in the one-legged stance compared with the Romberg stance.However, in a laboratory session (pre-test and post-test) × balance test (one-legged or Romberg) rmANOVA, both effects missed significance (F(1, 19) = 3.69, p = 0.073, η p 2 = 0.16, and F(1, 19) = 4.14, p = 0.056, η p 2 = 0.18 for laboratory session and balance test, respectively).The interaction was clearly not significant (F(1, 19) = 0.49, p = 0.494, η p 2 = 0.03).

Vestibular Stimulation
The left panel of Figure 6 shows the mean FMS-D ratings before and after vestibular stimulation (chair rotation) as a function for the pre-test and post-test.In a time × laboratory session rmANOVA, the effect of the time was significant (F(1, 19) = 72.83,p < 0.001, ηp 2 = 0.79), indicating a pronounced increase in dizziness due to vestibular stimulation.Descriptively, this increase was somewhat steeper in the pre-test compared with the post-test.However, the time × laboratory session interaction did not reach significance (F(1, 19) = 3.00, p = 0.102, ηp 2 = 0.14).The main effect of the laboratory session was clearly not significant (F(1, 19) = 0.09, p = 0.773, ηp 2 = 0.14).Lastly, we took a look at the duration of post-rotatory dizziness (right panel of Figure 6).After completion of the training period (post-test), we observed a significant reduction in dizziness duration by an average of 5.25 s (SD = 5.54, t(19) = −4.24,p < 0.001, d = 0.95) compared with the pre-test.

Vestibular Stimulation
The left panel of Figure 6 shows the mean FMS-D ratings before and after vestibular stimulation (chair rotation) as a function for the pre-test and post-test.In a time × laboratory session rmANOVA, the effect of the time was significant (F(1, 19) = 72.83,p < 0.001, η p 2 = 0.79), indicating a pronounced increase in dizziness due to vestibular stimulation.Descriptively, this increase was somewhat steeper in the pre-test compared with the post-test.However, the time × laboratory session interaction did not reach significance (F(1, 19) = 3.00, p = 0.102, η p 2 = 0.14).The main effect of the laboratory session was clearly not significant (F(1, 19) = 0.09, p = 0.773, η p 2 = 0.14).Lastly, we took a look at the duration of post-rotatory dizziness (right panel of Figure 6).After completion of the training period (post-test), we observed a significant reduction in dizziness duration by an average of 5.25 s (SD = 5.54, t(19) = −4.24,p < 0.001, d = 0.95) compared with the pre-test.

Vestibular Stimulation
The left panel of Figure 6 shows the mean FMS-D ratings before and after vestibular stimulation (chair rotation) as a function for the pre-test and post-test.In a time × laboratory session rmANOVA, the effect of the time was significant (F(1, 19) = 72.83,p < 0.001, ηp 2 = 0.79), indicating a pronounced increase in dizziness due to vestibular stimulation.Descriptively, this increase was somewhat steeper in the pre-test compared with the post-test.However, the time × laboratory session interaction did not reach significance (F(1, 19) = 3.00, p = 0.102, ηp 2 = 0.14).The main effect of the laboratory session was clearly not significant (F(1, 19) = 0.09, p = 0.773, ηp 2 = 0.14).Lastly, we took a look at the duration of post-rotatory dizziness (right panel of Figure 6).After completion of the training period (post-test), we observed a significant reduction in dizziness duration by an average of 5.25 s (SD = 5.54, t(19) = −4.24,p < 0.001, d = 0.95) compared with the pre-test.

Compliance with the Training Protocol
The data obtained from the questionnaires and protocols of all 20 participants suggest rather good compliance with the instructions when performing the training, although some participants skipped individual sessions for various reasons.On average, the participants completed 11.35 (SD = 2.62) training sessions during a 14 day period, with a range of 5-15 sessions.All participants stated that they followed the instructions of the training quite closely.They reported mean scores of 8.8-9.5 on the 11 point compliance scales (0 = "did not perform as instructed at all" and 10 = "performed exactly as instructed").The scales covered the quality of the exercises and their duration and speed, respectively; that is, the participants reported if they had performed the exercises and their duration accurately and if they had been able to maintain the required angular velocities during the 20 s rotation segments throughout the entire training session.Most participants reported a certain level of relief after they finished all the training sessions.Averaged across all sessions, each of the daily vestibular training sessions increased dizziness by 5.47 points on the FMS-D scale.The average baseline at the beginning of all sessions was 0.95 (SD = 1.47), and the average at the sessions' end was 6.41 (SD = 4.3).Thus, the participants carried out the training with sufficient intensity to create dizziness.More importantly, the training sessions created less dizziness from day to day as the training progressed.In the following, we report the relevant test-statistics comparing performances and ratings before and after the training.

Individual Vestibular Training
To further investigate the improvement in dizziness as the training progressed, we calculated the difference values for the FMS-D scores before and after each daily training session.We then averaged the difference scores for the first two training days (early training stage), the last two training days (late training stage), and the intermediate training days (middle training stage).Since the participants differed with regard to the number of completed training sessions, the number of FMS-D difference scores summarized as the middle training stage varied from 1 to 11. Figure 7 shows the distribution of these difference scores as well as the group means for these three training stages.Each data point represents one participant's mean FMS-D rating (averaged across the sessions per stage).The mean difference in the FMS-D ratings before and after each session decreased from the early (M = 8.18, SD = 3.64) to the middle training stage (M = 5.64, SD = 3.79) and continued to do so from the middle to the late training stages (M = 3.90, SD = 4.25).Note, however, that 4 out of the 20 participants (20%) did not respond to the training as intended and reported higher FMS-D differences in the late straining stage compared with the early training stage.An rmANOVA with the training stage (early, middle, or late) as the independent variable and the FMS-D difference scores as the dependent variable revealed a significant effect from the training stage (F(1,19) = 13.35,p < 0.001, η p 2 = 0.41.ε = 0.65).This finding indicates that the above differences in dizziness between the pre-test and post-test can indeed be attributed to a dizziness improvement throughout the course of the training.
Lastly, we calculated the correlations between the number of completed training sessions, FMS-D difference scores per training stage, measures from the post-test questionnaire regarding general vulnerability to dizziness, the subjective improvement due to vestibular training, and the laboratory tests.They are summarized in Table 1.They clearly indicate that the more training sessions a given participant completed, the less susceptible to dizziness he or she was in the late training stage.Also, participants who described themselves as generally more susceptible to dizziness did indeed show greater susceptibility to dizziness (higher mean FMS-D difference scores) in the middle and late training stages, which could indicate that these participants might have required additional training sessions to reduce their vulnerability to dizziness more effectively.Finally, the participants' subjective improvement in vulnerability to dizziness correlated negatively with training-induced dizziness in the middle and late training stages, which suggests that the participants were able to self-assess their training success in retrospect well.
training sessions and measures from the questionnaire and in the laboratory testing.

General Discussion
Exposure to self-administered vestibular training improved dizziness in the vast majority of our participants, as indicated by both subjective and objective measures.We Table 1.Correlations between the intervals of the FMS-D differences with the amount of completed training sessions and measures from the questionnaire and in the laboratory testing.

General Discussion
Exposure to self-administered vestibular training improved dizziness in the vast majority of our participants, as indicated by both subjective and objective measures.We discuss in turn the compliance with the training plan, the quality of the measures, potential limitations of this study, as well as avenues for further experimentation.

Subjective Improvement
An average increase in the reported FMS-D of approximately 5.5 over the course of the short training session suggests that it was taxing.It also suggests that the participants did not avoid dizziness but embraced it.They reported merely small general vulnerability to dizziness in everyday life (M = 2.63 on a 0-10 scale), and thus particular susceptibility to pirouetting movements can be ruled out.Notwithstanding the initial normal or low susceptibility, the average subjective improvement in vulnerability to dizziness amounted to 3.42 points on the 0-10 scale.This is quite considerable and a solid indication that the training was perceived to be successful.
As hypothesized, the eight 20 s segments of full-body rotations did produce a pronounced dizziness response during each training session.Over the course of the two-week period, this response was significantly attenuated; that is, the stimulus was effective, and the participants adapted to this provocative stimulus.The adaptation is also reflected in the global subjective improvement judgment, which correlated negatively with the FMS-D increase in the last interval.There was a significant negative correlation between the number of training sessions and the FMS-D increase in the last interval.More training sessions led to more resilience.This correlation is even more impressive when considering the lack of variance in the amount of completed training sessions.From our data, we cannot conclude, however, if the gain in resilience will asymptote or whether it is likely to continue at a substantial rate such that it might be possible to suppress dizziness altogether at some point.The expected negative relationship between the number of completed training sessions and the reported vulnerability to dizziness in everyday life was discernible but failed to reach significance.This failure might have been produced by a self-regulation bias; that is, those participants who were most prone to dizziness and thus could have benefited most from the training may have decided to put in fewer sessions.If this was the case, then the training would have shown even larger increases in resilience if all participants had completed the full 14 training sessions.For these participants, it might have been helpful to monitor the execution of the training sessions more closely than we did in the current study.

Comparing Pre-Test and Post-Test Measurements
The dizziness response to the strong vection-inducing visual stimulation presented via HMD was attenuated after the training compared with the initial assessment.This was the case for the objective as well as for the subjective dizziness measures.The post-rotatory sensations as well as dizziness ratings improved.Additionally, postural stability while exposed to the rotating drum stimulus was improved after training.The path lengths of the CoP of the force plate on which the participants stood were reduced both for mere vertical axis rotation as well as for the added pitch condition, albeit much less so in the pitch condition.This may have been the case due to the fixed order of the stimuli.Since the no-pitch condition was always presented first, the subjects could have become somewhat accustomed to the stimulus, leading to less sway on the balance board which in turn may have left smaller room for improvement in the post-test.We observed the same pattern when comparing the FMS-D ratings right after the visual stimulation in the pre-test and post-test.Again, the subjects improved less in the pitch condition.However, due to our current experimental design, it cannot be ruled out that the training simply had less of an effect on the visual yaw rotation when it was combined with the visual pitch motion.
The two stability measures taken with the Balance Error Scoring System, the one-leg stance and the Romberg stance, descriptively point toward an improved ability to maintain balance after the training, albeit not a significant one.We also found consistently larger postural sway values for the one-legged stance compared with the two-legged stance.A plausible explanation could lie in the differential amount of vestibular involvement between these conditions.The two-legged stance on the foam pad may have commanded less vestibular control.The surface of contact between the feet and the foam pad was twice as large in the Romberg condition, and the motor feedback loop, which keeps the body balanced, can rely much more on muscles and sinew sensors and is less dependent on vestibular information.The latter presumably played a more decisive role in the onelegged condition, which called for a more rigid posture.Note also that the Romberg test is mainly used to assess vestibular dysfunction and may have been consistently stable for our healthy participants.
The largest training-related improvement we could observe was in the duration of the post-rotatory dizziness after vestibular stimulation via the rotating chair.The sudden stop of the chair after having been rotated produces a pronounced post-rotatory sensation of spinning in the opposite direction.During the pre-test, this sensation lasted for approximately 10 seconds on average.During the post-test, it lasted for an average of merely five seconds.This is a remarkable decrease.The relative similarity between the rotations practiced during the training sessions and both the pre-test and post-test could be responsible for this impressive effect.The only difference between the spinning segments during training was that the participants were standing as opposed to being seated and spinning actively versus being spun passively.It appears that the training was specific to body rotation but generalized to visually induced feelings of rotation and to postural control.Interestingly, the intensity of dizziness right after the end of the stimulus remained unaltered.Why did the duration of the post-rotatory sensation adapt, whereas the magnitude of the dizziness did not?This may be indicative of the nature of the adaptation process.The adaptation of the duration might be attributable to sensory suppression of the response, whereas the subjective degree of dizziness may be a cognitive response which is affected by expectations and attribution in addition to sensory processes.

Limitations
We have no direct experimental comparison data to predict whether other forms of physical training might have had comparable effects.Given the time-consuming nature of the training, we chose not to introduce a control group which performed comparable training containing no spinning movements.It is thus impossible to say unequivocally how large the share of the vestibular stimulation was and what might have been due to the increase in general fitness.Note, however, that superior fitness is not necessarily related to less dizziness; the opposite is more likely, at least when inferring from the negative correlation of fitness and proneness to motion sickness [16].It is also conceivable that the improvements in dizziness after our training stemmed from the experience of the pre-test.When performing the test a second time during the post-test, the participants had been familiar with the test procedures.They might have merely learned to perform the tests more accurately.However, the rather brief test periods and the large interval of two weeks in between make it highly unlikely that significant learning had taken place due to mere retesting.Moreover, the pronounced improvement in the duration of post-rotatory motion sensation indicates that the vestibular stimulus provided by the training had been essential.Finally, the decrease in FMS-D values during the course of the training itself speaks in favor of a true training effect which cannot be attributed to the experience of the pre-test alone.
The majority of the participants were female psychology students.Could there be any concern that other samples would react differently to the vestibular training?First, we sought to test whether the training worked in principle.This was the case.In the present sample, there were already a few nonresponders.We do not know if they really did not respond to the training, if they did not challenge themselves sufficiently during the training, or if they might have required a more powerful stimulus.Given how well the training worked for the current sample, it is desirable to carry out continued testing with a larger group of male participants, who are often reported to be less susceptible to motion sickness.They may also be less susceptible to vestibular training.Given that successful vestibular training will be most beneficial for people who are prone to dizziness, however, it may be worthwhile to optimize a type of training for this subgroup.

Outlook
The current study is the first proof of concept that vestibular training involving complex full-body rotations can reduce subjective dizziness as well as improve balance and the degree to which passive rotations affect the sensorimotor system.A variety of future enhancements of this training are conceivable.Essentially, there are two approaches to achieving this.First, we can induce dizziness by spinning the subject, which creates a strong vestibular stimulus.Second, we can produce a visual stimulus strong enough to produce dizziness.This can be accomplished by placing a person into a rotating drum such that the entire visual world rotates.Such optokinetic drums used often for vestibular assessment are typically upright and rotate around a vertical axis.They induce vection, a strong illusion of being rotated, and are particularly powerful when tilted such that the axis of rotation deviates from the vertical body axis [17].However, it is impossible to create complex rotations with such drums.Therefore, we used the former method of spinning the participant or rather asked them to pirouette.
One might augment the vestibular training with specific breathing instructions to further increase the efficacy of the self-administered training.Yen Pik Sang et al. [18] found that controlled breathing was able to reduce nausea, which had been induced by pitch and roll head movements while seated on a rotating chair.Note, however, that they did not measure dizziness.One could also continue to enhance the training by performing it as before during the first week of training and making the rotations more challenging during the second week.During piloting, we noticed that the rotations in the three postures could be performed faster and for longer time intervals once the routine was established.Also, personalized trainings are conceivable which maintain a level of challenge rather than fixed durations and rotational velocities.Due to our maximum required training duration of 14 days, it remains unclear whether the improvements asymptotically approached a minimum of unavoidable dizziness or whether the development of such dizziness could be completely prevented if the training were only carried out long enough.Both of these venues seem promising when further exploring anti-dizziness training protocols.Also, at this point, we do not know how long the positive effects of the training will last without being refreshed.Further studies should look at the extinction rates of dizziness relief and, in particular, the ideal time intervals for booster sessions.

Conclusions
Compared with the running routine which involved full-body rotations more or less exclusively around the vertical body axis [9], the present vestibular training was more complex and involved rotations around all three major axes.The training was successful because it not only improved post-rotatory dizziness when rotated around a vertical axis on a rotating chair, but it also improved balance when standing upright on a soft surface in a posture which presented a challenge to maintaining balance.The subjective measures of dizziness and the objective measures of balance both bore witness to successful adaptation.The former were taken on a daily basis and revealed that there was continued improvement as the training progressed.This points to the potential for even stronger effects if the training were to extend beyond two weeks.The successful adaptation suggests that this training is suitable for building resilience toward dizziness.It remains to be investigated if the training additionally reduces motion sickness and whether training protocols can be spread out over longer time intervals while maintaining their efficiency.

Figure 1 .
Figure 1.Experimental set-up, showing the visual stimulation task on the right and the vestibular stimulation task on the left.

Figure 1 .
Figure 1.Experimental set-up, showing the visual stimulation task on the right and the vestibular stimulation task on the left.

Figure 2 .
Figure 2. Timeline of the experiment.

Figure 3
Figure3shows the mean path length of the center of pressure (CoP) for visual yaw rotation with (gray line) and without (black line) pitch in the pre-test and post-test.The mean path lengths were significantly larger in the pre-test compared with the post-test in the yaw rotation both without and with pitch, as confirmed by paired-samples t-tests, where t(19) = 3.25, p = 0.004, and d = 0.73 while t(17) = 2.45, p = 0.026, and d = 0.58, respectively.Note that two participants could not be tested in the condition with pitch due to technical issues.

Figure 2 .
Figure 2. Timeline of the experiment.

Figure 3
Figure3shows the mean path length of the center of pressure (CoP) for visual yaw rotation with (gray line) and without (black line) pitch in the pre-test and post-test.The mean path lengths were significantly larger in the pre-test compared with the post-test in the yaw rotation both without and with pitch, as confirmed by paired-samples t-tests, where t(19) = 3.25, p = 0.004, and d = 0.73 while t(17) = 2.45, p = 0.026, and d = 0.58, respectively.Note that two participants could not be tested in the condition with pitch due to technical issues.

Figure 3 .
Figure 3. Mean path length of the center of pressure for visual yaw rotation with and without pitch for the pre-test and post-test.Error bars indicate ± 1 SEM of the 20 participants in the condition without pitch (black line) and the 18 participants in the condition with pitch (gray line).

Figure 4 .
Figure 4. Mean FMS-D scores obtained before, during, and after visual yaw rotation, as measured during pre-and post-test.The left panel corresponds to the pure visual yaw rotation, and the right panel corresponds to the added pitch oscillation.Dizziness ratings were obtained just before (before = 0 s), in the middle of (during = 30 s), and right after 60 s of visual stimulation (after = 60 s).Error bars indicate ± 1 SEM of the 20 participants in the condition without pitch (left panel) and the 19 participants in the condition with pitch (right panel).

Figure 4 .
Figure 4. Mean FMS-D scores obtained before, during, and after visual yaw rotation, as measured during pre-and post-test.The left panel corresponds to the pure visual yaw rotation, and the right panel corresponds to the added pitch oscillation.Dizziness ratings were obtained just before (before = 0 s), in the middle of (during = 30 s), and right after 60 s of visual stimulation (after = 60 s).Error bars indicate ± 1 SEM of the 20 participants in the condition without pitch (left panel) and the 19 participants in the condition with pitch (right panel).

Figure 5 .
Figure 5. Mean error scores of the one-legged and Romberg stances obtained with the balance error scoring system (BESS) during pre-test and post-test.Error bars indicate ± 1 SEM of the 20 participants in each condition.

Figure 6 .Figure 5 .
Figure 6.Mean FMS-D ratings for vestibular yaw rotation before and after 30 s of exposure (left panel) and mean duration of post-rotatory dizziness (right panel) at pre-test and post-test.Before =

Figure 5 .
Figure 5. Mean error scores of the one-legged and Romberg stances obtained with the balance error scoring system (BESS) during pre-test and post-test.Error bars indicate ± 1 SEM of the 20 participants in each condition.

Figure 6 .Figure 6 .
Figure 6.Mean FMS-D ratings for vestibular yaw rotation before and after 30 s of exposure (left panel) and mean duration of post-rotatory dizziness (right panel) at pre-test and post-test.Before = Figure 6.Mean FMS-D ratings for vestibular yaw rotation before and after 30 s of exposure (left panel) and mean duration of post-rotatory dizziness (right panel) at pre-test and post-test.Before = 0 s and after = 30 s of vestibular rotation exposure time.Error bars indicate ± 1 SEM of the 20 participants in each condition.

Figure 7 .
Figure 7. Individual FMS-D difference scores before and after training sessions plotted for the three stages of the 14 day training.The points indicate individual difference scores, and the line connects the averages.The early training stage comprises the first two training sessions; the middle stage comprises a varying number of intermediate training sessions; and the late stage comprises the last two trainings sessions.Error bars indicate ± 1 SEM of the 20 participants.

Figure 7 .
Figure 7. Individual FMS-D difference scores before and after training sessions plotted for the three stages of the 14 day training.The points indicate individual difference scores, and the line connects the averages.The early training stage comprises the first two training sessions; the middle stage comprises a varying number of intermediate training sessions; and the late stage comprises the last two trainings sessions.Error bars indicate ± 1 SEM of the 20 participants.

Funding:
This research received no external funding.Institutional Review Board Statement:This study was conducted in accordance with the Declaration of Helsinki.Informed Consent Statement: Informed consent was obtained from all participants involved in the study.