Predicting individual acclimation to the cross-coupled illusion for artificial gravity

Abstract

BACKGROUND:

The cross-coupled (CC) illusion and associated motion sickness limit the tolerability of fast-spin-rate centrifugation for artificial gravity implementation. Humans acclimate to the CC illusion through repeated exposure; however, substantial inter-individual differences in acclimation exist, which remain poorly understood. To address this, we investigated several potential predictors of individual acclimation to the CC illusion.

METHODS:

Eleven subjects were exposed to the CC illusion for up to 50 25-minute acclimation sessions. The metric of acclimation rate was calculated as the slope of each subject’s linear increase in spin rate across sessions. As potential predictors of acclimation rate, we gathered age, gender, demographics, and activity history, and measured subjects’ vestibular perceptual thresholds in the yaw, pitch, and roll rotation axes.

RESULTS:

We found a significant, negative correlation (p = 0.025) between subjects’ acclimation rate and roll threshold, suggesting lower thresholds yielded faster acclimation. Additionally, a leave-one-out cross-validation analysis indicated that roll thresholds are predictive of acclimation rates. Correlations between acclimation and other measures were not found but were difficult to assess within our sample.

CONCLUSIONS:

The ability to predict individual differences in CC illusion acclimation rate using roll thresholds is critical to optimizing acclimation training, improving the feasibility of fast-rotation, short-radius centrifugation for artificial gravity.

Keywords

Human spaceflight vestibular perceptual threshold incremental acclimation short-radius centrifugation physiological countermeasure

1 Introduction

1.1 Artificial gravity as a countermeasure for long duration space exploration

Artificial gravity (AG) is a promising comprehensive countermeasure for astronauts during long-duration space exploration, replicating Earth-like gravitational loading in an effort to mitigate space-flight-induced deconditioning of multiple physiological systems. However, the "Coriolis" cross-coupled (CC) illusion has historically limited the feasibility of implementing fast-rotation, short-radius centrifugation for AG due to human tolerability concerns [31, 49]. The CC illusion is a provocative tilting or tumbling sensation experienced following an out-of-plane head tilt performed in a constantly rotating field [26 , 30]. It can cause substantial disorientation - as well as motion sickness – if head tilts are repeated in the rotating environment. Of operational concern for spaceflight AG implementation, the intensity of the CC illusion is greater for faster spin rates, and most individuals first experience the CC illusion at very slow spin rates (1-3 RPM) [9]. To produce 1g loading at the rider’s feet, this would require a space-based centrifuge on the order of 100+ meters in radius, both technologically and financially infeasible in the near term. The desire for using a more practical, short-radius centrifugation approach as a future, comprehensive spaceflight countermeasure has encouraged further investigation into the potential for individuals to acclimate to (i.e., become more tolerant of) the CC illusion.

1.2 The cross-coupled illusion limits the tolerability of fast-spin-rate rotation

Previous research has shown that acclimation to the CC illusion is feasible rather quickly (e.g., in two sessions of less than 1 hour in duration) [15, 28]. We have recently extended these previous studies to investigate acclimation in subjects exposed to 10 daily 25-minute sessions [9]. In this personalized, threshold-based, incremental acclimation protocol, all subjects began spinning at 1 RPM. During each session, all 10 subjects were exposed to unique sequences of increasing spin rates, determined in real time and based upon individual subject response following a pair of head tilts performed in the rotating environment (roll head tilt 40° down and back to the upright position). If the subject reported that he/she did not feel the CC illusion on both head tilts of one head tilt pair, the spin rate increased by 1 RPM, otherwise it was maintained.

In this previous study, we found that subjects reached an average spin rate of 17.7 RPM (SD:+/- 9.1) in which no CC illusion was felt following 10 sessions of exposure. Notably, all subjects exhibited a positive acclimation rate [9]. However, while many subjects showed rapid improvement in CC illusion tolerability, others acclimated at a much slower rate. By the end of the study, the spin rate at which no illusion was felt ranged across subjects from 3–30 RPM (n = 10), demonstrating substantial inter-individual differences in ability to acclimate to the CC illusion. Notably, prior investigations have provided no insight into the source of these substantial inter-individual differences. It is of both scientific and operational interest to be able to predict how quickly subjects (or astronauts) will be able to acclimate to the CC illusion to better inform required training protocols and future centrifuge designs.

1.3 Potential predictors of CC illusion acclimation rate

Previous studies investigating various aspects of vestibular function and balance provide some guidance for potential predictors of how rapidly a subject acclimates to the CC illusion. For example, earlier investigations have suggested that vestibular performance may be affected by demographics [2, 5], age [5 , 32], and frequency of performing certain physical activities [27]. The role of gender on inter-individual vestibular differences have been mixed [13 , 45].

It has previously been proposed that an individual’s vestibular perceptual threshold may be predictive of his/her ability to adapt to an altered gravity environment [16 –19]. It is thought that central adaptation to a novel environment is driven by an attempt to reduce sensory conflict between actual and expected sensory perception (e.g. vestibular signals). In the CC illusion, sensory conflict occurs from the unexpected semicircular canal stimulation when making head tilts in a rotating environment. However, sensory noise [21 , 40] may produce sensory conflict as well, even in a normal, non-rotating, 1g environment. This hypothesis suggests that sensory conflict must be substantial and sustained to induce adaptation, in order to avoid inappropriately adapting when the environment has not changed and sensory conflict is simply due to sensory noise. Thus, sensory noise, specifically in the vestibular system, may be a limiting factor in an individual’s capacity to adapt (e.g. rate of acclimation) to novel sensory environments. An individual’s vestibular sensory noise can be estimated by measuring vestibular perceptual thresholds [40] (i.e., the smallest self-motion that can reliably be recognized). Thus, we hypothesize that individuals with a lower vestibular perceptual threshold (i.e., less vestibular sensory noise) may be able to acclimate more quickly to the CC illusion.

Fig. 1

Human Eccentric Rotator Device (HERD) located in the Bioastronautics Laboratory at CU Boulder. HERD was used for all experiments outlined in the present study, either in the upright configuration (A) or the supine configuration (B).

2 Methods

We tested several potential predictors of an individual’s ability to acclimate to the CC illusion in an effort to explain the large degree of inter-individual differences present in previous findings. Data for the current study were collected from one group of subjects in two phases. First, we gathered demographic information, activities history, and tested subjects to determine their vestibular perceptual threshold in three rotational axes: yaw, pitch, and roll. Next, we quantified each subject’s ability to acclimate to the CC illusion using an extended, personalized, incremental acclimation protocol that exposed subjects to the CC illusion for up to 50 daily acclimation sessions, each with a duration of 25-minutes. Upon conclusion of both studies, data were analyzed to identify potential correlations between the prediction metrics collected and the rate at which subjects acclimated to the CC illusion. Finally, a cross-validation analysis was performed to quantify the ability to predict a new individual’s acclimation rate.

2.1 Subjects

Eleven subjects (10M/1F) volunteered to participate in this investigation. Subjects had an average age of 22.2 years old (range: 20-25), and none of the subjects reported a history of vestibular dysfunction. As this was part of a larger study [8], subjects were not specifically recruited based upon age, demographics, activity levels, or other attributes (see Discussion for limitations). Additionally, while subjects were not included or excluded based on susceptibility to motion sickness, all subjects completed the Motion Sickness Susceptibility Questionnaire (MSSQ) [43, 44] and, on average, scored in the 43rd percentile (SD: +/- 34.6). Subjects signed a written informed consent, and all protocols were approved by the University of Colorado Institutional Review Board.

2.2 Equipment

Experiments were conducted at the University of Colorado Boulder in the Bioastronautics Laboratory. The Human Eccentric Rotator Device (HERD) (Gyrostim, Colorado Springs, CO, USA) was used to evaluate subjects’ vestibular perceptual thresholds and each individual’s rate at which they acclimated to the CC illusion. Vestibular perceptual thresholds were tested with the subject configured on a supine/recumbent bed (Fig. 1B, pitch and roll rotation thresholds) and upright chair (Fig. 1A, yaw rotation thresholds) configurations, while acclimation was tested using only the upright chair configuration. In all experiments, subjects were spun in a dark room with no visual cues, and they were monitored at all times using infrared cameras. Two-way, wireless communication between the subjects and operators was provided, and subjects were also given wireless push buttons for redundant reporting of motion direction (e.g., left vs. right) and presence or absence of the CC illusion for the threshold and acclimation studies, respectively. During the vestibular perceptual threshold experiments, the subject’s head was comfortably secured in a headrest to couple HERD rotation to head rotation/vestibular stimulation. In the acclimation investigation, foam blocks were used to ensure consistency in head tilt angles between subjects and across acclimation sessions. Finally, white noise was introduced into the subjects’ headphones (mixed with operator communication) to help mask HERD motor noise. Between motion profiles in the threshold experiment, the white noise was turned off, such that the presence of white noise indicated when a motion was occurring.

2.3 Questionnaires

Three questionnaires were administered to all subjects. The first contained questions regarding demographic information; we asked subjects to describe themselves using the following: White, Asian, Black or African American, American Indian/Alaska Native, Native Hawaiian or Other Pacific Islander, or More than One Race, in addition to answering the question “Are you of Hispanic or Latino origin” with a binary “Yes” or “No”. The demographic questionnaire was prepared using the classifications outlined by the Federal Data on Race and Ethnicity Standards, also used by the National Institute of Health [54].

The second questionnaire sought to identify and categorize subjects’ vestibular history information. Subjects were to first asked to report their flight experience in terms of being a passenger on small or large aircraft and/or their status as a private pilot (if applicable), accompanied by years flying, acrobatic experience, and total recent flight time (i.e., both within the last year and within the last 30 days). Next, we asked subjects to report their prior centrifuge experiences (for how long, an account of any motion sickness experienced, etc., if applicable). Finally, the questionnaire asked how often subjects performed two different types of activities. These activities were categorized into “proprioceptive” and “bioenergetic” activities, adapted from Caillet [14]. Example proprioceptive activities listed included trampoline, gymnastics, climbing, martial arts, downhill skiing, waterskiing, sailing, fencing, and archery. Example bioenergetic activities included running, basketball, cycling, football, handball, swimming, volleyball, rugby, cross-country skiing, and canoe/kayaking. When asked how regularly subjects performed the individual activity, two time points were queried (within the last 12 months and during their peak activity level) and subjects were given 7 choices: “never”, “1 time”, “2-5 times”, “6-12 times”, “monthly”, “weekly”, or “daily”. These questionnaires were motivated by the hypothesis that individual differences in CC illusion acclimation may relate to activities an individual has previously or regularly participates in, as opposed to an innate difference. While speculative, we attempted to characterize potentially relevant aspects of activity history that may differ between individuals.

The third and final questionnaire was the commonly used Motion Sickness Susceptibility Questionnaire (MSSQ) [43, 44], asking subjects to report their age, their sex, and how often they felt sick or nauseated during exposure to various types of transportation or recreational activities both as a child and over the last 10 years. The MSSQ is scored to produce a percentile (0-100%. of susceptibility to motion sickness.

2.4 Procedure

2.4.1 Vestibular perceptual thresholds

Subjects’ self-motion thresholds were quantified in three axes of rotation: yaw, pitch, and roll, with the axis of rotation at the subject’s head and about an Earth-vertical axis, in all configurations. For the roll and pitch axes, subjects were in the supine and the left lateral decubitus position, respectively. To quantify the yaw rotational thresholds, subjects were seated upright.

Methods of quantifying vestibular perceptual thresholds were similar or identical to previous, recent approaches [24, 37]. Subjects completed 100 trials in each of the three rotational axes. Stimulus direction (e.g., for yaw, rotation to the left or right) and magnitude were randomized for each trial, with the magnitude selected using the method of constant stimuli from peak angular velocities of 0.2, 0.5, 0.9, 2, 4, or 7 degrees per second (deg/sec), with 5, 20, 30, 30, 10, and 5 trials at each, respectively. Each motion was two seconds in duration, with the profile defined by a 0.5 Hz single-cycle sinusoid in acceleration. Immediately after the rotation of the HERD stopped, the white noise signaled the end of the trial to the subjects, who then reported the direction in which they felt they had rotated (forward or backward for pitch, left or right for yaw and roll), in a two-alternative forced-choice task [7]. Perceived rotation direction was reported verbally, as well as with the wireless pushbuttons. After ensuring the subject was comfortable with the procedures and their reporting responsibilities, each subject was provided ∼10 training trials prior to completing the 100 test trials, which typically required a test duration of about 20 minutes. Each rotational axis (roll, pitch, and yaw) was tested in a separate block, often on separate days, in a counterbalanced order.

Binary responses (e.g., perceived left vs. right) were fit with a standard cumulative Gaussian psychometric function [38]. In this direction-recognition task, the fitted value for σ corresponds to the “1-sigma” threshold. A 95%.onfidence interval for each threshold estimate was approximated using a leave-one-out jackknife procedure [42, 51].

2.4.2 Extended duration CC illusion acclimation

The second phase of the investigation was an extended-duration CC illusion acclimation investigation. Further detail is provided elsewhere [8]. Briefly, subjects (n=11) were spun clockwise in yaw about an Earth-vertical axis while seated upright in the HERD chair configuration (Fig. 1A) for daily 25-minute acclimation sessions. During each session, subjects were incrementally exposed to the CC illusion to encourage benign acclimation (i.e., tolerance of higher spin rates without subjects experiencing substantial motion sickness). The CC illusion was induced by prompting subjects to perform a pair of roll head tilts –first a head tilt 40° right ear down, followed by a head tilt back to upright, each over one second.

At the start of the first acclimation session, subjects performed one head tilt pair while spinning at a supra-threshold stimulus of 10 RPM to become familiar with the CC illusion and provide a basis for future reporting of the presence or absence of the illusion. Following this initial head tilt pair, subjects were spun down to 1 RPM to begin the personalized, incremental training protocol. At each constant spin rate, subjects performed head tilts every 30 seconds, when prompted. After each head tilt, subjects kept their head in the tilted position while reporting “Yes, I felt the illusion” or “No, I did not feel the illusion” and pressing the corresponding “Yes” or “No” wireless pushbutton. This simple binary decision task allows acclimation to be quantified in terms of physical spin rate, such that direct comparisons can be made across subjects (as opposed to relative intensity ratings, which can be more difficult to compare due to inconsistent intensity baselines or spin rate tolerability between subjects [12, 52]). If subjects reported that they did not feel the illusion following both head tilts of one head tilt pair (i.e., head tilt down and head tilt up), the spin rate would be increased by 1 RPM at a rate in which the acceleration could not be perceived by the subject. If the subject instead reported that they did feel the illusion on either head tilt of the head tilt pair, the spin rate was maintained. The next head tilt pair began after 30 seconds at the constant spin rate to allow for semicircular canal stimulation to equilibrate. This procedure was repeated for 25 minutes based upon previous work, which indicated that the duration was sufficient to promote acclimation while not causing subject fatigue [9, 53]. Sessions were performed once per weekday. Similar to previous studies [9], the initial spin rate on each session (after the first) was the fastest spin rate at which no illusion was experienced at the beginning of the previous session.

Subjects completed the testing when they met one of three ending criteria: 1) reach a spin rate of 25 RPM in which they did not feel the CC illusion at the beginning of a session, 2) reach a plateau in acclimation, or not increase the spin rate at which they did not feel the CC illusion at the end of each session for ten consecutive sessions, or 3) complete 50 sessions of CC illusion acclimation training. This extensive protocol was aimed at capturing a more accurate acclimation rate for each subject.

Additionally, subjects reported a Motion Sickness Rating (MSR) on a scale of 0 (no motion sickness) to 20 (extreme nausea or vomiting) every 5 minutes within each testing session.

We fit a linear regression model to each subject’s CC illusion thresholds (i.e., the fastest spin rate at which each subject did not feel the CC illusion at the end of each session) across all testing days to calculate individual acclimation rates (the slope of the linear fit). This metric refers to the average increase in spin rate per training session that did not elicit the CC illusion (units of RPM/session).

2.5 Statistical analysis

The purpose of this investigation was to identify potential predictors of CC illusion acclimation rate. To assess candidate predictors (i.e., questionnaires and vestibular perceptual thresholds), we performed correlation tests between the predictor metrics of interest and each subject’s rate of acclimation. To determine the appropriate correlation test for continuous predictor variables we first used the small “n” Anderson-Darling and Shapiro-Wilks tests to assess normality. If the variable was consistent with being normally distributed, correlations were performed using a Pearson Product-Moment Correlation. If either of the two variables in the correlation were not continuous or were found to be not normally distributed, a Spearman’s Rank-Order Correlation was performed. All vestibular threshold data was log-transformed before correlation tests were conducted, as it has been previously shown that thresholds are lognormally distributed across subjects [3 , 50]. In cases when the appropriate predictor was categorical (such as when testing for a relationship between acclimation rate and race), we used a One-Way ANOVA to compare groups.

In addition to our primary statistical analysis to identify existing correlations between the potential predictor variables and CC illusion acclimation rate, we also conducted a supplementary analysis to evaluate the efficacy of using the statistically significant correlates as useful predictors of acclimation rate. To do so, we employed a leave-one-out cross-validation (LOOCV), in which the remaining 10 subjects’ data were used to predict the acclimation rate of the 11th subject, given his/her predictor variable. The performance of the model was evaluated using error-based metrics (root mean square error (RMSE), mean absolute error (MAE), and median absolute deviation (MAD) of the LOOCV predictions). Also, an R²-based metric, referred to as $Q_{cv}^{2}$ , relates the predictive residual sum of squares (PRESS) obtained via cross-validation to the sum of squares (SS) of the training set as follows [48]: $Q_{cv}^{2} = 1 - \frac{PRESS}{SS}$

All analyses were conducted using MATLAB R2020a and the R/RStudio packages.

3 Results

3.1 Vestibular perceptual thresholds and extended duration acclimation findings

Subjects’ (n = 11 for roll and yaw, n = 10 for pitch) range of vestibular perceptual thresholds were the following: roll rotation: 0.35–1.39 deg/sec, pitch rotation: 0.42–1.30 deg/sec, and yaw rotation: 0.88–3.60 deg/sec. One subject’s pitch threshold data was removed from the data set, as the headrest did not secure his/her head sufficiently.

Of the 11 subjects who began the extended acclimation investigation, 8 completed the study by reaching the ending criteria. The final 3 subjects chose to leave the study prematurely due to scheduling challenges (after 15, 26, and 35 days of testing). Acclimation in all 11 subjects, even those who were deemed “slow acclimators” or those who discontinued their participation in the investigation, continued fairly linearly with repeated exposure to the CC illusion. Thus, a linear acclimation rate was calculated for each subject, and all 11 subjects were included in the correlation analysis. Subjects’ range of acclimation rates were calculated to be 0.26–2.91 RPM/session (SD: 0.81 RPM/session, CV: 0.69).

3.2 Demographic information, activities history, and motion sickness not correlated to subjects’ acclimation rates

We investigated potential trends between subjects’ demographic information (age and race) and acclimation rates. No relationships were found between acclimation rate and A) age (Pearson Correlation, t(9) = –0.64, p = 0.54) or B) race (One-Way ANOVA, F(10) = 0.14, p = 0.87). We did not test gender effects nor effects of “Latino origin/not of Latino origin”, as we only had 1 female and 1 person who identified as “Latino” in the experiment.

To analyze the activities history information, we performed Spearman Rank Order Correlation tests and found no significant correlations between subjects’ acclimation rate and A) frequency of proprioceptive activities within the last year (Spearman Rank, t(9) = –0.65, p = 0.53), B) frequency of proprioceptive activities at the peak of proprioceptive activity frequency within their lifetime (Spearman Rank, t(9) = –0.04, p = 0.97), C) frequency of bioenergetic activities within the last year (Spearman Rank, t(9) = –0.98, p = 0.35), or D) frequency of bioenergetic activities at the peak of activity (Spearman Rank, t(9) = –0.49, p = 0.63).

Additionally, though not a primary objective of this investigation, we did perform statistical tests to identify any potential correlation between acclimation rate and motion sickness –both susceptibility reported before any acclimation training, and motion sickness reported during testing. We found no statistical significance between MSSQ and acclimation rate (Pearson Correlation, t(9) = –0.44, p = 0.67), or average reported MSR and acclimation rate (Spearman Rank, t(9) = –1.42, p = 0.19).

3.3 Roll rotation vestibular perceptual thresholds correlated with subjects’ acclimation rates

The final relationships tested were that of correlations between acclimation rate and vestibular perceptual thresholds (Fig. 2). We found a significant correlation between subjects’ acclimation rate and the log-transformed roll rotation thresholds (Pearson Correlation, t(9) = –2.68, r = –0.67, p = 0.025) (Fig. 2A). Correlations between acclimation rate and the other rotational threshold axes (log-transformed) were not significant: pitch (Pearson Correlation, t(8) = 0.15, p = 0.88) (Fig. 2B) nor yaw (Pearson Correlation, t(9) = 0.17, p = 0.87) (Fig. 2C).

Fig. 2

Correlations between acclimation rate and A) roll rotation threshold, B) pitch rotation threshold, and C) yaw rotation threshold. In all graphs, the x-axis is shown in a semi-log axis. Error bars show standard errors, which for the thresholds (x-axis) were determined using jackknife. On the semi-log axis, these appear asymmetric. The p-value from each statistical test is shown, while the line of best fit and r-value are displayed in panel A. The x-axis in panel A was constrained to improve resolution, truncating one error bar, which extends leftward to a value of 0.07 deg/sec.

Fig. 3

The relationship between predicted acclimation rate (calculated via LOOCV) and observed acclimation rate in RPM/session is shown. The proximity of the data to the overlaid unity line suggests that the predicted values reasonably represent those observed.

3.4 Leave-one-out cross-validation to assess roll rotation threshold as a predictor of acclimation rate

With the correlation between roll rotation thresholds and CC illusion acclimation rate, we used the LOOCV approach to determine the efficacy of using roll thresholds to predict acclimation rate. Figure 3 shows each predicted acclimation rate (produced with LOOCV) and the associated observed acclimation rate. These values were further analyzed to quantify the predictive capability of roll tilt thresholds, and the following metrics of the LOOCV residuals were calculated: RMSE = 0.726 RPM/session, MAE = 0.577 RPM/session, MAD = 0.593 RPM/session. Depending on the metric, this suggests that we would be able to reasonably predict acclimation rate based on roll rotation vestibular perceptual thresholds within 0.577–0.726 RPM/session of the subject’s actual rate. To provide non-dimensional metrics of model fit, we also normalized the RMSE, MAE, and MAD values by the range of the observed acclimation rates. These normalized values were calculated to be the following: nRMSE = 0.274, nMAE = 0.218, nMAD = 0.224.

Further, in relating the sum of squares of the predictive residuals and the sum of squares of the training set (i.e., observed acclimation rates), we obtained a $Q_{cv}^{2}$ value of 0.26, which suggests a reasonable predictive capability of the vestibular perceptual thresholds.

4 Discussion

4.1 Predicting acclimation to the CC illusion

From our results, demographics, vestibular history information, and motion sickness susceptibility do not appear to correlate with subjects’ CC illusion acclimation rates. However, the statistical power was limited by the diversity and size of our subject samples. We suggest future studies specifically aim to recruit a diverse subject pool, in terms of subjects’ age, gender, demographic, activity, and vestibular history. Nonetheless, with our current dataset, we have no evidence to suggest that any of these factors are related to inter-individual differences in CC illusion acclimation.

We did find a statistically significant correlation between subjects’ roll rotation vestibular perceptual threshold and their acclimation rate. Additionally, using a LOOCV analysis, we identified that roll rotation thresholds may provide a reasonable predictor of CC illusion acclimation rate. While an intriguing and novel finding, it should be interpreted cautiously. First, while a significant correlation, the error bars shown in Fig. 2A suggest substantial uncertainty in acclimation rates for faster adaptors (since stopping condition 1 was reached with fewer sessions to be used in the linear fit) and in thresholds for individuals with higher thresholds (since, in general, uncertainty in the threshold estimate scales with the estimate). Second, we note that due to the exploratory nature of this investigation, conservative Bonferroni or Šidák multiple comparisons correction were not applied for the three statistical tests performed for each rotational axis (for yaw, pitch, and roll). If applied, the correlation between roll rotation threshold and acclimation would fall just outside of statistical significance. Third, the $Q_{cv}^{2}$ was 0.26, where 0 corresponds to no predictive capability and 1 suggests a perfect prediction. However, it should be reiterated that the LOOCV analysis produces a true prediction (as opposed to an observed correlation/association), in that each subject’s acclimation rate is predicted using only the other 10 subjects’ data, such that any $Q_{cv}^{2}$ > 0 suggests some predictive capability. Ultimately, the relationship established here (dotted line in Fig. 2A) should be assessed in a new, independent sample of subjects to validate the use of thresholds as a predictor. Nonetheless, this finding is an important first step in understanding inter-individual differences in CC illusion acclimation rates.

It is critical to note that roll rotation thresholds significantly correlated with CC illusion acclimation rate, while yaw or pitch rotation thresholds did not. While speculative, roll rotation thresholds may be most closely associated due to the orientation in which subjects were exposed to the CC illusion. Subjects were spun in yaw about an Earth-vertical axis and performed roll head tilts. These roll head tilts resulted in illusory sensations of pitching either forward or backward. Perhaps better sensing of roll (as evident by lower roll rotation vestibular perceptual thresholds) facilitates expedited acclimation to the CC illusion resulting from roll head tilts. While the resulting CC illusion is predominantly in the pitch axis, better sensing of the actual roll head tilt may allow for better distinction of the illusory component of the sensation, enabling faster acclimation. Future studies should test this hypothesis by investigating acclimation in other axes of constant rotation and performed head tilts. For example, this hypothesis would conjecture the pitch rotation thresholds should best predict individual differences in acclimation to the CC illusion resulting from pitch head tilts.

Our rotational threshold tests were all conducted about an Earth-vertical axis located at the head. The head-centered rotational motions primarily stimulate the semicircular canals of the vestibular system and cause effectively no change in the stimulation to the otoliths (which sense gravity and linear acceleration). As we did not assess a motion threshold that stimulates the otoliths (e.g., tilt or translation thresholds), we cannot comment on whether these specific threshold tasks would correlate with acclimation to the CC illusion (which stimulates both the otoliths and canals). Instead, we just note that the 2-second motion duration (0.5 Hz) we used for rotation thresholds would yield a similar threshold for tilt about an Earth-horizontal axis (i.e., pitch or roll tilt) [37].

The significant correlation between roll rotation thresholds and CC illusion acclimation rate was in the direction hypothesized. Specifically, individuals with lower thresholds were faster acclimators. Correspondingly, individuals with higher thresholds were slower at acclimating. This is consistent with the hypothesis that vestibular sensory noise, as quantified with roll rotation thresholds, is a limiting factor to acclimating or adapting to unexpected sensory signals, such as the CC illusion stimulus. It further relates to the “credit assignment problem” [1, 46] where the brain has to associate an outcome (in this case, unexpected sensory information) with a cause (exposure to a novel, rotating environment) despite alternatives (the unexpected sensory information being simply due to sensory noise). While computational solutions have been proposed that would enable learning/adaptation [22], it is still reasonable to expect sensory noise may limit the rate at which acclimation to the CC illusion occurs. These findings may also be relevant to adapting in other scenarios with sensory conflict, such as gravity transitions experienced by astronauts. Furthermore, vestibular perceptual thresholds may predict individual susceptibility to adapting to sensory conflict not produced by the changes in the environment, such as patients who suffer acute vestibular dysfunction.

Vestibular perceptual thresholds have previously been found to relate to standing balance performance [32] and even manual control nulling ability [47], suggesting an important operational impact. However, these impacts were on static or steady-state performance, as opposed to acclimation to a novel sensory paradigm, as studied here. This suggests vestibular perceptual thresholds may be able to explain inter-individual differences in both vestibular performance and vestibular acclimation.

4.2 Artificial gravity as a countermeasure and implications for CC illusion acclimation

The findings from this investigation indicate that physiological factors may impact an individual’s ability to acclimate to the CC illusion, requiring either longer or shorter acclimation protocols to tolerably achieve acclimation to a desired operational spin rate. Additionally, we note that this required acclimation duration for spaceflight implementation may differ to that required in our ground-based approach. Previous research suggests that the CC illusion may be less provocative in microgravity [20 , 34–36]. These studies indicate a potential reduction in CC illusion intensity in both parabolic flight and on-orbit experiments during Skylab. However, though reduced, the CC illusion and/or disorientation were still experienced to some extent. It is important to note that these experiments were all conducted while subjects spun in yaw in an upright chair, with their head aligned with the rotation axis. We would hypothesize that if these experiments were repeated in microgravity in a conventional, off-axis centrifuge configuration with loading from centripetal acceleration on the otolith organs, this would create sensory conflict similar to that experienced in 1g. Further, AG may be required for mitigation of physiological deconditioning during a surface stay on the Moon or Mars. Any amount of planetary-based gravitational loading may induce the CC illusion during sustained head movements in AG. Therefore, the relevance of this investigation and the development of a protocol to acclimate humans to the CC illusion extends beyond our ground-based studies and may still be a critical requirement in spaceflight AG applications, despite these early findings in microgravity.

4.3 Limitations and future work

Additional limitations of the CC illusion acclimation experimental design and protocol can be found elsewhere [8, 9], while here we note a summary of these limitations and those within the current vestibular perceptual threshold arm of the study. First, our small sample sizes and limited diversity in the sample create difficulties in identifying potential correlations within the predictors, should they exist. This was part of another study [8], thus, recruiting specific subject ages or demographics was infeasible. Instead, this served as an opportunistic investigation leveraging our available dataset. Second, our subject group consisted largely of college-aged students, who are typically younger than the average age of astronauts during flight. We note that vestibular precision is known to degrade above the age of 40 [5, 6]; however, based upon the positive acclimation rate observed in all subjects - even those highly susceptible to motion sickness and those who acclimated at a much slower pace –we would expect similarly positive rates in astronaut-aged subjects.

In all of our acclimation studies published to date, we have investigated CC illusion acclimation to roll head tilts conducted during upright yaw rotation [8 –11]. We do not expect that acclimation in this axis would necessarily transfer to tolerance of head tilts made during supine rotation in a conventional centrifuge [23], though we have no evidence to suggest that supine acclimation would result in different findings from that reported here. We recommend that future studies investigate CC illusion acclimation in subjects exposed to supine, roll rotation, which would better replicate the configuration of spaceflight artificial gravity.

5 Conclusion

Predicting how quickly subjects will acclimate to the CC illusion is a critical step towards improved protocols to increase human tolerance of fast-spin-rate rotation necessary for short-radius, intermittent centrifugation for AG. This research begins to identify potential predictors of individual acclimation capacity.

Motivated by previous research on vestibular responses, we explored age, demographics, and history of activities as potential predictors. We hypothesized that subjects with lower vestibular perceptual thresholds (i.e., able to determine the direction of very small motions) would acclimate to the CC illusion more quickly. In 11 subjects, we found that roll rotation vestibular perceptual thresholds were significantly correlated to the rate at which subjects acclimated to the CC illusion. Additionally, we found relatively low RMSE, MAE, and MAD of predicted rates, suggesting that roll rotation thresholds may be able to reliably predict acclimation rate, improving the ability of investigators to pre-determine the duration of acclimation training that may be required for each individual to tolerably acclimate to a given, operationally-relevant spin rate. Ultimately, understanding the individual differences in CC illusion acclimation may enable more personalized acclimation procedures, further advancing the feasibility of using short-radius centrifugation as a countermeasure for spaceflight physiological deconditioning.

Footnotes

Acknowledgments

We would like to thank our experimental subjects for participation in the investigation, as well as Carson Brumley, Sebastian Metcalf, Varun Seth, and Marcos Mejia for assistance with data collection. This work was supported by a NASA Space Technology Research Fellowship, Grant Number 80NSSCK0085.

Declarations

References

Agogino

A.K.

and Tumer

, Unifying temporal and sturctural credit assignment problems, AAMAS. 4 (2004).

Agrawal

, Carey

, Hoffman

, Sklare

and Schubert

, The modified romberg balance test: normative data in U.S. adults, Otology & Neurotology 32 (2011), 1309–1311.

Benson

, Hutt

and Brown

, Thresholds for the perception of whole body angular movement about a vertical axis, Aviation, Space, and Environmental Medicine 60 (1989), 205–213.

Benson

, Spencer

and Stott

, Thresholds for the detection of the direction of whole-body, linear movement in the horizontal plane, Aviation, Space, and Environmental Medicine 57 (1986), 1088–1096.

Bermúdez Rey

M.C.

, T.K. Clark

M.C.

and Merfeld

D.M.

, Balance screening of vestibular function in subjects aged 4 years and older: a living laboratory experience, Frontiers in Neurology 8 (2017).

Bermúdez Rey

M.C.

, Clark

T.K.

, Wang

, Leeder

, Bian

and Merfeld

D.M.

, Vestibular perceptual thresholds increase above theage of 40, Frontiers in Neurology 7 (2016).

Bogacz

, Brown

, Moehlis

, Holmes

and Cohen

J.D.

, The physics of optimal decision making: a formal analysis of models ofperformance in two-alternative forced-choice tasks, Psychological Review 113 (2006), 700–765.

Bretl

K.N.

and Clark

T.K.

, Improved feasibility of astronaut short-radius artificial gravity through a 50-day incremental, personalized, vestibular acclimation protocol, Npj Microgravity (2020).

Bretl

K.N.

, McCusker

A.T.

, Sherman

S.O.

, Mitchell

T.R.

, Dixon

J.B.

and Clark

T.K.

, Tolerable acclimation to the cross-coupled illusion through a 10-day, incremental, personalized protocol, Journal of Vestibular Research 29 (2019), 97–110.

10.

Bretl

K.N.

, Mitchell

T.R.

, Sherman

S.O.

, McCusker

A.T.

, Dixon

J.B.

and Clark

T.K.

Retention of cross-coupled illusion training to allow for a short radius space centrifuge, in: IEEE Aerospace, Big Sky, MT, 2018.

11.

Bretl

K.N.

, Sherman

S.O.

, Dixon

J.B.

, Mitchell

T.R.

and Clark

T.K.

, A standardized, incremental protocol to increase human tolerance to the cross-coupled illusion, Journal of Vestibular Research 29 (2019), 229–240.

12.

Brown

E.L.

, Hecht

and Young

L.R.

, Sensorimotor aspects of high-speed artificial gravity: I. Sensory conflict in vestibular adaptation, Journal of Vestibular Research 12 (2002), 271–282.

13.

Bryant

E.C.

, Trew

M.E.

, Bruce

A.M.

, Kuisma

R.M.E.

and Smith

A.W.

, Gender differences in balance performance at the time of retirement, Clinical Biomechanics 20 (2004), 330–335.

14.

Caillet

, Bosser

, Gauchard

, Chau

, Benamghar

and Perrin

, Effect of sporting activity practice on susceptibility to motion sickness, Brain Research Bulletin 69 (2006), 288–293.

15.

Cheung

C.C.

, Hecht

, Jarchow

and Young

L.R.

, Threshold-basedvestibular adaptation to cross-coupled canal stimulation, Journal of Vestibular Research 17 (2007), 171–181.

16.

Clark

T.K.

, Predicting Individual Differences in Sensorimotor Adaptability to Altered Gravity using Measures of Sensory Noise: Validation and Operational Considerations, (2015).

17.

Clark

T.K.

, Galvan-Garza

R.C.

, Bermúdez Rey

M.C.

, Yi

and Merfeld

D.M.

, Perceptual Noise and Sensorimotor Adaptation, (2015).

18.

Clark

T.K.

and Merfeld

D.M.

, Vestibular Perceptual Noise and Adaptation to an Altered Gravity Environment (2016).

19.

Clark

T.K.

and Merfeld

D.M.

, Does Adaptation to Tilt Perception to Altered Gravity Related to Vestibular Perceptual Thresholds? (2017).

20.

DiZio

, Lackner

J.R.

and Evanoff

J.N.

, The influence ofgravitoinertial force level on oculomotor and perceptual responsesto coriolis, cross-coupling stimulation, Aviation, Space, andEnvironmental Medicine 58 (1987), 218–223.

21.

Faisal

A.A.

, Selen

L.P.J.

and Wolpert

D.M.

, Noise in the nervoussystem, Nature Reviews Neuroscience 9 (2008), 292–303.

22.

and Anderson

J.R.

, Solving the credit assignment problem:explicit and implicit learning of action sequences withprobabilistic outcomes, Psychological Research 72(2008), 321–330.

23.

Garrick-Bethell

, Jarchow

, Hecht

and Young

L.R.

, Vestibularadaptation to centrifugation does not transfer across planes of headrotation, Journal of Vestibular Research 18 (2008), 25–37.

24.

Grabherr

, Nicoucar

, Mast

F.W.

and Merfeld

D.M.

, Vestibular thresholds for yaw rotation about an earth-vertical axis as a function of frequency, Experimental Brain Research 186 (2008), 677–681.

25.

Graybiel

, Miller

E.F.

and Homick

J.L.

, Individual differences in susceptibility to motion sickness among six Skylab astronauts, Acta Astronautica 2 (1975), 155–174.

26.

Guedry

F.E.

and Montague

E.K.

, Quantitative evaluation of the vestibular coriolis reaction, Aerospace Medicine 32 (1961), 487–500.

27.

Hartmann

, Haller

, Moser

, Hossner

E.-J.

and Mast

F.W.

, Direction detection thresholds of passive self-motion in artistic gymnasts, Experimental Brain Research 232 (2014), 1249–1258.

28.

Hecht

, Brown

E.L.

and Young

L.R.

, Adapting to artificial gravity (AG) at high rotational speeds, in: Life in Space for Life on Earth, Stockholm, Sweden, 2002.

29.

Holly

J.E.

, Vestibular coriolis effect differences modeled with three-dimensional linear-angular interactions,Equilibrium & Orientation}, Journal of Vestibular Research {– 14 (2004), 443–460.

30.

Jones

G.M.

, Origin significance and amelioration of coriolisillusions from semicircular canals – a non-mathematical appraisal, Aerospace Medicine 41 (1970), 483–490.

31.

Joosten

B.K.

, Preliminary Assessment of Artificial Gravity Impacts to Deep-Space Vehicle Design, NASA Lyndon B. Johnson Space Center, Houston, TX, 2007.

32.

Karmali

, Bermúdez Rey

M.C.

, Clark

T.K.

, Wang

and Merfeld

D.M.

, Multivariate analyses of balance test performance, Vestibular Thresholds, and Age, Frontiers in Neurology 8 (2017).

33.

Karmali

and Merfeld

D.M.

, A distributed, dynamic, parallelcomputational model: the role of noise in velocity storage, Journal of Neurophysiology 108 (2012), 390–405.

34.

Lackner

J.R.

, Motion sickness: more than nausea and vomiting, Experimental Brain Research 232 (2014), 2493–2510.

35.

Lackner

J.R.

and Graybiel

, Influence of gravitoinertial force level on apparent magnitude of coriolis cross-coupled angular accelerations and motion sickness, Mechanisms, Prediction, Prevention, and Treatment (1984).

36.

Lackner

J.R.

and Graybiel

, The effective intensity of coriolis, cross-coupling stimulation is gravitoinertial force dependent: Implications for space motion sickness, Aviation, Space, and Environmental Medicine 57 (1986), 229–235.

37.

Lim

, Karmali

, Nicoucar

and Merfeld

D.M.

, Perception precision of passive body tilt is consistent with statistically optimal cue integration, Journal of Neurophysiology 117 (2017), 2037–2052.

38.

Merfeld

D.M.

, Signal detection theory and vestibular thresholds: I. basic theory and practical considerations, Experimental Brain Research (2011), 389–405.

39.

Mickle

K.J.

, Munro

B.J.

and Steele

J.R.

, Gender and age affect balance performance in primary school-aged children, Journal of Science and Medicine in Sport 14 (2010), 243–248.

40.

Nouri

and Karmali

, Variability in the vestibulo-ocular reflexand vestibular perception, Neuroscience 393 (2018), 350–365.

41.

Peterka

R.J.

, Black

F.O.

and Schoenhoff

M.B.

, Age-related changes in human vestibulo-ocular and optokinetic reflexes: pseudorandom rotation tests, Journal of Vestibular Research 1 (1990), 61–71.

42.

Quenouille

M.H.

, Notes on bias in estimation, Biometrika 43 (1956), 353–360.

43.

Reason

J.T.

, Relations between motion sickness susceptibility, the spiral after-effect and loudness estimation, British Journal of Psychology 59 (1968), 385–393.

44.

Reason

J.T.

and Brand

J.J.

, Motion Sickness, Academic Press London, 1975.

45.

Reschke

M.F.

, Cohen

H.S.

, Cerisano

J.M.

, Clayton

J.A.

, Cromwell

, Danielson

R.W.

, Hwang

E.Y.

, Tingen

, Allen

J.R.

and Tomko

D.L.

, Effects of sex and gender on adaptation to space: neurosensorysystems, Journal of Women’s Health 23 (2014), 959–962.

46.

Richards

B.A.

and Lillicrap

T.P.

, Dendritic solutions to the creditassignment problem, Current Opinion in Neurobiology 54(2019), 28–36.

47.

Rosenberg

M.J.

, Galvan-Garza

R.C.

, Clark

T.K.

, Sherwood

D.P.

, Young

Y.R.

and Karmali

, Human manual control precision depends on vestibular sensory precision and gravitational magnitude, Journal of Neurophysiology (2018), 3187–2197.

48.

Schüürmann

, Ebert

, Chen

, Wang

and Kühne

, External validation and prediction employing the predictive squaredcorrelation coefficient - test set activity mean vs training setactivity mean, Journal of Chemical Information and Modeling 48 (2008), 2140–2145.

49.

Stone

R.W.

, An overview of artificial gravity, in: Fifth Symposium on the Role of Vestibular Organs in Space Exploration, Pensacola, FL: Naval Aerospace Medical Center, 1973: pp. 23–33.

50.

Suri

, Clark

T.K.

, Human vestibular perceptual thresholds forpitch tilt are slightly worse than for roll tilt across a range offrequencies, Experimental Brain Research 238 (2020), 1499–1509.

51.

Tukey

J.W.

, Bias and confidence in not-quite large samples, Annals of Mathematical Statistics 29 (1958), 614.

52.

Young

L.R.

, Hecht

, Lyne

L.E.

, Sienko

, Cheung

C.C.

and Kavelaars

, Artificial gravity: head movements during short-radiuscentrifugation, Acta Astronautica 49 (2001), 215–226.

53.

Young

L.R.

, Sienko

, Lyne

L.E.

, Hecht

and Natapoff

, Adaptation of the vestibuo-ocular reflex, subjective tilt, andmotion sickness to head movements during short-radiuscentrifugation, Journal of Vestibular Research 13(2003), 65–77.

54.

Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity, Office of Management and Budget, 1997.