Abstract
The velocity-storage circuit participates in the vestibulopostural reflex, but its role in the postural reflex requires further elucidation. The velocity-storage circuit differentiates gravitoinertial information into gravitational and inertial cues using rotational cues. This implies that a false rotational cue can cause an erroneous estimation of gravity and inertial cues. We hypothesized the velocity-storage circuit is a common gateway for all vestibular reflex pathways and tested that hypothesis by measuring the postural and perceptual responses from a false inertial cue estimated in the velocity-storage circuit. Twenty healthy human participants (40.5 ± 8.2 years old, 6 men) underwent two different sessions of earth-vertical axis rotations at 120°/s for 60 s. During each session, the participants were rotated clockwise and then counterclockwise with two different starting head positions (head-down and head-up). During the first (control) session, the participants kept a steady head position at the end of rotation. During the second (test) session, the participants changed their head position at the end of rotation, from head-down to head-up or vice versa. The head position and inertial motion perception at the end of rotation were aligned with the inertia direction anticipated by the velocity-storage model. The participants showed a significant correlation between postural and perceptual responses. The velocity-storage circuit appears to be a shared neural integrator for the vestibulopostural reflex and vestibular perception. Because the postural responses depended on the inertial direction, the postural instability in vestibular disorders may be the consequence of the vestibulopostural reflex responding to centrally estimated false vestibular cues.
SIGNIFICANCE STATEMENT The velocity-storage circuit appears to participate in the vestibulopostural reflex, which stabilizes the head and body position in space. However, it is still unclear whether the velocity-storage circuit for the postural reflex is in common with that involved in eye movement and perception. We evaluated the postural and perceptual responses to a false inertial cue estimated by the velocity-storage circuit. The postural and perceptual responses were consistent with the inertia direction predicted in the velocity-storage model and were correlated closely with each other. These results show that the velocity-storage circuit is a shared neural integrator for vestibular-driven responses and suggest that the vestibulopostural response to a false vestibular cue is the pathomechanism of postural instability clinically observed in vestibular disorders.
Introduction
To maintain head and body stability, humans must update and adjust their head and body positions in a given reference frame, such as head and body positions in space or the position of the head on the body (Suzuki and Cohen, 1964; Goldberg and Cullen, 2011). Although multisensorial inputs, such as vestibular, visual, proprioceptive, and efference copy cues, are involved in this function, vestibular cues play a primary role in stabilizing head and body positions in space through the vestibulocollic reflex (VCR) and vestibulospinal reflex (VSR; Angelaki and Cullen, 2008; Cullen, 2011). This function also operates while standing upright through the righting reflex, a kind of VCR and VSR, which activates the neck and spine extensors to counteract the gravity force (Molina-Negro et al., 1980; Dichgans and Diener, 1989).
Despite the common vestibular inputs, the vestibulopostural reflex shows a response distinct from that resulting from the vestibulo-ocular reflex (VOR) and vestibular perception. Vestibular perception and the VOR operate sensitively with similar thresholds ∼1°/s2 (Seemungal et al., 2004; Cousins et al., 2013). However, the vestibulopostural reflexes have a higher threshold and a lower gain than the VOR (Mitchell et al., 2013). The VOR closely interacts with the saccadic system and stabilizes eye position in the orbit during head motion, thereby generating alternating slow and quick phases of vestibular nystagmus (Cullen and Roy, 2004; Leigh and Zee, 2015). The VCR also interacts with the cervicocollic reflex, which secures the head position above the body (Goldberg and Cullen, 2011). However, the head responses are rather damped during motion, which is dissimilar to vestibular eye movements. These differences among the vestibulopostural reflex, VOR, and vestibular perception largely depend on discrepancies in the behavioral goals and the biomechanical properties of the plants (i.e., eyes vs neck and trunk; Peng et al., 1996, 1999).
In addition, vestibulopostural reflexes need to be explored further in two aspects. One is the role of the velocity-storage circuit in generating vestibulopostural reflexes. The vestibular cues relayed from the end organs and nerves are known to undergo central processing through the velocity-storage circuit in the brainstem and cerebellum (Merfeld et al., 1999; Angelaki et al., 2004; Laurens and Angelaki, 2011; Choi et al., 2018). During step-velocity rotation, the vestibular motion perception and VOR decay with similar time constants (Cousins et al., 2013). The variabilities in eye movements and perceptual decisions increase proportionally to the rotation velocity and correlate with each other. Therefore, vestibular perception and the VOR share the velocity-storage circuit for central processing (Bertolini et al., 2012; Cousins et al., 2013; Nouri and Karmali, 2018). Experiments using optokinetic or vestibular stimulations imply that the velocity-storage circuit is also involved in postural reflexes (Dichgans et al., 1972; Keshner and Kenyon, 2000; Haggerty et al., 2017; Bonsu et al., 2021). In previous observations, however, the postural reflex was explained mainly through the mechanism of aligning the head and body toward a biased gravitational orientation because of visual or vestibular stimulation. Of interest, the velocity-storage circuit participates in estimating translation-induced inertial cues, in addition to estimating rotational velocity and gravity orientation (Laurens and Angelaki, 2011). However, the effect of inertial cues estimated from the velocity-storage circuit on the postural reflex has not been evaluated in detail, and this area of research still needs to be supplemented. In addition, it is uncertain whether the vestibulopostural responses share the velocity-storage circuit with the VOR and vestibular perception and how the velocity-storage circuit generates postural responses. The other aspect to explore is the role of vestibulopostural reflexes in generating the clinical features observed in vestibular disorders. Many patients with vestibular disorders suffer from postural symptoms such as unsteadiness, directional pulsion, and falls (Bisdorff et al., 2009). Although the clinical features and their mechanisms because of impaired VOR are well established in vestibular disorders, those from abnormal vestibulopostural reflexes require further elucidation (Fitger and Brandt, 1982; Andre et al., 2005; Thomke et al., 2005; Choi et al., 2015).
To address these issues, we hypothesized that the velocity-storage circuit shared in perceptual and ocular responses is also engaged in generating postural responses. We tested this hypothesis by adopting cross-coupled vestibular stimuli, in which the head position is changed at the end of earth-vertical whole-body rotation (Guedry and Benson, 1978). This task allows our brain to estimate a false inertial cue (Laurens and Angelaki, 2011), making it possible to test the effect of the inertial cue estimated in the velocity-storage circuit on the postural reflex. We also attempted to explain the mechanisms of postural manifestations observed in vestibular disorders.
Materials and Methods
Hypothesis
We hypothesized that the outputs of the common velocity-storage circuit generate postural responses. To test this hypothesis, we adopted cross-coupled rotational stimuli and first simulated the velocity-storage output according to the following theoretical background. The velocity-storage circuit is known to sort the otolith information (gravitoinertial acceleration,
Schematics of experiments and the inertia predicted by the velocity-storage model. For simulation, we adopted the right-handed coordinate system as the positive directions presented in the figure. A, Internal estimation of gravity and inertia using a rotational cue. Top, During actual counterclockwise head rotation, the internally estimated gravity rotates clockwise and matches the gravitoinertial acceleration cue from the otolith, resulting in zero inertia estimation. In contrast, with a false counterclockwise rotational cue, the internally estimated gravity rotates clockwise, but the otolith-provided gravitoinertial acceleration remains steady, resulting in nonzero inertia estimation. B, Schematic illustration of the experiments. C, The inertia predicted by the velocity-storage model. The red circle in the right lower area indicates the false inertia evoked by a counterclockwise false rotational cue during the test session. CCW, Counterclockwise; CW, clockwise; g, gravitational acceleration; GIA, gravitoinertial acceleration; i, inertial acceleration. Ω = rotational cue.
Subjects
This study followed the tenets of the Declaration of Helsinki, the Institutional Review Board of Seoul National University Bundang Hospital approved this prospective experimental study protocol (B-2007-627–304), and all participants submitted written informed consent. Twenty-one healthy human volunteers free from a medical history of neurologic, psychiatric, or vestibular disorders initially participated in this experiment. Among them, one was unable to complete the experiment, and finally 20 (6 men, age range, 22–57 years, mean = 40.5 ± 8.2 years) participants were included for analyses.
Experimental setup and protocols
A rotation chair operated with a trapezoid velocity profile was set to rotate at 120°/s for 60 s. The duration of rotation, therefore, was more than three times the human time constant of the VOR (∼15 s) during earth-vertical axis rotation (Cohen et al., 1981; Dai et al., 1999; Lee et al., 2017), enough to eliminate the per-rotational cue. Each participant sat in the rotatory chair and fastened a seat belt on his/her waist. Goggles with a cover were applied to eliminate any visual cues. An inertial measurement unit (IMU) with a sampling rate of 100 Hz (model LPMS-B2 Series, Life Performance Research) was attached on the participants’ lower occiput to record head and neck motion. The IMU incorporated different sensing units to measure the inertia: a three-axis gyroscope and a three-axis accelerometer. To assess perceptions of inertial motion, the participants were instructed to report the pulling sensation on the head and body in one of the three directions, that is, left, none, or right.
As stated above, we designed the experiment to involve two sessions. The control session consisted of four rotations in which the participants were rotated clockwise and then counterclockwise with two different starting head positions (head-down and head-up). After each rotation, the participants kept their head position steady. The test session consisted of another four rotations, but the participants lifted up or bent down their head just after the rotations (Fig. 1B). Before the experiment, the participants were instructed not to intentionally change their head and body positions. They had one practice trial before the experiment.
The experiment was conducted with the control session first and the test session section, and during each session, the rotation order was as follows: head-down clockwise, head-down counterclockwise, head-up clockwise, and head-up counterclockwise. Throughout the experiment, we monitored the participants’ head and body positions and acquired their perceptions of inertial motion. Between the rotations, the participants were allowed to have at least a 5 min break to relieve the postrotational motion sickness. The experiment took about an hour for each participant.
Data processing and statistical analyses
For the head position, we analyzed data on the linear acceleration and angular velocity off-line using MATLAB/Simulink software version 2020a (MathWorks). To measure the gravity orientation in the head reference frame before and after the rotations, we averaged the gravity orientation for 2 s before and 5 s after the rotations during the control and test sessions. The linear acceleration vector at the stationary head position before the rotation was set to the initial gravity vector. The stabilized linear acceleration vector after each rotation was set to the final gravity vector. We plotted the initial and final gravity vectors in three dimensions and calculated the change in the gravity vector before and after the rotations. After evaluating the normality of data distribution with the Kolmogorov–Smirnov and Shapiro–Wilk tests, the changes in the gravity vector along the interaural axis were compared between the paired rotations during the control and test sessions (e.g., head-down clockwise in each session) using the paired t test. The level of statistical significance was set at 0.05. For the inertial motion perceptions, we assumed that subjects would have an inertial motion sensation only during the test session. We compared the participants’ inertial perceptions between the control and test sessions with the Bowker test for symmetry. For the post hoc analysis, we dichotomized the participants’ three possible responses for inertial perceptions (left, none, and right) into two categories—in the simulated direction or not. Then, the frequency of inertial perception was compared between the control and test sessions using the McNemar test. The level of statistical significance was also set to 0.05. Finally, to test whether the head position change and perceptual response were correlated, the polyserial correlation test was adopted using R statistical software (version 4.2.2).
Data availability
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available because of privacy or ethical restrictions.
Results
Postural responses
The postural responses are summarized in Table 1 and Figure 2. Generally, the participants had little changes in head position throughout the experiment during the control session. The changes in gravitational acceleration along the interaural axis were small during the control session, with the absolute mean value ranging from 0.005 to 0.030 g. However, during the test session, the participants showed significant head position changes along the interaural axis. These changes in interaural gravity acceleration ranged from 0.057 to 0.308 g as absolute mean values.
Head position changes in the interaural axis during the experiments
The postural responses. The right-handed coordinate system is adopted in the figure; therefore, positive values indicate forward, leftward, and upward. A–D, Left, The three-dimensional vector plot indicates the changes in the gravity vector in the head-centered reference frame before and after rotations. The x, y, and z axes represent the naso-occipital, interaural, and rostrocaudal axes. Right, Only the changes of the gravity vector in the interaural axis before and after rotations are presented. The shaded area indicates the reference range obtained during the control session. CCW, Counterclockwise; CW, clockwise; IA, interaural; NO, naso-occipital; RC, rostrocaudal.
At the end of head-down clockwise rotation, the participants’ heads were tilted slightly to the left (Δ = 0.019 × g) during the control session, whereas their heads were tilted to the right (Δ = −0.309 × g) during the test session (p < 0.001; Table 1). At the individual level, the participants’ heads were tilted rightward during the test session in 18 participants (18/20, 90%), outside the reference head position defined during the control session (Fig. 2A). No participants showed head tilting toward the left. At the end of head-down counterclockwise rotation, the participants’ heads were minimally tilted to the right (Δ = −0.005 × g) during the control session but tilted significantly to the left (Δ = 0.221 × g) during the test session (p < 0.001; Table 1). During the test session, 14 participants showed leftward head tilt, whereas three showed rightward head tilt (Fig. 2B).
During head-up rotation, the head tilted minimally during the control session (Δ = −0.012 × g and 0.030 × g for clockwise and counterclockwise rotations; Table 1). During the test session, the amplitude and the direction of head tilt were larger and opposed (Δ = 0.100 × g and −0.058 × g for clockwise and counterclockwise rotations, p = 0.001 and p = 0.015). However, the difference was not as prominent as was observed during head-down rotations between the control and test sessions (Table 1). During the head-up clockwise rotation in the test session, 10 participants showed leftward head tilt, but none had rightward head tilt (Fig. 2C). During head-up counterclockwise rotation in the test session, nine participants showed rightward head tilt, whereas two had leftward head tilt (Fig. 2D).
Perceptual responses to the inertial motion
The perceptual responses are summarized in Figure 3. About 3–5 of 20 participants reported an inertial perception (pulling sensation) during the control session. However, 11–16 participants had an inertial perception during the test session. Specifically, four participants reported a pulling sensation in the head-down clockwise rotation during the control session. During the test session, 16 participants reported a pulling sensation, 14 to the right and 2 to the left (p = 0.0074, Bowker test for symmetry; Fig. 3A). Three participants reported a pulling sensation during head-down counterclockwise rotation during the control session, two to the left and one to the right. During the test session, however, 16 had a pulling sensation, 2 to the right and 14 to the left (p = 0.0114; Fig. 3A). The results were similar during the head-up rotation, although less significant. For the head-up clockwise rotation in the control session, five participants had a pulling sensation, three to the left and two to the right. During the test session, in contrast, 11 had a pulling sensation, 10 to the left and 1 to the right (p = 0.0869; Fig. 3B). Finally, 5 subjects reported a pulling sensation toward the left or right after the head-up counterclockwise rotation during the control session, whereas 13 had a pulling sensation, all toward the right (p = 0.0117), during the test session (Fig. 3B).
The perceptual responses. A, B, The perceptual response for inertial motion (pulling sensation) during head-down (A) and head-up rotations (B) is the response during the test session corresponding to that during the control session. For the statistical analysis, the Bowker test of symmetry was adopted. C, D, For the post hoc analysis, the frequency of inertial motion expected by the velocity-storage model during the test session was evaluated using the McNemar test. CCW, Counterclockwise; CW, clockwise.
In the post hoc analysis, we compared the frequency of the pulling sensation in the direction that we expected to occur during the test session based on the velocity-storage model. A rightward pulling sensation was the expected perception during the head-down clockwise and head-up counterclockwise rotations in the test session. In contrast, a leftward pulling sensation was expected during the head-down counterclockwise and head-up clockwise rotations in the test session. In all pairwise comparisons, the frequency of a pulling sensation in the expected direction was significantly higher during the test sessions (p < 0.05, McNemar test; Fig. 3C,D).
Correlations between perceptual and postural responses
When the pooled data from all rotations during the control and test sessions were analyzed, there was a significant correlation between the head position changes and perceptual responses for the inertial motion (ρ = 0.56, p < 0.001, polyserial correlation test; Fig. 4).
Relationship between posture and perceptions. Statistical analyses of all pooled data were performed using a polyserial correlation test. For visual convenience, the changes in interaural gravity, a continuous variable, are placed along the x-axis, and the nominally acquired perceptual response is placed along the y-axis. L, Left side; N, none; R, right side.
Discussion
The present study demonstrated that the head and neck positions changed along the direction of inertia simulated using the velocity-storage model. In addition, the head and neck positions were generally concordant with the perceptual responses for the inertial motion. Based on these observations, we could propose that the velocity-storage circuit in the brainstem and cerebellum serves as a common neural integrator for the vestibulopostural reflexes and vestibular perception.
The velocity-storage circuit has been suggested to participate in postural control. In experiments using optokinetic visual stimulation, the rotating visual cue in the roll plane induced a tilt of perceived gravity and body toward the direction of visual cue by the mechanism of visuo-vestibular interaction (Dichgans et al., 1972; Keshner and Kenyon, 2000; Tanahashi et al., 2007). Because a rotating visual stimulation gives rise to the illusion that the gravity is deviating in the direction of rotation (Dichgans et al., 1972; Previc, 1992; Previc and Neel, 1995), the postural response has been described as a compensatory behavior for the tilt of gravity perception (Keshner and Kenyon, 2000). Galvanic vestibular stimulation, which delivers a direct current to the vestibular nerve, also generates postural responses (Séverac Cauquil et al., 1998; Fransson et al., 2003). Of interest, a distinct difference has been shown in the dynamics of postural response between optokinetic and galvanic stimulations. An optokinetic stimulation produced a postural response (tilt) that slowly built up (Dichgans et al., 1972; Keshner and Kenyon, 2000; Tanahashi et al., 2007), whereas a step galvanic vestibular stimulation caused an immediate postural response (Séverac Cauquil et al., 1998; Fransson et al., 2003). When the galvanic stimulation was prolonged, the tilted body posture gradually returned to the primary position, which resembles the postural response observed during the optokinetic stimulation (Haggerty et al., 2017). Based on these findings, an earlier experiment has suggested that the velocity-storage circuit contributes to postural responses and has a filter-like function for sensory information of various frequencies (Haggerty et al., 2017).
In this context, the head tilt observed during the test session in the present study can be explained by combining the findings of previous experiments. The retinal optic information induced by the optokinetic visual stimulation in the roll plane is fed to the velocity-storage circuit like the vestibular signals induced by the head motion in the opposite direction (Waespe and Henn, 1977). For example, the clockwise optokinetic stimulus in the head-up position is analogous to the counterclockwise vestibular stimulation for the velocity-storage circuit, which in our experiment was simulated by the counterclockwise postrotatory cue during the head-down clockwise rotation of the test session. Given the actual stimulation in the present study was of vestibular origin, the immediate postural responses, unlike that of the optokinetic visual stimulation, could be explained by the mechanism of postural responses during galvanic vestibular stimulation.
One fundamental distinction of the present study, unlike previous research, was its consideration of internally estimated inertial acceleration as the element for head tilt. A previous model for VCR that predicted the head position during step-velocity yaw rotation incorporated a central processing step (Peng et al., 1996, 1999; Cullen and Roy, 2004; Goldberg and Cullen, 2011). Simply, the model simulated the head position in space during yaw rotation, in which the head deviated away from the direction of rotation with damped response characteristics. However, there have been few studies on the inertia generated by the velocity-storage circuit and resultant postural responses. Based on the results of the current study, we could conclude that the velocity-storage circuit participates in the integration of two different vestibular cues (the canal and otolith) to generate postural responses, as is known for ocular movements (Glasauer, 1992; Merfeld and Zupan, 2002; Choi et al., 2018). As the head moves in the direction of inertia, this motor behavior is clearly an example of sensorimotor transformation in the form of a vestibular-drive postural reflex (Suzuki and Cohen, 1964). Therefore, this postural response aims to maintain postural stability in space during motion but paradoxically causes postural instability when a false vestibular cue is present.
The change of interaural gravity orientation was oppositely signed depending on the head position during and after rotation (e.g., rightward tilt in head-up positioning after head-down clockwise rotation and leftward tilt in head-down positioning after head-up clockwise rotation). This finding suggests that changes in muscle activation stabilize the head and neck in space by inertial cues in the head-centered reference frame. From the perspective of the motor system of the vestibulopostural reflex, this finding is in accord with previous observations of the reorganization between vestibular cues and motor commands related to maintaining balance (Forbes et al., 2016).
Of interest, apart from the overall results, there was a slight difference in the postural responses among rotations during the test session. During the head-down clockwise rotation in the test session, which was performed first, most participants showed homogeneous postural responses. There were only two participants whose head position stayed within the reference range. Meanwhile, during the head-down counterclockwise rotation in the test session, which was performed second, three participants showed a head position within the reference range, and another three showed a head position tilted in the direction opposite to what we anticipated. This finding was also reproduced during the clockwise and counterclockwise head-up rotation in the test session. The former generally tilted the participants’ heads in one direction, whereas the latter did not. Learning may have played a role in this difference according to the order of rotation; that is, the head and body tilt in response to the inertia during the prior rotation may have made the participant instantly generate a motor response against the inertia during the following rotation in an attempt to prevent unwanted postural instability. Indeed, this kind of learning effect has been established in experiments on motion sickness adaptation (Dai et al., 2010; 2011). However, not randomizing the rotation protocol to minimize the learning effects is a limitation of this study. In future studies, a randomized rotation protocol may be warranted.
The present study also showed that the velocity-storage circuit is common for posture and perception. By acquiring the inertial perception from the participants, we could observe a significant difference in the inertial perception between the control and test sessions. Despite the limitation in evaluating perceptions because of the response error, participants’ inertial perceptions during the test session were generally consistent with the simulated response. This observation implies that an inertial cue as an output of the velocity-storage circuit involves both perceptual and postural responses. The correlation between the inertial perception and postural response further solidifies the shared velocity-storage circuit in generating perceptual and postural responses. Although we did not evaluate eye movements, given our observation and prior knowledge of the common velocity-storage circuit for vestibular perception and VOR pathways (Bertolini et al., 2012; Cousins et al., 2013; Nouri and Karmali, 2018), this velocity-storage neural integrator appears to be involved in all kinds of vestibular-driven responses.
Clinical implications
Based on our observations, we may be able to interpret the postural manifestations of vestibular disorders. Previously, postural instability has been explained in terms of impaired VCR and VSR (Shin et al., 2012; Fujimoto et al., 2014), but this explanation is rather abstract. Hence, we tested whether there was an inertial component from the velocity-storage circuit in the postural bias (i.e., ipsilesional veering) of patients with acute vestibulopathy. In acute left-side vestibulopathy that homogenously involves three semicircular canals and two otoliths (Fig. 5A), the patients would have a clockwise rotational cue in the yaw and roll axes. However, there would be no pitch rotational cue because of the equivalent loss of the anterior and posterior semicircular canals, which cancels out the opposing rotational cues in the pitch axis (Leigh and Zee, 2015). In addition, the patients would have a tilted gravitoinertial acceleration cue in which the vector tip is aligned toward the healthy (right) side (Kim, 2020). We first noted a false clockwise rotational cue in the roll axis (rotating toward the right shoulder) that would rotate the estimated gravity counterclockwise in the roll axis, resulting in a difference between the gravitoinertial acceleration cue from the otolith and estimated gravity. The difference is interpreted as the inertial acceleration (
Postural responses in clinical disorders predicted by the velocity-storage model. Aa–c, Postural instability in patients with acute left vestibulopathy in which three semicircular canals and two otoliths were uniformly involved is taken as an example. There are three sources generating inertial cues. The first two are the abrupt tilt of the gravitoinertial acceleration cue from damaged otoliths (a) and a clockwise rotational cue in the yaw axis on the tilted gravitation cue (b), which both generate inertial cues toward the right. The other one is a clockwise rotational cue in the roll axis upon the tilted gravitational cue, in which the generated inertial cue directs downward to the left (c). B, The velocity-storage model simulation showed results concordant with the prediction. The input of the first row is a false otolith cue only (tilt of gravitation, 10°), and a clockwise rotational cue (10°/s) in each yaw and roll axis is added stepwise as the inputs in the second and third rows. The model set the vestibular lesion to develop at 10 s. For a detailed explanation, see below. †g, Neurally represented gravity acceleration; GIA, gravitoinertial acceleration; †GIA, neurally represented GIA; †i, neurally represented inertial acceleration; VST, vestibulospinal tract; L, left; R, right; F, forward; B, backward. Ω = rotational cue; X = cross-product in mathematical terms.
However, in addition to a false clockwise rotational cue, there are additional sources of inertia generation. One is otolith imbalance. When the gravitoinertial acceleration cue from the otolith tilts abruptly, a vectorial discrepancy between gravitoinertial and gravitational accelerations will occur, consequently creating inertial acceleration. However, in this case, the inertia heading upward to the right dissipates soon (within seconds) because of somatogravic and rotational feedback, which function to realign the gravitational acceleration to the gravitoinertial acceleration (Laurens and Angelaki, 2011), and there is only a static tilt of estimated gravity, which is the source for ocular tilt reaction (Dieterich and Brandt, 2019). The other is a false clockwise rotational cue in the yaw axis (rotating toward the right ear), which would rotate the estimated gravity counterclockwise in the yaw axis and create inertial acceleration. In the normal gravity orientation, a rotational cue in the yaw axis would not change the gravity estimate (because the gravity and rotational axis vectors are parallel). However, with a static tilt of estimated gravity because of the otolith imbalance, the false clockwise rotational cue in the yaw axis is no longer parallel to the estimated gravity vector. The result would be the same as observed in the off-vertical axis rotation (Laurens et al., 2013). Again, the created inertia is aligned downward to the right.
Because the inertia associated with a clockwise rotational cue in the roll axis, as we first noted, and the inertia created by gravitoinertial acceleration tilt and a clockwise rotational cue in the yaw axis are different, we tested the inertia direction when these three components are combined through the velocity-storage model we adopted here. For a clear understanding, we simulated the inertia generated by the velocity-storage model, first by using a false otolith cue only and then by adding yaw and roll rotational cues stepwise. In the case of isolated left otolith function loss, a simulation using the velocity-storage model showed that the direction of inertia was initially toward the healthy (right) side but shortly afterward transitioned toward the lesion (left) and soon disappeared (Fig. 5B, first row). Adding a false clockwise rotational cue in the yaw axis from the loss of the left horizontal semicircular canal function made the velocity-storage model again show the inertia directed toward the healthy (right) side, which soon changed the direction toward the lesion (left side) and then dissipated (Fig. 5B, second row). Last, adding a false clockwise rotational cue in the roll axis gave rise to the velocity-storage model generating constant inertia directed toward the lesion (left) side (Fig. 5C, third row).
In conclusion, the velocity-storage model could successfully predict inertia heading downward to the left when the three sources of falsely generated inertia were combined. The estimated inertia direction is concordant with the postural leaning in patients with left-side vestibular neuritis. The inertial effect could be adjusted by visual or somatosensorial cues aiding in adjusting to earth verticality. However, when the inertial cue is excessive, or when the compensation from visual or somatosensory cues is insufficient or eliminated, as well as during the Romberg test, patients will fall toward the lesion side. This explanation may apply to other vestibular disorders with false rotational and inertial cues.
Conclusion
The velocity-storage circuit is a common neural integrator, estimating the velocity, gravity, and inertial acceleration that are essential for motion and gravity perception and oculomotor and postural controls. Understanding the function of the velocity-storage circuit may allow an interpretation of the abnormal eye movements and postural instability observed in various vestibular disorders.
Footnotes
This study was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (2020R1A2C4002281).
The authors declare no competing financial interests.
- Correspondence should be addressed to Jeong-Yoon Choi at saideiju{at}gmail.com
