|Home | About | Journals | Submit | Contact Us | Français|
Upright stance was perturbed using sinusoidal platform rotations to see how vestibular and somatosensory information are used to control segment and intersegmental dynamics in subjects with bilateral vestibular loss (BVL) and healthy controls (C). Subjects stood with eyes closed on a rotating platform (±1.2°) for frequencies ranging from 0.01–0.4 Hz in the presence and absence of light fingertip touch.
Trunk movement relative to the platform of BVLs was higher than Cs at higher platform frequencies whereas leg movement relative to the platform was similar for both groups. With the addition of light touch, both groups showed similar trunk and leg segment movement relative to the platform. Trunk-leg coordination was in-phase for frequencies below 1 Hz and anti-phase above 1 Hz. Interestingly, BVLs showed evidence of a “legs-leading-trunk” relationship in the shift from in-phase to anti-phase around 1 Hz. Controls showed no preference for either segment to lead the coordinative shift from in- to anti-phase.
The results suggest that the balance instability of BVL subjects stems from high variability of the trunk, rather than the legs. The high trunk variability may emerge from the “legs-leading” intersegmental relationship upon which BVLs rely. Because BVLs derive information about self-orientation primarily from the support surface when their eyes are closed, the legs initiate the shift to anti-phase trunk-leg coordination that is necessary for stable upright stance control. Higher trunk variability suggests that this strategy results in lower overall postural stability. Light touch substitutes for vestibular information, leading to lower trunk variability along with a trunk-leg phase shift similar to controls, without a preference for either segment to lead the shift. The results suggest that vestibulospinal control acts primarily to stabilize the trunk in space and to facilitate intersegmental dynamics.
Human postural control is often approximated as having two essential modes of control: a single-joint, inverted pendulum “ankle strategy” and a two-joint double-pendulum “hip strategy” [14,16,25]. As their names imply, the ankle and hip strategies are characterized by rotation primarily about the ankle and hip, respectively. During quiet stance and small perturbations, the nervous system is thought to employ primarily the single-joint ankle strategy, with active control of CoM motion derived primarily from ankle torque [13, 14,32,34]. With larger support surface translations, the hip strategy, in which CoM motion derives primarily from hip torque, is “recruited”. The generally accepted idea is that these basic patterns are centrally selected from a set of motor programs, arising from high-level neural strategies and implemented by complex sensori-motor control processes to most effectively counteract the physical characteristics of the perturbation [14,15].
Recent experimental and theoretical work, however, has questioned the distinction between these two modes of control. The current thinking is that both patterns are present to varying degrees all the time. For example, a re-analysis of body kinematics during support surface translations showed that both ankle- and hip strategies were observed at all translation frequencies [30,32]. Theoretical work from Alexandrov et al. [1,2] has also shown the co-existence of multiple patterns. They derived ankle, hip and knee eigenmovements from the mechanical properties of the body, where each eigenmovement is named after the joint that contributes most to the movement. The three patterns emerged as solutions to a three-joint model which co-exist with varying amounts of power during tasks such as voluntary bending, but that the ankle and hip eigenmovements constitute the biomechanical bases for the ankle and hip strategies described by Horak and Nashner [3,14]. Even during quiet, unperturbed stance, in-phase and anti-phase coordination patterns have been shown to co-exist [7,36]. At frequencies below 1 Hz, the trunk-leg segments clearly show an in-phase pattern. However above 1 Hz, an abrupt shift to an anti-phase strategy is observed. Such results unify the artificial distinction between quiet and perturbed stance, reflecting excitable modes or flexible synergies that predominate depending upon the task. Moreover, co-existing patterns discount the need for a selection process between pre-programmed patterns depending upon perturbation, environmental constraint or pathology [13,14,23].
A recent study has shown that the co-existing in-phase and anti-phase trunk-leg patterns during quiet stance respond differently to different types of sensory information . Changes in coherence due to additional sensory information were observed at frequencies below 1 Hz where the in-phase pattern predominates, but no discernable effect was observed above 1 Hz, where the anti-phase pattern predominates. The sensitivity of the in-phase pattern to sensory changes potentially reflects a greater degree of neural control than the anti-phase pattern, whose underlying basis may be primarily due to the biomechanics of the human body approximated as a double pendulum .
The role of the vestibular sensory system in coordination of the ankle and hip postural strategies is not well understood. Previous studies of patients with bilateral vestibular loss responding to discrete surface translations showed normal ankle strategy patterns but inability to use the hip strategy to balance across a narrow beam although vestibular loss subjects could add hip torque to ankle torque to recover equilibrium in response to large translations while standing on a large, firm surface [15,32]. Studies have also suggested that light fingertip touch on a stable surface may substitute as a vertical reference for patients with bilateral vestibular loss to reduce postural sway [6,18,24]. Here, we used spectral methods to investigate intersegmental dynamics during continuous translations of various frequencies between individuals with bilateral vestibular loss (BVL) and healthy controls. Our intent was to investigate how intersegmental dynamics adjust to platform perturbations of different frequencies in the presence or absence of vestibular information, and in the presence or absence of additional somatosensory information supplied by light touch fingertip contact .
The experimental protocol was approved by the Institutional Review Boards of the Oregon Health Sciences University and the University of Maryland and was performed in accordance with the 1964 Helsinki Declaration. All subjects gave their informed consent prior to participation in this study.
Five bilaterally deficient vestibular subjects, two females and three males (age = 54.8 ± 8.9 years), and five healthy control subjects matched by age and gender participated in the study. Vestibular function was assessed by a combination of methods to determine the degree of deficit, as shown in Table 1. Dynamic otolith function was determined from the modulation component of horizontal eye movements recorded during off-vertical axis rotations (OVAR) . Static otolith function was determined by measuring the peak-to-peak amplitude of counter-rolling eye movements during left and right slow-roll tilts (OC) . Horizontal semicircular canal function (HVOR) was determined by fitting an exponential function, A* exp (−t/T c), to the slow-component velocity of HVOR eye movements for constant velocity rotations about an earth-vertical axis . DC gain of the HVOR relative to acceleration was estimated from the product of the time constant, Tc, and the gain constant. The gain constant was determined by dividing the response amplitude, A, by the velocity of rotation. Pitch and yaw VOR gain were derived in a similar manner .
Subjects were instructed to stand on a variable pitch platform that rotated ±1.2° in the A-P direction, as shown in Fig. 1. Subjects stood in a standard parallel stance with their ankles directly above the platform axis of rotation. In the touch condition, subjects maintained contact with a touch plate with their right index fingertip. The touch plate consisted of a circular aluminum disk that was mounted on top of a 97.8 cm pedestal. Force sensors mounted between the disk and pedestal determined touch forces in the anterior-posterior, medial-lateral, and vertical directions. The pedestal stood on a supportive metal structure that straddled the platform in order to isolate the touch plate from platform vibrations. The center of the touch plate was adjusted so that it was directly in front of the subject’s right shoulder. The distance between the subject and the touch plate was adjusted so that the subject could reach the touch plate with their right index finger in a comfortable manner. Subjects wore a safety harness that was secured to a movable ceiling mount by a connecting strap. The connecting strap was adjusted to allow subjects to lower their body approximately 30 cm before becoming taut. The platform rotation signal was sampled at 120 Hz.
Kinematic data was recorded using a Motion Analysis data acquisition system (Motion Analysis Crop., Santa Rosa, CA) at a sampling rate of 30 Hz. Markers were located on the right side of the subjects’ bodies as shown in Fig. 1. The trunk segment was defined as the distance between the acromium process (shoulder) and the head of the greater trochanter (hip) while the leg segment was defined as the distance between the head of the greater trochanter and the lateral malleolus (ankle). Additionally, reference markers were placed on the pedestal and a stationary position on the non-rotating part of the platform. To quantify measurement noise, we computed the power spectral density (PSD) of the stationary platform marker using the same method applied to markers placed on the body (see below). The PSD of measurement noise was about 0.02 mm2/Hz, which was substantially less than the PSDs of hip and shoulder displacements up to about 3 Hz.
The study of posture is a complex problem partly because the human body consists of many segments that have been shown to be active during quiet stance e.g. [9, 20]. In the present experiment, we chose to study trunk and leg segments as a continuation of previous work which explored the manner in which these segments respond to changes in sensory information [7,36]. Trunk segment angular displacement was defined as the angle formed by the shoulder, the hip, and vertical. Leg segment angular displacement was defined as the angle formed by the hip, the ankle, and vertical. Angular displacements were measured in the sagittal plane. Positive angles refer to forward movements from upright vertical.
The platform rotation stimulus consisted of a single sinusoidal waveform at five frequencies: 0.01, 0.03, 0.10, 0.20, and 0.40 Hz. Two sensory conditions were employed: (1) Light (non-supportive) touch. Subjects were instructed to maintain contact of the center of the touch plate with their right index fingertip. An auditory alarm sounded when touch force exceeded 1 Newton; (2) No touch. Subjects were instructed to hold their right index finger in a stationary position in space, slightly above the point they believed to be the center of the touch plate.
Subjects were instructed to stand facing a visual target. Before the light touch trials, subjects were instructed to place their right index finger at the center of the touch bar. Before the no touch trials, subjects were instructed to place their finger directly above the center of the touch bar without making contact. This hand position mimicked the arm position and any biomechanical effects due to changes in the COM during light touch trials. After positioning the finger, but prior to the start of each trial, subjects were instructed to close their eyes. Trials were initiated with the movement of the support surface.
All trials were 100 seconds in duration. All subjects completed two, randomized 10-trial blocks, consisting of 5 frequency conditions × 2 sensory conditions. Subjects rested a minimum of 100 seconds between trials, but were allowed more rest if they desired. The total experiment lasted approximately 1.5 hours.
All of the control subjects and three of the BVL subjects completed all trials without falling. The remaining two BVL subjects fell close to the end of the trials (t = 90 seconds or later) on the 0.4 Hz platform condition. BVL subjects did not repeat trials in which they lost equilibrium. Data recorded before a subject lost equilibrium was included in the analysis.
A linear systems spectral analysis was performed on each trial by calculating the Fourier transforms of the platform and the subjects’ leg and trunk angles after subtracting the mean angle. Fourier transfer functions describing the response of the leg and trunk angles to platform movement were calculated as the Fourier transform of the response divided by the Fourier transform of the input at the driving frequency. Since the platform motion is deterministic, this calculation is consistent with the definition of the transfer function in terms of the power spectra. Within-subject gain and phase were calculated as the absolute value and the argument of the average transfer function across trials for each subject. Group averages for gain and phase were calculated as the arithmetic and circular means respectively. The circular mean of phase in radians is the argument of the mean of exp(i). A positive phase indicates that body sway led the platform movement.
One-sided power spectral densities (PSDs) and cross spectral density (CSD) were computed for the leg and trunk segment angles using Welch’s method with 100-second Hanning windows.
Velocity PSDs were calculated by multiplying the position PSD by (2πf)2, where f is frequency. Velocity variability was calculated as the square root of the integral from 0.01–3.0 Hz of the velocity PSD after subtracting the power at each of the five driving frequencies. Within-subject velocity variability was calculated for each trial, and then averaged. Group means were calculated as the arithmetic mean of the within-subject averages.
Additionally we calculated cophase and magnitude squared coherence from the complex coherence, where Pxy is the CSD between the trunk and legs, and Pxx and Pyy are the respective PSDs of the trunk and legs. Within-subject averages for complex coherence were calculated as the arithmetic mean across trials while group averages were calculated as the arithmetic mean across subjects. Cophase was calculated as the argument of the group average of the complex coherence using the trunk as the reference (positive is legs leading trunk) and magnitude squared coherence was calculated as the absolute value squared of the group average of the complex coherence.
Values for gain, phase, and variability were assessed using a group (2) × frequency (5) × condition (2) × segment (2) repeated-measures ANOVA with condition, frequency and segment as the repeated measures. The Greenhouse-Geiser correction for non-spherical data was applied. For all testing,p < 0.05 resulted in rejection of the null hypothesis.
In order to address the low statistical power associated with the small number of subjects, nonparametric pairwise comparisons for gain, phase and variability were made using the Wilcoxon Rank-Sum test. Multiplicity of testing for the five platform stimulus conditions was addressed by applying a Bonferroni correction.
Differences in magnitude squared coherence were tested for statistical significance using paired (touch condition) and unpaired (group) T tests on the Fisher’s z-transformed coherence values at each frequency step. Multiplicity of testing for the 300 frequency steps was addressed by controlling the false discovery rate .
Differences in cophase and complex coherence were tested for significance using paired (touch condition) and unpaired (group) F tests at each frequency step based on the assumption that the complex values of the transfer function were bivariate normally distributed in the complex plane . Cophase was tested using the argument of the complex coherence. Multiplicity of testing (cophase tested at 300 frequency steps) was addressed by controlling the false discovery rate .
Figure 2 compares the trunk segment, platform, and leg segment time series of a healthy control subject (black) with a BVL subject (gray) standing on the oscillating platform. The left plots show the no-touch condition and the right plots show the light-touch condition. For illustrative purposes, only the 0.01 Hz and the 0.4 Hz platform conditions are shown.
Figures 2A–B show how both the BVL and control subjects display similar sway characteristics for the 0.01Hz platform condition. Both subjects show similar trunk/leg sway without light touch and an attenuation of trunk/leg sway with light touch. Comparing these results to the 0.4 Hz platform condition in Figs 2C–D, the trunk/leg segments of both subjects contain motion at the platform frequency of 0.4 Hz, but the amplitude of the BVL subject is larger than that of the control subject. When compared to the 0.01 Hz platform condition, the BVL subject showed an increase in both trunk and leg sway, but the increase in trunk sway was noticeably greater than the increase in leg sway. When light touch was added, both subjects show an attenuation of body sway, although the BVL subject does not appear to use light touch as effectively at the higher platform frequency.
All BVL and control subjects were able to control fingertip contact force below the 1 N threshold for all platform stimulus-frequency conditions. The average (standard deviation) of the contact force across all platform conditions was 0.503 (0.167) N for BVL subjects (range 0.327 to 0.721 N) and 0.401 (0.166) N for control subjects (range 0.252 to 0.599 N).
Figure 3 shows the group data for gain, phase and velocity variability. The primary result was that BVL subjects displayed dramatic deficits in trunk control that were not observed in control subjects which can be seen in Fig. 3A&B as a significant, 4-way group × condition × frequency × segment interaction (p = 0.0097).
A significant group × segment effect was observed for gain (p = 0.023). BVL (gray) and control (black) subjects showed similar trunk gains for the no-touch condition (Fig. 3A) at the lowest platform frequency (0.01 Hz), but as the platform frequency increased, BVL trunk gains increased dramatically for the four highest platform frequencies (0.03, 0.1, 0.2, and 0.4 Hz) compared to control subjects (Bonferroni corrected p = 0.045). In fact, for the three highest platform frequencies, trunk gains for all of the BVL subjects were greater than all of the control subjects. There were no significant group differences in leg gain for the no-touch condition (Fig. 3B). For the light touch condition (solid lines) there were no significant group differences in trunk or leg gain at any of the platform frequencies.
A significant group × condition effect for gain was also observed (p = 0.018). With the addition of light touch, BVL subjects showed a decrease in trunk gain (Fig. 3A) at the four highest platform frequencies (0.03, 0.1, 0.2, and 0.4 Hz) (p = 0.010), whereas light touch caused a somewhat smaller decrease in trunk gain for control subjects at the 4 lowest platform frequencies (0.01, 0.03, 0.1, and 0.2 Hz) (p = 0.010). Light touch also caused a reduction in leg gain (Fig. 3B) for the BVL subjects at the three highest platform frequencies (0.1, 0.2 and 0.4 Hz) (p = 0.010), while the control subjects displayed a decrease in leg gain at the four lower platform frequencies (0.01, 0.03, 0.1, and 0.2 Hz) (p = 0.010).
A significant group × frequency effect for gain (p = 0.024) can also be seen in Fig. 3A&B. Differences were found to occur only for the no-touch condition. BVL subjects displayed an increase in trunk gain with increasing platform frequency for the four lowest platform frequencies (Fig. 3A, 0.01, 0.03, 0.1 and 0.2 Hz) (p = 0.008). In contrast, control subjects showed decreasing trunk gains with increasing platform frequency for the three highest platform frequencies (p = 0.008). Leg gains for BVL subjects (Fig. 3B) increased for the three lowest platform frequencies (p = 0.008), then decreased for the two highest frequencies (0.2 and 0.4 Hz, p = 0.010). Control subjects showed decreasing leg gains with increasing platform frequency for the four highest platform frequencies (p = 0.008).
Differences in phase were observed between trunk and leg segments relative to the platform (p = 0.0002). Trunk segment phase, shown in Fig. 3C, led the platform at lower frequencies and lagged the platform at frequencies above 0.1 Hz. By comparison, leg segment phases, shown in Fig. 3D, were not significantly different from zero. Additionally, there were no significant differences in phase for either segment based on group although there was a significant condition × segment interaction (p = 0.0133) that can be seen in Fig. 3C that shows slightly less negative phase angles for the light touch condition above 0.1 Hz.
While gain and phase characterize the postural response at the platform driving frequency, velocity variability reflects changes in body sway at all frequencies other than the driving frequency. There was a significant 4-way group × frequency × condition × segment interaction (p = 0.0052) as shown in Fig. 3E&F.
Figure 3E shows how BVL subjects displayed a dramatic increase in trunk variability for the no touch condition at the two highest platform frequencies (0.2 and 0.4 Hz) (p = 0.045), a result that was not observed in control subjects. BVL leg variability, Fig. 3F, showed a smaller, but significant increase at the highest platform frequency (p = 0.045). By comparison, notice that velocity variability for the control subjects did not appear significantly different for either the trunk or the legs for all platform frequencies and touch conditions.
Figure 4A–C shows the cophase, magnitude squared coherence, and complex coherence that describe the dynamic relationship between trunk and leg segments averaged across all platform frequencies. BVL and control subjects showed a similar in-phase coordinative relationship between the trunk and leg segments for frequencies below ~ 0.8 Hz and an anti-phase relationship for frequencies above~ 1.1 Hz (Fig. 4A). Note that the anti-phase relationship may be represented by either positive or negative 180 degrees.
The magnitude squared coherence in Figure 4B varied across the frequency spectrum and showed no significant condition differences. The only significant group difference occurred for the no touch condition between 2–3Hz (compare gray and black dashed lines).
Figure 4A (shaded region) shows that there are two possible paths from the in-phase pattern to the anti-phase pattern that occur at ~ 1 Hz. At the transition between in-phase and anti-phase, BVL subjects appear to adopt a legs-leading-trunk coordinative relationship (i.e. cophase goes from 0 to + 180 degrees), while control subjects appear to do the opposite. In Fig. 4C we show the complex form of the coherence corresponding to the frequency range defined by the shaded region shown in Fig. 4A, illustrating how the coordinative relationship between the trunk and legs changes in the complex plane. Complex coherence values that lie along the positive real axis represent the in-phase relationship between trunk and legs while values that lie along the negative real axis represent the anti-phase relationship. Likewise, complex coherence values with imaginary parts that are greater than 0 represent a “legs-leading” coordinative relationship (the trunk was used as the reference in the calculation of complex coherence) while complex coherence values with imaginary parts that are less than 0 represent a “trunk-leading” coordinative patterns. Complex coherence for BVL subjects in the no-touch condition (gray dashed line) were above the real axis (i.e. the imaginary part is greater than 0) indicating that the legs are leading the trunk while shifting from in-phase to anti-phase. For the light touch condition (gray solid line), the trajectory appears to have a positive imaginary part, but in fact the group values were not statistically different from 0. A frequency-by-frequency analysis of individual values showed that three of the BVL subjects displayed a “legs-leading” coordinative relationship while the remaining two did not. In the case of the control subjects, neither condition was significantly above or below the real axis indicating that control subjects displayed neither a “legs-leading” nor a “trunk-leading” relationship between in-phase and anti-phase coordinative patterns.
Trunk and leg segment dynamics and their coordination during continuous surface rotations showed a number of trunk-specific and frequency-specific effects due to the loss of vestibular information.
With a rotating support surface, several important differences emerge between BVLs and controls that demonstrate the influence of vestibular information on postural control. The most striking difference was the control of the trunk. Trunk gain for control subjects showed a slight decrease with increasing platform frequency in the no-touch condition while BVL subjects showed a large increase in trunk gain for the same condition. Unlike trunk gains, leg gains for all platform conditions appeared similar in the present results, suggesting that vestibular information plays more of a role in controlling trunk sway than leg sway. Furthermore, vestibular information seemed to have a less significant effect on leg segment gains than information derived from the platform, suggesting that control of the legs is affected by somatosensory information from the support surface to a much greater degree than other forms of sensory input.
These results are consistent with the conclusions of studies that have linked trunk control to vestibular sensory sensory information [5,17]. Horak and Hlavacka  found that body sway induced by 3 sec galvanic pulses caused greater increases in trunk sway than in center of mass sway. Buchanan and Horak  found that although BVL and control subjects behaved in a similar manner at low frequencies of platform translations by riding the sinusoidally oscillating platform, their postural strategies differed at higher frequencies. The present results similarly show that as the frequency of surface rotation increased, healthy subjects oriented their trunks in a stationary position in space while their legs moved with the platform underneath them, whereas, BVLs continued to follow the platform with both the trunk and legs, often resulting in a fall or temporary loss of stability. Such results suggest that healthy subjects use vestibular information to improve control of the trunk in space at higher frequencies, while somatosensory information from the platform is used to control the legs. BVL trunk sway was far more variable, presumably because platform somatosensory information alone was less effective for trunk control at the higher rates of velocity/acceleration.
In addition to larger gains, postural sway velocity variability increased at the highest two platform frequencies in BVLs with no touch, whereas controls showed no differences. Postural sway velocity variability measures sway power at frequencies other than the driving frequency, which is interpreted as a measure of overall postural stability (i.e., higher variability = less stability). Loss of vestibular information clearly affects not just the response to the driving stimulus, but overall postural stability as well, primarily at higher frequencies. Semicircular canals effectively convey angular velocity information to the CNS over a broad range of frequencies [8,10,11,28]. Consequently, deficits in BVLs are observed primarily at frequencies in which vestibular information is useful for minimizing body sway.
The addition of light touch fingertip contact caused a large reduction in trunk sway for BVL subjects while the decrease for control subjects was significant, but less pronounced. Previous studies [6,18,24] have demonstrated that light touch information can act as a substitute for sensory information in subjects with vestibular deficits. Results of the present study suggest that the benefit from light touch is primarily in trunk control. While both trunk and leg segment gains were reduced with the addition of light touch for both groups, the reduction was most dramatic for the trunk in BVLs who displayed large increases in trunk gain with increasing platform frequency.
Similar findings were observed for velocity variability with the addition of light touch in BVL subjects, which showed reductions at the two highest platform frequencies for the trunk and at the highest platform frequency for the legs. An important difference however, is that light touch had no apparent effect on velocity variability in the trunk and leg segments in controls. With a fully intact vestibular system, the benefit of light touch was observed as a reduction in gain, but not velocity variability, at the lower platform driving frequencies, while with vestibular loss, benefits of light touch are apparent at higher platform driving frequencies as a reduction in both gain and velocity variability.
The trunk-leg intersegmental coordinative relationship was in-phase for frequencies below ~1 Hz and anti-phase for frequencies above ~1 Hz for all touch and platform conditions in all subjects, similar to previous studies [7,36]. However, differences between the two groups occurred at the shift between in-phase and anti-phase behavior. While control subjects displayed no tendency for one segment to lead the other, the shift between the in-phase and anti-phase patterns showed a legs-leading-trunk relationship for BVL subjects for the no touch condition. When compared to the control subjects, it appears that BVL subjects are receiving all of their sensory information from the support surface, which relegates the trunk to follow the platform-driving signal via the legs. The presence of vestibular information in the control subjects has the effect of eliminating the “legs-leading” relationship perhaps because the postural control system is able to determine that the signal driving the legs and the vestibular system are one in the same and treats the stimulus as a single, synchronized signal. For the light touch condition, three of the BVL subjects displayed the “legs-leading” relationship while the remaining two appeared similar to the control subjects, consistent with varying abilities to compensate for their vestibular loss with remaining sensory information . Not only do BVL subjects become more reliant than controls upon proprioceptive information from the support surface, but this increased proprioceptive dependence shifts the coordinative strategy to one in which the legs could be interpreted as “driving” the coordinative relationship.
The heavy reliance on proprioceptive information through the support surface may explain why BVL subjects are extremely sensitive to changes in the support surface . BVLs compensate for support surface translations similarly to healthy individuals when the feet are in contact with the full length of the support surface. However, on a surface that is shorter than the length of the feet, vestibular patients lose equilibrium almost immediately after a perturbation, rather than changing to a "hip strategy" as observed with healthy subjects . Such observations are not confined to the laboratory. Vestibular loss patients commonly relate no difficulty walking on hard support surfaces, but report discomfort maintaining normal equilibrium on compliant surfaces (e.g., grass, sandy beach) that disrupt the interpretation of somatosensory information at the feet, severely hampering their range of functional mobility.
These results are consistent with the up- and down-channeling hypothesis of Mergner and colleagues which postulates that vestibular information is transmitted down to lower body segments where it is fused with ascending somatosensory information, resulting in a transformation of sensory coordinates that enables one to accurately estimate body position in space [26, 27].
The present results indicate that segments close to the sensor are influenced more strongly than more distal segments. Loss of vestibular information led to deficits in trunk control but had less effect on the legs. Light touch led to more precise control of the trunk, but also had less effect on the legs. Measurable trunk gains during platform rotation indicate that somatosensory information from the platform drove the trunk motion more effectively than vestibular or light touch information drove the leg motion. Alternatively, because the platform stimulus is not purely sensory, trunk gain due to platform movement may be the result of mechanical trunk-leg coupling as well.
This study illustrates that the effect of vestibular loss or the addition of light touch cannot be viewed relative to single segments in isolation. The trunk and legs are biomechanically coupled and the vestibular system and light touch influences all body segments through such coupling. The larger variability observed in the trunk due to loss of vestibular function is due not only to the proximity of the trunk to the vestibular sensor but as well to its massive size relative to other segments. Joint interactions lead to segment variability that may originate from distal forces. For example, flexion at the hip may result primarily from ankle plantarflexion combined with gravitational forces acting on the trunk . Hip motion during upright stance is necessary to stabilize the upper body segments due to forces from more distal joints and the fact that muscles in the lower limbs act at more than one joint [23,35]. From this perspective, it is the effect of vestibular loss on intersegmental dynamics that we find most interesting. The sensitivity of BVL subjects to the properties of the support surface clearly emerges from their “legs-leading” intersegmental coordination strategy (see Fig. 4C), resulting in the disruption of trunk segment control through interaction torques and overall loss of balance stability.
We are deeply indebted to Robert Peterka for providing assessment of bilateral vestibular loss individuals.
Support for this research was provided by NIH grant R01NS35070 (John Jeka, PI) and NIH grants RO1AG06457 and DC-01849 (Fay Horak, PI).