Modification of human postural responses to soleus muscle vibration by rotation of visual scene

1. Introduction 

Vision is an important sensory input integrated into the postural control system, along with somatosensory and vestibular information [1]. Absence of visual cues results in a slight increase of body sway. The best demonstration that the visual system improves postural control comes from the simple observation that subjects normally sway more when their eyes are closed than when open [2]. Sway velocity also increases when subjects stand with their eyes closed instead of open [3]. Supplementary visual biofeedback of shifts in the CoP (centre of pressure) on a display improves balance and reduces body sway [4].

The role of vision in postural control has often been investigated by measuring postural responses to linear movement and rotation of the visual scene. Linear motion of the visual scene induces postural inclination in the same direction as scene motion [5], [6]. Rotation of the visual scene biases subjects’ perception of verticality and induces a systematic postural deviation towards the direction of visual field rotation. Roll motion of the visual scene induces a stronger and more destabilising postural change than linear motion [7]. A large-visual-field motion in the roll plane during upright stance induces a sensation of apparent body motion (roll vection) and increases lateral body sway and a lateral shift of the CoP [8]. When roll or pitch rotation of a large disc was viewed by subjects with various combinations of horizontal eye-in-orbit and head-on-trunk deviations, it was found that the main direction of body sway was always reoriented to be parallel to the direction of disc motion [9].

When the vestibular system is impaired, vision appears to have a greater influence on standing postural control. Patients with vestibular disorders were found to have a significantly higher magnitude of postural sway than healthy subjects while viewing central optic flow stimuli [10]. Similarly, patients with somatosensory deficits in the lower extremities showed an increased influence of vision on the magnitude of body sway [11].

Proprioceptive input from postural muscles, particularly from leg muscles is important in balance control. In standing subjects, vibration of leg muscles results in a postural response known as vibratory-induced falling [12], which can be characterised as involuntary body lean in the direction of vibrated muscles [13]. Postural responses to calf muscle vibration were minimal when subjects’ eyes were open, but increased when subjects’ eyes were closed or when vision was inverted [14], [15].

In order to understand visual and proprioceptive interaction in human stance control, postural responses to paired visual scene rotation and bilateral vibration of soleus muscles were investigated. The aim of this study was to find out whether timing of visual and proprioceptive stimulation onsets could modulate the magnitude and the velocity of postural responses to lower leg vibration. We hypothesised that postural response to lower leg muscle vibration would be altered to a greater extent when visual scene motion was started earlier than muscle vibration as compared to when scene motion and vibration start in the same time.

2. Methods 

We tested 20 healthy subjects (eight men and 12 women, age range 24–60, mean 31 years) who gave their informed consent. The local Ethics Committee approved the study.

Proprioceptive stimulation was applied by two mechanical vibrators, consisting of small dc motors with excenters [16]. The vibrators were attached to the subject by elastic cuffs on both soleus muscles. The vibration stimulus had a frequency of 60Hz, an amplitude of 1mm, and a duration of 8s. The visual stimulus consisted of a rotating scene: a white disc with a diameter of 125cm and with randomly placed black spots of size from 3cm to 9cm. The visual scene was located on the right side of the subject at a distance of 50cm. The scene rotated either clockwise (backward visual stimulation) or counter clockwise (forward visual stimulation, VSF). The visual scene rotated with an angular velocity of 60°/s and a duration of either 8s or 11s.

Each experimental session consisted of 32 trials (four trials in each of eight conditions). The duration of each trial was 20s. The eight conditions consisted of two single vibration, two single visual rotation, and four combined proprioceptive-visual stimulation conditions. Single vibration of the soleus muscles during stance was given when subjects’ eyes were open (VIB) or closed (VEC). Single visual scene rotation was given in two directions, forward (VSF) or backward (VSB). The duration of each single stimuli was 8s. In the combined stimulation conditions, muscle vibration either started at the same time as forward or backward visual scene rotation (V0F, V0B) or was initiated with a 3s delay relative to the visual scene rotation (V3F, V3B). Conditions were presented in random order by dividing the 32 trials into four blocks of eight trials and presenting one trial from each condition in random order within each block. Each trial began with a 2s baseline period.

Subjects stood relaxed on the force platform, with their arms along the body and their feet parallel and 15cm from each other. During each trial, subjects kept their heads turned to the right, while their eyes followed an upper part of the disc. After each 20s trial, subjects relaxed with their heads in a straight-ahead position for 2min.

Postural responses to sensory stimuli were quantified by displacements of the centre of pressure (CoP), measured by a force platform. Before starting each trial, subjects were required to realign the location of the CoP to the initial position, which was checked on monitor. The initial CoP location was arbitrary assigned a value of zero.

Trunk tilts in the antero-posterior (AP) direction were measured in selected conditions by two strain-gauge, single axis, ±2G acceleration transducers, (San-ai, Japan, dimension 19mm×22mm, weight 20g, response frequency 0–30Hz). The acceleration output was low-pass filtered with a cut-off frequency of 5Hz [17] and the dc output (inclination) was calibrated in stationary conditions for a ±30° range of body tilt. The accelerometers were attached on the spinal column of the upper trunk at the level of fourth thoracic vertebra and on the lower trunk at the level of the fifth lumbar vertebra. The CoP signals, AP trunk inclination signals, and the stimuli were sampled at 100Hz.

The CoP parameters were evaluated only in the AP direction, because vibration of soleus muscles and visual forward or backward stimulation induced body tilts that were mainly in the antero-posterior direction. The CoP postural responses to sensory stimulation were quantified by the initial CoP slope in the first 2s after vibration onset, and by the final CoP position during last 2s before vibration offset. Postural responses in the upper and lower trunk were quantified in the same way as for the CoP responses. For each subject, the CoP displacement and trunk angle data were averaged across the four trials for each condition. Group averages were calculated from the individual subjects’ averages. All figures show group averages with standard errors of the mean (S.E.M.).

The differences in early velocities and final positions of the CoP in response to visual scene rotation were statistically analysed by repeated measures ANOVA with post-hoc Bonferroni tests: two vibration conditions×two visual direction conditions and two delay conditions. The level of significance was set at p<0.05.

3. Results 

The results showed that postural responses to lower leg vibration were modulated by visual scene rotation in forward and backward directions. Visual scene rotation significantly altered both the magnitude and the velocity of the vibration-induced CoP displacement and trunk tilt.

3.1. Single visual and proprioceptive stimulation 

Soleus muscle vibration induced an initial, rapid backward body tilt that was followed by a slower body lean. When vibration ceased, body alignment quickly returned to the initial position (Fig. 1). Subjects leaned further backward when their eyes were closed than when their eyes were open. During the final 2s before vibration offset, the shift in CoP location from the initial position was −2.13±0.50cm (VIB, Fig. 1A) for the eyes open and −2.75±0.51cm for the eyes closed (VEC, Fig. 1B) conditions. The CoP velocity in the early period, 2s after vibration onset, was also greater when subjects’ eyes were closed then when their eyes were open.

View Large Image Download to PowerPoint

Fig. 1. Group average CoP responses in the eight different conditions. (A) The muscle vibration with eyes open (VIB) evoked body sway in the backward direction, whereas the visual forward (VSF) and visual backward (VSB) stimulation induced postural lean to the forward or backward, respectively. When visual and vibratory stimulation were initiated in the same time (V0F and V0B), the visual scene motion modified the final CoP shift. (B) During muscle vibration with eyes closed (VEC), the fast CoP shift was observed to the backward direction. When muscle vibration was initiated with 3s delay to visual stimulation (V3F, V3B), there was a fast, unidirectional CoP shift to the backward during early period.

The single forward visual (VSF) and single backward visual (VSB) stimuli evoked smaller CoP shifts than the single vibration stimulus (Fig. 1), with final CoP displacements of 0.61±0.13cm for VSF and −0.08±0.20cm for VSB. At the onset of visual stimulation, subjects began to tilt in the forward direction. This was seen for both directions of visual stimulation. Likely it is a result of biomechanical structure of the lower limb which allowed us sway more in forward direction without the fear of falling. Based on this assumption as a first reaction to visual scene motion subject adapted a slight body inclination forward as more safety position. Beginning from the fifth second, the CoP shifts occurred in the same direction as the scene motion and increased slightly during the stimulation. When the visual rotation stopped, the location of the CoP returned to the initial position.

3.2. Combined visual and proprioceptive stimulation 

Postural responses to vibration of soleus muscles paired with visual scene rotation showed a clear modulation of the initial CoP velocity and the final CoP magnitude that depended on the direction of visual stimulation (Fig. 1A).

The repeated measures ANOVA that compared the final CoP position between the two directions of visual stimulation, with and without muscle vibration (Fig. 2B), showed significant effects of both vibration (F=143.83, P<0.001) and of the direction of scene rotation (F=28.90, P<0.001). The effects of vibration combined with forward visual stimulation (V0F) on the final CoP position were not significantly different from the effects of vibration alone. Vibration paired with backward visual stimulation significantly shifted the final CoP position further backward than vibration alone (t=2.565, P<0.02). This enhanced backward CoP shift occurred in all 20 subjects.

View Large Image Download to PowerPoint

Fig. 2. Group averages of early CoP velocity (V: left panel) and final CoP tilts (CoP: right panel) recorded under both visual forward and backward stimulation with vibration and without vibration conditions. Each value averages four trials for all subjects for each condition. The error bars represent standard errors of the mean.

Pairing of visual stimulation with soleus muscle vibration resulted in an increase in the velocity of the initial CoP response as compared to vibration alone. Repeated measures ANOVA, comparing the initial velocity of the CoP response during the two directions of visual stimulation with and without vibration (Fig. 2A), gave a significant effect of vibration alone (F=102.21, P<0.001). However, effect of the direction of scene rotation approached significance (F=4.06, P=0.0511).

When the onset of vibration was delayed 3s instead of simultaneous with respect to visual scene rotation (V3F, V3B, Fig. 1B), the velocity of the initial CoP response to vibration increased. We found a similar increase in the initial velocity of the CoP response when vibration was given alone while subjects’ eyes were closed (VEC, Fig. 1B). The repeated measures ANOVA that compared the initial CoP velocity during two direction of visual stimulation and two delays (0s or 3s) of vibration (Fig. 3A) showed a significant effect of the delay of vibration onset (F=8.295, P=0.007), with larger amplitude backward lean when the onset vibration onset was delayed with respect to scene rotation. No significant effect of the direction of scene rotation was found on the early CoP velocity (F=0.83, P=0.37).

View Large Image Download to PowerPoint

Fig. 3. Group averages of early CoP velocity (V: left panel) and final CoP tilts (CoP: right panel) recorded under both visual forward and backward stimulation with vibration onset delay 0s and delay 3s to visual stimulation conditions. Each value averages four trials for all subjects for each condition. The error bars represent standard errors of the mean.

In the paired-stimulation conditions, the final CoP position at the end of the postural response was affected both by the direction of visual scene motion and by the relative timing of the onset of vibration (Fig. 3B). The repeated measures ANOVA revealed that backward scene rotation significantly shifted the final CoP position further backward than forward scene rotation (F=10.94, P=0.002). When the onset of vibration was delayed by 3s with respect to scene rotation, the final CoP position shifted further backward than when the onsets of vibration and scene rotation were simultaneous (F=5.53, P=0.024).

The group-averaged records of upper and lower trunk tilt (Fig. 4) showed patterns of responses to vibration and visual stimulation that were similar to the CoP postural responses (Fig. 1A). Fig. 4 shows that the amount of lower trunk tilt that was evoked by muscle vibration alone (VIB=−1.40±0.45) shifted forward when vibration was simultaneously paired with forward scene rotation (V0F=−0.88±0.40) and shifted backward when vibration was simultaneously paired with backward scene rotation (V0B=−2.17±0.58). For the V0B condition, in which an increase in backward CoP shift had occurred in 19 of 20 subjects, the lower trunk tilted significantly more backward than when vibration was given alone (t=5.759, P<0.001). For the V0F condition, in which the CoP response to vibration was shifted forward of the vibration-alone response in only 12 of 20 subjects, there was no significant effect on the lower trunk final position.

View Large Image Download to PowerPoint

Fig. 4. Group average of trunk inclination registered by the accelerometers Acc1 and Acc2. The responses at the lower trunk level (Acc1) indicated some tendency to linear modulation of postural responses in the forward and backward direction (V0F, V0B in relation to VIB), depending on the direction of visual scene motion. The responses at the upper trunk level (Acc2) the visual and proprioceptive interaction was presented only to the backward direction (V0F similar to VIB, while V0B in relation to VIB was shifted to the backward).

At the upper trunk level, no significant difference was found between single vibration (VIB) and forward visual stimulation (V0F). Note that in this V0F condition, a forward shift of the CoP was found in only nine of 20 subjects. In contrast to the effects of forward visual rotation, backward visual rotation resulted in a significant backward shift of the final values of upper trunk tilt (t=4.801, P<0.001) and the backward shift occurred in 18 of 20 subjects. A stick figure on the right side of (Fig. 4) presents a simplified scheme of how forward and backward visual stimulation affected lower trunk tilt. The amplitude of the final trunk tilt increased significantly when the direction of visual rotation was in the same direction as the postural effect induced by vibration, in the backward direction.

4. Discussion 

The substantial finding of this study is that visual input plays a direction-specific role in the control of postural orientation (body tilt) to somatosensory perturbation during stance. When combined with muscle vibration, visual stimulation in the forward direction induced small forward CoP shifts and visual stimulation in the backward direction induced backward CoP shifts of the postural response to vibration alone. When the postural effects of both sensory stimulation were in the same direction, i.e. backwards, the longer-term influence of vision upon body tilt was significantly enhanced. In contrast, when the directions of proprioceptive and visual inputs were contradictory, the final CoP shift in response to vibration was smaller and not significantly different than vibration alone. This asymmetry of postural response to both stimulation does not support interpretation by simply addition of sensory inputs. Similar asymmetry in effects of combined visual and proprioceptive perturbations on standing postural responses have been reported when translations of the support surface are paired with visual surround motion [18]. Body tilt was greatest when platform translations were paired with visual surround motion in the same direction. The authors considered this situation as a conflict between platform and visual motion because, in normal real life condition the direction of visual surround motion is opposite to the body movement. Similarly in our experiment, when soleus muscle vibration induced a backward body tilt, backward visual disc motion should be considered as an unusual situation where two different sensory inputs are in conflict.

Our data confirmed that body tilt evoked by lower leg muscle vibration is more pronounced with eyes closed than with eyes open. When visual input is reversed by an inverting prism, body tilt evoked by leg muscle vibration has been shown to be even greater than when subjects eyes are closed [14], [15]. This suggests that a sensory conflict in proprioceptive-visual input created by the opposite, unusual direction of visual scene motion during body movement can have a strong destabilising effect on body tilt. Based upon our results and other findings [6], [14], [18], the long term influence of vision on postural reorientation to muscle vibration is on compensatory basis. Enhancing visual input reduced the final amplitude of body tilt, while removing or reversing visual input increased the final amplitude.

We may consider the compensatory action of visual input in the control of postural orientation as a control system with simple negative feedback. Enhancement of gain in the negative feedback loop decreases slightly the system's response to error input. This is our case when forward visual stimulation is acting in a way that is contradictory to the effect of muscle vibration. We cannot interpret the case of backward visual stimulation with a simple decreased gain of visual input, but instead must deal with the reversal of polarity in visual input. In a control system, this means a change from negative to positive feedback, with known destabilizing influences. This effect of reversal in polarity might explain the amplitude asymmetry of postural responses to muscle vibration induced by different direction of visual scene motion.

The measured postural response to vibration of both soleus muscles was observed as backward body tilt, characterised as a postural readjustment or reorientation. We refer to postural reorientation as the average shift of the CoP position or of trunk tilts from the mean position during pre-stimulus baselines [19]. Postural reorientation to a vertical body position is a continual process that uses an integrative body reference of the vertical derived from single sensory vertical references [20]. This long-term adaptive control of postural orientation is centrally initiated with the aim of keeping the body aligned with gravity and to minimise postural muscle activation during stance. From this point of view, the leg muscle vibration induces a new body vertical and body tilt is adjusted to align with this new body position and to decrease the tension in vibrated muscles. This approach is consistent with an interpretation that vibration-induced muscle stimulation can create a false signal evoking displacement of the referent vertical position. As a result, the body sways forward or backward and regulation takes place relative to the new position [21].

We hypothesised that the small effects of a single visual stimulus on posture control would be enhanced when combined with proprioceptive input. To support this idea, we showed that the difference in CoP displacements between the forward and backward visual rotations was greater when paired with vibration (V0B-V0F=1.64±0.77) than when given alone (VSB-VSF=0.69±0.29). The sensitivity of the postural control system to visual input increased about two times in the condition when somatosensory afferentation was altered by lower leg vibration.

It is interesting that the pattern of trunk postural responses to lower leg muscle vibration was similar to the pattern of responses observed in the CoP (see Fig. 1, Fig. 4). The similarity of the CoP displacement and trunk tilt patterns indicate that the body's response to soleus vibration is realized mainly by the leaning of the trunk segment as the largest mass-segment of the body. This fact has a major implication for the diagnosis of balance deficit using a measurement of trunk angular velocity [22]. The validity of trunk acceleration as a balance control parameter suitable for use in clinical settings has been suggested [17].

There were a differences in the patterns of trunk tilt at level of the lower trunk and the upper trunk (Fig. 4). The visual and proprioceptive interaction at the lower trunk level showed a tendency toward an additive influence, similar to the effects of combined foot tactile and proprioceptive information [23]. A different type of response was observed at the upper trunk level. Forward visual rotation stabilised the upper trunk's position during soleus vibration, while backward visual rotation increased the amount of backward trunk tilt. One possible explanation for different effects of paired stimulation at the upper and at the lower trunk could be that vision has a stronger effect on head-down postural control, while proprioception has a stronger effect on foot-up postural control. The findings of a study of postural responses to support surface translation paired with visual stimulation suggested that visual information was important for maintaining a fixed position of the head and trunk in space, whereas proprioceptive information was sufficient to produce stable coordinate pattern between the support and legs [24]. However, in our case, it is difficult to conclude that the influence of visual input is more pronounced at the upper level of trunk and the effect of proprioceptive input is more visible at lower level of trunk.

It is likely that the stance control system determines whether the visual scene around subject is stable, not moving, and therefore suitable as an input for control of body sway. Our data indicate that balance control during stance needs a stable visual field when another sensory input is altered. This conclusion is in accord with findings that vision plays an important role in the maintenance of postural stability, particularly under challenging conditions where proprioceptive information from the feet and ankles is altered or reduced [25].

Current findings show that visual inputs have a long-term, direction-specific, influence on the control of postural orientation during stance. Further, the influence of vision on how the postural control system responds to proprioceptive input depends on whether or not the visual surround is interpreted as a stable in space. We conclude that a stable visual surround is necessary for adequate exploitation of visual information by the postural control system when other sensory inputs are changed.