Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

BASIC AND CLINICAL ASPECTS OF VERTIGO AND DIZZINESS

Vestibular and Proprioceptive Contributions


to Human Balance Corrections
Aiding These with Prosthetic Feedback
C.G.C. Horlings,a M.G. Carpenter,b F. Honegger,a
and J.H.J. Alluma
a
Department of ORL, University Hospital, Basel, Switzerland
b
School of Human Kinetics, University of British Columbia, Vancouver, British
Columbia, Canada

Movement strategies controlling quiet stance and rapid balance corrections may have
common characteristics. We investigated this assumption for lower leg proprioceptive
loss (PL), peripheral vestibular loss (VL), and healthy controls. Our underlying hypoth-
esis was that changes in movement-strategy modulation following sensory loss would
improve with prosthetic biofeedback. Quiet stance was measured under different sen-
sory conditions and compared to corrections induced by multidirection support-surface
tilts. Response synergies were assessed using electromyography recordings from sev-
eral muscles. Biofeedback of trunk sway during gait and stance tasks used lower trunk
rotations to drive head-band-mounted vibro-tactile and auditory actuators. Strategies
of quiet stance were different for roll and pitch, depending on sensory conditions. Si-
multaneously acting strategies were observed for low- and high-frequency sway. PL
induced strategies different from those of VL and controls. VL strategies were identical
to those of controls but with greater amplitudes. Tilt perturbation movement strategies
were similar to high-frequency strategies of quiet stancemultisegmental. VL induced
increased trunk pitch and roll responses with hypermetric trunk muscle responses and
hypometric knee responses but unchanged synergies. Increasing PL up the legs caused
changed synergies. Biofeedback reduced stance body sway in VL and elderly subjects.
In conclusion, several movement strategies underlie quiet stance with high-frequency
strategies being common to those of perturbed stance. PL changes both movement
strategies and synergies, whereas VL only causes pathological changes to the modula-
tion depth. Thus, VL is more easily rectified using trunk sway positional biofeedback.

Key words: balance control; movement strategies; vestibular loss; leg proprioceptive
loss; biofeedback rehabilitation of imbalance

Introduction assumed that an appropriately scaled balance-


correcting response will not be properly gen-
Vestibular and proprioceptive sensory infor- erated and a fall will occur. This assumption
mation is considered essential for stable bal- raises a number of questions concerning the
ance. If one motion percept is not available effect of vestibular or proprioceptive loss (VL
when stance suddenly becomes unstable, it is or PL) on movement strategies and underlying
muscle response synergies.
Most evidence indicates that vestibular in-
Address for correspondence: Prof. Dr. Biomed. Eng. J.H.J. Allum, puts do not trigger balance corrections. Instead,
Department of Audiology and Neurootology, Universitatsspital Basel, CH-
4031 Basel, Switzerland. Voice: +41-61-265-2040; fax: +41-61-265-2750.
the role of vestibular-spinal inputs on balance
jallum@uhbs.ch corrections is to modulate the amplitude of

Basic and Clinical Aspects of Vertigo and Dizziness: Ann. N.Y. Acad. Sci. 1164: 112 (2009).
doi: 10.1111/j.1749-6632.2009.03872.x  c 2009 New York Academy of Sciences.

1
2 Annals of the New York Academy of Sciences

postural muscle responses,14 and is specific to tation concepts using biofeedback of body
the roll and pitch planes.5 In contrast, lower sway.23,24
leg proprioceptive feedback has long been con-
sidered critical for triggering and modulating
balance-correcting responses,69 although, re- Methods
cent evidence has indicated that proprioceptive
inputs from more proximal joints (such as the Subjects
hip joint) could fulfill this role.1012 For exam-
ple, when the support surface is servoed during For investigating movement strategies dur-
translation so that ankle rotations are negligi- ing quiet stance subjects included six VL pa-
ble and stretch reflexes in ankle muscles are tients (all male; mean age [range] 40.7 [21
absent,3,13 balance corrections are still present 48] years), six PL patients (three male; 57.5
and appropriately modulated to maintain sta- [2770] years) and 28 healthy, age-matched
bility. These results suggest that ankle proprio- controls (16 male; 46.2 [2869] years).33 Sim-
ceptive inputs are not always required for trig- ilar subject demographics and numbers were
gering and modulating balance corrections.13 used for experiments with support-surface per-
Furthermore, if proprioceptive inputs at other turbations.11 Criteria for subject selection are
joints are able to act as a substitute for the ab- listed in previous publications.1113,33 All sub-
sence of ankle inputs, then movements at other jects signed informed consent as approved by
joints need to have occurred, as has been ver- the ethical committees of the University Hos-
ified by observations of multisegmental move- pital Basel and the Radboud University Ni-
ment during both translations and rotations of jmegen Medical Center.
the support surface.10,16
Movements at joints other than at the ankle Procedure for Quiet Stance Experiments
during quiet stance or following a perturbation
to stance are essential if a distributed use of pro- Subjects stood without shoes with feet apart
prioceptive inputs is to contribute to balance at shoulder width and with their arms hanging
control.3 It has generally been assumed that at the sides of their body during two postural
the body sways primarily as an inverted pendu- tasks: standing on two legs with eyes closed on
lum during quiet stance with little movement 1) a firm support or 2) a foam support surface.
at the knee and hip joints.14,15 However, the The sequence was randomized for firm and
concept of movement as an inverted pendulum foam support surfaces to restrict influence of a
has recently been challenged, suggesting that training effect. Two spotters were positioned in
more than one mode of movement, including case balance was lost. All tasks lasted 3 min or
that like an inverted pendulum, occurs simul- until balance control was lost. In this case, the
taneously.17,18 Whether or not these modes of task was repeated; the longest of a maximum
control during quiet stance are affected by PL of three trials was used for analysis. However,
or VL is not known. Furthermore, these modes trials shorter than 20 sec were excluded from
of control have only been explored in the pitch the analysis.
plane. Two digital angular-rate gyroscopes
The aim of the current study was to com- (SwayStarTM , Balance International Innova-
pare the modes of control in healthy, VL, and tions GmbH, Switzerland) mounted on con-
PL subjects during quiet stance33 to those un- verted motorcycle kidney belts, were synchro-
derlying balance corrections to support-surface nized to measure angular velocity in the roll
perturbations.1113,27 Having established dif- (side-to-side) and pitch (fore-aft) planes at
ferences between these populations, our goal the pelvis and at the shoulder with a rate
was to apply this knowledge to rehabili- of 100 Hz. Pelvis and shoulder velocity data
Horlings et al.: Vestibular and Proprioceptive Contributions to Balance 3

were integrated to yield position data and were collected using a three-dimensional opti-
then partitioned into low-frequency and cal tracking system with 21 infrared-emitting
high-frequency sway, after removal of the very diodes (IREDs) (Optotrak, Northern Digi-
low-frequency trend using wavelet analysis. tal). Marker positions and joint angle anal-
The remaining swayoriginal data minus the ysis procedures are described in Visser and
low-frequency trendwas further analyzed colleagues.20
in two separate frequency bands. The first Head linear and angular accelerations were
band consisted of low-pass filtered sway with a computed from the outputs of four dual-axis
cutoff frequency of 0.7 Hz. The second band linear accelerometers (Entran), with ranges
consisted of high-pass filtered sway with a of 5 g, each mounted at 90 deg separation on
cutoff frequency of 3 Hz. Once these data were a lightweight, adjustable, tight head-band.5,11
low- or high-pass filtered, a weighted total All analog computed signals were sampled at 1
least-squares algorithm was used for fitting KHz.
a regression line to the shoulderpelvis x-y EMG recordings were obtained using pre-
plots. viously described techniques.4,5,11 To record
EMG signals, pairs of silversilver chloride elec-
Procedure for Perturbed Stance trodes were placed approximately 3 cm apart
along the muscle bellies of left tibialis anterior,
Balance control was examined in a simi- soleus, quadriceps, hamstrings, left and right
lar way as in previous studies.5,1120 Briefly, gluteus medius, left and right paraspinals at the
subjects stood barefoot with their feet lightly L1-L2 level of the spine, left biceps, and left
strapped into heel guides fixed to the surface of medial deltoid muscles.
a dual-axis rotating platform. Thus the subjects Following analog to digital conversion of the
used in-place balance-correcting strategy, not data, all biomechanical and EMG signals were
stepping reactions, to correct for the support- averaged offline across each perturbation direc-
surface tilt. Stance width was the same for all tion. Subject averages were pooled to produce
subjects (14 cm). Subjects were requested to population averages for a single direction. Zero
stand with their arms hanging by their sides. latency was defined as the onset of platform
Two handrails (adjusted to the height of the rotation measured with potentiometer systems
wrist of each subject) were located 40 cm from on platform axes.
the sides of the platform center. Two spotters
were positioned to lend support in case of a fall.
Perturbations to upright stance were deliv- Results
ered by rotating the platform randomly in one
of eight different directions with a constant ve- Movement Strategies during
locity of 60 deg/sec and constant amplitude Quiet Stance
of 7.5 deg. The perturbation directions were
forward (toes-down or 0 deg), backward (toes- In controls the low-frequency movements of
up or 180 deg), right (90 deg), left (270 deg) the shoulder and pelvis were found to be in
and four combinations of pitch and roll, in- phase in both the pitch and roll planes (Fig. 1B
cluding forward right (45 deg), backward right and G).33 When the pelvis and shoulder
(135 deg), backward left (225 deg), and forward movements were plotted against one another
left (315 deg). (Fig. 1C) control values for pitch slopes were
Recordings of all biomechanical and elec- approximately 52 deg, regardless of stance con-
tromyographical (EMG) data commenced 100 dition. This indicates an approximate 1-to-1
msec prior to perturbation onset and were movement of shoulder and pelvis, as expected
collected for 1 sec. Full body kinematic data from an inverted pendulum mode of sway. In
4 Annals of the New York Academy of Sciences

Figure 1. Modes of movement in the roll and pitch planes during quiet standing on foam
with eyes closed. The original traces of roll (A) and pitch (F) deviations of the shoulders and
the pelvis are shown for 50 sec of data with the trend of the drift estimated by a wavelet filter.
With the trend removed, the data was then low-pass filtered (at 0.7 Hz) to reveal deviations
shown in B for roll and G for pitch. A smaller section of the high-pass (3 Hz) filtered data is
shown in D for roll and I for pitch. Shoulder versus pelvis deviations over the complete 180
sec of each trial, corresponding to the traces shown in B, D, G, and I, are plotted in C, E, H,
and J, respectively as x-y plots. A regression line has been drawn through the x-y data. Note
the approximate in-phase movements of shoulder and pelvis in C and H for low-frequency
roll and pitch, respectively, and the out-of-phase movement in E and J for high frequencies of
sway. (Data reproduced from Ref. 33.)

the roll plane, greater shoulder movement rel- There was no difference in the pattern of
ative to pelvis movement was observed when low-frequency displacements of the shoulder
standing on firm compared to foam sup- and pelvis between controls and patients with
port surfaces (79.5 2.0 deg versus 60.2 VL or PL in either the pitch or roll planes.33 Al-
1.9 deg). though synchronization of the shoulderpelvis
Horlings et al.: Vestibular and Proprioceptive Contributions to Balance 5

(Fig. 2). In contrast, stance on foam surfaces


was controlled by antiphasic corrections in
the roll plane at both the shoulder and pelvis
(Figs. 1 and 2). There were no significant
differences between VL patients and con-
trols in the pattern of high-frequency displace-
ments of the shoulder and pelvis for either
plane or surface type. Significant high-
frequency differences were observed between
PL patients and controls during quiet stance
in the pitch plane.33 When standing on firm
and foam support, PL subjects pitch-plane
Figure 2. Eyes-closed movement-strategies corrections were accomplished almost ex-
changes as depicted by slopes of high-frequency
clusively through shoulder movements (with
modes of pelvis and shoulder motion. Changes
with support surface and with vestibular (VL) or slopes of pelvisshoulder plots being on
proprioceptive loss (PL) are apparent. The mean average 80 deg), implying that these subjects
and standard error of the mean of the slopes are attempted to keep their ankle joints as still as
shown by a column and vertical bar, respectively. possible. In contrast, PL patients had similar
As slopes of +90 and 90 are equal, we referred high-frequency patterns of shoulder and pelvis
all slopes to one sign (that of the majority) for
purposes of computing a mean. For roll, significant
displacements compared with controls in the
differences were noted with support surface. For the roll plane when standing on both firm and foam
higher-frequency pitch mode, significant differences support.
between PL and controls are marked ( ). There were
no significant differences in slope between VL and Movement Strategies following
control subjects, only amplitude differences. (Data Support-Surface Tilt: Unifying Findings
reproduced from Ref. 33.) on Vestibular Loss with Previous Findings
was unaffected by VL or PL, both groups had on Proprioceptive Loss
differences in the amplitude modulation of both During a pitch-plane, toe-up rotation of the
body segments during quiet stance compared to support surface, the body does not act as an
controls. Larger amplitude low-frequency dis- inverted pendulum. Instead the platform per-
placements of the shoulder and pelvis were ob- turbation induces-high frequency (>1 Hz) dis-
served in both VL and PL patients compared placement of the legs in the backward direction,
to controls in the pitch plane. Likewise, in the while the trunk is rotated forward to counter the
roll plane, larger amplitude low-frequency dis- backward displacement of the center of mass
placements were observed in PL subjects on (CoM).10 Compared with controls, VL sub-
a firm surface and in VL subjects on a foam jects responded to backward support-surface
surface. tilts with increased forward pitch of the trunk
High-frequency displacements in the shoul- followed by increased backward pitching of the
der and pelvis were controlled differently in trunk (Figs. 3 and 4).1,5,11 Underlying these
the pitch and roll planes.33 In the pitch plane, changes are, as shown in the lower part of
shoulder and pelvis displacements were uncor- Figure 4, an increased early activity in trunk
related and slightly antiphasic during stance on flexors followed by increased late activity in the
both firm and foam surfaces (Fig. 2) for VL trunk extensors, paraspinals. This instability is
and controls. In the roll plane, stance on a enhanced by decreased activity in the leg mus-
firm surface was dominated primarily by high- cles, tibialis anterior, and quadriceps, which
frequency corrections at the shoulder, with lit- thereby fail to rotate the body forward about
tle corrective movement observed at the pelvis the ankle joints.13
6 Annals of the New York Academy of Sciences

Figure 3. Stick figure representation of knee movements to right tilt of the support-surface
and trunk movements to backward tilt. A typical control is shown on the left and a typical VL
patient on the right. The 64 time frames recorded over 1sec, are indicated by the color code,
with blue colors marking the early frames and red colors the late frames. The support-surface
rotation of 7.5 deg started at frame 6. Note the differences in left (uphill) knee flexion and
right (downhill) knee extension between control and VL subjects.

Figure 4. Pitch and roll responses of control and VL subjects. The mean population traces
are shown for a pure left roll (270 deg direction) in the upper row of traces. Increased trunk
roll and uphill quadriceps responses but decreased hamstring responses were present for VL
patients. The lower row of traces shows responses to pure backward pitch (180 deg direction).
Increased early activity in external obliques and increased late activity in paraspinals can be
associated with the early increased pitch of the trunk and later backward-pitch trunk motion.
Horlings et al.: Vestibular and Proprioceptive Contributions to Balance 7

In contrast, differences that emerge with pro- to characterize the stimulus are intact (pelvis
prioceptive loss subjects are associated with de- muscle proprioceptive and head-roll accelera-
layed ankle torque and trunk displacements tion vestibular inputs). Indeed this appears to
with respect to controls.13 The delay in trunk be the case (Bloem and Allum, unpublished
motion was accompanied by a longer phase of observations, 2000).
backward trunk rotation compared to controls,
leading to instability. While there is a slight Effect of Trunk Sway Biofeedback
decrease in activity in tibialis anterior mus- on Balance Control
cles with PL, there is also decreased activity
in paraspinal muscles, and, with this pattern Given the similarity of balance-correcting
of activity, less instability compared with VL strategies between VL subjects and controls
patients.13 described above, it would seem logical that a
Normal responses to support-surface roll in- biofeedback system23 that aimed simply to re-
clude flexion of the uphill knee, extension of the duce sway would be effective for these two pop-
downhill knee, and trunk lateral rotation in the ulations. Presumably proprioceptive loss sub-
opposite direction to support-surface tilt. The jects would require a feedback device that
characteristic feature of an unstable reaction helped to change the strategy used to correct
to tilt in the roll direction is insufficient uphill balance instability. As Figure 5 shows, healthy
knee flexion and downhill knee extension.11,19 young and elderly subjects reduced their trunk
After the initial stimulus induced uphill trunk sway considerably when the source of feedback
roll, the trunk movement reverses direction in was trunk sway measured at the level of the
VL subjects.11 The motion of the CoM is in the lower trunk.24 The actual feedback is equiv-
same direction, and a fall is induced downhill alent to summed angular motion of the lower
(Fig. 4). This would explain the activity illus- trunk and pelvis for roll and the angular motion
trated in Figure 4, the uphill knee-flexing mus- of the lower trunk for pitch (Allum and Kueng,
cle, the hamstrings, responds more weakly than unpublished observations, 2008). The sway re-
normal after 200 msec, and the uphill knee- duction with feedback was greater than the re-
extending muscle, the quadriceps, extends the duction brought about by training alone.24,25
knee too much. Furthermore, as the plots in Figure 5 indi-
In contrast to VL subjects, a subject with cate, both young and elderly benefited from
total proprioceptive loss (TPL) in the legs the feedback, with those with the greatest sway
demonstrates a reversal of trunk roll to the prior to feedback use benefiting most. This im-
downhill direction far earlier.12 Two reasons plies that patients with balance disorders hav-
probably underlie this difference. First, the ing increased body sway due to predominantly
common characteristic of TPL is a change in vestibular problems would presumably benefit
the response synergy pattern with deactivation most from feedback use.
of trunk and leg muscle responses when nor- Figure 6 provides an example of such im-
mal responses are active.12 Second TPL sub- provement from ongoing trials with patients.
jects have enhanced background muscle activ- The patient in question had a unilateral
ity, which causes them to generally stiffen up. vestibular loss, which had not been compen-
Their trunk-roll responses are then very like sated (failure of low-frequency vestibulo-ocular
those of normal subjects whose bodies have reflex [VOR] gain to be normal) within the
been stiffened with a corset.21 It can well be normal recovery time of 3 months.25 The im-
imagined that patients with only lower leg pro- provement for the typical tests for which this
prioceptive loss (termed PL patients here) have type of patient is unstable26 was clearly present
less difficulty with rapid-roll tilts of the sup- after training with the vibrotactile feedback de-
port surface because sensory inputs required vice and using it to control sway during the
8 Annals of the New York Academy of Sciences

Figure 5. Relationship between pretest peak-to-peak trunk sway and the reduction in sway achieved
with feedback. Roll- and pitch-angle changes for two different tests are shown. The tests were standing feet
together eyes closed on a firm surface on the left (A, roll; C, pitch) and walking tandem steps eyes closed (B,
roll; D, pitch). Regression lines through the data have been drawn separately for elderly and young subjects.
Positive changes mean that sway was reduced with feedback (Data from Davis et al. 2008).

tests. In fact, measures of sway for the two tests (PL), we found that VL subjects suffer from
shown under feedback conditions were within a lack of appropriate modulation of their at-
normal bounds for the subjects age. tempted balance corrections. In contrast, PL
patients exhibited changed synergies in re-
Discussion sponse to imbalance. As we will expand on
below, this insight suggests it will be easier to
The aim of this paper was to compare strate- rehabilitate VL than PL subjects using artifi-
gies used in quiet stance to maintain a sta- cial feedback signals, because once the vestibu-
ble posture with those used when the sup- lar loss subjects are informed, via feedback, of
port surface is suddenly tilted. The strategies their inadequate response conditions, then pre-
during quiet stance could be split into low- sumably they would be able to modulate their
and high-frequency components.33 The high- responses adequately. On the other hand, pro-
frequency modes of movement during quiet viding feedback information alone may not en-
stance were multilink as were balance correc- able the PL subjects to change their response
tions to support-surface tilt. These latter bal- strategies.
ance corrections involved knee movements for The response strategies PL patients employ
both forward and lateral tilt, in addition to an- to control quiet stance appear to be different
kle, hip, and lumbosacral movements. When we from those correcting perturbed stance. Dur-
examined differences between patient groups ing quiet stance, PL subjects could execute slow
with vestibular loss (VL) or proprioceptive loss movements (those with frequencies less than
Horlings et al.: Vestibular and Proprioceptive Contributions to Balance 9

Figure 6. Improvements in trunk sway with biofeedback for an uncompensated unilateral


peripheral vestibular loss patient. The improvements for two different tests are shown; standing
on two legs eyes closed on foam (A and B) and walking eight tandem steps, eyes open in C
and D. On left the time traces with and without feedback are shown. To the right angle and
angular velocity plots before training and with no feedback compared with post training and
with feedback are shown as x-y plots of roll and pitch deviations. Notice the considerable
reduction with feedback. An envelope (the convex hull) has been drawn around the x-y
plots.

0.7 Hz) just like controls, moving the body responses, just as the elderly do,29 possibly be-
as an inverted pendulum. However for high- cause there are a number of other somatosen-
frequency (greater than 3 Hz) movements for sory signals that can replace the lower leg pro-
which relative movements between the lower prioceptive inputs. Only when the PL extends
and upper body segments are unavoidable, PL up the legs to encompass the knee and hip joints
subjects attempted to hold their legs as still as is the loss so profound that the changed re-
possible, a strategy that is different from that sponse strategies become destabilizing.12
used by VL and normal subjects. When, how- Traditionally, balance responses have re-
ever, the same high-frequency motion is im- ceived a different focus on two-legged humans
posed by stance perturbations as a result of and four-legged animals. Studies on the ef-
tilts of the support surface, PL subjects could fect of VL on humans have concentrated pri-
not avoid motion of the ankle joints.13 Further- marily on balance corrections that maintain
more, they need to adopt a response synergy pitch-plane stability.35,9,14,16,22 Animal studies
different from that of VL and healthy patients have focused more on responses to roll.28,30,31
because ankle stretch reflexes were absent, and In animal studies, it is generally accepted
therefore the biomechanical effect of these re- that vestibularspinal inputs contribute signif-
flexes, too. In fact, lower-leg PL patients adapt icantly to muscle responses flexing the uphill
well to the requirements for a change in the re- limbs and extending the downhill limbs.30,31 In
sponse strategy given delayed or absent reflex these studies it also appears that VL results in
10 Annals of the New York Academy of Sciences

lower-than-expected amplitudes of muscle ac- fective in improving balance control. We found


tivity, as if the CNS were responding to a tilt that that it was effective not only in the young but
was less than the actual tilt. Studies in humans also in the elderly and those with VL. These re-
have demonstrated a similar situation occurs sults imply that the CNS can utilize the types of
for roll and pitch in the leg muscles.5,11 That sensory information artificially provided here
is, leg muscle responses (such as ankle muscles) to optimize postural control. We have assumed
are hypometric as shown by the current and that this process occurs via a direct change
previous results.4,5,11 In contrast to the animal in the amplitudes of muscle activity. Whether
studies however, trunk and neck muscle activity this feedback can be used to change unstable
appears to be hypermetric in humans as shown balance-correcting strategies, as we have iden-
by the current and previous results.4,5,11 tified in proprioceptive loss and other patient
The reason for the difference between ani- groups (e.g. spinocerebellar ataxia19 ), remains
mal and human responses is probably biome- to be investigated.
chanical. For example, in humans when the The biofeedback signals provided in our de-
support surface tilts, the legs, pelvis, and center vice are all facilitated by supraspinal pathways:
of mass move in the direction of tilt, but the trigeminal nerve for vibrotactile informa-
the trunk and head tilt in the opposite direc- tion, the vestibular-cochlear nerve for auditory
tion.11,31 In animals all segments tilt downhill.28 and possibly otolithic stimulation, and at the ex-
This difference has three consequences. First, tremes of sway for the optic nerve visual infor-
in order to maintain stable stance in humans, mation. The proximity of these supraspinal to
proprioceptive information from both the an- cortical centers integrating sensory signals into
kle and lumbosacral joints is required. Second, balance commands may be beneficial in elim-
because the head movements to tilt are in the inating any errors in sensory integration that
opposite direction for bipedal stance of humans occur in spinal pathways, regardless of whether
and quadrupeds, the neural circuitry under- these are due to conduction delays in peripheral
lying the vestibulospinal-driven uphill knee nerves.
flexion response and downhill kneeextension The improvement we noted with biofeed-
responses must have opposite polarity too. back was proportional to initial balance perfor-
Third, if the same neurobiomechanical mech- mance of test subjects. That is, those with the
anisms underlie the control of quiet stance, largest initial balance deficits were most aided
then this would provide an explanation for by having biofeedback of their postural sway
the characteristics of low-frequency, inverted- available to them. It is possible that the modal-
pendulum mode of stance following VL. VL ity we used may have promoted this scalar im-
movements are characterized by greater shoul- provement. The visual feedback was set only to
der and pelvis motion in the pitch and roll be active at the extremes of stability for a given
planes, with both having greater amplitudes task and this rarely occurred, but when it did, it
than those of normals.32,33 However, the shoul- was very obvious to the subject. The vibrotac-
der motion coherent with the pelvis motion is tile feedback was set to be extremely sensitive to
greater than that of pelvis in VL subjects but less sway, and subjects immediately felt it. However,
in controls.33 This suggests that during quiet the auditory feedback was by nature more com-
stance excessive trunk muscle activity after VL plex as two signals need to be coded into four
tends to cause an overreaction of the trunk mo- directions of sway. By providing this feedback
tion instead of keeping it aligned with that of in the mid-range of sway and having its ampli-
the pelvis. tude continuously increase with sway, we hoped
The question arises whether, in groups with to overcome some of the processing problems
presumably near optimally functioning sensory that might occur if the auditory feedback was
modalities, biofeedback of trunk sway can be ef- constant. It may well be worthwhile to consider
Horlings et al.: Vestibular and Proprioceptive Contributions to Balance 11

replacing the auditory feedback with a graded 5. Carpenter, M.G., J.H.J. Allum and F. Honegger.
vibrotactile feedback. As vibrotactile feedback 2001. Vestibular influences on human postural con-
increases with intensity it is also perceived au- trol in combination of pitch and roll planes reveal
differences in spatio temporal processing. Exp. Brain
ditorily as a bone-conducted signal. Res. 140: 95111.
In this study we have explored differences 6. Nashner, L.M. 1976. Adapting reflexes controlling
in the balance-correcting strategies following the human posture. Exp. Brain Res. 26: 5972.
vestibular and proprioceptive loss. We have 7. Fitzpatrick, R., J.L. Taylor and D.I. McCloskey. 1992.
argued that there are common elements to Ankle stiffness of standing humans in response to im-
perceptible perturbation: reflex and task-dependent
these strategies for quiet and perturbed stance.
components. J. Physiol. 454: 533547.
The lack of adequate modulation of movement 8. Fitzpatrick, R. and D.I. McCloskey. 1994. Proprio-
strategies of balance instability following VL or ceptive, visual and vestibular thresholds for the per-
aging indicates that these are amenable to rec- ception of sway during standing in humans. J. Physiol.
tification using biofeedback that provides ad- 478: 173186.
ditional artificial sensory information on body 9. Horak, F.B., C.L. Shupert, V. Dietz and G.
Horstmann. 1994. Vestibular and somatosensory
sway. contributions to responses to head and body displace-
ments in stance. Exp. Brain Res. 100: 93106.
Acknowledgments
10. Carpenter, M.G., J.H.J. Allum and F. Honegger.
1999. Directional sensitivities of stretch reflexes and
The research reported here was supported
balance corrections for normal subjects in the roll
by a grant from the Swiss National Research and pitch planes. Exp. Brain Res. 129: 93113.
Foundation (No. 320000-117950) to J.H.J. Al- 11. Allum, J.H.J., L. Oude-Nijhuis and M.G. Carpenter.
lum. Some of the work described here draws 2008. Differences in coding provided by propriocep-
on findings established with our colleagues tive and vestibular sensory signals may contribute to
B.R. Bloem, J.R. Davis, U.M. Kueng, and L. lateral instability in vestibular loss subjects. Exp. Brain
Res. 1984: 391450.
Oude-Nijhuis whose contribution we gratefully 12. Bloem, B.R., J.H.J. Allum and M.G. Carpenter. 2002.
acknowledge. Triggering of balance corrections and compensatory
strategies in a patient with total leg proprioceptive
Conflicts of Interest loss. Exp. Brain Res. 142: 91107.
13. Bloem, B.R., J.H.J. Allum, M.G. Carpenter and F.
The authors declare no conflicts of interest. Honegger. 2000. Is lower leg proprioception essential
for triggering human balance corrections? Exp. Brain
Res. 130: 375391.
References 14. Nashner, L.M., F.O. Black and C. Wall III. 1982.
Adaptation to altered support and visual conditions
1. Allum, J.H.J. and C.R. Pfaltz. 1985. Visual and during stance in patients with vestibular deficits. J.
vestibular contributions to pitch sway stabilization Neurosci. 5: 536544.
in the ankle muscles of normals and patients with 15. Kuo, A.D. 1995. An optimal control model for ana-
bilateral peripheral vestibular deficits. Exp. Brain Res. lyzing human postural balance. IEEE Trans. Biomed.
58: 8290. Eng. 42: 87101.
2. Allum, J.H.J., F. Honegger and H. Schicks. 1994. The 16. Runge, C.F., C.L. Shepert, F.B. Horak and F.E. Zajac.
influence of a bilateral peripheral vestibular deficit on 1998. Role of vestibular information in initiation of
postural synergies. J Vestib Res 4: 4970. rapid postural responses. Exp. Brain Res. 122: 403
3. Allum, J.H.J. and F. Honegger. 1998. Interactions 412.
between vestibular and proprioceptive signals in trig- 17. Aramaki, Y., D. Nozaki, K. Masani, et al. 2001. Re-
gering and modulating human balance-correcting re- ciprocal angular acceleration of the ankle and hip
sponses differ across muscles. Exp. Brain Res. 121: joints during quiet standing in humans. Exp. Brain
478494. Res. 136: 463473.
4. Keshner, E.A., J.H.J. Allum and C.R. Pfaltz. 1987. 18. Creath, R., T. Kiemel, F. Horak, et al. 2004. A uni-
Postural coactivation and adaptation in the sway sta- fied view of quiet and perturbed stance: simultaneous
bilizing responses of normals and patients with bilat- co-existing excitable modes. Neurosci. Let. 377: 75
eral vestibular deficit. Exp. Brain Res. 69: 7792. 80.
12 Annals of the New York Academy of Sciences

19. Bakker, M., J.H.J. Allum, J.E. Visser, et al. 2006. Pos- recovery from an acute unilateral peripheral vestibu-
tural responses to multi-directional stance perturba- lar deficit. Audiol. Neuro-Otology 8: 286302.
tions in cerebellar ataxia. Exp. Neurol. 202: 2135. 27. Horak, F.B., L.M. Nashner. 1986. Central program-
20. Visser, J.E., J.H.J. Allum, M.G. Carpenter, et al. 2008. ming of postural movements: adaptation to altered
Subthalamic nucleus stimulation and levodopa- support-surface configurations. J. Neurophysiol. 55:
resistant postural instability in Parkinsons disease. 13691381.
J. Neurol. 255: 205210. 28. MacPherson, J.M., D.G. Everaert, P.J. Stapley and
21. Gruneberg, C., J.H.J. Allum, F. Honegger and B.R. L.H. Ting. 2007. Bilateral vestibular loss in cats leads
Bloem. 2004. The influence of artificially increased to active destabilization of balance during pitch and
hip and trunk stiffness on balance control in the pitch roll rotations of the support-surface. J. Neurophysiol.
and roll planes. Exp. Brain Res. 157: 472485. 97: 43574367.
22. Allum, J.H.J., F. Honegger and H. Acuna. 1995. 29. Allum, J.H.J., M.G. Carpenter, B.R. Bloem, et al.
Differential control of leg and trunk muscle activity 2002. Age-dependent variations in the directional
by vestibulo-spinal and proprioceptive signals dur- sensitivity of balance corrections. J. Physiol. (Lond.)
ing human balance corrections. Acta Otol. Laryngol. 542: 643663.
(Stockh) 115: 124129. 30. Pompeiano, O. 1984. Excitatory and inhibitory influ-
23. Allum, J.H.J., J.R. Davis, M.G. Carpenter, et al. 2008. ences on the spinal cord during vestibular and neck
Neue Ansatze zur Sturzpravention mittels einer reflexes. Acta Otolaryngol. (Stockh) Suppl. 406: 59.
multi-modalen Gleichgewichtsprothese. In Neues aus 31. Peterson, B.W., K. Fukushima, N. Hirgi, et al. 1980.
Forschung und Klinik 6. Hennig Symposium: Der Gle- Responses of vestibulospinal and recticulospinal neu-
ichgewichtssinn. H. Scherer, Ed.: 211221. Springer rons to sinusoidal vestibular stimulation. J. Neurophys-
Verlag. Berlin. iol. 43: 12361250.
24. Davis, J.R., M.G. Carpenter, R. Tschanz, et al. 2008. 32. Horlings, C.G.C., U.M. Kueng, F. Honegger, et al.
Trunk sway reduction in the young and elderly us- 2008. Identifying deficits in balance control follow-
ing multi-modal biofeedback to postural equilibrium. ing vestibular or lower leg proprioceptive loss using
Submitted. posturographic analysis of stance tasks. Clin. Neuro-
25. Allum, J.H.J. and T. Ledin. 1999. Recovery of physiol. doi:10.1016/j.clinph.2008.07.221.
vestibulo-ocular function in subjects with acute pe- 33. Horlings, C.G.C., U.M. Kueng, F. Honegger,
ripheral vestibular loss. J. Vest. Res. 9: 135144. et al. 2008. Vestibular and proprioceptive influ-
26. Allum, J.H.J. and A.L. Adkin. 2003. Improvements in ences on movement strategies during quiet standing.
trunk sway observed for stance and gait tasks during Submitted.

You might also like