Balance in a rotating artificial gravity environment

Exp Brain Res (2003) 148:266–271
DOI 10.1007/s00221-002-1300-9
Kazuhiro Soeda · Paul DiZio · James R. Lackner
Balance in a rotating artificial gravity environment
Received: 26 April 2002 / Accepted: 20 September 2002 / Published online: 22 November 2002
Springer-Verlag 2002
Abstract When subjects stand at the center of a fully
enclosed room that is rotating at constant velocity, their
natural postural sway generates Coriolis forces that
destabilize their center of mass and head. We quantitatively assessed how exposure to constant velocity rotation
at 10 rpm affected postural control. Twelve subjects stood
in a heel-to-toe stance in the rotating room. Each test
session involved three phases: (1) pre-rotation, (2) perrotation, and (3) post-rotation. In each phase, subjects
were tested in both eyes open and eyes closed conditions.
Four measures were used to characterize center of mass
movement and head movement: mean sway amplitude,
total power, mean power frequency, and frequency of
maximum power. Each measure was computed for
anterior-posterior and medial-lateral sway. Both anterior-posterior and medial-lateral head and center of mass
sway during rotation had significantly greater mean sway
amplitude and total power compared with pre- and postrotation values. Mean power frequency and frequency of
maximum power were little affected. Eyes open conditions were significantly more stable in all test phases than
eyes-closed, but vision did not completely suppress the
effects of rotation. The greatest effect of rotation was in
the eyes-closed condition with mean sway amplitude and
total power increasing more than twofold. Inverted
pendulum sway was maintained in all phases of both test
conditions. No aftereffects of rotation were present after
the four 25-s exposures each subject received. We expect
that with longer exposure periods and with active
generation of body sway subjects would both adapt to
rotation and exhibit post-rotary aftereffects.
K. Soeda · P. DiZio ()) · J.R. Lackner
Ashton Graybiel Spatial Orientation Laboratory and Volen Center
for Complex Systems, Brandeis University, Waltham,
MA 02454-9110, USA
Tel.: +1-781-7362033
Fax: +1-781-7362031
P. DiZio
Ashton Graybiel Spatial Orientation Laboratory, MS033,
Brandeis University, Waltham, MA 02454-9110, USA
Keywords Posture · Perturbation · Rotation · Vision ·
Vestibular · Artificial gravity
A prolonged interplanetary space flight to Mars may well
require the use of artificial gravity to prevent adverse
physiological changes. Exercise with bungee cord loading
or ergocycles in weightless conditions is not an adequate
countermeasure to prevent loss of bone and muscle mass
(Vernikos 1996). Living in a continuously rotating space
habitat and exposure to periodic rotation in a short-radius
centrifuge (Lackner and DiZio 2000a, 2000b; Young
1999) are effective ways to generate “artificial gravity”
but ones that have potentially adverse side effects,
especially in short-radius devices turning at above
10 rpm (Nicogossian and McCormack 1987; Stone and
Letko 1965). The side effects include the generation of
Coriolis forces, g-gradients and changes in effective
weight during locomotion (see Lackner and DiZio 2000a,
2000b for recent reviews).
Our concern here is with Coriolis forces and their
influence on postural control during rotation. Coriolis
forces (CF) are generated by movements made within a
rotating reference frame: CF=–2m(wv), where m is the
mass of the moving object, w is the angular velocity of the
rotating environment, and v is the linear velocity of the
object relative to the plane of rotation. Figure 1 shows the
Coriolis forces that would be generated on a subject’s
body by lateral postural sway in a counterclockwise,
rotating environment. If the subject sways rightward, a
Coriolis force will be generated acting in the backward
direction with a magnitude proportional to sway velocity
in the plane of room rotation. This force would act both
on the subject’s body, tending to displace its path, and on
the sensory organs of the inner ear, altering the normal
relationship between afferent otolith signals and body
motion/orientation. If postural sway occurs that includes
tilt of the head, the semicircular canals will receive crosscoupled stimulation from the rotating environment. Other
Visual input has an attenuating effect on postural sway in
non-rotating conditions (e.g., Allum and Pfaltz 1985) and
we expected conditions involving vision to be more stable
than eyes closed conditions. With passive stance and the
brief exposure periods involved (25 s), we did not expect
significant adaptive changes in postural control to occur
during rotation.
Materials and methods
Fig. 1 Schematic illustration of the influence of Coriolis forces on
postural control in the experimental situation. While a subject
stands at the center of the room rotating counterclockwise (w),
rightward body sway (solid double arrow) induces a backward
Coriolis force (broken double arrow) and leftward sway generates a
forward Coriolis force (single arrows)
directions of sway would generate orthogonal, velocity
dependent Coriolis perturbations of body displacement
and Coriolis, cross-coupled perturbations of the semicircular canals. In a rotating environment if the subject is
standing at the axis of rotation, at 10 rpm, there are no
significant centripetal forces acting on his or her body.
In the 1960s an extensive series of studies was carried
out to assess how well humans could adapt to different
rates of rotation in a slow rotation room. The primary
concern in these studies was the disorientation and motion
sickness elicited by head movements during rotation. The
rotation rates studied ranged from about 2.8 rpm to 10 rpm
and one of the studies lasted as long as 3 weeks, the others
lasting days or 2 weeks (Graybiel et al. 1960, 1965;
Guedry et al. 1964; Kennedy and Graybiel 1962). Several
tests of posture, balance, and locomotion, e.g., the FreglyGraybiel ataxia and rail tests, were carried out pre- and
post-rotation. These tests showed post-rotation degradations in performance following days of exposure to
rotation (cf. Fregly 1974, for a review). The technology
was not available at the time to obtain quantitative
assessments of how body sway was affected during
rotation and the pre- and post-rotation measurements also
did not involve quantitative characterization of sway and
locomotion kinematics.
Our goal here was to evaluate postural sway during
exposure to constant velocity rotation at 10 rpm. This is
the highest velocity likely to be employed to generate an
artificial gravity environment for unconstrained activities
in a long duration space mission although higher rates
might be used for intermittent centrifugation of restrained
subjects (e.g., Young 1999). We wanted to characterize
the motion of the body during quiet stance. If sway is
about the ankles, the linear velocity of the head will be
greater than that of the center of mass; consequently the
Coriolis acceleration will be larger at the head and the
otolith organs than at the body center of mass. Therefore,
we were especially interested in whether inverted pendulum sway would be maintained during rotation (Nashner
1971; Nashner and McCollum 1985; Stockwell 1983).
Twelve subjects (eight men and four women between the ages of 25
and 44 years) participated after giving informed consent to a
protocol approved by the Brandeis Committee for the Protection of
Human Subjects. They were without vestibular or sensorimotor
impairments that could have influenced their performance; all had
normal or corrected-to-normal vision.
Figure 1 shows the configuration of the test situation. Subjects
stood at the center of the slow rotation room (SRR), located in the
Ashton Graybiel Spatial Orientation Laboratory at Brandeis
University. The position of two IRED (infrared) markers attached
to the subject’s bodys were recorded by a WATSMART motion
monitoring system (Northern Digital, Inc.) mounted in the SRR.
Sampling rate was 100 Hz. One IRED was taped to the subject’s
chin as a head (HD) marker. The other IRED was mounted on a
waistband worn by the subject, positioned at a point approximately
at the level of the center of mass (CM) of the subject’s body. A
three-sided safety railing was available for subjects to grasp if they
felt they were going to fall.
Subjects were tested standing in stocking feet in a heel-to-toe
stance (tandem Romberg stance) at the center of the SRR. This
stance was maintained before, during, and after constant velocity
rotation at 10 rpm of the SRR in a counterclockwise direction. Each
subject performed four 25-s trials in three periods: pre-rotation, perrotation, and post-rotation. Subjects alternated in a counterbalanced
order between eyes-open and eyes-closed conditions. Acceleration
to constant velocity and deceleration to rest were at 1/s2. The perand post-rotation measurements were made 2 min after velocity
changes to allow potential semicircular canal activity to abate. In
the eyes-open conditions, subjects had full view of the rich visual
environment of the interior of the SRR and were given a target on
the wall of the SRR to fixate. For the eyes-closed conditions,
subjects were instructed to close their eyes in the illuminated room.
They were also instructed to right themselves as quickly as possible
by touching the safety rails if they felt themselves losing balance in
the middle of a trial. We kept the SRR illuminated as a safety
precaution so that we could respond if subjects started to lose their
balance. Subjects were allowed brief rests between trials; however,
they remained in place, leaning against the back bar of the safety
rails and holding the side bars for support. There were two on-board
experimenters; one operated the recording console and communicated with the external SRR control room; the other was positioned
behind the test subject to stabilize him or her in case of extreme
Data reduction
The sampled data were digitally low pass filtered (10 Hz) and the
first and last 2.5 s were trimmed off to eliminate transients, leaving
20 s for analysis. Computer algorithms were used to identify the
MSA (mean sway amplitude), TP (total power), FMP (frequency of
maximum power), and MPF (mean power frequency) of head (HD)
and center of body mass (CM) motions, in the anterior-posterior (AP) and medial-lateral (M-L) directions. The formula for MSA was:
jxi xj
N i¼1
xi forx ¼ CMAP ; CMML ; HDAP or HDML
N i¼1
The remaining three measures were computed on the results of a
fast Fourier transform (N=2,000 points at a 100 Hz sample rate) of
the detrended, windowed (Hanning window) times series. A binary
search algorithm found the frequency bin containing the maximum
squared spectral amplitude, or frequency of maximum power
(FMP). The equation for total power was:
TP ¼
where the xi are spectral amplitudes per frequency bin. The mean
power frequency was calculated according to:
x2i fi
where the fi are frequency bins.
We also evaluated the relationship of HD and CM sway in two
ways. The cross-correlation and time lag between sway of HD and
CM were computed. In our nomenclature, positive time lags mean
that the second variable of a pair is leading the first temporally. In
addition, the ratio of HD to CM sway amplitude was calculated.
High correlations and short time lags along with similar ratios of
the sway amplitude and height of HD relative to CM would indicate
inverted pendulum sway.
Statistical analysis
Our a priori questions were: (1) how rotation affects sway, (2)
whether differences between eyes-open and eyes-closed conditions
would be present, and (3) whether inverted pendulum sway would
be maintained during rotation. A repeated measures MANOVA was
performed to assess the effects on MSA, TP, FMP and MPF of five
factors: body segment (HD and CM), direction of sway (A-P and
M-L), rotation (pre-, per-, and post-), vision (eyes open and eyes
closed) and the order of trials. There was no multivariate effect of
trial order on the measures of sway, but there were significant main
effects of the body segment (F(4,8)=5.82, P=0.017), sway direction
(F(4,8)=5.29, P=0.022), rotation (F(8,40)=2.94, P=0.011) and vision
(F(4,8)=8.30, P=0.006). Post hoc pairwise comparisons, collapsed
across repetitions, were done with multiple, Bonferroni corrected ttests. The corrected significance level for all pairwise comparisons
reported below is 0.05.
sway amplitude and total power were greater pre-rotation
with the eyes closed than open, and their increases perrotation were greater for eyes closed conditions. The
destabilizing effects of rotation and eye closure were
more prevalent in the M-L than the A-P direction. Sway
frequency was higher with the eyes closed than open. The
following sections describe the effects of rotation and
vision on each dependent measure of HD and CM sway in
both directions.
Mean sway amplitude
Univariate tests indicated that rotation significantly
affected MSA (F(2,22)=3.84, P=0.037). Medial-lateral
MSA was more affected than A-P (F(1,11)=6.29,
P=0.029). In pairwise comparisons, per-rotation values
were significantly greater than pre-rotation for mediallateral CM sway and for both directions of HD sway. Preand post-rotation values of MSA did not differ significantly from one another. The pattern of significant
rotation effects on MSA was the same in both visual
conditions, but the magnitude of the visual effect was
significantly greater with eyes closed than open
(F(1,11)=20.91, P<0.0001). The per-rotation increase was
about twofold with eyes closed and about 50% with eyes
Total power
Total power (TP) of HD and CM approximately tripled
during rotation in the eyes closed condition. There was a
significant increase in TP during rotation (F(2,22)=7.13,
P=0.0041), and TP of HD sway was greater than of CM
sway (F(1,11)=11.65, P=0.0058). In pairwise comparisons
of HD sway, the per-rotation TP was greater than prerotation for both sway directions and both visual conditions, but the per-rotation increase in CM was only
significant for TP of M-L sway with eyes closed. Pre- and
post-rotation values did not differ significantly.
Frequency of maximum power
This measure also showed a tendency toward an increase
during rotation from about 0.1–0.11 Hz pre-rotation to
about 0.13 Hz on average, but the effect of rotation was
not significant. The univariate tests indicated that the peak
frequency of sway was higher in eyes closed than eyes
open conditions when the data were collapsed across
rotation conditions, body segments and sway direction
(F(1,11)=5.56, P=0.038).
The results are summarized in Fig. 2. MSA and TP
increased during rotation relative to the pre-rotation
baseline and returned to baseline post-rotation. Mean
Mean power frequency
Univariate tests showed that MPF of sway did not change
significantly as a function of either rotation or vision.
Fig. 2A–D Mean sway amplitude, total power, frequency of
maximum power, and mean
power frequency, of head and
center of mass movement in the
medial-lateral (M-L) and anterior-posterior (A-P) directions
in eyes open and eyes closed
trials in the three rotational
conditions (pre-, per-, and postrotation). The bar graphs represent the means and standard
deviations calculated across
subjects, N=12
Loss of balance
The experimenter counted the number of times the subject
had to use the railings in order to avoid losing balance.
Each contact involved a momentary push against the rail
with the back of the hand. Subjects had to use the rails
more frequently during rotation than pre- or post-rotation,
and more frequently when their eyes were closed than
open. Pre-rotation there were 0.25 touches per trial in the
dark and none with vision; per-rotation there were 5.04
touches per trial in the dark and 1.16 when the subject’s
eyes were open; post-rotation there were 0.125 touches
per trial in both visual conditions.
Inverted pendulum sway
There was no significant effect of rotation on either the
time lag or coefficient of correlation between the head
and center of mass. There was a slight lead of CM
fluctuations relative to HD sway in the pre-rotation
period. The time leads across the four combinations of
sway direction and vision conditions ranged from 53 ms
(eyes closed, M-L sway) to 9 ms (eyes open, A-P sway).
Per-rotation the timing ranged from a 47-ms lead of CM
relative to HD (eyes closed, A-P) to a 16-ms lag of CM
(eyes closed, M-L sway). The lead/lag values in all
conditions were not significantly different from zero, and
there were no differences between pre- and per-rotation
values. In other words, sway of the HD and CM remained
in phase. The minimum and maximum correlation
coefficients were very high, ranging between r=0.95
(eyes closed, A-P) to 0.69 (eyes open, M-L sway) prerotation and 0.95 (eyes closed, A-P) to 0.77 (eyes open,
M-L) per-rotation. The correlation values did not differ
across rotation conditions.
The average ratio of the height of the HD marker
relative to the CM marker was 1.51 (SD 0.06), which
represents the expected ratio of HD to CM sway in
inverted pendulum sway. Pre-rotation, the ratios of HD to
CM sway amplitude, averaged across the A-P and M-L
directions, were 1.36 (SD 0.07) and 1.39 (SD 0.09) in the
eyes open and eyes closed conditions, respectively.
During rotation, the ratios increased to 1.49 (SD 0.10)
in the eyes open condition and 1.46 (SD 0.14) in the eyes
closed condition. The pre-rotation ratios of HD to CM
sway are significantly smaller than the expected ratio for
inverted pendulum sway (P<0.05, Bonferroni adjusted ttests), but the per-rotation ratios are not significantly
Our experimental findings indicate a significant degradation of postural control during passive rotation at 10 rpm.
Mean sway amplitude and total power of sway are
primarily affected with frequency measures being little
affected. As anticipated, allowing visual input significantly stabilized the body relative to eyes closed conditions but did not completely suppress the effects of
rotation. Lateral sway amplitude and spectral power were
greater pre- and post-rotation than anterior-posterior sway
in the eyes closed conditions, which was expected given
that the heel-to-toe stance used is inherently more
unstable laterally. During rotation, the lateral head and
center of mass mean sway amplitude and total power
increased more than the anterior-posterior direction,
reflecting the Coriolis forces and the cross-coupled
vestibular stimulation generated by sway. The destabilizing effects of rotation are underestimated in this study
because subjects used the hand rails to limit their body
Despite the increased amplitude of sway during
rotation, inverted pendulum sway was maintained as
closely as pre-rotation. The magnitudes of head and center
of mass sway were similarly affected by rotation in both
sway directions and in both visual conditions. In each
sway direction, HD and CM sway tended to be phase
locked, pre- and per-rotation. Eyes closed conditions, pre-,
per-, and post-rotation most closely approximate inverted
pendulum sway.
An important feature of the results is the absence of a
difference between post-rotation and pre-rotation sway
amplitude and frequency. This absence of aftereffects
means that significant adaptation to the Coriolis forces
generated by sway did not occur during the four, 25-s
rotation exposure periods. The lack of a difference
between trial repetitions within each rotation period is
also evidence that adaptation did not occur.
In other studies involving pointing movements to
targets, we have shown that subjects initially are unable to
execute accurately their intended arm movement during
rotation. However, when permitted to make repeated
movements, subjects quickly become more accurate and
are able to perform near normally within 10–15 reaches
(Lackner and DiZio 1994). Such adaptation means that
the subjects’ nervous systems have planned anticipatory
muscle innervations which cancel the consequences of the
Coriolis forces. Aftereffects are present post-rotation with
movement paths being mirror image to the initial ones
during rotation.
We have also shown that head movements made in a
rotating room are deviated by Coriolis forces (Lackner
and DiZio 1998) and adaptation is possible if the subject
makes the same head movement over and over (DiZio and
Lackner 1995). Head movement adaptation is slower than
arm adaptation, and the present findings indicate that
postural adaptation may be slower still. It is likely that
adaptation would occur with longer than four 25-s
exposure periods. In an early study, subjects living in a
rotating room for a period of 12 days showed postural and
locomotory aftereffects, indicating they had adapted to
the rotating environment (cf. Fregly 1974; Graybiel et al.
1965). The aftereffects were strongest on the walking test,
weak and transient in the test of standing with the eyes
closed, and non-existent when quiet stance with eyes open
was evaluated. The slow rate of adaptation of head
movements and postural sway to rotation may be because
these movements generate aberrant vestibular signals
which are not generated by arm movements. The aberrant
vestibular signals are from linear Coriolis forces on the
otoliths and cross-coupling of angular velocity of the
semicircular canals.
The bizarre vestibular signals generated by head and
body sway would make it more difficult for the nervous
system to compensate for Coriolis perturbations for
several reasons. For example, the center of mass of a
simple inverted pendulum in the rotating room would
precess in an oblate cycloidal pattern due to Coriolis
forces. However, attempting to stabilize such a system
with feedback from vestibular signals would instead
create a more chaotic motion because such feedback is
aberrant in a rotating environment. Such vestibular
signals would also affect gaze stability, making visual
information less helpful.
The rotating room paradigm for perturbing posture
makes it possible to experimentally manipulate the
amount and direction of vestibular, visual and other
sensory feedback relative to the Coriolis forces generated
on the body. Coriolis forces on a body segment moving in
a particular way are governed by rotation speed of the
environment, but the Coriolis, cross-coupled vestibular
signals about a particular body motion at a particular
rotation speed can be experimentally manipulated by
taking advantage of the dynamics of sensory systems and
central sensory interactions (Guedry and Benson 1978;
Watanuki et al. 2000). In future research, it will be
important to study the patterns of initial disruption of
posture and of adaptive recovery using different combinations of exposure schedules and sensory feedback. With
passive stance, adaptive corrections may be more reflexive without the need for conscious intervention and
planning. Adaptation likely requires the specification of
movement goals (path, trajectory, and kinematics) and
efferent/afferent feedback about the ongoing state of the
limb. For example, we are finding rapid adaptation of leg
trajectory during treadmill locomotion in the SRR. We
expect that if subjects voluntarily sway during rotation,
they will exhibit adaptive accommodation to the resulting
Coriolis forces. We further expect that such adaptation
will show substantial if not complete transfer to passive
stance. These possibilities are currently being explored.
Acknowledgement This research was supported by NASA grant
Allum JHJ, Pfaltz CR (1985) Visual and vestibular contributions to
pitch sway stabilization in the ankle muscles of normals and
patients with bilateral peripheral vestibular deficits. Exp Brain
Res 58:82–94
DiZio P, Lackner JR (1995) Effects of Coriolis, cross-coupled
stimulation on head movement control. Soc Neurosci Abstr
Fregly AR (1974) Vestibular ataxia and its measurement in man.
In: Kornhuber HH (ed) Handbook of Sensory Physiology,
vol VI. Springer, New York, pp 321–360
Graybiel A, Clark B, Zarriello JJ (1960) Observations on human
subjects living in a “slow rotation room” for periods of two
days. Arch Neurol 3:55–73
Graybiel A, Kennedy RS, Knoblock EC, Guedry FE Jr, Mertz W,
McLeod MW, Colehour JK, Miller EF II, Fregly AR (1965)
The effects of exposure to a rotating environment (10 rpm) on
four aviators for a period of twelve days. Aerospace Med
Guedry FE, Benson AJ (1978) Coriolis, cross-coupling effects:
disorienting and nauseogenic or not? Aviat Space Environ Med
Guedry FE, Kennedy RS, Harris CS, Graybiel A (1964) Human
performance during two weeks in a room rotating at three RPM.
Aerospace Med 35:1071–1082
Kennedy RS, Graybiel A (1962) Symptomatology during prolonged exposure in a constantly rotating environment at a
velocity of one revolution per minute. Aerospace Med 33:817–
Lackner JR, DiZio P (1994) Rapid adaptation to Coriolis force
perturbations of arm trajectory. J Neurophysiol 72:299–313
Lackner JR, DiZio P (1998) Adaptation in a rotating artificial
gravity environment. Brain Res Rev 28:194–202
Lackner JR, DiZio P (2000a) Human orientation and movement
control in weightlessness and artificial gravity environments.
Exp Brain Res 130:2–26
Lackner JR, DiZio P (2000b) Artificial gravity as a countermeasure
in long duration spaceflight. J Neurosci Res 62:169–176
Nashner LM (1971) A model describing vestibular detection of
body sway. Acta Otolaryngol 72:429–436
Nashner LM, McCollum G (1985) The organization of human
postural movements: a formal basis and experimental synthesis.
Behav Brain Sci 8:135–172
Nicogossian AE, McCormack PD (1987) Artificial gravity – a
countermeasure for zero-gravity. IAF/IAA-87–533, Proceedings of the 38th Congress of the International Astronautical
Federation, Brighton, UK
Stockwell CW (1983) Conceptual models of human postural
control. In: Vestibular and visual control on posture and
locomotor equilibrium. 7th Int Symp Int Soc Posturography,
Houston, TX, pp 22–28
Stone RW, Letko W (1965) Some observations on the stimulation
of the vestibular system of man in a rotating environment. The
role of the vestibular organs in the exploration of space. NASA
SP-77, pp 263–278
Vernikos J (1996) Human physiology in space. Bioessays 18:1029–
Watanuki K, Takahashi M, Ikeda T (2000) Perception of
surrounding space controls posture, gaze, and sensation during
Coriolis stimulation. Aviat Space Environ Med 71:381–387
Young LR (1999) Artificial gravity considerations for a Mars
exploration mission: otolith function in spatial orientation and
movement. Ann N Y Acad Sci 871:367–378