Exp Brain Res (1998) 120:202±216
RESEARCH ARTICLE
I.B.M. van der Fits ´ A.W.J. Klip ´ L.A. van Eykern
M. Hadders-Algra
Postural adjustments accompanying fast pointing movements in standing,
sitting and lying adults
Received: 9 April 1997 / Accepted: 9 October 1997
Abstract The present study evaluated the effect of differ-
ent positions, which varied in the amount of bodily sup-
port, on postural control during fast pointing movements.
Fourteen adult subjects were studied in standing, various
sitting and lying positions. Multiple surface electromyo-
grams (EMGs) of arm, neck, trunk and upper leg muscles
and kinematics were recorded during a standard series of
unilateral arm movements. Two additional series, consist-
ing of bilateral arm movements and unilateral arm move-
ments with an additional weight, were performed to assess
whether additional task-load affected postural adjustments
differently in a sitting and standing position. Two pointing
strategies were used ± despite identical instructions. Seven
subjects showed an elbow extension throughout the move-
ments. They used the deltoid (DE) as the prime mover (DE
group). The other seven subjects performed the movement
with a slight elbow flexion and used the biceps brachii
(BB) as the prime mover (BB group). The two strategies
had a differential effect on the postural adjustments: pos-
tural activity was less and substantially later in the BB-
group than in the DE group. Anticipatory postural muscle
activity was only present in the DE group during stance. In
all positions and task-load conditions the dorsal postural
muscles were activated before their ventral antagonists.
The activation rate, the timing and ± to a lesser extent ±
the amplitude of the dorsal muscle activity was position
dependent. The position dependency was mainly found
in the caudally located lumbar extensor (LE) and ham-
strings (HAM) muscles. The EMG amplitude of LE and
HAM was also affected by body geometry (trunk and pel-
vis position). Position and body geometry had only a mi-
I.B.M. van der Fits ´ A.W.J. Klip ´ L.A. van Eykern
M. Hadders-Algra
Department of Medical Physiology, University of Groningen,
Groningen, The Netherlands
)
M. Hadders-Algra ( )
University Hospital Groningen, Developmental Neurology,
CMC-IV, 3rd floor, Hanzeplein 1, 9713 GZ Groningen,
The Netherlands
e-mail: m.hadders-algra@med.rug.nl,
Tel.: +31-50-3614247, Fax: +31-50-3636905
 Springer-Verlag 1998
nor effect on the activity of the neck and thoracic extensor
muscles. This difference in behaviour of lower and upper
postural muscles suggests that they could serve different
postural tasks: the lower muscles being more involved in
keeping the centre of mass within the limits of the support
surface, and the upper ones in counteracting the reaction
forces generated by movement onset. Increasing task-load
by performing bilateral movements and ± to a minor extent
± during loaded unilateral movements affected the tempo-
ral and quantitative characteristics of the postural adjust-
ments during standing and sitting in a similar way. The ef-
fect was present mainly during the early part of the re-
sponse (within 100 ms after prime mover onset). This sug-
gests that feedforward or anticipatory mechanisms play a
major role in the task-specific modulation of postural ad-
justments.
Key words Anticipatory postural adjustments ´
Arm movements ´ Load ´ Standing ´ Sitting
Introduction
Movements are accompanied by a complex of fine-tuned
postural adjustments in order to maintain equilibrium.
The origin of these postural adjustments is twofold. First,
the movement causes a change in body geometry. This
may result in a shift of the centre of gravity (COG), which
causes a disequilibrium. Second, the forces generated to
initiate a movement induce reaction forces in the support-
ing segments of the body (Massion 1992).
The majority of studies on the nature of postural con-
trol during voluntary movements have been carried out in
standing adults performing various types of arm move-
ments (e.g. Lee 1980; Bouisset and Zattara 1981; Cordo
and Nashner 1982; Friedli et al. 1984; Horak et al.
1984; Brown and Frank 1987; Lee et al. 1987; Gurfinkel
et al. 1988; Aruin and Latash 1995). From these studies it
can be concluded that the basic organization of postural
adjustments can be described by three parameters: a spa-
tial, temporal and quantitative one. The spatial distribu-

tion of postural muscle activation is guided by direction
specificity. Forward-directed movements produce a for-
ward shift of the COG, which results in a primary activa-
tion of the postural muscles at the dorsal side of the body.
On the other hand, movements inducing a backward shift
of the COG evoke an activation of the ventral postural
muscles (Cordo and Nashner 1982; Friedli et al. 1984;
Aruin and Latash 1995). The temporal parameter is in-
volved in the timing of postural muscle activation. In spe-
cific conditions, such as during fast movements, the pos-
tural muscles are activated in an anticipatory manner, that
is, prior to the muscle that initiates the movement. The
timing of postural muscle activation is related to, for in-
stance, arm movement velocity (Horak et al. 1984; Lee
et al. 1987; Massion 1992), the predictability of the forth-
coming movement (Horak et al. 1984; Brown and Frank
1987), or the amount of postural support provided during
the movement (Cordo and Nashner 1982; Friedli et al.
1984). Furthermore, increasing forward displaced mass,
by adding a weight to the moving arm or by performing
bilateral arm movements instead of unilateral ones,
changes the timing of postural muscle activity substantial-
ly (Bouisset and Zattara 1981; Friedli et al. 1984; Horak
et al. 1984; Zattara and Bouisset 1988; Benvenuti et al.
1990; Aruin and Latash 1995). The quantitative parame-
ter is involved in electromyogram (EMG) amplitude mod-
ulation. The EMG amplitude of the postural muscles de-
creases when additional support is provided during the
movement (Cordo and Nashner 1982; Friedli et al.
1984), or increases when the arm is loaded by a weight
(Friedli et al. 1984; Benvenuti et al. 1990; Aruin and
Latash 1995). These findings support the idea of Forss-
berg and Hirschfeld (1994) that postural adjustments are
organized in two functional levels. At the first level a ba-
sic, spatially defined or direction-specific muscle activa-
tion pattern is generated. At the second level, additional
spatial properties (the amount of antagonistic activity)
and the temporal and quantitative characteristics of this
basic, direction-specific pattern can be fine-tuned on the
basis of task-related parameters (Forssberg and Hirschfeld
1994; Hadders-Algra and Forssberg 1997).
Little is known about the organization of postural ad-
justments during voluntary movements in non-standing
positions. Moore and coworkers (1992) and Tyler and
Hasan (1995) demonstrated that adults who perform arm
movements in a sitting position also show direction-specif-
ic activity in the trunk muscles. In both studies the record-
ing of the postural muscles was restricted to the trunk mus-
cles (abdominal and paraspinal muscles), thereby preclud-
ing conclusions on the order of activation within the direc-
tion-specific response. Anticipatory activity of the postural
muscles was not consistently present in the sitting position
as the activity of the deltoid muscle usually preceded the
activity of the postural paraspinal muscle (Moore et al.
1992). Other studies on arm movements in sitting posi-
tions focussed on movement velocity. They demonstrated
that movement velocity decreases when contact with the
support surface increases (Lino and Bouisset 1994; Goutal
et al. 1994). Furthermore, Bazalgette et al. (1986) found
203
that the duration of an arm movement is shorter in a stand-
ing than in a sitting position.
The above data indicate that position might affect the
organization of postural control, but precise data on the
nature of position-dependent effects on postural control
are lacking. However, such knowledge is a prerequisite
for the understanding of postural adjustments during vol-
untary movements in those subjects who are unable to use
the standing position, either because they suffer from a
(neurological) disorder or because they are too young.
Therefore we assessed the effect of different positions
on the organization of postural adjustments accompany-
ing arm movements. To this end 14 healthy adults per-
formed fast unilateral pointing movements in five differ-
ent positions that varied in the amount of bodily support:
standing, three sitting positions (semi-reclined sitting,
ªlong-legº sitting, and upright sitting on the edge of a ta-
ble), and lying supine. To further improve our insight into
position-specific modulation of postural adjustments, the
subjects performed additional trials with increased task-
load (unilateral arm movements with a weight in the hand
of the moving arm and bilateral arm movements). In all
conditions we simultaneously recorded EMG activity of
arm, neck, trunk, and upper leg muscles and kinematics.
We centred our questions on three issues, which are
based on the three parameters of postural organization.
The first one concerned the spatial distribution of postural
muscle activation, i.e. the postural activity representing
the primary level of postural control (the direction speci-
ficity) and part of the second level (the amount of antag-
onistic activity). We hypothesized that direction-specific
responses would be present in all positions, but that the
amount of antagonistic activity would vary with the
amount of bodily support (negative relationship) and the
load of the task (positive relationship). The second and
third issues, which dealt with the temporal and quantita-
tive characteristics of postural activity, also addressed
postural mechanisms at the second, modulatory control
level. We hypothesized that the absolute and relative tim-
ing of the postural response would depend on the position
and task-load, in the sense that a decrease in bodily sup-
port and an increase in task-load would result in an earlier
activation of the postural muscles, and also would induce
an increase in the EMG amplitude of the postural mus-
cles. The kinematic data allowed us to address two addi-
tional questions: (a) whether eventually position or task-
load dependent modulatory effects could be attributed to
changes in arm movement velocity or specific changes
in body geometry, and (b) whether arm velocity and body
geometry affected postural adjustments in the different
positions to a similar extent.
Subjects and methods
Subjects
Fourteen healthy adults (7 men and 7 women) between 22 and 36
years of age participated in the study. Their heights ranged from
204
Fig. 1 Five recording positions.
Fast pointing movements were
made during stance (ST), upright
sitting (US), long-leg sitting
(LL), semi-reclined sitting (SR)
and lying supine (LY). The end
position of the arm is displayed
by the dashed arms

160 to 195 cm, and their weights from 65 to 95 kg. All subjects gave
informed consent following the guidelines of the Medical Ethics
Committee of the University Hospital Groningen.

Protocol

The subjects performed pointing movements towards a red dot (di-

ameter 4 cm) placed in the midline of the body at shoulder height
and outstretched arms length away (Fig. 1). Using a simple reaction
time set-up, the subjects were instructed to initiate the movements
immediately after an auditory ªgoº signal, and to execute the move-
ment as quickly as possible. Accuracy of the movement was not em-
phasized. One second after ending the movement the subjects re-
turned their arm(s) to the starting position. Before the start of the ex-
periment two practice trials were allowed. Series of ten unilateral,
right-handed pointing movements were performed in five different
positions: (1) lying supine (LY), (2) sitting semi-reclined with ex-
tended legs and with the back leaning against the back of the chair,
which was at an angle of 45 (SR), (3) non-reclined sitting with ex-
tended legs and a non-supported back (ªlong-legº sitting, LL), (4)
upright sitting on the edge of a table, with the legs flexed and back
and feet unsupported (US), and (5) stance with the feet about 25 cm
apart (ST) (Fig. 1). Task-load effects were studied by adding two se-
ries of ten pointing movements in the upright sitting position and in
the standing position. During the first additional series the subjects
performed bilateral pointing movements ± holding the hands clasped
together in the midline while raising both arms simultaneously (UB
and SB, for the sitting and standing position respectively). The sec-
ond additional series consisted of unilateral pointing movements, Fig. 2A, B Schematic representation of the kinematic marker posi-
which were performed while holding a load of 500 g in the hand tions and angles. A Representation of the position of the kinematic
of the moving arm (UL and SL, respectively). markers: 1 processus condylaris mandibulae, 2 1 cm in front of an-
The time between the movements within the series was 15 s and gulus mandibulae, 3 processus spinosus at C7, 4 processus spinosus
that between the series 5 min. The duration of the whole experiment at T10, 5 processus spinosus at L5, 6 spina iliaca anterior superior, 7
was 1 h. To see whether fatigue might play a role, protocol order proximal edge of the trochanter major, 8 acromion, 9 epicondylis ra-
was reversed in five of the 14 subjects. dialis, 10 processus styloideus radii. B Representation of the calcu-
lated angles in the sagittal plane: head and pelvis rotation in relation
to the X-axis, and the trunk extension angle
EMG recordings

Pilot recordings revealed that upper and lower leg muscles were sel-
domly recruited in the non-standing positions. As our equipment on-
ly allowed the recording of a limited number of EMG signals we de-
cided to omit the assessment of lower leg muscle activity, which im-
plies that we missed a part of postural activity in the standing con-
ditions. Disposable bipolar surface electrodes (interelectrode dis-
tance 13 mm; Twente Medical Systems) were placed over the sur-
face of the deltoid (DE), pectoralis major (PM), biceps brachii
(BB), triceps brachii (TB), neck flexor (NF, sternocleidomastoide-
us), neck extensor (NE, at the C7 level), rectus abdominis (RA), tho-
racic extensor (TE, at the T10 level), lumbar extensor (LE, at the L5
level), the rectus femoris (RF), and the hamstrings (HAM) muscles
at the right side of the body. The EMGs were recorded continuously
during the whole session using POLY, a dedicated software program
for long-lasting polygraphic recordings (sampling rate 1000 Hz; In-
spector Research Systems, Amsterdam, The Netherlands).
Kinematics
Kinematics were recorded synchronously with the EMGs using a
two-camera ELITE system (BTS, Milan, Italy). The data were sam-
pled at 50 Hz for 3 s, starting about 1 s before movement onset. Re-
flective markers were placed at the right side of the body on the fol-
lowing landmarks: (1) processus condylaris mandibulae, (2) 1 cm in
front of angulus mandibulae, (3) processus spinosus at C7, (4) pro-
cessus spinosus at T10, (5) processus spinosus at L5, (6) spina iliaca
anterior superior, (7) proximal edge of the trochanter major, (8) ac-
romion, (9) epicondylis radialis, and 10) processus styloideus radii
(Fig. 2A). In the LY and SR positions the marker at the T10 level
was excluded.

Video recordings
Continuous video registrations, time-coupled to the EMG record-
ings, were made during each session. These registrations were used
to score the onset of the pointing movements with a dedicated com-
puter program (TAPETIME). The accuracy of the scoring of the
movement onset was evaluated by scoring a series of ten trials five
times each. This procedure revealed that the differences in the onset
times between the five scores was 3.1  3.6 ms. The behavioural da-
ta were merged with the EMG data.
Data analysis
EMG analysis
The first step in the EMG analysis was to determine which muscles
participated in the activation pattern accompanying the movement.
To this end we used a semi-automatic computer algorithm, which
was developed for the detection of significant phasic muscle activ-
ity. The algorithm, on the basis of the root mean square (RMS) of
a full wave rectified signal (200 ms moving window), was construct-
ed as follows. First, baseline activity, determined by a 3.7 s moving
window, was subtracted. Next, the signal was enhanced by adding
the first derivative of the signal to the RMS, and a fixed detection
level (0.4 times baseline activity) was set. Bursts were considered
to be present when the activity exceeded the detection level for at
least 30 ms within a time-window from 1 s prior to the onset of a
movement as scored on the video, until 2 s after movement onset
(Fig. 3). An interactive assessment offered the possibility of distin-
guishing the EMG bursts from cardiac activity. The latter was fre-
quently present in the PM and the RA muscles. Second, the latencies
of muscle activation were defined as the time interval between the
onset of activation of the first arm muscle (prime mover) and the on-
set of responses in the other muscles. Finally, EMG amplitude was
evaluated by calculating the mean amplitude of the signal after sub-
traction of baseline activity in three different time windows. T1 cov-
ered the 100 ms prior to the activation of the first arm muscle, there-
Fig. 3A±D Representation of the EMG analysis procedure. A Raw
electromyogram (EMG) of the deltoid muscle during a fast pointing
movement during stance. B The RMS of the same EMG; the hori-
zontal line indicates the baseline activity. C The enhanced EMG sig-
nal and the detection level (horizontal line). D The detection of the
start and end of an EMG burst. The raw EMG (as in A) together with
the signal indicating the onset (upward deflection) and the offset
(downward deflection) of phasic EMG activity
205
by covering anticipatory postural muscle activity. T2 and T3 ranged
from prime mover activation to 100 ms and 400 ms respectively af-
ter arm muscle activation, the longer T3 interval including longer-la-
tency and feedback components of the postural adjustments.
Rates of activation (calculated by dividing the number of trials
during which a muscle was activated by the total number of trials),
mean latencies and mean amplitudes were calculated for each sub-
ject in each position. The non-parametric Wilcoxon and Mann-
Whitney tests were used to analyse within-group and between-sub-
ject differences in these values of the various muscles in the differ-
ent positions.
Kinematic analysis
Off-line analysis consisted of calculation of spatial angles for the
head (by a vector between markers 1 and 2) and pelvis (by a vector
between markers 6 and 7) in relation to the horizontal axis. During
long-leg sitting, upright sitting and stance the trunk angle was de-
fined by calculating the angle of the two intersecting vectors be-
tween markers 3 and 4 and markers 4 and 5 (Fig. 2B). The ELITE
data were transformed into ASCII files and further analysed with the
help of Datamonster 2.0 (E. Otten, Dept. of Medical Physiology,
University of Groningen). Arm movement onset was defined as
the moment at which the velocity of the wrist increased, while the
moment at which wrist velocity became zero was considered as
the end of the movement. The analysis focused on: (1) the maximum
velocity of elbow and wrist (markers 9 and 10) during the move-
ment, (2) the angular positions at movement onset, and (3) the angle
displacements (a) between 100 ms prior to movement onset and
movement onset (d1), and (b) during the movement (d2). The veloc-
ity, angle positional and angle displacement values were compared
for each position using the Wilcoxon and Mann-Whitney tests. A
multiple linear regression analysis was used to correlate the kine-
matic data with the EMG data.
In general, differences with a P value < 0.05 were considered to
be statistically significant, but in the large number of correlations we
decided to accept only P values < 0.01 as non-chance findings.
Results
Preliminary data analysis revealed that two different strat-
egies were used to perform the pointing movements, not-
withstanding identical instructions. The type of pointing
strategy affected the postural adjustments considerably.
Seven subjects used one strategy. They started the move-
ment with flexion of the shoulder while keeping the arm
extended throughout the movement, and activated the del-
toid (DE) as the first arm muscle (DE group; Figs. 4, 5).
The other seven initiated the movement with a slight flex-
ion of the elbow after which an elevation and extension of
the arm followed. These subjects activated the biceps bra-
chii (BB) as the prime mover (BB group; Figs. 4, 5). The
subject-specific strategy and the preference for the prima-
ry activation of the DE in the DE group and the BB in the
BB group were consistently present in all trials and all po-
sitions. Only in the lying supine position did the DE sub-
jects often activate the BB before the DE muscle (Fig. 5).
The addition of task-load (unilateral with weight or bilat-
eral arm movements) did not change the movement strat-
egy.
Consequently, the onset latencies of the arm and pos-
tural muscles were related to the DE in the DE group
and to the BB in the BB group, and, from this point on-
wards, we will discuss the results of the two groups sep-
206
Fig. 4 Stick diagrams of a fast
pointing movement in a stand-
ing DE subject and BB subject.
A representative fast pointing
trial of a subject in the DE group
(upper part), in the BB group
(lower part). For marker identi-
fication see the legends of Fig. 2.
The sticks are represented every
20 ms
Fig. 5 Latencies to the deltoid (DE) and biceps brachii (BB) muscle
in the DE and BB groups. Group data on the mean recruitment la-
tencies of the BB related to the DE in the two groups (DE group,
thick lines; BB group, thin lines) in various positions. The data
are presented as the ranges (horizontal bars) and median values
(vertical bars). The 0 line indicates the onset of the DE. Asterisks
indicate significant differences between the recruitment latencies
of the two arm muscles (Wilcoxon: *P £ 0.05)
arately. Protocol order, age and sex did not affect the re-
sults.
Spatial distribution of postural adjustments
Pointing movements in the standing condition were al-
ways accompanied by a primary activation of the dorsal
postural muscles NE, TE, LE, and HAM (Fig. 6). Also
in the other positions and in the different task-load condi-
tions the dorsal muscles were recruited before their ven-
tral antagonists, thereby confirming the basic nature of
the direction-specific organization of postural adjust-
ments. However, the extent to which the dorsal postural
muscles participated was position dependent in both
groups (Fig. 6). LE was frequently activated in the stand-
ing and upright sitting position, but only occasionally in
the other positions (P < 0.05). Substantial HAM muscle
activity was present only during stance. A similar posi-
tion-dependent recruitment of the more caudally located
dorsal muscles was found in the load conditions.
The neck muscles (NE and NF) usually showed a pat-
tern of co-activation in both groups and in all positions
(Figs. 7, 8). In the trunk and leg muscles, however, tripha-
sic activation patterns were found (TE-RA-TE, LE-RA-
LE, and HAM-RF-HAM; Figs. 7, 8). The rate of the tri-
phasic pattern showed a cranial-to-caudal gradient, with
the TE-RA-TE pattern occurring most frequently
(Fig. 8). The TE-RA-TE pattern was more often present
in the DE group than in the BB group, but did not vary
with position. In contrast, the triphasic activation patterns
of the more caudally located muscle pairs were affected
by the position of the subject, but not by pointing strategy
(Fig. 8). Here, the triphasic patterns were most frequently
present during stance (LE-RA-LE and HAM-RF-HAM),
and during upright sitting (LE-RA-LE). The decrease in
the amount of triphasic LE-RA-LE pattern was related
to a significant decrease in LE activation during long-
leg sitting, semi-reclined sitting and lying, whereas the
drop in HAM-RF-HAM activation occurred because the
two leg muscles were rarely activated in the non-standing
conditions.
These results demonstrate that the amount of agonistic
activity of the dorsal postural muscles is negatively relat-
ed to the support surface, and not, as we expected, the ac-
tivation rate of the antagonists. The addition of task-load,
by means of bilateral movements or loaded unilateral
movements, did not affect the rate and type of postural ac-
tivation patterns. Thus, our hypothesis of the existence of
a positive relation between the amount of antagonistic ac-
tivity and task-load was rejected.
Temporal characteristics
Since the dorsal muscles were always recruited before
their ventral antagonists, we focussed on the temporal
and quantitative characteristics of the dorsal postural mus-
cles.
During stance the dorsal postural muscles in the DE
group were recruited in a caudal-to-cranial order
(Fig. 9). Only the HAM muscle was recruited prior the
prime mover DE (P < 0.05; median latency ±36 ms). In
the BB group the dorsal postural muscles were in general
activated later than in the DE group (Fig. 9). This was
most pronounced for the LE muscle. Anticipatory activa-

Fig. 6 Frequencies of dorsal muscle activation in the various posi-
tions. Group data on the frequencies of trials during which the dorsal
muscles (NE neck extensor, TE thoracic extensor, LE lumbar exten-
sor, HAM hamstring) were activated. For each subject in each con-
dition the rate of activation of the various muscles was calculated by
dividing the number of trials during which a muscle was activated
by the total number of trials, resulting in an muscle-activation rate
per individual. The vertical bars denote the range of the individual
values and the horizontal bars the median values of the group of
subjects. Thick lines represent the results of the DE group, thin lines
those of the BB group. Asterisks indicated the differences between
the standing (ST), various sitting (US, LL, SR), and lying (LY) posi-
tions (Wilcoxon: *P £ 0.05). No differences were present between
the DE and the BB group
tion of the postural muscles was not found in this group.
In the BB group the caudally located HAM muscle was
activated before the cranial NE and TE muscles. LE, how-
ever, was activated much later than the other postural
muscles, thereby interfering with the appearance of a real
caudal-to-cranial recruitment order as was present in the
DE group. In both groups the more caudally located LE
and HAM muscles in particular were recruited later when
the support provided in the various positions increased
(Fig. 9). This resulted in a reversed (top-down) order of
activation in the various sitting and lying positions. In
these positions no anticipatory activation of the postural
muscles was present.
Increasing task-load by performing the pointing task
with two arms simultaneously, resulted in an earlier acti-
207
vation of LE, TE and, to a lesser extent, NE (Fig. 10).
This effect was found in both groups (DE and BB) and
in both positions (stance and upright sitting; P < 0.05 dur-
ing sitting). The changes in the onset of muscle activity
induced a reversal of the recruitment order in the sitting
position: instead of the top-down order that was present
during unilateral pointing, a caudal-to-cranial sequence
was present. In the standing position such a bottom-up re-
cruitment was already found during the unilateral arm
movements. This order persisted during bilateral move-
ments. In standing DE group, the earlier postural muscle
activity resulted in an anticipatory activation of LE and
HAM (7 ms and 16 ms respectively before the onset of
the prime mover DE). The other task-load condition, the
unilateral arm movements with additional weight, did
not affect the temporal characteristics of the postural re-
sponse.
The temporal data confirmed our hypothesis that a de-
crease in bodily support and an increase in task-load by
means of bilateral arm movements induces an earlier ac-
tivation of the postural muscles. This effect was mainly
found in the more caudally located LE and HAM muscles.
Quantitative characteristics
In both groups the mean amplitudes of the dorsal neck
(NE) and trunk muscles (TE and LE) were low during
208
Fig. 7 EMG patterns during fast
pointing in the ST and US po-
sitions of the DE and BB
groups. Averaged EMG patterns
of a DE subject during stance
and upright sitting (upper part),
of a BB subject in the same
positions (lower part). The ver-
tical lines denote the onset of
the DE in the DE subject, of the
BB in the BB subject (DE del-
toid, PM pectoralis major, BB
biceps brachii, TB triceps bra-
chii, NF neck flexor, NE neck
extensor, RA rectus abdomis, TE
thoracic extensor, LE lumbar
extensor, RF rectus femoris,
HAM hamstring)

the 100 ms prior to activation of the prime mover (T1) in
all recorded positions. Only leg muscle activity during the
T1 interval showed position-dependent differences in
both groups of subjects. HAM activity was low in the sit-
ting and lying positions, but during standing it was signif-
icantly increased (P < 0.05). During the first 100 ms after
prime mover activation (T2) the mean amplitudes of NE,
TE and LE were in general higher than during T1, and the
amplitudes were affected by position and ± to a lesser ex-
tent ± pointing strategy. As during T1, HAM muscle ac-
tivity during the T2 interval was again significantly high-
er during stance than in the other positions (P < 0.05).
This was true for both groups. For the other postural mus-
cles, the position dependency of the EMG amplitudes var-
ied for the two groups. In the DE group the NE and LE
amplitudes were higher in upright sitting and stance than
in the other positions (P < 0.05), whereas in the BB group
such an effect was absent. During the longer interval after
prime mover activation (T3) the amplitudes of NE, TE,
LE and HAM were lower than during T2. Only in the
DE group did NE show a position dependency, which
was similar to that found during T2.
The addition of task-load particularly affected the am-
plitudes of the dorsal postural muscles during the 100 ms
before and after prime mover activation (T1 and T2). In
the DE group the effects varied with amplitude interval
(T1 or T2), position, and the type of task-load. Bilateral
arm movements during standing and sitting induced sig-
nificantly higher amplitudes in the T1 interval of NE,
TE and LE (P < 0.05), whereas the T1 amplitude of
HAM was not affected (Fig. 11). Unilateral load did not
change EMG amplitudes during T1. The effects of load
on the T2 amplitudes were restricted to the standing posi-
tion. Both bilateral and loaded unilateral movements re-
sulted in a decreased NE amplitude and an increased am-
plitude of TE, LE and, especially, HAM (Fig. 11). In con-
trast, the addition of task-load in the BB group had little
effect on postural EMG amplitudes.
The quantitative data support the idea that a decrease
in support surface and an increase in task-load induce
an increase in postural EMG activity. This modulatory ef-
fect, present in the periods just prior to and after prime
mover onset, was mainly present in the DE group.
Kinematics
Arm movement velocity was clearly affected by pointing
strategy, the position of the subjects, and additional task-
load. In general, the BB group showed a significantly
higher maximum velocity of the wrist during the standing
and sitting positions than did the DE group (during US
and LL, P < 0.01). In line with the results of other studies
(Lino et al. 1994; Goutal et al. 1994), the maximum ve-
locity of the wrist decreased step-wise from standing to
upright sitting, long-leg sitting and semi-reclined sitting
in both groups (P < 0.05). However, during lying supine
wrist velocity increased again ± the DE group reaching
the highest wrist velocities in this position (P < 0.05). El-
bow velocities showed a similar position dependency. In
 209

Fig. 8 Activation patterns of dorsal and ventral postural muscles.

Group data on the frequencies of trials during which co-activation
patterns were present in the neck (NE-NF; difference in recruitment
latency < 50 ms), during which triphasic patterns were found in
trunk and leg muscles (TE-RA-TE, LE-RA-LE, HAM-RF-HAM)
in both groups. Thick lines represent the DE group, thin lines the
BB group (see legend to Fig. 6). The data are presented as the ranges
(vertical bars) and the median values of the group (horizontal bars).
Asterisks indicate the frequency differences between the various po-
sitions (Wilcoxon: *P £ 0.05), and the black boxes indicate the dif-
ferences between the DE and BB groups (Mann-Whitney: n P £ 0.05)

contrast, the duration of the arm movement did not vary

with position. Increasing task-load by performing the
movement bilaterally did not affect arm movement veloc-
ity and duration. On the other hand, loaded unilateral
movements were associated with significantly lower wrist
velocities (P < 0.05 in sitting and standing BB group) and
longer movement durations (P < 0.05 in the standing BB

Fig. 9 Latencies to dorsal postural muscle activation in the DE and c

BB group. Group data on the latencies of the various muscles of the
DE group (thick lines) and the BB group (thin lines). The vertical
line (t = 0) denotes the onset of EMG activity in the prime mover.
The data are presented as the ranges (horizontal bars) and the medi-
an values (vertical bars). Asterisks indicate the differences between
the latency of the various dorsal postural muscles, between dorsal
muscles and the prime mover (Wilcoxon: *P £ 0.05), the black box-
es indicate the differences between the DE and BB groups (Mann-
Whitney: n P £ 0.05, n n P £ 0.01)
210

group and sitting DE group) than non-loaded unilateral

Fig. 10 Latencies to dorsal postural muscle activation in the DE and
BB groups during bilateral arm movements. Group data on the laten-
cies of the dorsal postural muscles in the DE group (thick lines) and
the BB group (thin lines) during stance (SB) and upright sitting
(UB). Details as for Fig. 9
movements.
During the 100 ms before movement onset (d1) very
little angle displacements occurred in either group or
any of the positions (Figs. 12, 13). Therefore, we will re-
port only the angle positions at movement onset and the
angle displacements during the movement (d2). In both
groups the position of the head at movement onset was
similar in all positions. Only in the long-leg sitting posi-
tion did the head tend to be more anteflexed compared
with the other positions. Within a single subject head dis-
placement was consistently small: < 7 in all subjects,
and in 80% of the subjects even < 3. But, a considerable
interindividual variation was present, ranging, for exam-
ple, during upright sitting from a mean head anteflexion
of 3.7 in one subject to a mean head extension of 4.1
in another subject. The trunk was slightly anteflexed at
movement onset in the standing, upright sitting and
long-leg sitting positions. The arm movement was accom-
panied by a trunk extension of about 7 (median value) in

Fig. 11 Mean EMG amplitudes

of the dorsal postural muscles
during the load conditions in DE
subjects. Group data on mean
amplitudes of the various mus-
cles for the intervals T1 and T2
(100 ms prior to activation of
the first arm muscle, 100 ms
after prime mover activation
respectively) of the DE group
during stance (upper part) and
upright sitting (lower part) dur-
ing unilateral movements (ST
and US), bilateral movements
(SB and UB), and loaded uni-
lateral movements (SC and UC).
The data are presented as the
ranges (vertical bars) and me-
dian values (horizontal bars).
Asterisks indicate the difference
between the amplitude of the
muscles in the various positions
(Wilcoxon: *P £ 0.05,
**P £ 0.01)
 211

Fig. 12 Kinematic data of eight representative trials of one DE
group subject. Displacements of the angles of the head, the trunk
the pelvis during stance (left) and upright sitting (right). The traces
are aligned according to the start of the arm movement, which is in-
dicated by the vertical line. The interval of 100 ms prior to move-
ment onset is denoted by d1. The d2 interval denotes the interval be-
tween the onset of the movement and the mean end of the eight
movements
these positions. Of course, the pelvis angle differed signif-
icantly between the various positions (P < 0.05), and was
most extended during stance (Fig. 13). During the arm
movement the pelvis anteflexed slightly (about 1; medi-
an value) in all positions (Fig. 13). The addition of task-
load did not affect the position of head, trunk, and pelvis
position at arm movement onset, or the angular displace-
ments during the pointing movement (d2) in all standing
and sitting subjects.
Kinematics and EMG
A multiple linear regression was performed with the
groups and subjects as independent co-variables, since ki-
nematic and EMG differences were present both between
groups and between subjects.
Fig. 14A±C Schematic representation of the results of the multiple
regression analysis of EMG amplitudes and kinematic parameters,
subject and group. Filled squares denote a significant effect
(P < 0.01) on the EMG amplitudes of A maximum wrist velocity
(range of R2: 0.18±0.55, P < 0.01), B trunk position at movement
onset, increased trunk angle denoting extension (range of R2:
0.23±0.47, P < 0.01), C pelvis position at movement onset, in-
creased pelvis angle reflecting anteflexion (range of R2: 0.14±0.43,
P < 0.01). The colour of the squares indicate the direction of the re-
lations (black squares signifying a positive relation, grey squares a
negative relation). T1, T2 and T3 correspond to the amplitude during
the various time intervals
The maximum wrist velocity was not related to the on-
set latencies of the dorsal postural muscles, but it showed
a positive correlation with the EMG amplitude in the dif-
ferent time intervals (P < 0.01; Fig. 14A). This positive
relation was most frequently present in the NE and TE
muscles. Occasionally, a negative relation between
EMG amplitude and wrist velocity was present [in TE
and HAM during stance (T1), and in TE (T2 and T3) dur-
ing upright sitting]. Elbow velocity correlated with pos-
tural EMG activity in the same way as did wrist velocity.
The multiple regression analysis did not show a relation
between the duration of the arm movement and postural
EMG characteristics.

Fig. 13 Pelvis position and pel-

vis displacement in the DE
group. DE group data on mean
pelvis position at the onset of
the movement, mean pelvis dis-
placement 100 ms before
movement onset (d1), mean
pelvis displacement during the
movement (d2). Data are pre-
sented as the ranges (vertical
bars) and median values (hori-
zontal bars) of the group. As-
terisks indicate the differences
between the various positions
(Wilcoxon: *P £ 0.05)
212
Fig. 15 Schematic representation of the results of the multiple re-
gression analysis of the position of head, trunk and pelvis, subject
and group. Presence of squares denotes a significant effect
(P < 0.01) of head, trunk or pelvis position on EMG amplitude
(range of R2: 0.18±0.54, P < 0.01). The colour of the squares indi-
cates the direction of the relations (black squares signifying a pos-
itive relation, grey squares a negative relation). T1, T2 and T3 cor-
respond to the amplitude during the various time intervals. Increased
angle of the head denotes anteflexion, of the trunk denotes exten-
sion, of the pelvis reflects anteflexion
To see whether the different positions or an increase in
task-load affected input-output relations, the effects of po-
sition and load on relations between initial body geometry
and EMG characteristics were evaluated. The head, trunk
and pelvis positions at movement onset were not related
to the recruitment latencies of the dorsal postural muscles,
with the exception of an earlier activation of LE when the
trunk position was more extended during long-leg sitting
(P < 0.01). However, the initial positions of the trunk and
pelvis did affect the amplitudes of trunk and leg muscle
activity. Trunk position at movement onset was positively
correlated with the amplitude of TE and LE muscles dur-
ing T1 and T2 (Fig. 14B). The amplitudes increased when
the trunk was more extended at movement onset. The pel-
vis angle was positively related to the LE muscle and neg-
atively related to HAM activity in the different positions
(Fig. 14C).
Task-load addition affected the relationships between
the kinematic and EMG parameters, but the effects were
different in the two conditions and positions. The rela-
tions between arm movement velocity and EMG ampli-
tudes were only affected during bilateral pointing in the
standing position. In this condition the relations between
increased arm movement velocity and larger EMG ampli-
tude were strengthened in all postural muscles. Additional
task-load especially affected the relations between body
geometry and EMG amplitudes. Bilateral arm movements
in both positions usually resulted in a disappearance of the
relation between the initial position of the trunk and pel-
vis and the amplitudes of the postural muscles (Fig. 15).
This was also found during loaded unilateral movements
in the sitting position, but during stance the relations be-
tween trunk and position and EMG amplitude remained
virtually unchanged. Furthermore, loaded unilateral
movements induced a relation between head anteflexion
at movement onset and NE amplitudes: during stance in-
creased head anteflexion resulted in higher NE ampli-
tudes, whereas during upright sitting it was related to low-
er NE amplitudes (Fig. 15).
Discussion
The present study demonstrated that postural adjustments
accompanying fast pointing movements depend on the
pointing strategy, the position of the subject, the load of
the task, and the body geometry.
Different pointing strategies
Each subject consistently selected one of two pointing
strategies, notwithstanding the fact that all subjects re-
ceived identical instructions. Half the subjects, who used
DE as the prime mover, initiated the movement with a
shoulder flexion. They kept the arm extended throughout
the pointing movement. The other subjects, who activated
BB as the prime mover, started the movement with a
slight flexion of the elbow, which was followed by an
arm elevation and extension. Other studies on postural
control during voluntary arm movements never reported
the presence of different strategies (Bouisset and Zattara
1981; Lee et al. 1987; Gurfinkel et al. 1988; Moore et
al. 1992; Aruin and Latash 1995; Latash et al. 1995).
Two studies on the ªcentral setº effect on postural control
did mention differences between subjects (Brown and
Frank 1987; Horak et al. 1989). Brown and Frank
(1987) reported that in half their subjects preparatory
set affected the reaction time, whereas such a set effect
was absent in the other half. Likewise, Horak and cowork-
ers (1989) demonstrated set-dependent differences in the
EMG amplitudes of the postural muscles after platform
perturbations in only half the group tested. The finding
of subject-specific strategies reflects the ability of the ner-
vous system to generate different motor patterns to reach
an identical goal.
The pointing strategy affected the postural adjustments
substantially, having its main effect on the temporal char-
acteristics. In general, postural activity occurred later in
the BB group than in the DE group. This difference could
not be attributed to a difference in arm velocity as maxi-
mum wrist and elbow velocity of the DE group never ex-
ceeded that of the BB-group (cf. Horak et al. 1984; Lee et
al. 1987). However, the arm movement trajectories dif-
fered considerably between the two groups. The arm tra-
jectory consisted of a single segmented movement in the
DE group, and of a ªtwo-stepº movement in the BB
group. The single segmented movement probably induced
higher inertial moments, which in turn resulted in an ear-
lier activation of the postural muscles in the DE group (cf.
Horak et al. 1984; Friedli et al. 1984).

Spatial characteristics: agonist and antagonist activity
Direction-specific responses were present in the postural
muscles in all positions and task-load conditions, indicat-
ing that the dorsal postural muscles were always activated
before their ventral antagonists. This preservation of di-
rection specificity in all conditions fits with the idea that
direction specificity is a basic characteristic of postural
adjustments (cf. MacPherson 1988; Forssberg and Hirsch-
feld 1994; Hadders-Algra and Forssberg 1997). But even
at this basic level of direction specificity, postural re-
sponses can be adapted to task-specific conditions, as
the number of muscles recruited increased with the pos-
tural challenge imposed by the task (cf. Friedli et al.
1984; Hadders-Algra et al. 1997). The position dependen-
cy of agonist activity was restricted to the more caudally
located postural muscles, suggesting that these muscles
are especially involved in the maintenance of equilibrium,
i.e. in keeping the centre of mass within the limits of the
support surface. The position-independent activation of
the upper trunk and neck muscles probably serves to op-
pose the reaction forces induced by the arm movement.
The trunk and leg muscles were often activated in a tri-
phasic, agonist-antagonist-agonist pattern. This was also
reported in other studies on postural adjustments (Friedli
et al. 1984; Aruin and Latash 1995). We found a similar
triphasic activation pattern in the arm muscles, pointing to
the ballistic nature of these arm movements (Hallett et al.
1975). The early part of the triphasic pattern of ballistic
movements is thought to reflect a pre-programmed motor
control (Hallett et al. 1975; Sanes and Jennings 1984).
The finding that triphasic patterns were present both in
the arm and postural muscles supports the notion of a cou-
pled feedforward control of posture and movement (cf.
Friedli et al. 1984; Massion 1992; Palmer et al. 1994;
Aruin and Latash 1995). The later part of the triphasic ac-
tivation pattern (the antagonist activity and the second
burst of the agonist) is affected by feedback mechanisms,
the concurrent afferent information playing a role in the
fine-tuning of the pattern to task-specific parameters
(Forget and Lamarre 1987; Dietz 1992; Massion 1992).
In this respect it is interesting to note that the amount
of antagonistic activity in the postural response patterns
did not vary with position or task-load condition. This
might be interpreted as an additional piece of evidence
for a major role of feedforward mechanisms controlling
postural adjustments during voluntary arm movements.
Interestingly, the neck flexor and extensor muscles
were co-activated during the arm movement (cf.
Gurfinkel et al. 1988). Hogan (1984) argued that co-acti-
vation in antagonists promotes joint stabilization. Head
stabilization is one of the major goals in postural control
(Pozzo et al. 1990). Pozzo and coworkers (1990) reported
that during various locomotor tasks head movements do
not exceed a 20 rotation. In our study head movements
were even smaller (less than 7 in all subjects), possibly
reflecting the need for head stabilization during goal-di-
rected arm movements (cf. Herman et al. 1981; Jeannerod
1988).
213
Temporal characteristics
The temporal characteristics of the postural adjustments
were dependent on the position of the subject and on
the task-load conditions. During stance the dorsal postural
muscles were activated in a caudal-to-cranial order, a
finding which is in agreement with previous reports
(Cordo and Nashner 1982; Friedli et al. 1984; Horak et
al. 1984; Zattara and Bouisset 1988). In the sitting and ly-
ing positions the order was reversed due to the delayed
activation of LE and HAM (cf. Friedli et al. 1984). This
suggests that position had more impact on the recruitment
time of the lower postural muscles than on the timing of
the more cranially located ones. A similar effect on re-
cruitment order was found by increasing the task-load
by performing bilateral pointing movements. Bouisset
and coworkers, who recorded the activity of paraspinal
and leg muscles, including muscles involved in hip ab-
duction [gluteus maximus (GM) and tensor fasciae latae]
and hip adduction (semitendinosus, ST), showed that bi-
lateral movements do not induce an earlier activation of
all postural muscles (Bouisset and Zattara 1981; Zattara
and Bouisset 1988). They reported an earlier activation
of the ipsilateral paraspinal muscles and the contralateral
GM and ST, and a delayed recruitment of the ipsilateral
GM and ST muscles during bilateral arm movements.
This means that muscles which mainly operate in the sag-
ittal plane, such as LE and TE, are activated earlier when
task-load is increased by means of bilateral arm move-
ments. The situation for the hip-stabilizing muscles,
which play a role in the counteraction of the torsional
forces, is different. The torsional reaction forces are larger
during unilateral arm movements than during bilateral
movements. This probably explains why ipsilateral hip
muscles are recruited earlier during unilateral than bilater-
al arm movements.
Our data indicate that during the most destabilizing
conditions ± during stance with or without additional load
and during sitting with load ± the most caudally located
muscles are recruited first. These findings underline the
notion of body-segment-specific postural behaviour, and
point to a primary role of the lower postural muscles in
the maintenance of equilibrium.
Anticipatory postural activity was only present during
stance in the DE group, notwithstanding the fact we used
a reaction time set-up which is known to reduce anticipa-
tory postural activity (Horak et al. 1984; Lee et al. 1987).
Our kinematic data did not show angular displacements
prior to movement onset, but probably more sensitive re-
cordings with the help of accelerometers would have been
able to demonstrate head, trunk or pelvis displacements
before the start of the movement (cf. Bouisset and Zattara
1981; Zattara and Bouisset 1988). The EMG data did re-
veal an anticipatory activation of the HAM muscle, i.e.
HAM activation before the activation of the prime mover.
This anticipatory activation occurred at similar latencies
as has been described by others using reaction time para-
digms (Bouisset and Zattara 1981; Friedli et al. 1984; Lee
et al. 1987; Zattara and Bouisset 1988). In general, we did
214
not find an anticipatory activation of the trunk muscles.
Such an early activation of LE was only found during bi-
lateral movements in standing DE group (cf. Friedli et al.
1984). Zattara and Bouisset (1988) reported that anticipa-
tory trunk muscle activity could be found during unloaded
unilateral movements as well, but only at the side contra-
lateral to the arm movement. During sitting and lying an-
ticipatory activation of the postural muscles was absent
(cf. Moore et al. 1992). Our results indicate that anticipa-
tory activation of the postural muscles is only called into
play when the support surface is small and a large effort
has to be performed to keep the centre of mass within the
stability limits (Massion 1992).
Quantitative characteristics
The amplitudes of the postural adjustments were clearly
related to body geometry at movement onset, and only
to a minor extent to position. A more extended position
of the trunk and pelvis resulted in a higher activity of
the dorsal trunk muscles, whereas an increase in pelvis
extension was associated with a decreased activation of
the pelvis extensor HAM. Furthermore, EMG amplitudes
were positively related to arm movement velocity (cf. Lee
et al. 1987), a parameter which showed position-depen-
dent variations. The lowest arm movement velocities
were found in the two sitting positions with a relatively
large support area. These findings, as well as the higher
velocities during upright sitting and standing, are in
agreement with those of others reporting lower velocities
when the support surface increases (Bazalgette et al.
1986; Lino and Bouisset 1994; Goutal et al. 1994). A larg-
er support surface induces a decrease in the posturo-kinet-
ic capacity of the body, thereby limiting the body to coun-
teract the perturbation induced by the dynamics of the
movement (cf. Bouisset and Zattara 1987). The finding
of high movement velocities during lying seems at vari-
ance with this body dynamics hypothesis. It is possible
that the subjects, especially those of the DE group, exert-
ed a thrust on the supporting surface in the lying position
as they showed a reversal in the onset of the prime mover
(starting with BB instead of DE in this position).
Our study indicated that bilateral pointing movements
induce an increase in postural muscle activity. This was
not reported in earlier studies (Bouisset and Zattara
1981; Zattara and Bouisset 1988). The way in which bilat-
eral arm movements modulated postural EMG activity re-
sembled the modulatory effect of unilaterally applied
weight reported by others (Bouisset and Zattara 1981;
Horak et al. 1984; Zattara and Bouisset 1988; Benvenuti
et al. 1990; Aruin and Latash 1995). In the present study
we were, however, unable to demonstrate such a clear-cut
effect of unilateral load. This discrepancy can be attribut-
ed to the weights used: ³ 900 g in the other studies and
only 500 g in ours. This relatively small weight (0.5±
0.8% of body weight) was chosen because of our ontoge-
netic interests: the weight was proportional to the maxi-
mal weight that can be used in infants. The results suggest
that an increase in task-load modulates postural muscle
activity in proportion to the weight of the forward dis-
placed mass. Not surprisingly, the load effects were more
pronounced in the DE group, in which the onset of the
single-segmented movement induces higher inertial forc-
es than does the two-step movement of the BB group.
Increasing the task-load affected the EMG amplitude
during the 100 ms before and after prime mover activa-
tion (T1 and T2), and not during the longer 400-ms inter-
val after activation of the prime mover (T3). This early
adaptation suggests that the contribution of feedback
mechanisms in the fine-tuning of postural muscle activity
during pointing movements is limited (cf. Friedli et al.
1984; Horak et al. 1989). Furthermore, the finding that
not only an extrinsically induced increase in the task-load
(addition of weight) causes an increase in the postural
EMG activity, but also an intrinsically evoked task-load
(bilateral arm movements), suggests that feedforward or
anticipatory mechanisms play a major role in the modula-
tion of postural activity to task-specific constraints (cf.
Massion 1997). Possibly, such an anticipatory modulation
of postural activity is comparable to a form of central
ªsetº ± the capacity of the central nervous system to pre-
pare the sensory and motor systems for a particular task
(Prochazka 1989). But usually the term ªsetº is used for
short-term adaptation based on prior experience (Friedli
et al. 1984; Keshner et al. 1987; Horak et al. 1989; Bur-
leigh and Horak 1996).
The addition of load had a different effect on the rela-
tions between EMG parameters and kinematics in the sit-
ting and standing positions. In stance the relations be-
tween body geometry and EMG amplitude disappeared
when task-load was increased sufficiently. Apparently,
the modulatory effects of body geometry were overruled
by the movement parameters. This is supported by the
finding that bilateral arm movements facilitated the rela-
tion between wrist movement velocity and EMG ampli-
tude. In the sitting position the effect of load differed
for the upper and lower postural muscles (cf. Aruin and
Latash 1995). In the lower postural muscles the relation
between body geometry and EMG amplitude disappeared,
whereas in the upper postural muscles a new relation be-
tween body geometry and EMG amplitude appeared.
Again, this finding underlines the hypothesis that the up-
per postural muscles are involved in opposing the reaction
forces induced by the movement, since the inertial forces
are larger when task-load increases.
Concluding remarks
Our data confirm the idea of Forssberg and Hirschfeld
(1994) that postural adjustments are organized in two
functional levels. The basic postural response during fast
pointing movements consist of an activation of the direc-
tion-specific dorsal muscles. Fine-tuning of this basic re-
sponse can be achieved by regulating the number of dor-
sal muscles that participate in the response, and the timing
and quantity of the dorsal muscle activity. The fine-tuning

of postural adjustments to various positions occurs by
changing the number and timing of the recruited dorsal
postural muscles, whereas the modulation of the postural
adjustments to task-load is achieved with the help of the
temporal and quantitative characteristics of dorsal muscle
activity.
The early onset of postural muscle activity and the
presence of triphasic activation patterns of trunk and leg
muscles during fast pointing movements in various posi-
tions point to the pre-programmed character of these types
of postural adjustments (Massion 1997). This pre-pro-
grammed or anticipatory postural control does not imply
an activation of the postural muscles prior to prime mover
activity. Such an anticipatory activation of the postural
muscles occurs only when the centre of mass is expected
to move beyond the margins of the support surface. The
latter can be caused by a small support surface (e.g. stand-
ing), or by large inertial forces generated by the arm
movement due to a large arm acceleration and/or a large
mass. The pre-programmed character of the postural ad-
justments does not preclude participation of feedback
mechanisms, which particularly contribute to the fine-tun-
ing of the later parts of these adjustments (Dietz 1992;
Massion 1992).
The present study indicates that the task-specific flex-
ibility of postural adjustments during pointing movements
is mainly controlled by a modulation lower trunk and leg
muscle activity. The finding that trunk and pelvis position
were involved in EMG modulation suggests that segmen-
tal somatosensory information promotes the appropriate
fine-tuning of lower trunk and leg muscle activity. Prop-
rioceptors in the lumbovertebral column (Gurfinkel et al.
1981) and ªgraviceptorsº located in the kidney region
(Mittelstaedt 1997) could be mediators of this segmental
sensory information.
Acknowledgements We thank Ria Willemse for her help during
the recordings and data analysis, Dr. E. Otten for his technical assis-
tance during the analysis of the kinematic data, and Prof. Dr. B.C.L.
Touwen, Prof. Dr. A. Gramsbergen and Prof. Dr. D. Kernell for crit-
ically reading previous drafts of this manuscript.
References
Aruin AS, Latash ML (1995) Directional specificity of postural
muscles in feed-forward postural reactions during fast voluntary
arm movements. Exp Brain Res 102: 323±332
Bazalgette D, Zattara M, Bathien N, Bouisset S, Rondot P (1986)
Postural adjustments associated with rapid voluntary arm move-
ments in patients with Parkinson©s disease. Adv Neurol 45: 371±
374
Benvenuti F, Panzer V, Thomas S, Hallett M (1990) Kinematic and
EMG analysis of postural adjustments associated with fast el-
bow flexion movements. In: Brandt T, Paulus W, Bles W, Diete-
rich M, Krafczyk S, Straube A (eds) Disorders of posture and
gait. Thieme, Stuttgart, pp 72±75
Bouisset S, Zattara M (1981) A sequence of postural movements
precedes voluntary movements. Neurosci Lett 22: 263±270
Bouisset S, Zattara M (1987) Biomechanical study of the program-
ming of anticipatory postural adjustments associated with volun-
tary movements. J Biomech 20: 735±742
215
Brown JE, Frank JS (1987) Influence of event anticipation on pos-
tural actions accompanying voluntary movement. Exp Brain
Res 67: 645±650
Burleigh AL, Horak FB (1996) Influence of instruction, prediction,
afferent sensory information on the postural organization of step
initiation. J Neurophysiol 75: 1619±1627
Cordo PJ, Nashner LM (1982) Properties of postural adjustments as-
sociated with rapid arm movements. J Neurophysiol 47: 287±
302
Dietz V (1992) Human neuronal control of automatic functional
movements: interaction between central programs and afferent
input. Physiol Rev 72: 33±69
Forget R, Lamarre Y (1987) Rapid elbow flexion in the absence of
proprioceptive and cutaneous feedback. Hum Neurobiol 6: 27±
37
Forssberg H, Hirschfeld H (1994) Postural adjustments in sitting hu-
mans following external perturbations: muscle activity and kine-
matics. Exp Brain Res 97: 515±527
Friedli WG, Hallett M, Simon SR (1984) Postural adjustments asso-
ciated with rapid voluntary arm movements. 1. Electromyo-
graphic data. J Neurol Neurosurg Psychiatry 47: 611±622
Goutal L, Lino F, Bouisset S (1994) ModalitØs de l©appui corporel et
vitesse du mouvement de pointage (abstract). Arch Int Physiol
Biochim 102(5)
Gurfinkel VS, Lipshits MI, Mori S, Popov KE (1981) Stabilization
of body position as the main task of postural regulation. Hum
Physiol 7: 155±165
Gurfinkel VS, Lipshits MI, Lestienne FG (1988) Anticipatory neck
muscle activity associated with rapid arm movements. Neurosci
Lett 94: 104±108
Hadders-Algra M, Forssberg H (1998) Patterns of muscle coordina-
tion underlying posture. Gait Posture (in Press)
Hadders-Algra M, Brogren E, Forssberg H (1998) Development of
postural control: differences between ventral and dorsal mus-
cles? Neurosci Biobehav Rev (in press)
Hallett M, Shahani BT, Young RR (1975) EMG analysis of stereo-
typed voluntary movements in man. J Neurol Neurosurg Psychi-
atry 38: 1154±1162
Herman R, Herman R, Maulucci R (1981) Visually triggered eye-
arm movements in man. Exp Brain Res 42: 392±398
Hogan N (1984) Adaptive control of mechanical impedance by co-
activation of antagonistic muscles. IEEE Trans Automatic Con-
trol 29: 681±690
Horak FB, Esselman E, Anderson ME, Lynch MK (1984) The
effects of movement velocity, mass displacement, task certain-
ty on associated postural adjustments made by normal and
hemiplegic subjects. J Neurol Neurosurg Psychiatry 47:
1020±1028
Horak FB, Diener HC, Nashner LM (1989) Influence of central set
on human postural responses. J Neurophysiol 62: 841±855
Jeannerod M (1988) The neural and behavioural organization of
goal-directed movements. Clarendon Press, Oxford
Keshner EA, Allum JHJ, Pfaltz CR (1987) Postural coactivation and
adaptation in the sway stabilizing responses of normals and pa-
tients with bilateral vestibular deficit. Exp Brain Res 69: 77±92
Latash ML, Aruin AS, Neyman I, Nicholas JJ, Shapiro MB (1995)
Feedforward postural adjustments in a simple two-joint synergy
in patients with Parkinsons disease. Electroencephalogr Clin
Neurophysiol 97: 77±89
Lee WA (1980) Anticipatory control of postural and task muscles
during rapid arm flexion. J Mot Behav 12: 185±196
Lee WA, Buchanan TS, Rogers MW (1987) Effects of arm acceler-
ation and behavioral conditions on the organization of postural
adjustments during arm flexion. Exp Brain Res 66: 257±270
Lino F, Bouisset S (1994) Is velocity of a pointing movement per-
formed in a sitting posture increased by upper body instability?
J Biomech 27:733
Macpherson JM (1988) Strategies that simplify the control of quad-
rupal stance. II. Electromyographic activity. J Neurophysiol 60:
218±231
Massion J (1992) Movement, posture and equilibrium: interaction
and coordination. Prog Neurobiol 38: 35±56
216
Massion J (1998) Postural control systems in developmental per-
spective. Neurosci Biobehav Rev (in press)
Mittelstaedt H (1998) Origin and processing of postural information.
Neurosci Biobehav Rev (in press)
Moore SP, Brunt D, Nesbitt ML, Juarez T (1992) Investigation of
evidence for anticipatory postural adjustments in seated subjects
who performed a reaching task. Phys Ther 72: 335±343
Palmer E, Cafarelli E, Ashby P (1994) The processing of human bal-
listic movements explored by stimulation over the cortex. J Phy-
siol (Lond) 481: 509±520
Pozzo T, Berthoz A, Lefort L (1990) Head stabilization during var-
ious locomotion tasks in humans. I. Normal subjects. Exp Brain
Res 82: 97±106
Prochazka A (1989) Sensorimotor gain control: a basic strategy of
motor systems? Prog Neurobiol 33: 281±307
Sanes JN, Jennings VA (1984) Centrally programmed patterns of
muscle activity in voluntary behavior of humans. Exp Brain
Res 54: 23±32
Tyler AE, Hasan Z (1995) Qualitative discrepancies between trunk
muscle activity and dynamic postural requirements at the initia-
tion of reaching movements performed while sitting. Exp Brain
Res 107: 87±95
Zattara M, Bouisset S (1988) Posturo-kinetic organisation during the
early phase of voluntary upper limb movement. 1. Normal sub-
jects. J Neurol Neurosurg Psychiatry 51: 956±965