Neuroscience Letters 253 (1998) 103106

Effects of movement direction upon kinematic characteristics

of vertical arm pointing movements in man
Charalambos Papaxanthis*, Thierry Pozzo, Paul Stapley
Groupe dAnalyse du Mouvement (G.A.M), U.F.R. S.T.A.P.S., Campus Universitaire,
Universite de Bourgogne, B.P. 138, 21004 Dijon, France
Received 27 April 1998; received in revised form 22 July 1998; accepted 23 July 1998

Abstract
Vertical arm pointing movements in two directions (upwards and downwards), imposing two different loads (unload and 0.5 kg)
and speeds (normal and fast) have been studied in six subjects. Movements were recorded using an optoelectronic system.
Data analysis concentrated upon finger-tip kinematics. Significant effects of movement direction were recorded upon velocity
profiles. The acceleration time, computed relative to total movement time, was greater for downward movements than for upward
movements. In contrast however, no effects of load or speed were observed. Movement time was not affected by movement
direction or load, for both speeds tested. These results suggest different planning processes, for movements with and against
gravity and indicate that gravitational force influences the processes controlling movement execution. 1998 Published by
Elsevier Science Ireland Ltd. All rights reserved

Keywords: Direction; Gravitational force; Arm kinematics; Velocity profile; Reaching; Planning; Human

It has previously been demonstrated that upward and

downward arm movements are characterized by similar
velocity profiles when executed at different speeds and
with different hand-held loads [1]. This invariance of velocity profiles primarily suggested that movement planning
takes into consideration kinematic constraints [2,5]. However, the invariance of velocity profiles could also have a
dynamic significance. Similar velocity profiles of various
arm vertical movements can be obtained by separating gravitational from dynamic torques (related to speed and acceleration) [1,4]. Such a strategy could facilitate dynamic
computations during the implementation of the motor plan.
Nevertheless, from experimental evidence it is not yet
completely clear which parameters (kinematics, dynamics
or both), are incorporated in the motor plan for vertical arm
movements. Additionally, prior explanations have underestimated the role of gravity, considering it only as a minor
component of the motor program. Apart from providing the

* Corresponding author. Tel.: +33 3 80318705; fax: +33 3

80396702; e-mail: tpozzo@satie.ubourgogne.fr

vertical direction, gravity can either accelerate downward or

decelerate upward vertical arm movements and consequently could play a more operative role in their motor
control. Indeed, recent experimental data, recorded in normal and in micro-gravity environments [6], has suggested
that movement direction with respect to gravity is a significant factor in arm movement planning. The purpose of the
present study was to investigate the importance of gravitational force in planning and control processes of vertical
arm movements and also to debate the existence of kinematic and dynamic constraints in the formulation of the
motor plan.
Experiments were conducted using six right-handed male
subjects ranging from 23 to 30 years of age. All subjects
gave their consent to participate and experiments were conducted in accordance with legal requirements and international norms. Subjects stood upright and were asked to point
using their preferred arm, between two targets (two markers
attached to wooden dowels). The positions of both the targets relative to each subjects body were as follows: both
were placed at forearm length in front of the subjects along
the vertical axis (one target 60 cm above the other). Targets

0304-3940/98/$19.00 1998 Published by Elsevier Science Ireland Ltd. All rights reserved
PII S0304- 3940(98) 00604- 1

104

C. Papaxanthis et al. / Neuroscience Letters 253 (1998) 103106

were positioned such that the midpoint between the two (30
cm from the bottom target) corresponded to each subjects
shoulder level. Movements were executed in two different
directions (upwards and downwards), at two different
speeds (normally paced and as fast as possible), and with
two different loads (the hand empty, and holding a 0.5 kg
load in the palm of the hand). Accuracy was not the primary
constraint placed on subjects during experimentation and no
instructions were given concerning the desired paths form.
Each subject performed six movements in each experimental condition (48 trials). Thus, a total of 288 trials for all
subjects and experimental conditions were recorded. The
experimental protocol consisted in four blocks. Subjects
executed pointing movements in the following order: (1)
normal speed without load, (2) rapid speed without load,
(3) normal speed with load and (4) rapid speed with load.
In each block, movement direction was altered (upward
followed by downward). At the beginning of each trial,
subjects visually positioned their finger in front of the target
(upwards or downwards) and waited for the start signal. We
suggested to subjects to avoid tactile contact with the targets. Immediately preceding a verbal go signal, vision of
the whole arm and target was blocked in order to prevent
subjects from using visual feedback to correct their movements. Following the go signal, subjects performed discrete and uncorrected movements towards the subsequent
target. Following the termination of the movement subjects
relaxed their arm at their sides (with no vision being allowed
during this period). After a short rest period (approximately
20 s), they were asked to visually position their finger in
front of the next target. A rest period of 2 min was permitted
between movement blocks. Subjects had no verbal feedback
as to their performance (accuracy or paths form) during the
experiment. Subjects performed 32 visually guided practice
trials before movement acquisition, four in each experimental condition. Movements were recorded and analyzed using
an optoelectronic ELITE system. Two TV-cameras (sampling frequency 100 Hz) were placed at a 45 angle, 4 m
from the experimental apparatus. Six markers (plastic
spheres of 0.4 cm in diameter) covered with reflecting material, were placed on the shoulder (acromion), elbow (lateral
epicondyle), wrist (cubitus styloid process and radius styloid process), hand (metacarpophalangeal joint) and the tip
of the index finger. Analysis of the data was based only
upon index finger-tip kinematics. For this marker, movement time (MT) was defined as the time at which linear
velocity exceeded and was inferior to 10% of its peak
values. Kinematic parameters (position, velocity and acceleration) in three dimensions were computed for the fingertip marker. The effects of movement direction, speed and
load upon velocity profiles were tested using the ratio of
acceleration time to total movement time or relative time to
peak velocity (AT/MT). A ratio greater than 0.5 indicated
an acceleration duration longer than the deceleration duration, and vice versa. Invariances in velocity profiles across
experimental conditions, were qualitatively evaluated by

normalizing them in time and amplitude, using procedures

previously described by Gielen et al. [3]. All measures
showed normal distribution following a Kolmogorov-Smirnov test. Measures were subjected to an analysis of variance
(ANOVA). The factors examined included direction, load
and speed.
As may be mechanically predicted by the motion of the
arm and forearm between the two targets in the saggital
plane, during upward movements values of gravitational
torque exerted around the shoulder and elbow joints
increased and decreased, respectively. An opposite pattern
has been observed for the downward direction movements.
Furthermore, the addition of load increased values of both
shoulder gravitational torque (SGT) and elbow gravitational
torque (EGT).
Although movement accuracy was not the primary variable tested in our experiment, subjects executed movements
in all experimental conditions with similar amplitudes.
Average values and standard deviations for unloaded and
loaded movements were respectively: 59.43 1.22 cm and
59.77 0.73 cm (downward normal speed); 60.84 1.30
and 62.42 2.78 cm (downward rapid speed); 59.33 1.00
and 60.00 0.70 cm (upward normal speed); 61.93 3.19
and 62.52 1.45 cm (upward rapid speed).
Mean values of finger-tip kinematics for all subjects are

Fig. 1. Histogram showing the effects of movement direction, speed

and load upon acceleration time (AT), deceleration time (DT) and
movement time (MT). Data are the mean SE of all subjects. U,
upward direction without load; UL, upward direction with a 0.5 kg
load; D, downward direction without load; UL, downward direction
with a 0.5 kg load. *P , 0.05, **P , 0.01.

C. Papaxanthis et al. / Neuroscience Letters 253 (1998) 103106

shown in Fig. 1. Movement time was roughly equivalent for

the two directions and loads in both speed conditions.
Movement direction, load and interactions between two
and three factors showed no significant effects (P . 0.05).
As expected however, due to task constraints, movement
time decreased highly significantly, with speed [F(1,5) =
186.06, P , 0.001)]. An important characteristic of kinematics was that acceleration and deceleration times were
unequal, indicating an asymmetry of movement production.
Indeed, a t-test statistical analysis (dependent samples)
between acceleration and deceleration times revealed that
they differed significantly for all experimental conditions
(see Fig. 1). Another interesting characteristic was that
downward movements showed greater AT/MT ratios than
upward movements for both loads and speeds. Average
values for unloaded and loaded movements were respectively: 0.53 and 0.55 (downward normal speed); 0.54 and
0.54 (downward rapid speed); 0.46 and 0.44 (upward normal speed); 0.46 and 0.47 (upward rapid speed). A threeway ANOVA (2 directions 2 loads 2 speeds) conducted
upon values of the AT/MT ratios, revealed a highly significant main effect only for movement direction [F
(1,5) = 114.95, P , 0.001)]. There were no two or three
factor interaction effects (P . 0.05). Differences in relative
time to peak velocity (AT/MT ratio) between the two directions were compensated by differences in their values of
peak acceleration and deceleration. Downward directional
movements showed smaller peak acceleration and higher
peak deceleration values compared to upward directions
movements. Average values of acceleration and deceleration peaks were respectively: 3.87 and 2.85 m/s2 (upward
unloaded normal speed); 2.79 and 3.28 m/s2 (downward
unloaded normal speed); 3.47 and 2.61 m/s2 (upward loaded
normal speed); 2.82 and 3.62 m/s2 (downward loaded normal speed); 57.59 and 44.44 m/s2 (upward unloaded rapid
speed); 50.50 and 60.91 m/s2 (downward unloaded rapid
speed); 48.49 and 37.84 m/s2 (upward loaded rapid
speed); 48.22 and 56.32 m/s2 (downward loaded rapid
speed). Fig. 2 qualitatively illustrates the effects of movement direction upon finger-tip velocity profiles, averaged
for all subjects.
An important finding of the present study was that velocity profiles did not remain invariant regardless of movement direction. This result contrasts greatly with previous
experimental data [1] and suggests that gravity influences
planning and programming processes. The question is, however, in what manner does gravitational force affect the
control of vertical arm movements? The hypothesis of a
misrepresentation of gravitational force, which could have
produced differences between upward and downward
movements must be excluded when considering results of
the present study. Both movement directions showed similar
movement times and no effects of load and speed were
observed upon relative times to peak velocity. This would
suggest that the CNS possesses accurate knowledge of limb
dynamics and kinematics. If the brain was unable to cor-

105

rectly anticipate gravitational torques, then load (which produces modifications in gravitational torque and limb
inertia), speed (which modifies acceleration and velocity
dependent torques), and movement direction should have
given significant interaction effects of velocity profiles.
One possible explanation for the kinematic differences
between upward and downward movements could be that
gravitational force is included into the motor plan and consequently modifies the processes controlling movement programming and execution. This hypothesis proposes the
existence of both kinematic and dynamic constraints into
the formulation of the motor plan. For instance, a purely
kinematic plan [2], would not predict differences between
upward and downward movements but it would infer similar
velocity profiles for differing speeds and limb inertia, as
indeed has been found in the present study. Alternatively,
purely dynamic [8,9] criteria could in fact predict asymmetric velocity profiles for movements in the two directions
but could also lead to modifications according to movement
speed and limb inertia. Thus, the importance either of kinematic requirements or of dynamic constraints on the planning processes of vertical arm movements cannot be
discounted. Similar results have also been obtained for
drawing movements executed in different directions in the
frontal plane [7]. Upward direction drawing movements
compared to downward direction ones showed smaller relative times to peak velocity. This difference in kinematics,
which can be extended to various motor tasks, emphasizes

Fig. 2. Means (thick lines) 1 standard error (thin lines) of finger-tip

tangential velocity profiles normalized for all subjects. Upward movements are represented by positive and downward movements by
negative velocity profiles. Small open circles crossing velocity curves
indicate acceleration times.

106

C. Papaxanthis et al. / Neuroscience Letters 253 (1998) 103106

the role of gravitational force in movement planning. In

conclusion, from present and previous results we postulate
that gravitational force, which provides the vertical direction and produces torques around the joints must be taken
into account for models of arm trajectory formation.
This work was supported by CNES (Centre National
dEtudes Spatiales).
[1] Atkeson, C.G. and Hollerbach, J.M., Kinematic features of
unrestrained vertical arm movements, J. Neurosci., 5 (1985)
23182330.
[2] Flash, T. and Hogan, N., The coordination of arm movements:
an experimentally confirmed mathematical model, J. Neurosci.,
7 (1985) 16881703.
[3] Gielen, C.C.A.M., Van den Oosten, K. and Pull ter Gunne, F.,
Relation between EMG activation patterns and kinematic properties of aimed arm movements, J. Motor Behavior, 17 (1985)
421442.

[4] Hollerbach, J.M. and Flash, T., Dynamic interactions between

limb segments during planar arm movement, Biol. Cybern., 44
(1982) 6777.
[5] Ostry, D.J., Cooke, J.D. and Munhall, K.G., Velocity curves of
human arm and speech movements, Exp. Brain Res., 68
(1987) 3746.
[6] Papaxanthis, C., Pozzo, T., Popov, K. and McIntyre, J., Hand
trajectories of vertical arm movements in one-G and zero-G
environments: evidence for a central representation of gravitational force, Exp. Brain Res., 120 (1998) 496502.
[7] Papaxanthis, C., Pozzo, T., Vinter, A. and Grishin, A., The
representation of gravitational force during drawing movements
of the arm, Exp. Brain Res., 120 (1998) 233242.
[8] Soechting, J.F., Bueno, C.A., Herrman, U. and Flanders, M.,
Moving effortlessly in three dimensions: Does Donders law
apply to arm movement?, J. Neurosci., 15 (1995) 62716280.
[9] Uno, Y., Kawato, M. and Suzuki, R., Formation and control of
optimal trajectory in multijoint arm movement: minimum torquechange model, Biol. Cybern., 61 (1989) 89101.